INSIGHT INTO THE GENETIC CONTROL OF COOKING TIME AT THE TRANSCRIPTOMIC AND GENOMIC LEVELS By Hannah Raye Jeffery A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Plant Breeding, Genetics, and Biotechnology – Crop and Soil Sciences – Doctor of Philosophy 2024 ABSTRACT Dry beans provide a source of plant-based protein, key micronutrients such as iron, zinc, and folate, and fiber for millions of people around the world. However, dry beans have long and highly variable cooking times that limit their consumer appeal. Long cooking times can be further exacerbated by high temperature and humidity post-harvest storage conditions prevalent in some bean production environments. The goal of this research was to 1) identify quantitative trait loci (QTL), genes, and genetic mechanisms that have been associated with cooking time in dry beans in the literature; 2) quantify gene expression patterns related to genetic variability for cooking time in beans during soaking (a common method used by home cooks to lower dry bean cooking times), identify candidate genes for fast cooking time via joining of RNA-seq and QTL data, and investigate the phenotypic effects of candidate genes via biochemical analyses such as enzyme and substrate (i.e., phytate and total nutrient) assays; 3); understand the effects of high temperature and humidity post-harvest storage on dry beans on a transcriptomic and biochemical level; and 4) validate molecular and phenotypic markers for cooking time in multiple populations. The literature review revealed that a total of 60 unique QTL have been delineated for cooking time, several of which overlap at key regions in the genome. Transcriptomic analysis was conducted on dry bean varieties with contrasting cooking times exposed to different soaking and storage conditions. Patterns in gene expression and candidate genes for cooking time related to cell wall modification (i.e., pectin interconversions, pectin demethylesterfication, and phenolic compound deposition) were identified in both studies, several of which were found within QTL for cooking time. A gene related to cell wall modification, pectin methylesterase, was found to be more active in slow-cooking dry beans, and calcium levels were found to be higher in slow- cooking genotypes across planting locations and years. Kompetitive allele-specific PCR (KASP) markers were designed that target candidate genes and QTL for cooking time. The results of this work elucidated multiple genetic mechanisms for cooking time in dry beans stored in both temperate and hot, humid environments and identified several markers that could be useful for developing faster cooking germplasm. Copyright by HANNAH RAYE JEFFERY 2024 To my family, who always believed in me and supported me, no matter what. v ACKNOWLEDGEMENTS Many thanks to my family and friends for their support throughout this journey. Thanks to my dissertation committee for the many hours of their time and teaching, and for giving me the opportunity to succeed. A special thank you to my major advisor, Dr. Karen Cichy, and my lab colleagues, for providing me with unending financial, professional, academic, and emotional support. I could not have done this without all of you. Thanks, also, to my co-authors and anyone who took the time to review my manuscripts. It is because of you that science can continue to thrive. Lastly, my sincere thanks for the funding and mentorship provided to me by the National Science Foundation Research Traineeship Program (DGE-1828149). vi TABLE OF CONTENTS LIST OF ABBREVIATIONS……………………………………………………….….............viii CHAPTER 1: A COMPREHENSIVE REVIEW OF THE GENETIC CONTROL OF COOKING TIME IN DRY BEANS (PHASEOLUS VULGARIS L.)………………………………………...1 BIBLIOGRAPHY…………………………………………………………………..…....33 CHAPTER 2: GENE EXPRESSION PROFILING OF SOAKED DRY BEANS (PHASEOLUS VULGARIS L.) REVEALS CELL WALL MODIFICATION PLAYS A ROLE IN COOKING TIME………………………………………………………………………………………..........49 BIBLIOGRAPHY………………………………………………………………………..92 CHAPTER 3: HOT, HUMID ENVIRONMENTS INDUCE THE EXPRESSION OF CELL WALL REMODELING GENES IN WHOLE MATURE DRY BEANS (PHASEOLUS VULGARIS L.) AND DISPROPORTIONATELY AFFECT THE GERMINATION RATES OF FAST-COOKING GENOTYPES……………………………………………………………....101 BIBLIOGRAPHY……………………………………………………………………....144 CHAPTER 4: NEW (BIO)MOLECULAR RESOURCES TO IMPROVE THE COOKING TIMES OF DRY BEANS (PHASEOLUS VULGARIS L.)…………………………………....152 BIBLIOGRAPHY……………………………………………………………………....178 APPENDIX A: SUPPLEMENTARY INFORMATION FROM CHAPTER 1...………….…..183 APPENDIX B: SUPPLEMENTARY INFORMATION FROM CHAPTER 2...………….…..187 APPENDIX C: SUPPLEMENTARY INFORMATION FROM CHAPTER 3...………….…..224 APPENDIX D: SUPPLEMENTARY INFORMATION FROM CHAPTER 4...………….…..241 APPENDIX E: MISCELLAENOUS SUPPLEMENTARY INFORMATION (NOT IN ANY CHAPTERS)………………………………………………………………….…………...........253 vii LIST OF ABBREVIATIONS BP CC Biological process Cellular component CGIAR Consultative Group for International Agricultural Research DEG Differentially expressed gene EXPA1 Expansin-like A1 FCBC Fast-cooking backcross GATK Genome analysis toolkit GO Gene ontology GWAS Genome wide association study G×E HTC Genotype-by-environment Hard-to-cook ICAP Inductively coupled argon plasma ISTA International Seed Testing Association KASP Kompetitive allele-specific PCR KO Kyoto Encyclopedia of Genes and Genomics Orthology LFDE Log-fold difference in expression LPA MDS MF PAE PCR PHA Low phytic acid Multidimensional scaling plot Molecular function Pectin acetylesterase Polymerase chain reaction Phaseolin protein viii PME QTL RH RIL Pectin methylesterase Quantitative trait loci Relative humidity Recombinant inbred line RNA-seq RNA sequencing RPKM Reads per kilobase million S SCC SEM SNP Soaked Seed coat check Scanning electron microscopy Single nucleotide polymorphism TAIR The Arabidopsis Information Resource TTRIL Brown bean recombinant inbred line US UV Unsoaked Ultraviolet USDA-NPGS US Department of Agriculture National Plant Germplasm System WAK Wall-associated kinase WGCNA Weighted gene co-expression network analysis XTH Xyloglucan endotransglycosylase/hydrolase YYRIL Yellow bean recombinant inbred line ix CHAPTER 1: A COMPREHENSIVE REVIEW OF THE GENETIC CONTROL OF COOKING TIME IN DRY BEANS (PHASEOLUS VULGARIS L.) This manuscript has been submitted to Plants, People, Planet and is currently awaiting reviewer scores. 1 SOCIETAL IMPACT STATEMENT Dry beans (Phaseolus vulgaris L.) are the most important grain legume globally for human consumption. Cooking is an essential processing step needed to unlock the nutrient richness of dry beans, and to make them palatable and safe for consumption. Here, we have identified genetic factors in the literature that contribute to cooking times across a range of environments to aid breeding efforts to develop reliably fast-cooking varieties in different dry bean market classes, thereby making dry beans more accessible to consumers around the world. SUMMARY Dry beans generally require long cooking times, and this influences their utilization and acceptability as a food source. Fuel usage and convenience are two constraints associated with cooking beans. The cooking times of dry beans are dependent on many factors including growing conditions, post-harvest handling and storage, cooking methods, and genotype. While there have been significant research endeavors to elucidate how post-harvest storage and handling influence cooking time, there has been comparatively less research on the genetic regulation of cooking time. This review provides an in-depth examination of the genetic control of cooking time in dry beans, including genetic mapping studies and its implications for breeding purposes. Further, the available physiological and post-harvest research is summarized and evaluated in the context of understanding genotypic and genotypic-by-environmental mechanisms controlling cooking time. The current evidence suggests that cooking time is increased by genetic factors that decrease the amount of soluble compounds in the cotyledons, including but not limited to 1) genes that increase the abundance of calcium-binding storage proteins and/or de-methylesterfied pectins paired with high inter- or intra-cellular calcium levels (higher calcium levels can be caused by reduced levels of phytate) and 2) genes that increase 2 insoluble condensed tannin/lignin content in the cotyledon cell walls. INTRODUCTION Dry beans, the dry seeds of common bean (Phaseolus vulgaris) are available in a diversity of colors, shapes, and flavors, and offer nutritional, health, and sustainability benefits that appeal to modern societal needs (Havemeier & Slavin, 2020). Nutritionally, dry beans are rich in protein, dietary fiber, iron, zinc, and folate (Mudryj et al., 2014). Beans are especially appealing as a plant-based source of protein, with levels ranging from 20 to 28% of their total dry weight (Katuuramu et al., 2018; USDA, 2024). Positive health results related to blood lipid profile, blood pressure, inflammation markers, and body composition have been realized with the consumption of 150 g (~1 ½ servings) of cooked beans and other pulses per day according to a broad review of the literature (Ferreira et al., 2021). Moreover, environmental sustainability aspects associated with growing dry beans include their capacity for symbiotic nitrogen fixation and low greenhouse gas emissions (Uebersax et al., 2022). For these reasons, dry beans are a globally important food that can be found in nearly all world cuisines (Amin & Borchgrevink, 2022). Beans are a dietary staple in Latin America, the center of origin of the Phaseolus species. Central America, Mexico, and Brazil are top consumers in the region (Rawal & Navarro, 2018). The highest per capita bean consumption in the world is in sub–Saharan Africa, where over 200 million people (especially in Eastern and Southern Africa) rely on them as a major source of calories and protein (Akibode & Maredia, 2012; Farrow & Muthoni-Andriatsitohaina, 2020). Dry beans are a minor food in the U.S. with per capita consumption levels rising slightly in recent years at 3.8 kg on average per person from 2015-2020 (Sadohara et al., 2024). Overall grain legume consumption is low in Europe but with large variation among countries, with U.K., Spain, and France consuming the most (Schneider, 3 2002). Dry beans are generally cooked whole in boiling water until they achieve a soft texture. The actual time needed for a sample to achieve an acceptable texture depends on many factors, including genetic aspects, storage conditions, age, ability to absorb water, pre-soaking, cooking method, and cooking elevation (Didinger et al., 2023; Perera et al., 2023). That said, cooking instructions for dry beans in the U.S. generally recommend pre-soaking followed by boiling in water for 45 min to 120 min (Garden-Robinson & Rachey, 2022). The major cooking fuels used in the U.S. are electricity and natural gas (Woodward & McNary, 2018). Dry bean cooking instructions in Uganda are similar with a recommended 60 to 120 min of cooking, only the pre- soaking step is often omitted due to regional preferences, and charcoal is instead used as the major cooking fuel (Asiimwe et al., 2024). Due to the economic and social importance of dry beans, the time it takes to cook them affects the consumption habits and nutritional status of large populations of the world. Cooking fuel may be cost-prohibitive to some bean consumers, notably the urban poor demographic in East Africa who rely on charcoal for cooking (Asiimwe et al., 2024; Boy et al., 2000). Hence, many consumers in Africa and Latin America prefer beans with shorter cooking times and may pay a premium for them (Asiimwe et al., 2024; Kilima & Bolle, 2020; Zapata, 2019). According to a lifecycle assessment (LCA) of dry beans produced and consumed in the U.S., home cooking electricity usage is the major source of environmental impact for dry beans across all environmental indicators except land use. Both the cooking times and the energy use efficiency of the cooking process were found to contribute the most to overall electricity use (Bandekar et al., 2022). Incorporating cooking time as a breeding objective into dry bean cultivar improvement can help address cooking fuel constraints. Furthermore, shorter cooking times 4 could make dry beans more accessible to end-users around the world, as modern consumers have shown consistent, growing interest in convenient, low-cost food, owing in part to a growing lack of time for meal preparation (Karlsen et al., 2016; Ribeiro et al., 2019; Röös et al., 2020; Winham et al., 2019, 2020). The goal of this review is to summarize dry bean cooking physiology, research methodologies used to measure cooking time, environmental factors influencing cooking time, and current knowledge of the genetic control, genotypic variability, breeding progress for, and nutritional tradeoffs of the cooking time trait. PHYSIOCHEMICAL PROCESSES THAT OCCUR DURING SOAKING AND COOKING Dry beans are commonly cooked whole without prior removal of any of the components of the seed. Their cooking time is defined as the amount time required for the bean to become soft enough to become palatable and digestible, as well as to develop a desirable aroma and flavor (Wainaina et al., 2021; Wood, 2017). Before the cooking process begins, beans can be soaked in ambient-temperature water to lower their cooking times (Didinger et al., 2023; Munthali et al., 2022). The generally recommended soaking procedure is immersing the seeds for 8-12 hr in room temperature water or for 1-4 hr in pre-boiled water (Amin & Borchgrevink, 2022). Typically, the median amount of water absorbed by seeds relative to their initial dry weight ranges from 86-106% (Bassett, Kamfwa, et al., 2021; Cichy, Wiesinger, et al., 2015; Diaz et al., 2021; Sadohara et al., 2022). Water uptake percentage depends on several factors, though, including growing location, postharvest management practices, and genotype (Sadohara et al., 2022). During soaking, water first enters dry bean seeds via the micropyle, hilar fissure, lens, or a combination of these orifices (Deshpande & Cheryan, 1986; Kikuchi et al., 2006; Mikac et al., 5 2015; Soltani et al., 2021; Varriano-Marston & Jackson, 1981) (Fig. 1.1). The hydrophobic cuticle and sclereid layers of the seed coat generally limit the uptake of water, but cracks or pores in these layers can allow water to enter the seed (Agbo et al., 1987; Palmer et al., 2022; Soltani et al., 2021). Similarly, lower lipid and total phenolic content in the seed coats decrease the hydrophobicity of the cuticle, resulting in higher water uptake (Ross et al., 2010). The seed slowly absorbs water and stores it between its cotyledons and its seed coat so as not to damage any delicate internal tissues like the embryo (Kikuchi et al., 2006). When water reaches the radicle, it is absorbed into the gap between the cotyledons and, finally, into the cotyledons and the embryo, activating the germination process (Kikuchi et al., 2006; Mikac et al., 2015). 6 Figure 1.1: Simplified image of a dry bean imbibing water. The red outer layer represents the seed coat, also known as the testa. The testa consists of multiple layers which are enumerated on the left of the image (the wax cuticle is the outermost layer) (Kikuchi et al., 2006). The thin blue arrows represent water uptake routes. The thick blue arrow indicates an increase in water uptake. The thick green arrows represent an increase/decrease in factors that lower cooking time. Water can be taken up via the micropyle, the hilar fissure, the lens, openings in the seed coat, or a combination of these orifices.Water taken up via the micropyle, hilar fissure, and lens is transported to the parenchyma layer of the cell wall and the outside of the embryo via a complex (vasculature network (Kikuchi et al., 2006). The size of the space between the seed coat and the embryo (i.e., cotyledons, hypocotyl, and radicle) is exaggerated to enable easier visualization of water movement in this space. The hypothesized mechanism linking water uptake and cooking time in dry beans is shown in the bottom right corner. The soaking process decreases the cooking times of dry beans significantly by pre- solubilizing cellular components in the cotyledon cells such as pectins, starches, and proteins. This gives the cotyledon material a uniform texture, ensuring the even transfer of heat throughout the seed, and increases cell separation by increasing the amount of osmotic pressure within the cotyledon cells (Martínez-Manrique et al., 2011; Mendoza et al., 2018; Njoroge et al., 2015) (Fig. 1.1). Soaking may also lead to the expression of protein-coding transcripts and, consequently, the differential expression of enzymes that help to further break down the cellular 7 components of the cotyledons (H. R. Jeffery et al., 2023; Martínez-Manrique et al., 2011; Toili et al., 2022). During commercial processing, food-grade additives are used in the soak water to accelerate water uptake, which include ethylene-diamine-tetraacetic acid (EDTA), sodium bicarbonate, sodium hexametaphosphate, sodium chloride, calcium chloride (Munthali et al., 2022). Dry beans can be cooked by way of boiling, pressure cooking, or canning (Wood, 2017). During cooking, heat transfer through beans in a wet medium causes many biochemical changes, including the denaturation of enzymes and storage proteins. The beans undergo two transitions during thermal cooking. During the first transition, the low-molecular-weight protein fractions denature, and the starches gelatinize, whereas the heat-resistant protein fractions of the beans denature during the second transition (Sánchez-Arteaga et al., 2015). Heating also induces the breakdown of pectins found in an area between the cells called the middle lamella, leading to cell separation in the cotyledons (Salunkhe & Kadam, 1989). This process is what ultimately makes the seeds edible and gives them a soft texture. Among these processes, starch gelatinization is considered a major indicator that a raw bean has transitioned to a cooked bean (Kyomugasho et al., 2023; Ren et al., 2021). Although cooking recommendations may differ depending on the variety, studies performed in the U.S. have found that it generally takes between 45-120 minutes for pre-soaked beans to completely cook in boiling water on a traditional stovetop, whereas it takes 5-10 minutes and 12-25 minutes for pre-soaked and unsoaked beans to cook in a pressure cooker, respectively (Amin & Borchgrevink, 2022; H. Jeffery & Cichy, 2023). Commercially, canning of beans is carried at retort temperatures between 240 ºF to 250 ºF, and the process time varies according to the can size (Howard et al., 2018). 8 RESEARCH METHODS TO MEASURE COOKING TIME Different destructive and nondestructive approaches for measuring cooking time have been developed, although there is no agreed upon reference method (Wood, 2017). Common destructive methods to quantify cooking time include finger pressing (Fig. 1.2a), texture analysis (Fig. 1.2b), and the use of specialized equipment such as the Mattson cooker apparatus (Fig. 1.2c). Nondestructive approaches include near infra-red (NIR) spectroscopy and hyperspectral imaging (Fig. 1.2d). The finger press method consists of applying pressure to a bean with the index and thumb fingers (Bressani & Chon, 1996; Yeung et al., 2009). This is repeated at pre- defined times, until the sample is soft enough to be considered as cooked (Kinyanjui et al., 2015; Wafula et al., 2020). Tactile measurements have been used by diverse authors to decide when a batch of beans has been cooked (Jones & Boulter, 1983; Kaur et al., 2005; Kinyanjui et al., 2015; Sangani et al., 2014; Wani et al., 2017; Williams et al., 1983). This method is also employed as an official testing methodology in some countries, including Costa Rica, to determine cooking time (RTCR 384:2004 Frijol en Grano. Especificaciones, Decreto Ejecutivo 32149., 2005). 9 Figure 1.2: Common techniques used to measure the cooking times of dry beans. A) Finger press method after a set period; B) texture analyzer that measures the hardness of the beans after a set period; C) the Mattson cooker attached to a recording device. The pins of the Mattson cooker are shown piercing cooked dry beans; and D) spectral data from near infrared spectroscopy (NIRS). The images of the pot and the stopwatch are public domain. Image of NIRS spectral data is used with permission from the corresponding author (Mendoza et al., 2018). Machine-based texture analysis can provide more objective evaluations of cooking time. The principle of texture analysis is that as cooking time increases, the amount of force needed to compress a bean decreases (Bourne, 1972; Sánchez-Arteaga et al., 2015). This parameter has been used to estimate the amount of time needed for different dry bean genotypes to become palatable in response to different softening treatments (Ávila et al., 2015; Marconi et al., 2000). The main advantage of this method is the ability to generate data from samples as small as a single bean (Ávila et al., 2015). On the other hand, the disadvantages are that it requires 10 expensive, specialized, and occasionally hard-to-procure texture analyzers (Bourne, 1972; Kinyanjui et al., 2015). The Mattson pin drop cooker method can make quantifying the cooking times of dry beans much more efficient and automated. This device was invented by Dr. Sante Mattson to measure the cooking times of yellow peas. The original design consists of a rectangular box containing 100 steel plungers (each with the same weight and dimensions) ending in a needle- like tip. The tip of each plunger is intended to be placed vertically on top of a sample. Then, hot water steam is used to cook the samples. Finally, the time required for each needle to pierce through the sample is recorded as the cooking time of that sample (Mattson, 1946). The number of plungers was later reduced to 25-30 for convenience, and the beans were submerged in boiling water on a circular platform compatible with common cooking pots to better emulate the domestic cooking process (Proctor & Watts, 1987). Newer designs include sensors and automatic recording units, avoiding the need for constant surveillance to determine the exact time when a needle has pierced through a bean (Wang & Daun, 2005). A drawback of this method is that no large-scale manufacturers produce this equipment, so specialized workshops must be commissioned to create each new unit, or the researchers must build them themselves (Cichy, Porch, et al., 2015; Paredes-López et al., 1989). The texture analysis technologies described above require the sample to be cooked, but novel non-destructive methods such as hyperspectral imaging technology and visible or near- infrared reflectance spectroscopy could make it possible to determine the cooking time of a sample without any prior processing steps (Mendoza et al., 2018; Wafula et al., 2020). These methods measure the chemical composition of seeds which can then, in turn, be used to predict the cooking times of the seeds. However, the accuracy of these methods is lower than direct 11 measurements of cooking time (Tahmasbian et al., 2021). Regardless, non-destructive technologies could be useful if the integrity of a dry bean sample must be preserved for future use. ENVIRONMENTAL FACTORS INFLUENCING COOKING TIME This topic has been covered extensively in recent reviews (Perera et al., 2023; Wainaina et al., 2021). The purpose of this section is to provide a summary of this work to facilitate a better understanding of potential genetic mechanisms for cooking time. Together and separately, growing, storage, and preparation environment impose a strong influence on the cooking times of dry beans. The altitude above sea level at which beans are cooked, for example, is positively correlated with cooking duration. One study reported a 186- minute increase in the cooking time of an unsoaked bean variety when prepared at 2256 meters above sea level (m.a.s.l) relative to 0 m.a.s.l. (Bressani & Chon, 1996). After soaking, the average cooking times of beans rose by 98 minutes when cooked at 3125 m.a.s.l. as compared to 263 m.a.s.l (Didinger et al., 2023). Different cooking times at varying altitudes are the result of the inverse relationship between pressure and altitude, wherein lower pressure makes it more difficult for water to reach and remain at boiling temperature (Bressani & Chon, 1996). For this reason, some researchers have proposed correcting for differences in altitude so that cooking times collected at different altitudes can be compared more easily (García et al., 2001). Cooking time is also influenced by seed moisture. Dry beans that lose too much moisture (< 10% seed moisture content) can develop a defect called ‘hard shell’. Seeds with hard shell do not take up any water as a result of the closure of all possible entrance points for water (Deshpande & Cheryan, 1986). Hard shell can generally be reversed by placing seeds in a cool, high relative humidity environment for about a week to open the lens or by scarifying the seed 12 coat (Deshpande & Cheryan, 1986). The hardness of the soaking and cooking water influence cooking time. Water hardness is a measure of the calcium and magnesium in water. Bean cooking time increases as water hardness, whether for the soak or cooking water increases (Munthali et al., 2022). By contrast, adding sodium chloride or sodium bicarbonate to cooking water reduces cooking times (Didinger et al., 2023). Poor storage conditions increase cooking times, as beans stored even for short periods of time in adverse conditions tend to take longer to cook than freshly harvested beans (Reyes- Moreno et al., 2000). The phrase ‘hard-to-cook’ (HTC) is generally used to describe dry beans that do not soften after cooking for extended periods of time, even though they may appear to imbibe water (Gloyer, 1921; Hohlberg & Stanley, 1987). The HTC phenomenon is often caused by long-term storage or storage in high temperature (≥35°C) and/or relative humidity (≥83%) environments (Abay & Tolesa, 2023). The hard-to-cook phenomenon increases the cooking times of dry beans by altering the composition of the inner tissues of the seeds (Chigwedere et al., 2018; Shehata, 2009). The cooking times of fresh dry beans can be dramatically different from dry beans with HTC, even if they are the same genotype. One study reported a 6-fold increase in the cooking times of red haricot beans following storage (Demito et al., 2019), while another study reported that the cooking time of a dry bean variety increased from 13 to 150 minutes due to 80 days of storage at 36 °C (Chen et al., 2021). POTENTIAL GENETIC MECHANSIMS FOR COOKING TIME Studies of storage induced hardening (i.e., HTC) mechanisms have provided much insight into the genetic mechanisms controlling cooking times in fresh beans. Briefly, the 13 available evidence suggests that mutations affecting the phytic acid content, degree of hardshell, and/or condensed tannin content of seed coats also indirectly affect the water uptake percentage and cooking times of freshly harvested beans (Erfatpour et al., 2018; Islam et al., 2020; Panzeri et al., 2011; Pérez-Vega et al., 2010) (Table 1.1). Thus, the following section elaborates on three main theories behind the HTC phenomenon. The pectin-cation-phytate theory The most accepted explanation for HTC is the pectin-cation-phytate theory (Yi et al., 2016). This theory states that in HTC beans, calcium-binding compounds called phytate/phytic acid are hydrolyzed by phytase enzymes to create inositol-phosphate precursors (Moscoso et al., 1984; Silva et al., 2021). Consequently, calcium ions previously bound to the phytates are released which can, then, form complexes between adjacent de-methylesterfied pectin chains in the middle lamellae and/or with storage proteins (Chigwedere et al., 2018; Cominelli et al., 2020; Galiotou-Panayotou et al., 2007; Wang et al., 2017). Insoluble, crosslinked pectin chains are hypothesized to lengthen the cooking time of beans by interfering with cell separation in the cotyledons and impeding solubilization of inter- and intracellular components (Chen et al., 2021; Yousif et al., 2007). This could explain why changes in cooking time in HTC dry beans have been associated with pectin interconversions (Chigwedere et al., 2018). Other enzymes that are thought to be involved in this process include pectin methylesterase enzymes (PME) and exordium-like 2 which catalyze the de-methylesterfication of pectin to form pectate and the formation of pectate-calcium bonds, respectively (H. R. Jeffery et al., 2023; Toili et al., 2022). One study directly supported this hypothesis by showing that both pectin methylesterase transcripts and pectin methylesterase activity were significantly upregulated in a dry bean with a longer cooking time during early seed development (Toili et al., 14 2022) (Table 1.1). Structural variation may also contribute to differences in HTC susceptibility in dry beans. For instance, methylesterified pectin chains and the PHA-E isoform of phaseolin- type storage proteins bind fewer calcium ions than de-methylesterified pectin chains and the PHA-L isoforms of phaseolin (Chen et al., 2021; Cominelli et al., 2020; Njoroge et al., 2015). In addition, previous studies have generally found a strong positive relationship between low phytate content and cooking time in dry beans. Multiple dry bean varieties with unusually low levels of phytic acid have been identified and characterized genetically (Campion et al., 2009; Cominelli et al., 2018; Panzeri et al., 2011). Both genes known to cause the lpa phenotype [lpa-280-10 (Phvul.001G165500) and lpa-PvMrp2 (Phvul.007G153800) on chromosomes 1 and 7, respectively] encode ABC transporters that transport phytic acid across plasma membranes. These genes individually confer a 90% and a 75% reduction in seed phytic acid content, respectively (Cominelli et al., 2018; Panzeri et al., 2011). These mutations were later found to greatly increase the cooking times of dry beans, which could explain why these regions of the genome (particularly Phvul.001G165500) appear in multiple QTL for cooking time (Cominelli et al., 2020) (Fig. 1.3; Table SI 1.1). Mutations in distal sequences could also modify the expression of lpa genes, resulting in more nuanced variations in phytic acid levels and cooking times in diverse populations. This could explain why the lpa genes have not yet been identified as candidate genes for cooking time in RNA-sequencing studies (H. R. Jeffery et al., 2023; Toili et al., 2022). Genes related to inositol (a subunit of phytate) catabolism and oxygenase activity, on the other hand, were upregulated in slow-cooking beans, which further supports the idea that inositol breakdown can increase dry bean cooking times by decreasing phytate content (Toili et al., 2022). 15 Figure 1.3: Circos plot of all quantitative trait loci (QTL) for cooking time (red), water uptake (blue), and other traits associated with cooking time and/or water uptake (orange) †. The lengths of the dry bean chromosomes (in base pairs) are on the outside track. QTL markers from genome-wide association studies (GWAS) studies are depicted in the second track as filled circles ‡. QTL intervals from QTL mapping studies are depicted in the third track as filled rectangles. Normalized QTL intervals are depicted on the plot with ± 1 Mbp confidence intervals. The first and the eighty-first QTL (i.e., the first GWA and QTL mapping loci, respectively) are labeled “1” and “81” on the plot. Regions of the genome where two or more of these QTL overlapped are labeled “C” for cluster. †Perez-Vega et al., 2010 did not specify the linkage groups of their QTL for water uptake, so these QTL were not included in the figure (Pérez-Vega et al., 2010). ‡If part of a QTL exceeded the boundary of its respective chromosome, the QTL was cut off. If the entire QTL was located outside the boundary of the chromosome, the QTL was not included in the figure. Cotyledon cell wall theory Long cooking times have repeatedly been associated with an increase in the strength (i.e., the amount of force needed to compromise the structural integrity) of the cell walls and the 16 middle lamellae in dry bean cotyledons using a hammer mill or a hydrothermal treatment (i.e., cooking) (Chigwedere et al., 2018, 2019; Njoroge et al., 2015; Siqueira et al., 2018). In addition, increases in cell wall and middle lamella integrity in both fresh and aged beans have been associated with biochemical factors such as pectin interconversions, increased pectin interconnections, increased phytase activity, decreased protein solubility, and higher insoluble dietary fiber content in the cotyledons (Bassett, Hooper, et al., 2021; Cominelli et al., 2020; Wainaina, Kyomugasho, et al., 2022; Wainaina, Lugumira, et al., 2022). It is thought that these biochemical factors could increase cotyledon hardness by slowing the rate of water uptake, cell separation, starch gelatinization, and protein hydrolysis, all of which are traits that have been associated with differences in the cooking times of fresh dry beans (Bassett, Hooper, et al., 2021; Do et al., 2019; Sefa-Dedeh et al., 1979; Wang & Daun, 2005). It is expected that genes that affect cell wall strength/integrity or thickness would, therefore, be differentially expressed in fast- and slow-cooking dry beans. Indeed, the expression levels of several cell wall modification enzymes found within QTL for cooking time and/or water uptake were found to be upregulated in slow-cooking dry bean genotypes compared to fast-cooking genotypes during soaking, thus lending support to this hypothesis (H. R. Jeffery et al., 2023) (Table 1.1). 17 Table 1.1: Biochemical mechanisms hypothesized to control cooking time in dry bean (Phaseolus vulgaris L.) and putative candidate genes for each mechanism. Putative candidate gene(s) Phytases (Galiotou-Panayotou et al., 2007) ; galactanase, rhamnogalacturonases, polygalacturonases (Martínez-Manrique et al., 2011) ; lpa- 280-10, lpa-PvMrp2 (i.e., ABC transporter proteins) (Cominelli et al., 2018; Panzeri et al., 2011); pectin methylesterases/pectinesterases, pectin methylesterase/pectinesterase inhibitor proteins (Toili et al., 2022); calcium- binding EF-hand family proteins, expansins, xyloglucan endotransglycosylase/hydrolases, EXORDIUM like 2 proteins (H. R. Jeffery et al., 2023) Peroxidases (E. G. Alves et al., 2021; Rivera et al., 1989) ; J gene (i.e., R2R3- MYB-type transcription factor protein) (Erfatpour & Pauls, 2020) ; Psd gene (i.e., basic helix-loop-helix (Bhlh) domain transcription factor P (pigment) protein) (Islam et al., 2020) ; superoxide dismutases (Bento et al., 2020) ; polyphenoloxidases (E. G. Alves et al., 2021) ; HXXXD-type acyl-transferase family proteins (i.e., phenolic glucoside malonyltransferase 1 proteins) (H. R. Jeffery et al., 2023) Pectin acetylesterase 8 (Palmer et al., 2022; Soltani et al., 2021) None Biochemical mechanism Pectin-cation- phytate hypothesis Condensed tannins/lignins Seed coat permeability Seed surface area How the mechanism is predicted to affect cooking time Pectin chains in the middle lamellae form more glue-like bonds between adjacent cell walls in the presence of increasing intercellular calcium concentrations. Calcium can also bind to storage proteins in cotyledon cells, rendering them insoluble and less denaturable in hydrothermic conditions. Phytate/phytic acid can chelate calcium ions, preventing the formation of intercellular bonds. Phytase is an enzyme that cleaves phytate molecules, decreasing the amount of calcium bound to phytate. Cell wall expansion may facilitate the formation of bonds between intercellular pectin chains. Proanthocyanidins are synthesized by enzymes into long, insoluble polymers called condensed tannins, which are then deposited into dry bean cell walls. Specific flavanols, by contrast, are synthesized by enzymes into long, insoluble lignins. Both condensed tannins and lignins bind to polysaccharides and proteins (Cominelli et al., 2020; E. Garcia et al., 1998). These polymers increase the tensile strength of cell walls, decrease cell wall permeability, increase cell wall thickness, increase dietary fiber content, and decrease the solubility of cellular components (Bassett, Hooper, et al., 2021; Hincks & Stanley, 1987). These factors are thought to make it more difficult for dry beans to achieve a soft, palatable texture during cooking. Seed coat microcracks, macrocracks, pores, low seed coat thickness, degree of seed coat adhesion to the cotyledons, and/or lack of a prominent waxy cuticle increase seed coat permeability, leading to faster initial water uptake (Bassett, Hooper, et al., 2021; Palmer et al., 2022; Saba et al., 2016; Sandhu et al., 2018; Soltani et al., 2021). Rapid hydration of the cotyledon has been associated with fast cooking times (Sofi et al., 2022). Increased seed size is directly related to increases in seed coat and cotyledon surface area (Deshpande & Cheryan, 1986). Altering the texture of the seed coat surface likewise is hypothesized to increase seed coat surface area (Cichy et al., 2014). Increased surface area is associated with increased water uptake, which is associated with decreases in cooking time (Konzen & Tsai, 2014; Sandhu et al., 2018). 18 Condensed tannins/lignins theory Condensed tannins/lignins have long been suspected to play a role in cooking time and water uptake in dry beans based on observations that beans that developed darker seed coats also developed severe HTC while in storage (Hincks & Stanley, 1987). Postharvest darkening has since been found to be caused by the accumulation of compounds called proanthocyanidins in the seed coat either before or during storage and their subsequent condensation into condensed tannins (Islam et al., 2020; Marles et al., 2008; Wiesinger et al., 2021). Furthermore, it was found that proanthocyanidins accumulate more slowly in the seed coats of slow-darkening mutants and very little and not at all in non-darkening mutants, causing their seeds to have brighter seed coat colors even after prolonged storage (Islam et al., 2020). The complete lack of seed coat darkening (i.e., non-darkening) is regulated by the J gene (Phvul.010G130600) which is an R2R3-MYB-type transcription factor found on chromosome 10 (Erfatpour & Pauls, 2020). The master regulator of slow-darkening, on the other hand, is the Psd gene, which is a basic helix-loop-helix (Bhlh) domain transcription factor P (pigment) protein (Phvul.007G171333) found on chromosome 7 (Islam et al., 2020). It is hypothesized that proanthocyanidins cause slow cooking times by binding to cell wall components and proteins, thereby rendering the water-soluble components of the cotyledon insoluble and, thus, more stable in the presence of high heat (Elia et al., 1997; Hincks & Stanley, 1987; Stanley, 1992). Key genes responsible for the synthesis of condensed tannins in P. vulgaris are provided in this review (Table 1.1). Minor genes involved in condensed tannin biosynthesis can be found in other publications (Freixas Coutin et al., 2017; Islam & Dhaubhadel, 2023). Significant relationships have also been found between lignin content and seed hardness in P. vulgaris (Hincks & Stanley, 1987; Nasar-Abbas et al., 2008). Lignins are similar to 19 condensed tannins in terms of molecular function, in that they also form insoluble complexes with cell walls and storage proteins (E. Garcia et al., 1998; Mafuleka et al., 1993; Marles et al., 2008). The activity of lignin biosynthesis enzymes such as superoxide dismutases, peroxidaes, and polyphenoloxidaes have likewise been associated with changes in seed hardness in dry beans (Table 1.1) (Bento et al., 2020; E. G. Alves et al., 2021; Rivera et al., 1989). WATER UPTAKE IN RELATION TO THE GENETICS OF COOKING TIME Both physical and genetic elements contribute to the effect of water uptake on cooking time. For example, it has been shown that seeds with a larger surface area imbibe water at a faster rate, allowing for more complete hydration of the cotyledon prior to cooking (Deshpande & Cheryan, 1986). Also, genotypes that cook faster after soaking usually have thinner cotyledon cell walls, as well as decreased levels of insoluble cell wall components, soluble and total fiber (Bassett, Hooper, et al., 2021). Lastly, seeds with permeable seed coats tend to have longer cooking times, all of which indicate that seed coat integrity plays a role in the control of cooking time, possibly by restricting water uptake during presoaking (Palmer et al., 2022; Soltani et al., 2021). Microcracks on the seed coat increase seed permeability by increasing the ability of the cotyledons to absorb water (Lahijanian & Nazari, 2017; Ma et al., 2004). One gene called pectin acetylesterase 8 (PAE8) is known to create microcracks in seed hila by weakening the molecular integrity of the seed coat (Fig. 1.1) (Palmer et al., 2022; Soltani et al., 2021). The presence of a single, non-functional allele in a PAE8 ortholog is sufficient to increase water uptake in domesticated beans relative to wild accessions (Palmer et al., 2022; Soltani et al., 2021). Although not tested yet, the recessive, mutated version of this allele could shorten cooking times in an additive manner. However, while seed coat permeability is associated with imbibition and 20 water accumulation in tissues, faster-cooking genotypes do not always exhibit higher total seed water uptake levels, most likely because there are other physiochemical factors that can control cooking time (Bassett, Hooper, et al., 2021). Water uptake could also influence cooking times by activating enzymatic activity in dry beans. The activities of two pectin modifying enzymes, polygalacturonase and rhamnogalacturonase, for instance, have been positively associated with cooking times in dry beans following soaking (Martínez-Manrique et al., 2011). In addition, significant relationships have been found between presoaking, the increased expression of cell wall- strengthening/restructuring enzymes and calcium-binding proteins, and increased cooking times (H. R. Jeffery et al., 2023). It has been postulated that cell wall modification enzymes could increase the cooking times of dry beans by expanding the cell walls, thereby exposing pectin chains to cell wall modifying enzymes during water uptake and making it easier for calcium, condensed tannins, and/or lignin to bind to and rigidify the cell wall and middle lamella (Chen et al., 2023; Samalova et al., 2024; Yi et al., 2016). Ultimately, it is suggested that a genetically regulated connection between water uptake and cotyledon solubility determines the cooking time of fresh dry bean varieties (Berry et al., 2020; Hincks & Stanley, 1987). GENOTYPIC AND GENETIC VARIABILITY ASSOCIATED WITH COOKING TIME P. vulgaris is a diploid with 11 chromosomes (2n = 22) and a genome size of approximately 587 Mb (Schmutz et al., 2014). P. vulgaris originated in Mexico and dry beans were domesticated independently in Mesoamerica and the Andes (Bitocchi et al., 2013). Domesticated beans from Mesoamerica are smaller seeded with black, navy, and pinto as important market classes, whereas Andean beans, which are larger seeded, feature the kidney and cranberry classes (Singh et al., 1991). As an important consumer trait, cooking time has been 21 studied in dry beans in the context of germplasm diversity, heritability, growing and post-harvest conditions, as well as genotype × environment (G × E) interactions on cooking time and related traits. Germplasm diversity panels Diversity panels are collections of germplasm that represent broad genetic diversity within a defined ontological or geographical boundary and have been used to evaluate cooking time ranges within P. vulgaris (Turner-Hissong et al., 2020). Andean market classes, including light red, dark red, and white kidneys, cranberry, red mottled, yellow, and brown varieties, were collected from seven regions of the world and demonstrated wide variability for cooking time, ranging from 16.5-160 minutes (Table 1.2). Mixed panels such as the African bean panel, PABRA collection, and Yellow Bean Collection (YBC) have consisted of red, red mottled, small red, white, sugar, yellow, purple, black, navy, kidney, and cranberry beans. The cooking times of these varieties ranged from 15-122 minutes. Accessions domesticated in Mesoamerica have mostly been represented in mixed diversity panels, although one study that differentiated the cooking times of Andean and Mesoamerican varieties found that the Mesoamerican lines had longer cooking times on average (Diaz et al., 2021). 22 Table 1.2: Cooking times (mins) and water uptake capacities of dry bean diversity panels determined after 12 hours of presoaking at room temperature using a Mattson cooker. Population & Reference # of genotypes Environment al conditions Andean gene pool; from Eastern and Central Africa (Elia et al., 1997) Andean Diversity Panel (Cichy, Wiesinger, et al., 2015) Andean Diversity Panel (Bassett, Kamfwa, et al., 2021) African Bean Panel (Saradadevi et al., 2021) Andean (VEF) breeding lines and Middle American (MIP) breeding lines (Diaz et al., 2021) PABRA from E. Africa and CIAT (Mughi et al., 2021) 16 206 430 358 380 VEP lines; 222 MIP lines Short- and long-rain conditions in Morogoro, Tanzania; Montcalm, MI, USA (irrigated) Hawassa, Ethiopia, Kabwe and Lusaka, Zimbabwe Kawanda and Kachwekano, Uganda, and Kagera, Tanzania Palmira, Colombia Kwanda, Uganda 121 bush beans; 31 climbers; 152 total 295 Yellow Bean Collection (Sadohara et al., 2022) Michigan and Nebraska, USA Cooking time range (mins) 31 to 47 min † Traits correlated with cooking time (r2) Tannin content (mg/100 g) (0.77); water uptake (%) in short-rain conditions (-0.78) and long-rain conditions (-0.87) Heritability for cooking time 0.97** in short-rain conditions ‡; 0.90** in long-rain conditions ‡ 16 to 36 min Seed weight (g) (0.10); water uptake (%) (-0.13); market class (0.29); region of origin (0.28); cultivation status (0.16) Cotyledon texture (-0.12); seed coat percentage (0.20); bitter flavor (0.12); sweet flavor (-0.34); starchiness (-0.36); beany flavor (0.23); total flavor intensity (-0.16) Iron content (-0.57); zinc content (-0.67) 16.7-68.9 in Hawassa; 17.8-75.5 in Kabwe; 21.0-85.8 in Lusaka 46.1-98.5 . 0.73 0.50** 22.0-62.9 for the VEF population; 31.24-87.96 for the MIP population Water uptake (%) (-0.28) and seed weight (g) (-0.33) for the VEF population; seed weight (g) (-0.13) for the MIP population 0.80 for VEF; 0.65 for MIP 35-100 during the Apr-Jul season; and 43-122 during the Sept-Dec season 15-64 . . Water uptake (%) (-0.82); days to flower (0.16); growing environment (0.49- 0.61) 0.80 †Cooking times were measured after 52% of the pins on the Mattson cooker fell. All other cooking times were measured after 80% of the pins fell. ‡Narrow-sense heritability was calculated. All others calculated broad-sense heritability. 23 Germplasm diversity panels have repeatedly been used to identify associations between cooking time and multiple phenotypic traits in dry beans that further support the genetic mechanisms discussed in the previous section. Associations that have been found across environments and years include seed coat color, condensed tannin content, water uptake percentage, region of origin, days to flowering, days to maturity, seed size, seed flavor, and iron and zinc content (Table 1.2). Differences in seed coat color have been associated with differences in cooking time because beans with white seed coats tend to have shorter cooking times, whereas genotypes with red/purple mottled and cranberry patterned seed coats take longer to cook (Cichy, Wiesinger, et al., 2015). A possible explanation could be the higher concentrations of condensed tannins in colored seed coats, since higher condensed tannin content has been associated with reduced water uptake, decreased water uptake rates, higher seed coat darkness, and prolonged cooking times (Bassett et al., 2020; Elia et al., 1997; Wiesinger et al., 2021). Region of origin and domestication traits, such as days to flowering and days to maturity, were also significantly correlated with differences in cooking time. Beans from Africa, for example, tended to have shorter cooking times when compared to beans from North America (Cichy, Wiesinger, et al., 2015). The correlation between region of origin and cooking time could be the result of regional prioritization of cooking time as a quality trait (Cichy, Wiesinger, et al., 2015). Some of the largest differences in cooking time (up to two-fold or greater) between genotypes from the same market class have been attributed to differences in the physical characteristics of the seeds, some of which only become apparent if the beans are soaked prior to cooking (Bassett, Hooper, et al., 2021; Berry et al., 2020; Hooper et al., 2017). For example, 24 higher water uptake was found to lead to the softening of seed structures when the beans were soaked, while seed coat thickness and absorption speed were found to be critical for determining the cooking time of the seeds if no soaking phase was involved (Bassett, Hooper, et al., 2021; Devkota et al., 2022). However, the cooking times of unsoaked and presoaked dry beans are also weakly correlated, suggesting that the physiochemical factors controlling the cooking times of unsoaked beans are retained to some extent during presoaking (Bassett, Hooper, et al., 2021; Cichy et al., 2019; Hincks et al., 1987). The relationship between other seed characteristics such as seed size (i.e., seed weight or mass) and cooking time is inconsistent both within and between the Andean and Mesoamerican gene pools. Seed weight was determined in one study to be weakly yet positively correlated with cooking time in an Andean diversity panel consisting of 206 genotypes (r2 = 0.10, p = 0.043) (Cichy, Wiesinger, et al., 2015). In an Andean MAGIC (multiparent advanced generation intercross) population, seed weight and cooking time were determined to not be significantly correlated (Diaz et al., 2021). On the other hand, seed weight was negatively correlated with cooking time across an Andean diversity panels consisting of 380 genotypes (r2 = -0.33, p < 0.001) (Diaz et al., 2021). In an Andean recombinant inbred population, strong negative correlations (r2 = -0.30 to -0.80, no p-values reported) were found across multiple populations and years (Berry et al., 2020). In a Mesoamerican diversity panel consisting 222 genotypes, a weak negative correlation (r2 = -0.13, p < 0.05) was identified with seed weight (Diaz et al., 2021). In mixed diversity panels (Andean and Mesoamerican), no significant correlation was found between seed weight and cooking time (Sadohara et al., 2022; Wahome et al., 2023). Negative correlations between seed size and cooking time have been attributed to differences in water uptake between the smaller and larger seeded genotypes (Sadohara et al., 2022). It is 25 suspected that larger seed surface area is associated with increased water uptake, which, in turn, is associated with decreased cooking time (Konzen & Tsai, 2014; Sandhu et al., 2018). Heritability The heritability of water uptake during soaking and cooking time have been calculated in both the broad and the narrow sense. Briefly, broad sense heritability was reported to account for all forms of genetic variation that affect the cooking times of dry beans, including dominant and epistatic variation. Narrow sense heritability, by contrast, only accounts for phenotypic differences caused by additive genetic variation, or alleles that contribute to cooking time in an additive manner (Covarrubias-Pazaran, 2021). Broad and narrow sense heritability estimates for cooking time across multiple studies were 0.50-0.98 and 0.90-0.97, respectively (Bassett, Kamfwa, et al., 2021; Diaz et al., 2021; Elia et al., 1997; Sadohara et al., 2022; Saradadevi et al., 2021). For water uptake, broad and narrow sense heritability estimates range from 0.76-0.88 and 0.77, respectively (Bassett, Kamfwa, et al., 2021; Diaz et al., 2021; Elia et al., 1997; Sadohara et al., 2022). These relatively high heritability estimates suggest that the genetic components of cooking times and water uptake in different genotypes are not easily affected by or are equally affected by different growing environments and years (Bassett, Kamfwa, et al., 2021; Cichy et al., 2019; Elia et al., 1997; Sadohara et al., 2022). There is no conclusive evidence yet regarding the dominance of fast-cooking over the slow-cooking trait, or vice versa. One study that measured the dominance of cooking time in dry bean cultivars grown in Cameroon suggested that the fast-cooking trait may be governed by recessive alleles (Yadji Haman et al., 2020). However, another study that investigated the inheritance of cooking time in cowpeas (Vigna unguiculata)—a pulse widespread in Africa with cooking dynamics comparable to those of dry beans—found that the accumulation of dominant 26 alleles lead to shorter cooking times (Addy et al., 2020). Since cooking time is a polygenic trait, variable reports regarding dominance are to be expected. Genotype × environment (G × E) interactions Different authors have drawn different conclusions regarding the extent to which genotype-by-environmental effects affect cooking time in dry beans. In most studies where beans were grown at multiple locations but cooked at one central location, genotype by environment interactions were not observed, suggesting that dry bean varieties do not need to be planted at multiple locations to obtain representative results for cooking time (Cichy et al., 2019; Katuuramu et al., 2018, 2020). In addition, although significant G × E interactions have been found among the segregating descendants (i.e., RILs) of genotypes with contrasting cooking times, the reported heritability coefficients for cooking time were still relatively high across planting locations (R. A. V. Garcia et al., 2012). But, in a study where the germplasm was both grown and cooked in the same environment, G × E was prevalent for both water uptake and cooking time (Sadohara et al., 2022). Most measurements of G × E have been conducted using dry beans from different growing environments, not from different storage environments. One study, though, found a significant G × E interaction (p < 0.05) for cooking time among dry bean genotypes stored in a cool, humid environment and in an increasingly hot, humid environment for 90 days (Mvile et al., 2023). Taken together, these results suggest that not all dry bean genotypes have the same metabolic reaction to different growing/storage environments. More work is needed to understand the genetic mechanisms responsible for different reactions to long-term storage between genotypes. 27 Genetic mapping studies Quantitative trait loci (QTL) mapping and genome-wide association studies (GWAS) enable investigators to associate phenotypic characteristics with genetic variation in specific regions of the genome. This is particularly useful in plant breeding programs that employ molecular marker assisted selection or fine mapping to elucidate the relationship between individual gene(s) and phenotype(s) of interest (González et al., 2017). There have been multiple QTL studies for cooking time and traits related to cooking time such as water uptake, phytate content, and postharvest accumulation of condensed tannins. Three biparental population and five GWA studies have been conducted for cooking time (Table 1.2). In these studies, 56 total QTL for water uptake and 79 total QTL for cooking time were identified (Tables SI 1.2a-b). Based on positions of these QTL, it was determined that 48 unique QTL for water uptake and 60 unique QTL for cooking time have been identified (Figure 1.2; Tables SI 1.2a-b). A previous meta-QTL analysis that collected QTL for cooking time and water absorption capacity found 55 QTL for cooking time and 72 for water absorption capacity (Badji et al., 2022). However, this study grouped together QTL for water absorption capacity and cooking time with traits that are thought to indirectly affect these traits such as slow-darkening seed coats, making it challenging to delineate genes that are associated with specific genetic mechanisms. Hence, this study separated QTL that are thought to indirectly affect cooking time or water uptake from QTL that were directly associated with differences in these traits (Fig. 1.3) (Bassett, Kamfwa, et al., 2021; Bassett, Katuuramu, et al., 2021; Berry et al., 2020; Cichy, Wiesinger, et al., 2015; Cominelli et al., 2018; Diaz et al., 2021; R. A. V. Garcia et al., 2012; Islam et al., 2020; Panzeri et al., 2011; Sadohara et al., 2022; Wahome et al., 2023). QTL for cooking time and water uptake detected in multiple studies are especially useful 28 for understanding the genetic control of this trait and for molecular breeding. To find co- localizing QTL, all published QTL for cooking time and water uptake were normalized as previously described to a physical map of the dry bean genome and plotted using MapChart software (H. R. Jeffery et al., 2023; Voorrips, 2002) (Fig. 1.3). A total of 18 clusters consisting of at least three QTL were identified on every chromosome except 7 and 11. Two clusters containing the most QTL were found on chromosome 10. The cluster associated with the most studies, though, was found on chromosome Pv01 and contained a low phytic acid (lpa-280-10) gene. The identification of different QTL in different germplasm suggests that novel genetic variants controlling cooking time may exist exclusively in specific germplasm, meaning that crossbreeding may be necessary to introgress favorable genetic combinations into individual dry bean varieties. Molecular markers for cooking time The identification of major-effect genes could facilitate the transfer of desirable genetic effects into elite germplasm using marker-assisted selection (MAS) or genomic selection (GS). Using MAS would involve the creation of genetic markers for use in high-throughput genotyping assays such as traditional PCR or kompetitive allele specific PCR (KASP) (Cortés et al., 2011). Genotypes would be selected based on the results of these assays, so that germplasm with favorable genes would be used to develop more fast-cooking germplasm. By contrast, GS does not make use of individual markers, but instead enables breeders to make selections based on genetic variation that is associated with cooking time from across the entire genome, including low-effect variants (Meuwissen et al., 2001). Some special considerations are needed to successfully implement GS in a dry bean breeding program. Firstly, since the success of the selection is limited to the amount of genetic 29 diversity available in the parental population, wider crosses may be needed to increase genetic gain for cooking time. This explains why Andean x Mesoamerican crosses lead to greater increases in cooking time in breeding simulations (Saradadevi et al., 2021). Secondly, the predictive capabilities of GS models are often limited to the population that was originally genotyped, with low to detrimental effects on cooking time on unrelated populations (Diaz et al., 2021). However, the inclusion of genes known to affect cooking time as a fixed variable in a genomic selection model has been shown to improve its performance, though not always by a considerable amount in dry beans (Izquierdo et al., 2024). Effective markers could be especially useful for improving the average cooking times in wide genetic pools such as diversity panels and closely related legume species (Blair et al., 2007). Importance of cooking time in variety releases Generally, cooking times are not reported or required to be reported as part of new dry bean variety release bulletins, but cooking times are more likely to be highlighted if they are notably fast. In the U.S. and Canada, traits related to canning quality have been the most reported end-use quality characteristics highlighted in variety releases. For example, recent germplasm releases such as North Dakota State University’s pinto variety ‘ND Rodeo’, Michigan State University’s navy variety ‘AuSable’, and Agriculture and Agri-Food Canada’s cranberry variety ‘AAC Cranford’ provided information about water uptake during soaking and canning quality appearance data, but did not provide cooking times (Balasubramanian et al., 2019; Gomez et al., n.d.; Osorno et al., 2024). There are a few instances where cooking times were presented in specialty Flor de Mayo, Flor de Junio, and Otebo bean market classes releases (Kelly et al., 2015, 2016). In East Africa, where bean cooking times have historically mattered more to consumers than in the U.S. and Canada, cooking times are also not always presented as part of 30 variety releases, but beans with shorter cooking times are often highly promoted. For instance, in the 2022 Tanzania bean variety release catalog, varieties with shorter cooking times are noted, but actual cooking data is not presented for all releases (Ndimbo et al., 2022). The current trend in the Americas and Africa is that more bean breeders are evaluating cooking time, and some are integrating cooking time measurements into their breeding programs. Therefore, it is expected that reports of cooking time associated with variety releases will become more widespread. COOKING TIME AND NUTRITIONAL VALUE Dry beans provide many nutritional benefits such as complex carbohydrates, plant-based protein, antioxidant activity, vitamins, and essential minerals like iron and zinc (Wainaina et al., 2021). However, prolonged cooking can result in the leaching of protein and minerals (Candela et al., 1997; Carrasco-Castilla et al., 2012; Oliveira et al., 2018; Xu & Chang, 2011). Beans with that require shorter cooking times tend to have greater nutrient retention than beans that take longer to cook (Wiesinger et al., 2016, 2018). Beans with shorter cooking times tend to have less total dietary fiber and less insoluble fiber but are still a rich source of fiber (Bassett, Hooper, et al., 2021; Sadohara et al., unpublished). CONCLUSIONS Shortening the cooking times of dry bean varieties means direct time, cost, and energy savings for consumers regardless of the cooking method used. Beans with shorter cooking times have also been shown to have higher nutritional value due to their reduced processing requirements and lower condensed tannin levels. Water uptake affects the cooking times of dry bean primarily by solubilizing proteins and pectin within the cotyledons of dry bean seeds. In addition, separation of the cotyledon cells appears to play a key role in determining cooking time. To understand the genetic mechanisms controlling cooking times and identify sources of 31 genetic diversity for this trait, this review collected information about phenotypic and genotypic variation for cooking time and water uptake available for various dry bean market classes. Across multiple populations, the heritability of cooking time and water uptake were high, ranging from 0.65-0.98 and 0.76-0.88, respectively. One-hundred-and-thirty-five quantitative trait loci for cooking time and water uptake from the literature were displayed on the same linkage map to highlight regions of the genome that have been strongly associated with differences in cooking time and water uptake. Seventeen genetic hotspots were found across the genome, suggesting broad genetic diversity exists for these traits. The cluster associated with the most studies contained a gene that affects up to 90% of phytate production in dry beans. The available genetic evidence provides support not only for the pectin-cation-phytate hypothesis, but also for a relationship between cotyledon solubility and cooking time that may be affected by a variety of physicochemical factors, including calcium, condensed tannin, and lignin content. Transcription factors and enzyme-encoding genes that may affect cotyledon solubility were also identified for the purpose of speeding up the development of faster cooking dry bean varieties. Future research is needed on biomolecular and genetic marker-assisted interventions to ascertain whether this is a legitimate and effective breeding strategy. 32 BIBLIOGRAPHY Abay, M. M., & Tolesa, G. N. (2023). Effects of storage temperature and relative humidity on cooking time and moisture uptake of selected common bean (Phaseolus vulgaris L.) varieties grown in Ethiopia. Cogent Food & Agriculture, 9(1), 2163577. https://doi.org/10.1080/23311932.2022.2163577 Addy, S. N. T. T., Cichy, K. A., Adu-Dapaah, H., Asante, I. K., Emmanuel, A., & Offei, S. K. (2020). Genetic studies on the inheritance of storage-induced cooking time in cowpeas [Vigna unguiculata (L.) Walp]. Frontiers in Plant Science, 11. https://doi.org/10.3389/fpls.2020.00444 Agbo, G., Hosfield, G., Uebersax, M., & Klomparens, K. (1987). Seed microstructure and its relationship to water uptake in isogenic lines and a cultivar of dry beans (Phaseolus vulgaris L.). Food Structure, 6(1). https://digitalcommons.usu.edu/foodmicrostructure/vol6/iss1/12 Akibode, C. S., & Maredia, M. K. (2012). Global and Regional Trends in Production, Trade and Consumption of Food Legume Crops. Michigan State University. Amin, S., & Borchgrevink, C. (2022). A Culinology ® perspective of dry beans and other pulses. In Dry Beans and Pulses (2nd ed., pp. 453–480). John Wiley & Sons Ltd. https://doi.org/10.1002/9781119776802.ch18 Asiimwe, R., Katungi, E., Marimo, P., Mukankusi, C., Rubyogo, J. C., & Anthony, V. (2024). Evaluating consumer preferences for reduced cooking time, taste and colour of beans in rural and urban communities in Uganda. Agriculture & Food Security, 13(1), 19. https://doi.org/10.1186/s40066-024-00466-4 Ávila, B. P., Santos dos Santos, M., Nicoletti, A. M., Alves, G. D., Elias, M. C., Monks, J., & Gularte, M. A. (2015). Impact of different salts in soaking water on the cooking time, texture and physical parameters of cowpeas. Plant Foods for Human Nutrition, 70(4), 463–469. https://doi.org/10.1007/s11130-015-0504-7 Badji, A., Rubaihayo, P., & Edema, R. (2022). The promise of a fast-cooked healthy meal: Common genomic regions for cooking time and iron and zinc content in beans revealed by meta-analysis. African Journal of Rural Development, 7(2), Article 2. Balasubramanian, P., Chatterton, S., & Conner, R. L. (2019). AAC Cranford cranberry dry bean. Canadian Journal of Plant Science, 99(4), 572–576. https://doi.org/10.1139/cjps-2019- 0010 Bandekar, P. A., Putman, B., Thoma, G., & Matlock, M. (2022). Cradle-to-grave life cycle assessment of production and consumption of pulses in the United States. Journal of Environmental Management, 302(Pt B), 114062. https://doi.org/10.1016/j.jenvman.2021.114062 33 Bassett, A., Dolan, K. D., & Cichy, K. (2020). Reduced retort processing time improves canning quality of fast-cooking dry beans (Phaseolus vulgaris L.). Journal of the Science of Food and Agriculture, 100(10), 3995–4004. https://doi.org/10.1002/jsfa.10444 Bassett, A., Hooper, S., & Cichy, K. (2021). Genetic variability of cooking time in dry beans (Phaseolus vulgaris L.) related to seed coat thickness and the cotyledon cell wall. Food Research International, 141, 109886. https://doi.org/10.1016/j.foodres.2020.109886 Bassett, A., Kamfwa, K., Ambachew, D., & Cichy, K. (2021). Genetic variability and genome- wide association analysis of flavor and texture in cooked beans (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 134(3), 959–978. https://doi.org/10.1007/s00122-020- 03745-3 Bassett, A., Katuuramu, D. N., Song, Q., & Cichy, K. (2021). QTL mapping of seed quality traits including cooking time, flavor, and texture in a yellow dry bean (Phaseolus vulgaris L.) population. Frontiers in Plant Science, 12, 670284. https://doi.org/10.3389/fpls.2021.670284 Bento, J. A. C., Lanna, A. C., Bassinello, P. Z., Oomah, B. D., Pimenta, M. E. B., Carvalho, R. N., & Moreira, A. S. (2020). Aging indicators for stored carioca beans. Food Research International, 134, 109249. https://doi.org/10.1016/j.foodres.2020.109249 Berry, M., Izquierdo, P., Jeffery, H., Shaw, S., Nchimbi-Msolla, S., & Cichy, K. A. (2020). QTL analysis of cooking time and quality traits in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 133(7), 2291–2305. https://doi.org/10.1007/s00122- 020-03598-w Bitocchi, E., Bellucci, E., Giardini, A., Rau, D., Rodriguez, M., Biagetti, E., Santilocchi, R., Spagnoletti Zeuli, P., Gioia, T., Logozzo, G., Attene, G., Nanni, L., & Papa, R. (2013). Molecular analysis of the parallel domestication of the common bean (Phaseolus vulgaris) in Mesoamerica and the Andes. New Phytologist, 197(1), 300–313. https://doi.org/10.1111/j.1469-8137.2012.04377.x Blair, M. W., Fregene, M. A., Beebe, S., & Ceballos, H. (2007). Marker-Assisted Selection in common beans and cassava. In E. P. Guimarães, J. Ruane, B. D. Scherf, A. Sonnino, & J. D. Dargie (Eds.), Marker-assisted selection: Current status and future prospectives in crops, livestock, forestry, and fish (pp. 82–109). Food and Agriculture Organization of the United Nations. https://www.gov.uk/research-for-development-outputs/marker- assisted-selection-in-common-beans-and-cassava Bourne, M. C. (1972). Texture measurement of individual cooked dry beans by the puncture test. Journal of Food Science, 37(5), 751–753. https://doi.org/10.1111/j.1365- 2621.1972.tb02742.x Boy, E., Bruce, N., Smith, K. R., & Hernandez, R. (2000). Fuel efficiency of an improved wood- burning stove in rural Guatemala: Implications for health, environment and development. Energy for Sustainable Development, 4(2), 23–31. https://doi.org/10.1016/S0973- 0826(08)60239-2 34 Bressani, R., & Chon, C. (1996). Effects of altitude above sea level on the cooking time and nutritional value of common beans. Plant Foods for Human Nutrition, 49(1), 53–61. https://doi.org/10.1007/BF01092522 Campion, B., Sparvoli, F., Doria, E., Tagliabue, G., Galasso, I., Fileppi, M., Bollini, R., & Nielsen, E. (2009). Isolation and characterisation of an lpa (low phytic acid) mutant in common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 118(6), 1211– 1221. https://doi.org/10.1007/s00122-009-0975-8 Candela, M., Astiasaran, I., & Bello, J. (1997). Cooking and warm-holding: Effect on general composition and amino acids of kidney beans (Phaseolus vulgaris), chickpeas (Cicer arietinum), and lentils (Lens culinaris). Journal of Agricultural and Food Chemistry, 45(12), 4763–4767. https://doi.org/10.1021/jf9702609 Carrasco-Castilla, J., Hernández-Álvarez, A. J., Jiménez-Martínez, C., Jacinto-Hernández, C., Alaiz, M., Girón-Calle, J., Vioque, J., & Dávila-Ortiz, G. (2012). Antioxidant and metal chelating activities of peptide fractions from phaseolin and bean protein hydrolysates. Food Chemistry, 135(3), 1789–1795. https://doi.org/10.1016/j.foodchem.2012.06.016 Chen, D., Ding, A., Zhu, L., Grauwet, T., Van Loey, A., Hendrickx, M., & Kyomugasho, C. (2023). Phytate and mineral profile evolutions to explain the textural hardening of common beans (Phaseolus vulgaris L.) during postharvest storage and soaking: Insights obtained through a texture-based classification approach. Food Chemistry, 404, 134531. https://doi.org/10.1016/j.foodchem.2022.134531 Chen, D., Pham, U. T. T., Van Loey, A., Grauwet, T., Hendrickx, M., & Kyomugasho, C. (2021). Microscopic evidence for pectin changes in hard-to-cook development of common beans during storage. Food Research International, 141, 110115. https://doi.org/10.1016/j.foodres.2021.110115 Chigwedere, C. M., Nkonkola, C. M., Rai, S., Kyomugasho, C., Kermani, Z. J., Pallares Pallares, A., Van Loey, A. M., Grauwet, T., & Hendrickx, M. E. (2019). Cotyledon pectin molecular interconversions explain pectin solubilization during cooking of common beans (Phaseolus vulgaris). Food Research International, 116, 462–470. https://doi.org/10.1016/j.foodres.2018.08.062 Chigwedere, C. M., Olaoye, T. F., Kyomugasho, C., Jamsazzadeh Kermani, Z., Pallares Pallares, A., Van Loey, A. M., Grauwet, T., & Hendrickx, M. E. (2018). Mechanistic insight into softening of Canadian wonder common beans (Phaseolus vulgaris) during cooking. Food Research International, 106, 522–531. https://doi.org/10.1016/j.foodres.2018.01.016 Cichy, K. A., Fernandez, A., Kilian, A., Kelly, J. D., Galeano, C. H., Shaw, S., Brick, M., Hodkinson, D., & Troxtell, E. (2014). QTL analysis of canning quality and color retention in black beans (Phaseolus vulgaris L.). Molecular Breeding, 33(1), 139–154. https://doi.org/10.1007/s11032-013-9940-y Cichy, K. A., Porch, T. G., Beaver, J. S., Cregan, P., Fourie, D., Glahn, R. P., Grusak, M. A., Kamfwa, K., Katuuramu, D. N., McClean, P., Mndolwa, E., Nchimbi-Msolla, S., Pastor- 35 Corrales, M. A., & Miklas, P. N. (2015). A Phaseolus vulgaris diversity panel for Andean bean improvement. Crop Science, 55(5), 2149–2160. https://doi.org/10.2135/cropsci2014.09.0653 Cichy, K. A., Wiesinger, J. A., Berry, M., Nchimbi-Msolla, S., Fourie, D., Porch, T. G., Ambechew, D., & Miklas, P. N. (2019). The role of genotype and production environment in determining the cooking time of dry beans (Phaseolus vulgaris L.). Legume Science, 1(1), e13. https://doi.org/10.1002/leg3.13 Cichy, K. A., Wiesinger, J. A., & Mendoza, F. A. (2015). Genetic diversity and genome-wide association analysis of cooking time in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 128(8), 1555–1567. https://doi.org/10.1007/s00122-015-2531-z Cominelli, E., Confalonieri, M., Carlessi, M., Cortinovis, G., Daminati, M. G., Porch, T. G., Losa, A., & Sparvoli, F. (2018). Phytic acid transport in Phaseolus vulgaris: A new low phytic acid mutant in the PvMRP1 gene and study of the PvMRPs promoters in two different plant systems. Plant Science, 270, 1–12. https://doi.org/10.1016/j.plantsci.2018.02.003 Cominelli, E., Galimberti, M., Pongrac, P., Landoni, M., Losa, A., Paolo, D., Daminati, M. G., Bollini, R., Cichy, K. A., Vogel-Mikuš, K., & Sparvoli, F. (2020). Calcium redistribution contributes to the hard-to-cook phenotype and increases PHA-L lectin thermal stability in common bean low phytic acid 1 mutant seeds. Food Chemistry, 321, 126680. https://doi.org/10.1016/j.foodchem.2020.126680 Cortés, A. J., Chavarro, M. C., & Blair, M. W. (2011). SNP marker diversity in common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 123(5), 827–845. https://doi.org/10.1007/s00122-011-1630-8 Covarrubias-Pazaran, G. E. (2021, June 22). Heritability: Meaning and computation (V. Wimmer, E. Ziemke, J. Martini, & S. Storr, Eds.). Excellence in Breeding Platform. https://excellenceinbreeding.org/toolbox/tools/eib-breeding-scheme-optimization- manuals Demito, A., Ziegler, V., Goebel, J. T. S., Konopatzki, E. A., Coelho, S. R. M., & Elias, M. C. (2019). Effects of refrigeration on biochemical, digestibility, and technological parameters of carioca beans during storage. Journal of Food Biochemistry, 43(7), e12900. https://doi.org/10.1111/jfbc.12900 Deshpande, S. S., & Cheryan, M. (1986). Microstructure and water uptake of Phaseolus and winged beans. Journal of Food Science, 51(5), 1218–1223. https://doi.org/10.1111/j.1365-2621.1986.tb13089.x Devkota, L., He, L., Midgley, J., & Haritos, V. S. (2022). Effect of seed coat microstructure and lipid composition on the hydration behavior and kinetics of two red bean (Phaseolus vulgaris L.) varieties. Journal of Food Science, 87(2), 528–542. https://doi.org/10.1111/1750-3841.16030 36 Diaz, S., Ariza-Suarez, D., Ramdeen, R., Aparicio, J., Arunachalam, N., Hernandez, C., Diaz, H., Ruiz, H., Piepho, H.-P., & Raatz, B. (2021). Genetic architecture and genomic prediction of cooking time in common bean (Phaseolus vulgaris L.). Frontiers in Plant Science, 11, 622213. https://doi.org/10.3389/fpls.2020.622213 Didinger, C., Cichy, K., Urrea, C. A., Scanlan, M., & Thompson, H. (2023). The effects of elevation and soaking conditions on dry bean cooking time. Legume Science, 1–13. https://doi.org/10.1002/leg3.207 Do, D. T., Singh, J., Oey, I., & Singh, H. (2019). Modulating effect of cotyledon cell microstructure on in vitro digestion of starch in legumes. Food Hydrocolloids, 96, 112– 122. https://doi.org/10.1016/j.foodhyd.2019.04.063 E. G. Alves, N., J. C. Gomes, M., M. Vasconcelos, C., C. Lima, A., L. S. de Lima, S., S. Brito, E., Z. Bassinello, P., & S. D. Martino, H. (2021). Six months under uncontrolled relative humidity and room temperature changes technological characteristics and maintains the physicochemical and functional properties of carioca beans (Phaseolus vulgaris L.). Food Chemistry, 342, 128390. https://doi.org/10.1016/j.foodchem.2020.128390 Elia, F. M., Hosfield, G. L., Kelly, J. D., & Uebersax, M. A. (1997). Genetic analysis and interrelationships between traits for cooking time, water absorption, and protein and tannin content of Andean dry beans. Journal of the American Society for Horticultural Science, 122(4), 512–518. https://doi.org/10.21273/JASHS.122.4.512 Erfatpour, M., Navabi, A., & Pauls, K. P. (2018). Mapping the non-darkening trait from ‘Wit- rood boontje’ in bean (Phaseolus vulgaris). Theoretical and Applied Genetics, 131(6), 1331–1343. https://doi.org/10.1007/s00122-018-3081-y Erfatpour, M., & Pauls, K. P. (2020). A R2R3-MYB gene-based marker for the non-darkening seed coat trait in pinto and cranberry beans (Phaseolus vulgaris L.) derived from ‘Wit- rood boontje.’ Theoretical and Applied Genetics, 133(6), 1977–1994. https://doi.org/10.1007/s00122-020-03571-7 Farrow, A., & Muthoni-Andriatsitohaina, R. (Eds.). (2020). Atlas of common bean production in Africa (2nd ed.). PABRA/CIAT. https://alliancebioversityciat.org/publications-data/atlas- common-bean-production-africa-2nd-ed Ferreira, H., Vasconcelos, M., Gil, A. M., & Pinto, E. (2021). Benefits of pulse consumption on metabolism and health: A systematic review of randomized controlled trials. Critical Reviews in Food Science and Nutrition, 61(1), 85–96. https://doi.org/10.1080/10408398.2020.1716680 Freixas Coutin, J. A., Munholland, S., Silva, A., Subedi, S., Lukens, L., Crosby, W. L., Pauls, K. P., & Bozzo, G. G. (2017). Proanthocyanidin accumulation and transcriptional responses in the seed coat of cranberry beans (Phaseolus vulgaris L.) with different susceptibility to postharvest darkening. BMC Plant Biology, 17(1), 89. https://doi.org/10.1186/s12870- 017-1037-z 37 Galiotou-Panayotou, M., Kyriakidis, N. B., & Margaris, I. (2007). Phytase–phytate–pectin hypothesis and quality of legumes cooked in calcium solutions. Journal of the Science of Food and Agriculture, 88(2), 355–361. https://doi.org/10.1002/jsfa.2973 Garcia, E., Filisetti, T. M. C. C., Udaeta, J. E. M., & Lajolo, F. M. (1998). Hard-to-cook beans (Phaseolus vulgaris): Involvement of phenolic compounds and pectates. Journal of Agricultural and Food Chemistry, 46(6), 2110–2116. https://doi.org/10.1021/jf970848f Garcia, R. A. V., Rangel, P. N., Bassinello, P. Z., Brondani, C., Melo, L. C., Sibov, S. T., & Vianello-Brondani, R. P. (2012). QTL mapping for the cooking time of common beans. Euphytica, 186(3), 779–792. https://doi.org/10.1007/s10681-011-0587-7 García, V. E., Molina Arias, R., & Benavides López, F. A. (2001). Factor de corrección para el tiempo de cocción del frijol según la altitud. Universidad de Costa Rica. https://vinv.ucr.ac.cr/sigpro/web/projects/98521 Garden-Robinson, J., & Rachey, J. (2022). A pocket guide to preparing pulse foods (FN2068). North Dakota State University Extension. https://www.ndsu.edu/agriculture/extension/publications/pocket-guide-preparing-pulse- foods Gloyer, W. O. (1921). Sclerema and hardshell, two types of hardness of the bean. Proceedings of the Association of Official Seed Analysts of North America, 12/13, 60–62. Gomez, F. E., Kelly, J. D., Wright, E. M., Awale, H. E., & Bales, S. (n.d.). Registration of ‘AuSable’ navy bean. Journal of Plant Registrations, n/a(n/a). https://doi.org/10.1002/plr2.20374 González, A. M., Yuste-Lisbona, F. J., Fernández-Lozano, A., Lozano, R., & Santalla, M. (2017). Genetic mapping and QTL analysis in common bean. In M. Pérez de la Vega, M. Santalla, & F. Marsolais (Eds.), The Common Bean Genome (pp. 69–107). Springer International Publishing. Havemeier, S. M., & Slavin, J. (2020). Pulses and legumes: Nutritional opportunities and challenges. Cereal Foods World, 65(2). https://doi.org/10.1094/CFW-65-2-0021 Hincks, M. J., McCannel, A., & Stanley, D. W. (1987). Hard-to-cook defect in black beans. Soaking and cooking processes. Journal of Agricultural and Food Chemistry, 35(4), 576– 583. https://doi.org/10.1021/jf00076a034 Hincks, M. J., & Stanley, D. W. (1987). Lignification: Evidence for a role in hard-to-cook beans. Journal of Food Biochemistry, 11(1), 41–58. https://doi.org/10.1111/j.1745- 4514.1987.tb00111.x Hohlberg, A. I., & Stanley, D. W. (1987). Hard-to-cook defect in black beans. Protein and starch considerations. Journal of Agricultural and Food Chemistry, 35(4), 571–576. https://doi.org/10.1021/jf00076a033 38 Hooper, S., Wiesinger, J. A., Echeverria, D., Thompson, H. J., Brick, M. A., Nchimbi-Msolla, S., & Cichy, K. A. (2017). Carbohydrate profile of a dry bean (Phaseolus vulgaris L.) panel encompassing broad genetic variability for cooking time. Cereal Chemistry, 94(1), 135– 141. https://doi.org/10.1094/CCHEM-04-16-0126-FI Howard, L. R., White, B. L., Uebersax, M. A., & Siddiq, M. (2018). Dry beans processing, quality evaluation, and nutrition. In Uebersax, M. A. & M. Siddiq (Eds.), Handbook of vegetables and vegetable processing (2nd ed., Vol. 1, pp. 559–587). John Wiley & Sons Ltd. 10.1002/9781119098935.ch24 Islam, N. S., Bett, K. E., Pauls, K. P., Marsolais, F., & Dhaubhadel, S. (2020). Postharvest seed coat darkening in pinto bean (Phaseolus vulgaris) is regulated by Psd, an allele of the basic helix-loop-helix transcription factor P. PLANTS, PEOPLE, PLANET, 2(6), 663– 677. https://doi.org/10.1002/ppp3.10132 Islam, N. S., & Dhaubhadel, S. (2023). Proanthocyanidin biosynthesis and postharvest seed coat darkening in pinto bean. Phytochemistry Reviews. https://doi.org/10.1007/s11101-023- 09895-8 Izquierdo, P., Sadohara, R., Wiesinger, J., Glahn, R., Urrea, C., & Cichy, K. (2024). Genome- wide association and genomic prediction for iron and zinc concentration and iron bioavailability in a collection of yellow dry beans. Frontiers in Genetics, 15. https://doi.org/10.3389/fgene.2024.1330361 Jeffery, H., & Cichy, K. (2023). The energy-saving potential of fast-cooking bean varieties. Annual Report of the Bean Improvement Cooperative, 66, 82–83. http://www.bic.uprm.edu/?page_id=77 Jeffery, H. R., Mudukuti, N., Buell, C. R., Childs, K. L., & Cichy, K. (2023). Gene expression profiling of soaked dry beans (Phaseolus vulgaris L.) reveals cell wall modification plays a role in cooking time. The Plant Genome, 16(3), e20364. https://doi.org/10.1002/tpg2.20364 Jones, P. M. B., & Boulter, D. (1983). The cause of reduced cooking rate in phaseolus vulgaris following adverse storage conditions. Journal of Food Science, 48(2), 623–626. https://doi.org/10.1111/j.1365-2621.1983.tb10804.x Karlsen, M. C., Ellmore, G. S., & McKeown, N. (2016). Seeds—Health benefits, barriers to incorporation, and strategies for practitioners in supporting consumption among consumers. Nutrition Today, 51(1), 50. https://doi.org/10.1097/NT.0000000000000135 Katuuramu, D. N., Hart, J. P., Porch, T. G., Grusak, M. A., Glahn, R. P., & Cichy, K. A. (2018). Genome-wide association analysis of nutritional composition-related traits and iron bioavailability in cooked dry beans (Phaseolus vulgaris L.). Molecular Breeding, 38(4), 44. https://doi.org/10.1007/s11032-018-0798-x Katuuramu, D. N., Luyima, G. B., Nkalubo, S. T., Wiesinger, J. A., Kelly, J. D., & Cichy, K. A. (2020). On-farm multi-location evaluation of genotype by environment interactions for 39 seed yield and cooking time in common bean. Scientific Reports, 10(1), 3628. https://doi.org/10.1038/s41598-020-60087-2 Kaur, M., Singh, N., & Sodhi, N. S. (2005). Physicochemical, cooking, textural and roasting characteristics of chickpea (Cicer arietinum L.) cultivars. Journal of Food Engineering, 69(4), 511–517. https://doi.org/10.1016/j.jfoodeng.2004.09.002 Kelly, J. D., Trapp, J., Miklas, P. N., Cichy, K. A., & Wright, E. M. (2015). Registration of ‘Desert Song’ Flor de Junio and ‘Gypsy Rose’ Flor de Mayo common bean cultivars. Journal of Plant Registrations, 9(2), 133–137. https://doi.org/10.3198/jpr2014.05.0028crc Kelly, James. D., Varner, G. V., Hooper, S., Cichy, K. A., & Wright, E. M. (2016). Registration of ‘Samurai’ Otebo bean. Journal of Plant Registrations, 10(2), 109–114. https://doi.org/10.3198/jpr2015.09.0051crc Kikuchi, K., Koizumi, M., Ishida, N., & Kano, H. (2006). Water uptake by dry beans observed by micro-magnetic resonance imaging. Annals of Botany, 98(3), 545–553. https://doi.org/10.1093/aob/mcl145 Kilima, F., & Bolle, E. (2020). Factors underlying consumers’ choices of dry common beans in Dar es Salaam, Tanzania. African Journal of Food, Agriculture, Nutrition and Development, 20(02), 15619–15636. https://doi.org/10.18697/ajfand.90.18350 Kinyanjui, P. K., Njoroge, D. M., Makokha, A. O., Christiaens, S., Ndaka, D. S., & Hendrickx, M. (2015). Hydration properties and texture fingerprints of easy- and hard-to-cook bean varieties. Food Science & Nutrition, 3(1), 39–47. https://doi.org/10.1002/fsn3.188 Konzen, E., & Tsai, S. (2014). Seed coat shininess in Phaseolus vulgaris: Rescuing a neglected trait by its screening on commercial lines and landraces. Journal of Agricultural Science, 6. https://doi.org/10.5539/jas.v6n8p113 Kyomugasho, C., Wainaina, I., Grauwet, T., Van Loey, A., & Hendrickx, M. E. (2023). Bean softening during hydrothermal processing is greatly limited by pectin solubilization rather than protein denaturation or starch gelatinization. Food Research International, 165, 112471. https://doi.org/10.1016/j.foodres.2023.112471 Lahijanian, S., & Nazari, M. (2017). Increasing germination speed of common bean (Phaseolus vulgaris) seeds by ultrasound treatments. Seed Technology, 38(1), 49–55. Ma, F., Cholewa, E., Mohamed, T., Peterson, C. A., & Gijzen, M. (2004). Cracks in the palisade culticle of soybean seed coats correlate with their permeability to water. Annals of Botany, 94(2), 213–228. https://doi.org/10.1093/aob/mch133 Mafuleka, M. M., Ott, D. B., Hosfield, G. L., & Uebersax, M. A. (1993). The role of phytase and lignin decorticated dry bean (Phaseolus vulgaris) hardening during storage. Journal of Food Processing and Preservation, 17(1), 1–20. https://doi.org/10.1111/j.1745- 4549.1993.tb00222.x 40 Marconi, E., Ruggeri, S., Cappelloni, M., Leonardi, D., & Carnovale, E. (2000). Physicochemical, nutritional, and microstructural characteristics of chickpeas (Cicer arietinum l.) and common beans (Phaseolus vulgaris L.) following microwave cooking. Journal of Agricultural and Food Chemistry, 48(12), 5986–5994. https://doi.org/10.1021/jf0008083 Marles, M. A. S., Vandenberg, A., & Bett, K. E. (2008). Polyphenol Oxidase Activity and Differential Accumulation of Polyphenolics in Seed Coats of Pinto Bean (Phaseolus vulgaris L.) Characterize Postharvest Color Changes. Journal of Agricultural and Food Chemistry, 56(16), 7049–7056. https://doi.org/10.1021/jf8004367 Martínez-Manrique, E., Jacinto-Hernández, C., Garza-García, R., Campos, A., Moreno, E., & Bernal-Lugo, I. (2011). Enzymatic changes in pectic polysaccharides related to the beneficial effect of soaking on bean cooking time. Journal of the Science of Food and Agriculture, 91(13), 2394–2398. https://doi.org/10.1002/jsfa.4474 Mattson, S. (1946). The cookability of yellow peas. A colloid-chemical and biochemical study. Acta Agriculturae Suecana, 2, 185–231. Mendoza, F. A., Wiesinger, J. A., Lu, R., Nchimbi-Msolla, S., Miklas, P. N., Kelly, J. D., & Cichy, K. A. (2018). Prediction of cooking time for soaked and unsoaked dry beans (Phaseolus vulgaris L.) using hyperspectral imaging technology. The Plant Phenome Journal, 1(1), 180001. https://doi.org/10.2135/tppj2018.01.0001 Meuwissen, T. H., Hayes, B. J., & Goddard, M. E. (2001). Prediction of total genetic value using genome-wide dense marker maps. Genetics, 157(4), 1819–1829. Mikac, U., Sepe, A., & Serša, I. (2015). MR microscopy for noninvasive detection of water distribution during soaking and cooking in the common bean. Magnetic Resonance Imaging, 33(3), 336–345. https://doi.org/10.1016/j.mri.2014.12.001 Moscoso, W., Bourne, M. C., & Hood, L. F. (1984). Relationships between the hard-to-cook phenomenon in red kidney beans and water absorption, puncture force, pectin, phytic acid, and minerals. Journal of Food Science, 49(6), 1577–1583. https://doi.org/10.1111/j.1365-2621.1984.tb12848.x Mudryj, A. N., Yu, N., & Aukema, H. M. (2014). Nutritional and health benefits of pulses. Applied Physiology, Nutrition, and Metabolism, 39(11), 1197–1204. https://doi.org/10.1139/apnm-2013-0557 Mughi, I., Ochwo-Ssemakula, M., Edema, R., & Mukankusi, C. (2021). Variability in cooking time, iron and zinc content in common bean (Phaseolus vulgaris L.) genotypes. Journal of Scientific Agriculture, 6–11. https://doi.org/10.25081/jsa.2021.v5.6559 Munthali, J., Nkhata, S. G., Masamba, K., Mguntha, T., Fungo, R., & Chirwa, R. (2022). Soaking beans for 12 h reduces split percent and cooking time regardless of type of water used for cooking. Heliyon, 8(9), e10561. https://doi.org/10.1016/j.heliyon.2022.e10561 41 Mvile, A. J., Nchimbi-Msolla, S., & Tryphone, G. M. (2023). Elucidating the effect of genotype x environment interaction and storage period on cooking time of selected common bean (Phaseolus vulgaris L.) genotypes. East African Journal of Science, Technology and Innovation, 4. https://doi.org/10.37425/eajsti.v4i3.693 Nasar-Abbas, S. M., Plummer, J. A., Siddique, K. H. M., White, P., Harris, D., & Dods, K. (2008). Cooking quality of faba bean after storage at high temperature and the role of lignins and other phenolics in bean hardening. LWT - Food Science and Technology, 41(7), 1260–1267. https://doi.org/10.1016/j.lwt.2007.07.017 Ndimbo, M., Shida, N., Mbiu, J., Kilango, M., Madata, C., Binagwa, P., & Kasuga, R. (2022). Variety catalogue of common beans (Phaseolus vulgaris) in Tanzania (1st ed.). Tanzania Agricultural Research Institute. Njoroge, D. M., Kinyanjui, P. K., Christiaens, S., Shpigelman, A., Makokha, Anselimo. O., Sila, D. N., & Hendrickx, M. E. (2015). Effect of storage conditions on pectic polysaccharides in common beans (Phaseolus vulgaris) in relation to the hard-to-cook defect. Food Research International, 76, 105–113. https://doi.org/10.1016/j.foodres.2015.03.005 Oliveira, A. P. de, Mateó, B. dos S. O., Fioroto, A. M., Oliveira, P. V. de, & Naozuka, J. (2018). Effect of cooking on the bioaccessibility of essential elements in different varieties of beans (Phaseolus vulgaris L.). Journal of Food Composition and Analysis, 67, 135–140. https://doi.org/10.1016/j.jfca.2018.01.012 Osorno, J. M., Erfatpour, M., Simons, K. J., Maisonneuve, M., Posch, J., & Vander Wal, A. J. (2024). Seed yield improvements in slow-darkening pinto bean: Registration of ‘ND Rodeo.’ Journal of Plant Registrations, 18(2), 270–278. https://doi.org/10.1002/plr2.20368 Palmer, J. P., Pajak, A., Robson, B., Zhang, B., Joshi, J., Diapari, M., Pauls, K. P., & Marsolais, F. (2022). Pectin acetylesterase 8 influences pectin acetylation in the seed coat, seed imbibition, and dormancy in common bean (Phaseolus vulgaris L.). Legume Science, 4(3), e130. https://doi.org/10.1002/leg3.130 Panzeri, D., Cassani, E., Doria, E., Tagliabue, G., Forti, L., Campion, B., Bollini, R., Brearley, C. A., Pilu, R., Nielsen, E., & Sparvoli, F. (2011). A defective ABC transporter of the MRP family, responsible for the bean lpa1 mutation, affects the regulation of the phytic acid pathway, reduces seed myo-inositol and alters ABA sensitivity. New Phytologist, 191(1), 70–83. https://doi.org/10.1111/j.1469-8137.2011.03666.x Paredes-López, O., Maza-Calviño, E. C., & González-Castañeda, J. (1989). Effect of the hardening phenomenon on some physicochemical properties of common bean. Food Chemistry, 31(3), 225–236. https://doi.org/10.1016/0308-8146(89)90060-5 Perera, D., Devkota, L., Garnier, G., Panozzo, J., & Dhital, S. (2023). Hard-to-cook phenomenon in common legumes: Chemistry, mechanisms and utilisation. Food Chemistry, 415, 135743. https://doi.org/10.1016/j.foodchem.2023.135743 42 Pérez-Vega, E., Pañeda, A., Rodríguez-Suárez, C., Campa, A., Giraldez, R., & Ferreira, J. J. (2010). Mapping of QTLs for morpho-agronomic and seed quality traits in a RIL population of common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 120(7), 1367–1380. https://doi.org/10.1007/s00122-010-1261-5 Proctor, J. R., & Watts, B. M. (1987). Development of a modified mattson bean cooker procedure based on sensory panel cookability evaluation. Canadian Institute of Food Science and Technology Journal, 20(1), 9–14. https://doi.org/10.1016/S0315- 5463(87)70662-2 Rawal, V., & Navarro, D. K. (2018). The global economy of pulses (1st ed.). FAO. 10.4060/I7108EN Ren, Y., Yuan, T. Z., Chigwedere, C. M., & Ai, Y. (2021). A current review of structure, functional properties, and industrial applications of pulse starches for value-added utilization. Comprehensive Reviews in Food Science and Food Safety, 20(3), 3061–3092. https://doi.org/10.1111/1541-4337.12735 Reyes-Moreno, C., Okamura-Esparza, J., Armienta-Rodelo, E., Gómez-Garza, R. M., & Milán- Carrillo, J. (2000). Hard-to-cook phenomenon in chickpeas (Cicer arietinum L): Effect of accelerated storage on quality. Plant Foods for Human Nutrition, 55(3), 229–241. https://doi.org/10.1023/A:1008106229189 Ribeiro, N., Casagrande, C., Mezzomo, H., Klasener, G., & Steckling, S. (2019). Consumer preference and the technological, cooking and nutritional quality of carioca beans. Semina: Ciências Agrárias, 40, 651. https://doi.org/10.5433/1679-0359.2019v40n2p651 Rivera, J. A., Hohlberg, A. I., Aguilera, J. M., Plhak, L. C., & Stanley, D. W. (1989). Hard-to- cook defect in black beans—Peroxidase characterization and effect of heat pretreatment and storage conditions on enzyme inactivation. Canadian Institute of Food Science and Technology Journal, 22(3), 270–275. https://doi.org/10.1016/S0315-5463(89)70394-1 Röös, E., Carlsson, G., Ferawati, F., Hefni, M., Stephan, A., Tidåker, P., & Witthöft, C. (2020). Less meat, more legumes: Prospects and challenges in the transition toward sustainable diets in Sweden. Renewable Agriculture and Food Systems, 35(2), 192–205. https://doi.org/10.1017/S1742170518000443 Ross, K. A., Zhang, L., & Arntfield, S. D. (2010). Understanding water uptake from the induced changes occurred during processing: Chemistry of pinto and navy bean seed coats. International Journal of Food Properties, 13(3), 631–647. https://doi.org/10.1080/10942910902718220 RTCR 384:2004 Frijol en Grano. Especificaciones, Decreto Ejecutivo 32149. (2005). Saba, I., Sofi, P. A., Zeerak, N. A., Bhat, M. A., & Mir, R. R. (2016). Characterization of a core set of common bean (Phaseolus vulgaris L.) germplasm for seed quality traits. SABRAO Journal of Breeding and Genetics, 48(3), 359–376. 43 Sadohara, R., Cichy, K. A., & Uebersax, M. A. (2024). Market trends of new pulse-based products launched during 2012-2021 in the U.S. – Diversifying demands, challenges, and opportunities. Annual Report of the Bean Improvement Cooperative, 67. Sadohara, R., Izquierdo, P., Couto Alves, F., Porch, T., Beaver, J., Urrea, C. A., & Cichy, K. (2022). The Phaseolus vulgaris L. Yellow Bean Collection: Genetic diversity and characterization for cooking time. Genetic Resources and Crop Evolution, 69(4), 1627– 1648. https://doi.org/10.1007/s10722-021-01323-0 Salunkhe, D. K., & Kadam, S. (1989). CRC handbook of world food legumes: Nutritional chemistry, processing technology, and utilization (Vol. 1). Agricultural and Food Sciences. https://api.semanticscholar.org/CorpusID:92739299} Samalova, M., Melnikava, A., Elsayad, K., Peaucelle, A., Gahurova, E., Gumulec, J., Spyroglou, I., Zemlyanskaya, E. V., Ubogoeva, E. V., Balkova, D., Demko, M., Blavet, N., Alexiou, P., Benes, V., Mouille, G., & Hejatko, J. (2024). Hormone-regulated expansins: Expression, localization, and cell wall biomechanics in Arabidopsis root growth. Plant Physiology, 194(1), 209–228. https://doi.org/10.1093/plphys/kiad228 Sánchez-Arteaga, H. M., Urías-Silvas, J. E., Espinosa-Andrews, H., & García-Márquez, E. (2015). Effect of chemical composition and thermal properties on the cooking quality of common beans (Phaseolus vulgaris). CyTA - Journal of Food, 13(3), 385–391. https://doi.org/10.1080/19476337.2014.988182 Sandhu, K. S., You, F. M., Conner, R. L., Balasubramanian, P. M., & Hou, A. (2018). Genetic analysis and QTL mapping of the seed hardness trait in a black common bean (Phaseolus vulgaris) recombinant inbred line (RIL) population. Molecular Breeding, 38(3), 34. https://doi.org/10.1007/s11032-018-0789-y Sangani, V. P., Patel, N. C., Bhatt, V. M., Davara, P. R., & Antala, D. K. (2014). Optimization of enzymatic hydrolysis of pigeon pea for cooking quality of dhal. International Journal of Agricultural and Biological Engineering, 7(5), 123–132. https://doi.org/10.25165/ijabe.v7i5.922 Saradadevi, R., Mukankusi, C., Li, L., Amongi, W., Mbiu, J. P., Raatz, B., Ariza, D., Beebe, S., Varshney, R. K., Huttner, E., Kinghorn, B.…Cowling, W. A. (2021). Multivariate genomic analysis and optimal contributions selection predicts high genetic gains in cooking time, iron, zinc, and grain yield in common beans in East Africa. The Plant Genome, 14(3). https://doi.org/10.1002/tpg2.20156 Schmutz, J., McClean, P. E., Mamidi, S., Wu, G. A., Cannon, S. B., Grimwood, J., Jenkins, J., Shu, S., Song, Q., Chavarro, C., Torres-Torres, M., Geffroy, V., Moghaddam, S. M., Gao, D., Abernathy, B., Barry, K., Blair, M., Brick, M. A., Chovatia, M., … Jackson, S. A. (2014). A reference genome for common bean and genome-wide analysis of dual domestications. Nature Genetics, 46(7), 707–713. https://doi.org/10.1038/ng.3008 Schneider, A. V. C. (2002). Overview of the market and consumption of pulses in Europe. British Journal of Nutrition, 88(S3), 243–250. https://doi.org/10.1079/BJN2002713 44 Sefa-Dedeh, S., Stanley, D. W., & Voisey, P. W. (1979). Effect of storage time and conditions on the hard-to-cook defect in cowpeas (Vigna unguiculata). Journal of Food Science, 44(3), 790–795. https://doi.org/10.1111/j.1365-2621.1979.tb08503.x Shehata, A. (2009). Hard‐to‐cook phenomenon in legumes. Food Reviews International, 8, 191– 221. https://doi.org/10.1080/87559129209540938 Silva, V. M., Putti, F. F., White, P. J., & Reis, A. R. D. (2021). Phytic acid accumulation in plants: Biosynthesis pathway regulation and role in human diet. Plant Physiology and Biochemistry, 164, 132–146. https://doi.org/10.1016/j.plaphy.2021.04.035 Singh, S. P., Gepts, P., & Debouck, D. G. (1991). Races of common bean (Phaseolus vulgaris, Fabaceae). Economic Botany, 45(3), 379–396. https://doi.org/10.1007/BF02887079 Siqueira, B. S., Fernandes, K. F., Brito, P. V. A., & Santos, F. C. A. (2018). Histochemical and ultrastructural characterization of easy-to-cook and hard-to-cook carioca bean genotypes. LWT, 97, 117–123. https://doi.org/10.1016/j.lwt.2018.06.048 Sofi, P., Mir, R., Zargar, S., Rani, S., Fatima, S., Shafi, S., & Zaffar, A. (2022). What makes the beans (Phaseolus vulgaris L.) soft: Insights into the delayed cooking and hard to cook trait. Proceedings of the Indian National Science Academy, 88. https://doi.org/10.1007/s43538-022-00075-4 Soltani, A., Walter, K. A., Wiersma, A. T., Santiago, J. P., Quiqley, M., Chitwood, D., Porch, T. G., Miklas, P., McClean, P. E., Osorno, J. M., & Lowry, D. B. (2021). The genetics and physiology of seed dormancy, a crucial trait in common bean domestication. BMC Plant Biology, 21(1), 58. https://doi.org/10.1186/s12870-021-02837-6 Stanley, D. W. (1992). A possible role for condensed tannins in bean hardening. Food Research International, 25(3), 187–192. https://doi.org/10.1016/0963-9969(92)90136-S Tahmasbian, I., Morgan, N. K., Hosseini Bai, S., Dunlop, M. W., & Moss, A. F. (2021). Comparison of hyperspectral imaging and near-infrared spectroscopy to determine nitrogen and carbon concentrations in wheat. Remote Sensing, 13(6), 1128. https://doi.org/10.3390/rs13061128 Toili, M. E. M., de Koning, R., Kiekens, R., Ndumba, N., Wahome, S., Anami, S., Githiri, S. M., & Angenon, G. (2022). Transcriptome-guided identification of pectin methyl-esterase- related enzymes and novel molecular processes effectuating the hard-to-cook defect in common bean (Phaseolus vulgaris L.). Foods, 11(12), 1692. https://doi.org/10.3390/foods11121692 Turner-Hissong, S. D., Mabry, M. E., Beissinger, T. M., Ross-Ibarra, J., & Pires, J. C. (2020). Evolutionary insights into plant breeding. Current Opinion in Plant Biology, 54, 93–100. https://doi.org/10.1016/j.pbi.2020.03.003 Uebersax, M. A., Cichy, K. A., Gomez, F. E., Porch, T. G., Heitholt, J., Osorno, J. M., Kamfwa, K., Snapp, S. S., & Bales, S. (2022). Dry beans (Phaseolus vulgaris L.) as a vital 45 component of sustainable agriculture and food security—A review. Legume Science, 5(1), e155. https://doi.org/10.1002/leg3.155 USDA. (2024). FoodData Central. USDA-ARS. https://fdc.nal.usda.gov/ Varriano-Marston, E., & Jackson, G. M. (1981). Hard-to-cook phenomenon in beans: Structural changes during storage and imbibition. Journal of Food Science, 46(5), 1379–1385. https://doi.org/10.1111/j.1365-2621.1981.tb04179.x Voorrips, R. E. (2002). MapChart: Software for the graphical presentation of linkage maps and QTLs. Journal of Heredity, 93(1), 77–78. https://doi.org/10.1093/jhered/93.1.77 Wafula, E. N., Wainaina, I. N., Buvé, C., Nguyen, N.-D.-T., Kinyanjui, P. K., Saeys, W., Sila, D. N., & Hendrickx, M. (2020). Application of near-infrared spectroscopy to predict the cooking times of aged common beans (Phaseolus vulgaris L.). Journal of Food Engineering, 284, 110056. https://doi.org/10.1016/j.jfoodeng.2020.110056 Wahome, S. W., Githiri, M. S., Kinyanjui, P. K., Toili, M. E. M., & Angenon, G. (2023). Genome-wide association study of variation in cooking time among common bean (Phaseolus vulgaris L.) accessions using Diversity Arrays Technology markers. Legume Science, 5(4), e184. https://doi.org/10.1002/leg3.184 Wainaina, I., Kyomugasho, C., Delbaere, S., Wafula, E., Van Loey, A., Sila, D., & Hendrickx, M. (2022). (Bio)chemical reactions associated with ageing of red kidney beans (Phaseolus vulgaris) during storage probed by volatile profiling: The role of glass transition temperature. Food Research International, 162, 112102. https://doi.org/10.1016/j.foodres.2022.112102 Wainaina, I., Lugumira, R., Wafula, E., Kyomugasho, C., Sila, D., & Hendrickx, M. (2022). Insight into pectin-cation-phytate theory of hardening in common bean varieties with different sensitivities to hard-to-cook. Food Research International, 151, 110862. https://doi.org/10.1016/j.foodres.2021.110862 Wainaina, I., Wafula, E., Sila, D., Kyomugasho, C., Grauwet, T., Van Loey, A., & Hendrickx, M. (2021). Thermal treatment of common beans (Phaseolus vulgaris L.): Factors determining cooking time and its consequences for sensory and nutritional quality. Comprehensive Reviews in Food Science and Food Safety, 20(4), 3690–3718. https://doi.org/10.1111/1541-4337.12770 Wang, N., & Daun, J. K. (2005). Determination of cooking times of pulses using an automated Mattson cooker apparatus. Journal of the Science of Food and Agriculture, 85(10), 1631– 1635. https://doi.org/10.1002/jsfa.2134 Wang, N., Hou, A., Santos, J., & Maximiuk, L. (2017). Effects of cultivar, growing location, and year on physicochemical and cooking characteristics of dry beans (Phaseolus vulgaris). Cereal Chemistry, 94(1), 128–134. https://doi.org/10.1094/CCHEM-04-16-0124-FI 46 Wani, I. A., Sogi, D. S., Wani, A. A., & Gill, B. S. (2017). Physical and cooking characteristics of some Indian kidney bean (Phaseolus vulgaris L.) cultivars. Journal of the Saudi Society of Agricultural Sciences, 16(1), 7–15. https://doi.org/10.1016/j.jssas.2014.12.002 Wiesinger, J. A., Cichy, K. A., Glahn, R. P., Grusak, M. A., Brick, M. A., Thompson, H. J., & Tako, E. (2016). Demonstrating a nutritional advantage to the fast-cooking dry bean (Phaseolus vulgaris L.). Journal of Agricultural and Food Chemistry, 64(45), 8592– 8603. https://doi.org/10.1021/acs.jafc.6b03100 Wiesinger, J. A., Cichy, K. A., Tako, E., & Glahn, R. P. (2018). The fast cooking and enhanced iron bioavailability properties of the Manteca yellow bean (Phaseolus vulgaris L.). Nutrients, 10(11), 1609. https://doi.org/10.3390/nu10111609 Wiesinger, J. A., Osorno, J. M., McClean, P. E., Hart, J. J., & Glahn, R. P. (2021). Faster cooking times and improved iron bioavailability are associated with the down regulation of procyanidin synthesis in slow-darkening pinto beans (Phaseolus vulgaris L.). Journal of Functional Foods, 82, 104444. https://doi.org/10.1016/j.jff.2021.104444 Williams, P. C., Nakoul, H., & Singh, K. B. (1983). Relationship between cooking time and some physical characteristics in chickpeas (Cicer arietinum L.). Journal of the Science of Food and Agriculture, 34(5), 492–496. https://doi.org/10.1002/jsfa.2740340510 Winham, D. M., Davitt, E. D., Heer, M. M., & Shelley, M. C. (2020). Pulse knowledge, attitudes, practices, and cooking experience of midwestern US university students. Nutrients, 12(11), 3499. https://doi.org/10.3390/nu12113499 Winham, D. M., Tisue, M. E., Palmer, S. M., Cichy, K. A., & Shelley, M. C. (2019). Dry bean preferences and attitudes among midwest hispanic and non-hispanic white women. Nutrients, 11(1), 178. https://doi.org/10.3390/nu11010178 Wood, J. A. (2017). Evaluation of cooking time in pulses: A review. Cereal Chemistry Journal, 94(1), 32–48. https://doi.org/10.1094/CCHEM-05-16-0127-FI Woodward, M., & McNary, B. (2018). Thanksgiving week: EIA data highlight how energy is used in the kitchen. U.S. Energy Information Administration (EIA). https://www.eia.gov/todayinenergy/detail.php?id=37552 Xu, B., & Chang, S. K. C. (2011). Reduction of antiproliferative capacities, cell-based antioxidant capacities and phytochemical contents of common beans and soybeans upon thermal processing. Food Chemistry, 129(3), 974–981. https://doi.org/10.1016/j.foodchem.2011.05.057 Yadji Haman, T., Aladji Abatchoua, M. M., Noubissié Tchiagam, J. B., & Ibrahima, A. (2020). Genetic analysis of cooking time in five common bean (Phaseolus vulgaris L.) cultivars grown in Cameroon. EAS Journal of Biotechnology and Genetics, 2(4), 58–66. 47 Yeung, H., Ehlers, J. D., Waniska, R. D., Alviola, J. N., & Rooney, L. W. (2009). Rapid screening methods to evaluate cowpea cooking characteristics. Field Crops Research, 112(2), 245–252. https://doi.org/10.1016/j.fcr.2009.03.010 Yi, J., Njoroge, D. M., Sila, D. N., Kinyanjui, P. K., Christiaens, S., Bi, J., & Hendrickx, M. E. (2016). Detailed analysis of seed coat and cotyledon reveals molecular understanding of the hard-to-cook defect of common beans (Phaseolus vulgaris L.). Food Chemistry, 210, 481–490. https://doi.org/10.1016/j.foodchem.2016.05.018 Yousif, A. M., Kato, J., & Deeth, H. C. (2007). Effect of storage on the biochemical structure and processing quality of adzuki bean (Vigna angularis). Food Reviews International, 23(1), 1–33. https://doi.org/10.1080/87559120600865172 Zapata, C. (2019, February 14). Guatemalan iron beans are a “blessing from God.” HarvestPlus. https://www.harvestplus.org/guatemalan-iron-beans-are-a-blessing-from-god/ 48 CHAPTER 2: GENE EXPRESSION PROFILING OF SOAKED DRY BEANS (PHASEOLUS VULGARIS L.) REVEALS CELL WALL MODIFICATION PLAYS A ROLE IN COOKING TIME Reproduced with permission from Wiley Online Library. Source: This chapter has been published in The Plant Genome. Jeffery, H. R., Mudukuti, N., Buell, C. R., Childs, K. L., & Cichy, K. (2023). Gene expression profiling of soaked dry beans (Phaseolus vulgaris L.) reveals cell wall modification plays a role in cooking time. The Plant Genome, 16(3), e20364. https://doi.org/10.1002/tpg2.20364 49 ABSTRACT Dry beans (Phaseolus vulgaris L.) are a nutritious food, but their lengthy cooking requirements are barriers to consumption. Pre-soaking is one strategy to reduce cooking time. Soaking allows hydration to occur prior to cooking and enzymatic changes to pectic polysaccharides also occur during soaking that shorten the cooking time of beans. Little is known about how gene expression during soaking influences cooking times. The objectives of this study were to 1) identify gene expression patterns that are altered by soaking and 2) compare gene expression in fast-cooking and slow-cooking bean genotypes. RNA was extracted from four bean genotypes at five soaking time points (0, 3, 6, 12, and 18 hrs) and expression abundances were detected using Quant-seq. Differential gene expression analysis and weighted gene co- expression network analysis were used to identify candidate genes within quantitative trait loci for water uptake and cooking time. Genes related to cell wall growth and development as well as hypoxic stress were differentially expressed in the fast- and slow-cooking beans due to soaking. Candidate genes identified in the slow-cooking beans included enzymes that increase intracellular calcium concentrations and cell wall modification enzymes. The expression of cell wall strengthening enzymes in the slow-cooking beans may increase their cooking time and ability to resist osmotic stress by preventing cell separation and water uptake in the cotyledon. INTRODUCTION Dry beans (Phaseolus vulgaris L.) are a staple crop rich in protein, iron, zinc, and dietary fiber (Mullins and Arjmandi, 2021). Around the world, there are at least 62 recognized dry bean market classes, including black, brown, kidney, navy, pinto, and yellow bean. While market classes have strict definitions in the bean processing industry, individual cultivars within a market class can have different genetic characteristics. For this reason, dry beans can have long 50 and highly variable cooking times that can deter greater utilization. Cooking time is a quantitative trait that displays variability both within and between market classes, making it possible to breed faster cooking beans (Cichy et al., 2015; Cichy et al., 2019). The identification of genetic targets for cooking time could help advance breeding efforts to decrease cooking times. The genetic mechanisms controlling cooking time, however, have yet to be fully elucidated. Pre-soaking is one of the major methods used by consumers to reduce cooking times. Once submerged in water, a bean first takes up water through its lens in anticipation of germination. The amount of time it takes the lens to open in the presence of water can vary from 2.5 to >10 hours, depending upon the presence of micro-cracks in the lens (Soltani et al., 2021). The water flows to the testa (seed coat) of the bean first, giving it a wrinkled appearance. Once water reaches the radicle (after about 3 hours of soaking), water begins moving into the cotyledons at a rapid rate. The cotyledons swell to accommodate the water at a biologically controlled rate, preventing damage to the internal seed tissues (Kikuchi et al., 2006). Inside the cotyledons, water solubilizes hydrophilic molecules like pectin and hydrolyzes insoluble proteins and starches, making it easier to break down these macromolecules during cooking (Rockland and Jones, 1974; Chigwedere et al., 2018; Cominelli et al., 2020). There are several other ways that water uptake affects cooking time. Firstly, water can facilitate leaching of elements and compounds that increase cooking times, including divalent cations like calcium and magnesium as well as condensed tannins. It can also cause the leaching of compounds such as phytic acid that decrease cooking times (Helbig et al., 2003). Divalent cations like calcium and magnesium make it more difficult for cotyledon cells to separate during cooking by crosslinking pectin chains located between adjacent cell walls (Chigwedere et al., 51 2018; Chigwedere et al., 2019). Phytic acid decreases cooking times by chelating divalent cations, preventing crosslinking. Phytic acid will also chelate these elements in the human body, so increasing phytic acid levels in beans to reduce cooking times would make it difficult for humans to obtain essential nutrients from their food (Cominelli et al., 2020). Condensed tannins, on the other hand, increase cooking times by insolubilizing pectin and other cell wall components, thereby preventing water uptake (De Valle and Stanley, 1995). Secondly, soaking increases the activity of enzymes like rhamanogalaturonase, polygalacturonase, and galactanase that can digest pectin chains in the cell wall (Martinez-Manríque et al., 2011). Water uptake ultimately enables the cell walls to separate more easily from the pectin-rich middle lamella during cooking, providing the beans with a soft texture (Chigwedere et al., 2018). Knowing how water uptake affects cooking time provides valuable insight into the genetic control of cooking time in dry beans, but the underlying genes responsible for genotypic variation in cooking time remain unknown. The genetic control of water uptake during soaking and cooking time has been investigated via diversity panels using genome wide association studies (GWAS) and bi-parental populations using quantitative trait locus (QTL) mapping (Garcia et al.,2011; Cichy et al., 2015; Berry et al., 2020; Bassett, Kamfwa, et al., 2021; Bassett, Katuuramu, et al., 2021; Diaz et al., 2021). These studies found genomic regions associated with water uptake and cooking time on every chromosome and identified candidate genes within these QTL associated with cooking time (Martinez-Manríque et al., 2011; Berry et al., 2020; Sadohara et al., 2022; Wahome et al., 2023). The candidate genes identified include enzymes that modify pectin solubility such as galacturan 1,4-alpha-galacturonidase, pectinesterase/pectin methylesterase, polygalacturonase, and pectate lyase, as well as enzymes that indirectly impact cell wall solubility such as 52 peroxidase (synthesizes condensed tannins) and phytase (removes phytic acid). In addition, some QTL for water uptake and cooking time overlapped with GWAS markers for flooding tolerance and seed coat color (Garcia et al., 2012; Cichy et al., 2015; Bassett, Kamfwa, et al., 2021; Diaz et al., 2021; Sadohara et al., 2022). Significant changes in enzyme activity paired with changes in cell structure during soaking could indicate that soaking directly affects gene expression in beans, and that gene expression affects cooking time. However, few studies have measured the entire transcriptome in P. vulgaris L. seeds during soaking, and even fewer have used transcriptomic data as a tool to understand how gene expression affects cooking time and phenotypes related to cooking time (O’Rourke et al., 2014; Astudillo-Reyes et al., 2015; Parreira et al., 2018; Perez de Souza et al., 2019; Toili et al., 2022). Thus, the aim of this study was to 1) contrast and compare transcript expression in fast- and slow-cooking beans within market classes; 2) characterize the physiology of cooking time; 3) identify candidate genes for marker-assisted selection. To accomplish this, we performed RNA sequencing (RNA-seq) on pairs of bean genotypes from two market classes with contrasting cooking times at five different soaking time points (0, 3, 6, 12, 18 hr). MATERIALS AND METHODS Germplasm origin and description A fast- and slow-cooking bean genotype from two market classes (brown and yellow) were evaluated in this study. The brown bean pair included TZ-27 (slower cooking) and TZ-37 (faster cooking), and the yellow bean pair included PI527538 (slower cooking) and Ervilha (faster cooking). Both brown beans were large seeded, kidney-shaped beans from the USDA- NPGS Phaseolus Germplasm Collection. TZ-27 (a.k.a. Incomparable) was collected in Tanzania in 1943. TZ-37 (a.k.a. W6 16488) was collected from Lushoto, Tanzania in 1994. PI527538 is 53 from USDA-NPGS and is an Njano type with a green-yellow seed coat. It was originally collected from Burundi in 1985. Ervilha is a Manteca type with a pale-yellow seed coat that was collected in Angola in 2010 (Cichy et al., 2015). Field design For RNA-sequencing, water uptake, and cooking time analyses, the four genotypes were planted at the Michigan State University Montcalm Research Station in summer 2017. The seeds were planted in a randomized complete block design (RCBD) with two field replications. The soil types were Eutric Glossoboralfs (coarse-loamy, mixed) and Alfic Fragiorthods (coarse- loamy-mixed, frigid), and supplemental rainfall was provided when needed in the form of overhead irrigation. Recommended field maintenance practice was followed in weed and insect control with no additional fertilizer application. At maturity, the seeds were harvested and threshed with a Hege 140 plot combine harvester. Debris was manually removed, and the seeds were stored at 4°C and 70-75% relative humidity pending analysis. The four genotypes were planted at the Michigan State University Montcalm Research Station in 2018 with two field reps and at the Michigan State University Saginaw Valley Research and Extension Center in 2019 with a single field rep. Total minerals and protein (for all genotypes) as well as pectin methylesterase activity (for TZ-27 and TZ-37 only) were measured in the seeds from 2018 and 2019. Phytic acid content was measured in the seeds from 2019 (for all genotypes). Nutrient and enzyme activity data were not collected across all years due to limited seed availability. Water uptake, cooking time, nutrient content, pectin methylesterase activity, and phytic acid determination Seed moisture content was determined using a John Deere Moisture Check Plus. When 54 the moisture content was between 10-14%, Thirty-five seeds per sample were weighed and soaked in distilled water at 5 soaking treatments (0, 3, 6, 12, and 18 hrs) in triplicate (4 genotypes x 5 soaking times x 3 replicates = 60 samples). The water uptake percentage of the seeds was calculated on a dry weight basis using the following equation: [(seed weight after soaking – seed weight before soaking)/seed weight before soaking]x 100. Ten of the soaked seeds were flash-frozen in liquid nitrogen and stored in a -20°C freezer for RNA isolation. Cooking time was measured with the remaining 25 seeds. Cooking time was determined using a Mattson bean cooker (Customized Machining 246 and Hydraulics Co., 247 Winnipeg, Canada), wherein the seeds are placed beneath steel pins and boiled in distilled water until 80% of the pins fall through the beans, at which point the cooking time is recorded. Beans were cooked in East Lansing, MI approximately 898 ft above sea level within six months of harvest in a lab with a room temperature of 21°C. Whole seeds soaked from the 0 and 12 hr time points were assayed for total nutrient content, pectin methylesterase activity, and phytic acid content. For nutrient content determination, 10 seeds were freeze-dried and milled into a fine powder using a Polymix mill (PX-MFC 90D, Kinematica, Lucerne, Switzerland) (0.5 mm sieve). Samples were sent to A & L Great Lakes (Fort Wayne, Indiana) for total nutrient content analysis. Total nitrogen was measured using the Dumas method, and mineral analysis was conducted using inductively coupled argon plasma (ICAP) spectroscopy with a Thermo Icap 6500 (Thermo Electron Manufacturing Ltd., Hemel Hempstead, UK) (AOAC, 2000). Nitrogen values were converted to percent protein by multiplying by 6.25. The moisture content of the powder was determined using the low constant temperature oven method (AOAC, 2000). To measure pectin methylesterase activity, enzymes were extracted from one gram of 55 cotyledon material following seed coat removal by homogenizing bean cotyledons in sodium acetic acid (pH 5.5) with an IKA Kinematica Ultra-Turrax (IKA Works Inc., Wilmington, NC). Extraction was performed as previously described using Spectra/Por pre-treated dialysis tubing, 8 Kd (Spectrum Chemical Mfg. Corp., New Brunswick, NJ) (Martìnez-Manrique et al., 2011). D-galacturonic acid concentration was measured spectrophotometrically as previously described using a Biotek Synergy HT plate reader (Agilent Technologies, Santa Clara, CA), except that 0.00250% bromothymol blue was used to improve visualization (Cameron et al., 1991). The concentration of the D-galacturonic acid was divided by time (minutes) and the amount of protein in the sample (mg per 500 mg of liquid enzyme extract) to determine the specific activity of the pectin methylesterase extract. Phytic acid content was determined using a Megazyme Phytic Acid Assay Kit according to the kit protocol (Megazyme Ltd., Bray, Ireland). RNA extraction Total RNA was extracted from whole seeds subjected to 60 treatment combinations (four genotypes, five soaking times, and three replications) using the Spectrum Plant Total RNA kit (Sigma Aldrich, Germany). Between fifty and one hundred milligrams of soaked seed were ground in liquid nitrogen using a coffee grinder. Residual genomic DNA was removed using the Sigma On-Column Dnase digestion kit and the Turbo DNA-free kit (Thermo Scientific, Wilmington, DE). RNA quantity was determined with a Qubit Nanodrop (Thermo Scientific, Wilmington, DE) and RNA integrity was checked with 1.5% agarose gel electrophoresis. RNA sequencing Approximately 18 Μl of total RNA (100-250 ng/Μl) was submitted to the Cornell University Sequencing facility for RNA-seq library preparation. Tag-Seq or 3’Quant-seq 56 libraries were robotically constructed according to the Lexogen QuantSeq FWD kit manual. OligodT priming was performed first, followed by cDNA synthesis, RNA removal, and PCR to amplify the library. RNA-seq analysis pipeline Sequence read quality was initially checked using FastQC (v. 0.11.9), and CutAdapt (v. 2.3) was used to remove poor quality base pairs, adapter sequences, and poly-A tails (Andrews, 2010; Martin, 2011). PCR duplicates were not removed. After cleaning, data quality was checked again using FastQC. HiSat2 (v. 2.1.0) was used to align the cleaned reads to the P.vulgaris L. genome (v. 2.1) (Kim et al., 2019; Schmutz et al., 2014). The number of duplicate reads, mitochondrial, and chloroplastic sequences were checked, but these sequences were not removed because this would have introduced bias into the RNA-seq data (Guo et al., 2007; Fu et al., 2018; Bi et al, 2020). HTSeq (v. 0.11.2) was then used to quantify gene expression levels (Anders et al., 2014; Kim et al., 2015). DESeq2 (v. 1.28.0) and edgeR (v. 3.28.1) were performed in Rstudio to identify differentially expressed genes (Love et al., 2014; The Rstudio Team, 2018). Data from the brown and yellow beans were analyzed separately so that differences within the market classes could be easily distinguished. Prior to performing the analysis, genes that were not expressed in any of the libraries were removed from the raw count data, the remaining reads were transformed using the rlog function, and replicates were compared using Pearson’s and Spearman’s correlation coefficients. A multidimensional scaling plot was generated using the top differentially expressed genes detected between pairs of samples (Ritchie et al., 2015). Differentially expressed genes (DEGs) were detected by comparing gene expression at 0 hrs of soaking to expression in the same genotype at a later soaking time (i.e., 0v3, 0v6, 0v12, and 0v18). 57 Differences in expression between genotypes were found by comparing gene expression at the same time point in two genotypes (TZ-27 vs. TZ-37, and PI527538 vs. Ervilha). DEGs with a Benjamini-Hochberg adjusted p-value lower than 0.05 were kept for further analysis. Weighted gene co-expression network analysis (WGCNA v. 1.13) was performed in Rstudio to identify clusters of co-expressed genes (a.k.a. modules) (Langfelder and Horvath, 2008; Langfelder and Horvath, 2012). Raw gene counts were checked for heteroskedasticity using the voom function from the limma package in R (v. 3.42.2), filtered for low gene expression counts, and converted to RPKM counts (Law et al., 2014). Genes with ≤ 10 cpm in at least six libraries were excluded from the analysis. The duration of soaking (hrs) and the cooking times of both genotypes were used as weights for WGCNA. The top 30 hub genes were detected in each module using the exportNetworkToVisANT function. Connections between the hub genes were visualized with visANT software (Hu et al., 2004). The molecular functions of the hub genes were determined using the map module function on the KEGG Orthology (KO) database. DESeq2 and WGCNA produced lists of genes that were significantly associated with soaking time and cooking time. Genes and their associated p-values were annotated using the topGO package (v. 3.12) in R (Alexa and Rahnenführer, 2021). Gene annotations were obtained from Ensembl (v. 52) using the biomaRt package (v. 2.46.2; Durinck et al., 2005; Durinck et al., 2009; Kinsella et al., 2011). All three GO ontologies (biological process, cellular component, and molecular function) were accessed. The ‘elimKS’ method was used to detect overrepresented GO terms within each of the three ontologies: biological process (BP), cellular component (CC), and molecular function (MF) (ɑ=0.05). 58 Identification of candidate genes for cooking time and water uptake TZ-27 and TZ-37 are the parents of a recombinant inbred line (TTRIL) population consisting of 146 individuals that were evaluated for water uptake and cooking time over two years and two locations in Tanzania (Berry et al., 2020). PI527538 and Ervilha are the parents of the YYRIL population consisting of 240 individuals that were evaluated for water uptake and cooking time over two years in one location in the USA (Bassett, Kamfwa, et al., 2021). Quantitative trait loci (QTL) for cooking time and water uptake were available from both populations. Since QTL from the YYRIL population were originally prepared using a genetic map with a low marker density, the QTL boundaries were re-calculated using the formula CI=163/(N*𝑅2), where CI is the confidence interval, N is the population size and 𝑅2 is the proportion of phenotypic variance attributed to the QTL (Darvasi and Soller, 1997). Then, the boundaries (in Cm) of the new QTL intervals were converted to physical distances (in bps) using a higher density genetic map, wherein SNP markers at the inside edges of the new QTL intervals were used as outer boundaries (Berry et al., 2020). The new QTL were overlaid on the TTRIL physical distance map using MapChart (v. 2.32) (Berry et al., 2020; Voorrips, 2002). Two QTL [CT.8.1 (Mi_C) and WU.9.1 (Mi_C)] were removed from further analyses due to large size. Genes within QTL for cooking time and water uptake percentage were annotated using the latest version of the Arabidopsis-derived annotation file from Phytozome (Pvulgaris_442_v2.1.annotation_info.txt). Differentially expressed genes from DESeq2 were aligned to lists of module genes from WGCNA in Rstudio using the merge function. These filtered genes (DESeq2 + WGCNA) were further filtered with genes in QTL for cooking time and water uptake. Throughout the analysis, data from the brown and yellow beans were analyzed and filtered separately. For instance, genes differentially expressed in the brown beans were only 59 filtered using WGCNA data and QTL data collected from brown beans. The genes that survived all three filtering steps are considered candidate genes for cooking time. Since it is hypothesized that water uptake initiates differential gene expression that leads to different cooking times, genes expressed because of soaking (i.e., genes differentially expressed both in soaked seeds relative to unsoaked seeds and in slow-cooking genotypes relative to fast-cooking genotypes) are considered highly probable candidate genes. To obtain the latest functional annotation data for each candidate gene, their CDS sequences or open reading frames were submitted to TAIR (Araport 11 3’ UTRs DNA), UniprotKB/Swiss-Prot, and PfamScan, and the BLAST results with the lowest E score were kept for analysis. This information affirmed the annotation from Phytozome and provided further information about the functions of the candidate genes. Statistical Analysis The cooking times of the genotypes were compared using linear modeling. Each experiment was a factorial CRD and the factors to be tested were seed treatment (0, 3, 6, 12, and 18 hrs of soaking), genotype, and market class. Tests for significant differences and comparison testing were done in R using the lm function in the stats package (v. 3.6.2) (ɑ=0.05) (R Core Team, 2022). Water uptake and cooking time RESULTS Four dry bean genotypes, including two fast- and slow-cooking genotypes from the yellow and brown market classes, were evaluated for water uptake and cooking time after 0, 3, 6, 12, and 18 hrs of soaking (Fig. 2.1). In all the genotypes, seed weight increased by ~50% due to water uptake within 3 hrs of soaking, by >80% within 6 hrs, and by > 100% within 12 hrs. After 12 hrs, seed weight plateaued in all the genotypes except TZ-37 which increased to 129 % of its 60 dry weight by hour 18 of soaking. Figure 2.1: Left: Fast- and slow- cooking bean varieties that were used for RNA-sequencing. Right: Water uptake percentages and average cooking times (n=3) of the samples used for RNA sequencing. Dotted lines represent water uptake percentages and solid lines represent cooking times. Error bars represent the standard deviation of the mean. Error bars could not be calculated for water uptake percentages due to a lack of replicates. The cooking times of the beans changed in response to water uptake. Unsoaked beans took the longest amount of time to cook, followed by beans soaked for 3, 6, and 12 hrs, respectively. The cooking time of TZ-27 was significantly longer than that of TZ-37 at every time point. The cooking time of TZ-27 did not significantly change after 6 hrs of soaking, whereas the cooking time of TZ-37 did not change after 12 hrs of soaking. After 12 hrs of soaking, the brown beans, TZ-27 and TZ-37, cooked in 30 min and 19 min, respectively. The cooking time of Ervilha was shorter than that of PI527538 after 0, 12, and 18 hrs of soaking. After 3 hrs of soaking, Ervilha had a significantly longer cooking time than PI527538. After 6 hrs of soaking, Ervilha and PI527538 took the same amount of time to cook. Like TZ-27, 61 the cooking time of the slow-cooking bean, PI527538, did not change after 6 hrs of soaking, whereas the cooking time of the fast-cooking bean Ervilha did not change after 12 hrs of soaking. After 12 hrs of soaking, the yellow beans, PI527538 and Ervilha, cooked in 23.5 min and 18 min, respectively. RNA sequencing The number of duplicated and unique reads generated per sample is reported in the supplementary data (Table SI 2.1). The number of non-unique reads represented 4-21% of the total reads in the samples, and unique reads represented 62-85% of the total reads. After trimming and filtering, 382 million reads remained (an average of 6.4 million reads per sample). The average length of these reads was 30-74 nt, and their average reference genome alignment rate was 87%. Between 1.7-15.8% (an average of 9.3%) of the reads in each sample aligned with sequences present in the mitochondrial or the chloroplast genomes (Table SI 2.1). Differential expression (DE) of genes in four dry bean genotypes during soaking Gene expression profiles of the four dry bean genotypes during soaking were visualized using an MDS plot (Fig. 2.2; Figs. SI 2.1a-b). The location of each sample on the MDS plot was determined based on gene expression levels in the bean samples. As expected, most of the biological replicates were tightly clustered in multidimensional space, wherein the distance between the samples increased as the dissimilarity between their gene expression profiles increased. While the brown and yellow bean samples mostly clustered on opposite sides of the plot, there was some overlap between the TZ-37 and Ervilha samples both at hours 0 and 3 of soaking. PI527538 samples also overlapped with Ervilha samples at hour 3 of soaking. At hour 6, however, the PI527538 and Ervilha samples diverged into separate clusters. By contrast, the TZ-27 and TZ-37 samples clustered far from each other at most soaking times, except at hour 12 62 of soaking; no genes were differentially expressed between TZ-27 and TZ-37 at hour 12 of soaking. Figure 2.2: Multidimensional scaling (MDS) plot showing RNA-seq data collected from two pairs of dry beans with contrasting cooking times. Each point represents a biological replicate. The shapes indicate the duration of soaking, and the colors indicate the genotype. Total gene expression was not correlated with the number of DEGs (𝑟2=0.046) (Fig. 2.3a). However, the total number of genes and the number of DEGs expressed both generally increased as soaking progressed from 0 to 18 hrs (Fig. 2.3b). TZ-37 expressed more DEGs than slow-cooking counterpart, TZ-27 at 3 hrs of soaking, and Ervilha expressed more DEGs than slow-cooking counterpart, PI527538 at 12 hrs of soaking (relative to 0 hrs of soaking), whereas both slow-cooking beans expressed more DEGs than either of the fast-cooking beans at hours 6 63 and 18 (relative to 0 hrs of soaking). A complete list of the genes expressed during soaking is available in the Supplementary Data (Table SI 2.2). The number of GO terms associated with these DEGs changed depending on genotype and soaking time (Figs. SI 2.2a-b; 2.3a-d). These GO terms were examined in more detail. Figure 2.3: a) Total number of expressed genes at different soaking times. B) Number of genes that are differentially expressed at a specific time point during soaking relative to gene expression levels in the unsoaked seeds of the same genotype. The black arrow points out an increase in differential gene expression that only occurs in the slow-cooking beans. TZ-27 is the slow-cooking brown bean. TZ-37 is the fast-cooking yellow bean. PI527538 is the slow-cooking yellow bean. Ervilha is the fast-cooking yellow bean. Predicted gene expression pathways in four dry bean genotypes during soaking Weighted gene co-expression network analysis (WGCNA) was used to compare gene expression patterns in TZ-27 vs. TZ-37 and in PI527538 vs. Ervilha. WGCNA identified 25 gene co-expression networks (a.k.a. modules) within the brown beans and 32 in the yellow beans following soaking (Figs. SI 2.4a-b). In the brown beans, six putative gene expression networks/modules were significantly associated with changes in water uptake during soaking, six modules were significantly associated with changes in cooking time, and four modules were significantly associated with both soaking and cooking time (Fig. 2.4; Table 2.1; Figs. SI 2.4a, c; Tables SI 2.3-2.4). In the yellow beans, seven modules were significantly associated with 64 soaking time, five modules were significantly associated with cooking time, and five modules were significantly associated with both soaking and cooking time (Fig. 2.5; Table 2.1; Figs. SI 2.4b, d; Tables SI 2.5-2.6). Figure 2.4: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the turquoise gene co-expression module found in brown beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co- expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the TurB module. 65 Table 2.1: List of all modules identified using WGCNA. The brown bean modules are indicated with a ‘B’ and yellow bean modules are indicated with a ‘Y’. The third and fourth columns show whether the expression of these genes increases (+) or decreases (-) in relation to soaking and cooking time, respectively (p≤0.05 is indicated by an *; p≤0.01 is indicated by an **; p≤0.001 is indicated by an ***). Market class Brown Module name TurB Number of genes 2958 Correlation with soaking time 0.96*** Correlation with cooking time -0.73* Brown BroB Brown RedB Brown PnkB Brown GylB Brown MblB Brown G60B Brown LtGB Yellow TurY Yellow BluY Yellow Yellow BroY RedY Yellow MagY Yellow LtGY Yellow WhiY 963 483 381 311 198 124 66 2585 1396 906 669 295 103 54 0.75** 0.78** -0.68* -0.75** -0.74** -0.38 -0.49 -0.64* 0.69* -0.89*** 0.75* 0.47 0.41 -0.62* -0.88*** -0.83** 0.50 0.57 0.66* 0.65* 0.77** 0.93*** -0.94*** 0.68* -0.69* -0.69* -0.67* 0.64* 66 Figure 2.5: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the turquoise gene co-expression module found in yellow beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co- expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the TurY module. The largest module found in the brown beans consisting of 2,958 genes was labeled ’turquoise brown’ (TurB), and the largest module in the yellow beans consisting of 2,585 genes was labeled ‘turquoise yellow’ (TurY) (Figs. 2.4-2.5). Interestingly, there was significant overlap in terms of gene content between the TurB and TurY modules, the BroB and BluY modules, the MblB and BroY modules, and the RedB and RedY modules, even though the brown and yellow bean data were analyzed independently (Figs. SI 2.5a-g). All the modules produced at least one significant GO term during topGO analysis. Both TurB and TurY produced GO terms related to cell wall expansion, including 67 response to auxin (GO:0009733; TurB: p=<0.001; TurY: p=0.020), cell wall biogenesis (GO:0042546; TurB: p=0.011; TurY: p=0.040), and carbohydrate metabolic process (GO:0005975; TurB: p=0.018; TurY: p=0.030). The TurB module produced additional GO terms related to cell wall growth and development, including ‘cell wall polysaccharide metabolic process,’ ‘cell wall organization,’ and ‘xyloglucan:xyloglucosyl transferase activity.’ The TurY module, though, produced more terms related to ATP synthesis, including ‘ATP generation from ADP.’ Both the TurB and TurY modules produced terms that could be indirectly connected with cell wall or ATP synthesis, including ‘protein serine/threonine kinase activity,’ ‘integral component of membrane,’ ‘glycolytic process,’ ‘glucosyltransferase activity,’ and ‘protein dimerization activity.’ Both the TurB and TurY gene expression pathways were also strongly, positively correlated with soaking time (ST) and moderately, negatively correlated with cooking time (CT) (TurB ST: 𝑟2=0.96, p≤0.001; TurB CT: 𝑟2=-0.73, p≤0.05; TurY ST: 𝑟2=0.93, p≤0.001; TurY CT: 𝑟2=-0.64, p≤0.05). Highly co-expressed genes, or hub genes, were identified in all the modules found in the brown and yellow beans (Figs. SI 2.6a-i). Hub genes may function as central regulators of gene expression pathways since they are highly correlated not just with phenotypic expression but also with the expression rates of the other genes in the module. For this reason, the functions of the hub genes were investigated using the KEGG Orthology (KO) database (Table 2.2). The hub genes in the TurB module participate in pyruvate synthesis and monolignol biosynthesis. TurY hub genes could not be annotated by KO. 68 Table 2.2: KEGG module annotations from the KO database for each set of hub genes. There were thirty hub genes per WGCNA module. The number of hub genes found within a WGCNA and a KEGG module is shown in parentheses. Module name Hub gene modules (KEGG) TurB BroB RedB MblB TurY BluY BroY RedY M00002 Glycolysis, core module involving three-carbon compounds (1) M00001 Glycolysis (Embden-Meyerhof pathway), glucose => pyruvate (1) M00039 Monolignol biosynthesis, phenylalanine/tyrosine => monolignol (1) M00565 Trehalose biosynthesis, D-glucose 1P => trehalose (1) M00086 beta-Oxidation, acyl-CoA synthesis (1) M00854 Glycogen biosynthesis, glucose-1P => glycogen/starch (1) M00154 Cytochrome c oxidase (1) M00366 C10-C20 isoprenoid biosynthesis, plants (1) M00364 C10-C20 isoprenoid biosynthesis, bacteria (1) M00160 V-type ATPase, eukaryotes (1) n/a n/a M00125 Riboflavin biosynthesis, plants and bacteria, GTP => riboflavin/FMN/FAD (2) M00840 Tetrahydrofolate biosynthesis, mediated by ribA and trpF, GTP => THF (1) M00546 Purine degradation, xanthine => urea (1) M00157 F-type ATPase, prokaryotes and chloroplasts (1) M00097 beta-Carotene biosynthesis, GGAP => beta-carotene (1) WhiY M00307 Pyruvate oxidation, pyruvate => acetyl-CoA (1) Several hub genes were differentially expressed in fast- and slow-cooking beans (Figs. SI 2.6a, e). The DE hub gene in the TurB module (Phvul.001G029100) was annotated as uncharacterized conserved protein UCP031279. This gene was more expressed in the slow- cooking brown bean, TZ-27, at hour 6 of soaking [log fold difference in expression (LFDE) = - 0.41] (Table SI 2.7). Three hub genes were found in the TurY module, but only one of these genes (Phvul.008G095400) was differentially expressed between the two yellow beans. This gene was annotated as thiamine pyrophosphate dependent pyruvate decarboxylase family protein, and it was more expressed in the slow-cooking yellow bean, PI527538, at hour 6 of soaking (LFDE = -0.39). 69 The second largest modules in the brown and yellow beans (BroB and BluY, respectively) produced GO terms related to protein metabolism (Figs. SI 2.5a, d). The top three most significant GO terms produced by the BroB module, for instance, were ‘peptide transport,’ ‘amide transport,’ and ‘alpha-amino acid metabolic process.’ The BluY module also produced terms like ‘cellular protein metabolic process,’ ‘protein kinase activity,’ and ‘endoplasmic reticulum,’ although these were not the most significant terms. Unlike BluY, BroB produced terms related to carbohydrate metabolism, including ‘hydrolase activity, hydrolyzing O-glycosyl bonds’ and ‘cellular carbohydrate metabolic process’ as well as terms related to macromolecule transport, including ‘macromolecule localization’ and ‘organic substance transport.’ Unlike BroB, BluY produced GO terms related to transcription like ‘helicase activity,’ ‘chromosome,’ and ‘nucleic acid binding,’ terms related to protein modification like ‘ubiquitin-like protein transferase activity,’ ‘ubiquitin-protein transferase activity,’ ‘catalytic activity, acting on a protein,’ etc., and terms related to protein kinases like ‘protein kinase activity,’ ‘ATP binding,’ and ‘protein phosphorylation.’ Hub genes in the BroB module were involved in processes related to energy metabolism, including ATP and starch biosynthesis. BluY hub genes could not be annotated by KO. The MblB and BroY modules produced GO terms related to translation (Figs. SI 2.5c, e). These modules were positively correlated with soaking time and negatively correlated with cooking time in the brown and yellow beans. Both modules produced the terms ‘translation’ and ‘ribosomal subunit modules’ as well as terms related to interactions with RNA, including ‘RNA binding’ and ‘acting on RNA.’ Of note, BroY was the most strongly correlated with cooking time out of all the modules. MlbB hub genes are involved in generating a proton gradient in organelles to increase ATP synthesis, whereas BroY hub genes were involved in the synthesis of 70 riboflavin, tetrahydrofolate, and urea. Smaller modules of note include RedY, RedB, and WhiY. The RedY module produced GO terms like ‘mitochondrion’ and ‘generation of precursor metabolites and energy’ (Fig. SI 2.5f). This module was negatively correlated with soaking time and positively correlated with cooking time. The RedB and WhiY modules did not produce enough GO terms for the function of the module to be discerned (Figs. SI 2.5b, g). The functions of the hub genes in these modules, however, were able to be discerned. RedY hub genes are involved in generating a proton gradient in chloroplasts to increase ATP synthesis and beta-carotene synthesis. RedB hub genes are involved in generating a proton gradient to increase ATP synthesis and isoprenoid synthesis. WhiY hub genes participate in aerobic respiration. Candidate Genes Identified in QTL Associated with Fast Cooking Times This study overlapped QTL from previous studies with novel transcriptomic data to find intersections between genetic variation for cooking time and gene expression during soaking. In this study, the sizes of the YYRIL QTL intervals were normalized to a high-density linkage map to more accurately identify genes that affect cooking time. The TTRIL and YYRIL cooking time QTL used in this study physically co-localized on chromosomes 2 and 5. By overlapping genes within QTL for cooking time and water uptake with lists of genes from DESeq2 and WGCNA outputs using the merge function in base R (v. 3.6.2), the candidate genes for the genetic control of cooking time and water uptake were narrowed down considerably (Fig. SI 2.7). DEGs detected in slow- and fast-cooking beans using DESeq2 were filtered using genes in WGCNA modules and genes in QTL for both water uptake and cooking time yielding 109 candidate genes. The functions of these candidate genes for cooking time were subsequently determined. This list of genes is provided as Supplementary Data (Table SI 2.7). 71 The QTL for cooking time with the highest R2 values—CT.3.1_Mo17, CT.6.1_Ar17, CT.11.1_Ar16, CT.8.2_Mi16, and CT.10.1_Mi17—contained 1, 4, 4, 0, and 0 candidate genes, respectively. The expression levels of 76 out of the 109 candidate genes significantly changed during soaking, showing that these genes were being actively transcribed in the beans in response to water uptake (Table SI 2.8). Of these 76 genes, 54 of them were differentially expressed between the fast- and slow-cooking genotypes as a direct result of water uptake. Four genes that were only found within the two QTL that were discarded [CT.8.1 (MiC) and WU.9.1 (MiC)] were removed from the final data set (Table 2.3). These genes were placed on a physical map of the P. vulgaris genome beside the QTL for water uptake and cooking time (Fig. SI 2.8). 72 Table 2.3: Log-fold differences in expression of up and downregulated genes in brown and yellow beans. These genes were differentially expressed during soaking as well as between fast and slow cooking beans in the same market class. Hour of DE shows the hour that the gene was differentially expressed during soaking and was differentially expressed between fast and slow cooking beans. Gene counts are expressed in RPKM. Log-fold gene expression values were calculated as log(fast/slow). Significant log-fold differences in gene expression are shown in bold text. Market class Higher expression Annotation Gene QTL Brown Fast SKU5 similar 5 Brown Fast Brown Fast Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Responsive to abscisic acid 28 Auxin response factor 16 Salt tolerance zinc finger Uncharacterized conserved protein UCP031279 Plant protein of unknown function (DUF868) Plant stearoyl- acyl-carrier- protein desaturase family protein Calcium-binding EF-hand family protein Phvul. 011G025400 Phvul. 005G035800 Phvul. 005G134500 Phvul. 001G026700 Phvul. 001G029100 CT.11.1 (Ar_16) WU.5.1 (Ar_16) CT.5.2 (Mo_17) WU.1.1 (Mo_17) WU.1.1 (Mo_17) Phvul. 001G046800 CT.1.2 (Ar_17) Phvul. 001G064100 CT.1.2 (Ar_17) Phvul. 001G067400 CT.1.2 (Ar_17) Brown Slow Senescence- related gene 3 Phvul. 001G091000 CT.1.2 (Ar_17) Hour of DE during soaking 3 6 6 6 6 6 6 6 6 71 0 3 6 12 18 Slow (RPKM) Fast (RPKM) 0.52 0.68 -0.29 0.1 -0.10 2 8 0.10 0.11 0.45 0.02 -0.08 156 443 -0.03 0.12 0.45 0.03 0.00 2 0.01 0.05 -0.28 -0.01 -0.02 143 6 76 -0.01 0.1 -0.41 0.21 -0.08 329 127 0.53 -0.22 -0.74 -0.18 -0.39 7 1 -0.05 0.24 -0.52 0.41 -0.03 184 56 0.25 -0.05 -0.79 0.17 0.00 85 14 0.04 -0.10 -0.36 0.31 0.07 152 67 Table 2.3 (cont’d): Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Ortholog of sugar beet HS1 PRO-1 2 Hypoxia- responsive family protein Xyloglucan endotrans- glycosylase/hydro lase 15 Leucine-rich repeat protein kinase family protein Xyloglucan endotrans- glycosylase 6 Unknown function Xyloglucan endotrans- glycosylase 6 Xyloglucan endotrans- glycosylase/hydro lase protein Xyloglucan endotrans- glycosylase/hydro lase protein Phvul. 001G241300 CT.1.1 (Mo_17) Phvul. 001G248400 CT.1.1 (Mo_17) Phvul. 001G265300 CT.1.1 (Mo_17) Phvul. 001G265500 CT.1.1 (Mo_17) Phvul. 003G137600 WU.3.2 (Ar_16) Phvul. 003G140426 Phvul. 003G147400 WU.3.2 (Ar_16) WU.3.2 (Ar_16) Phvul. 003G147600 WU.3.2 (Ar_16) Phvul. 003G147700 WU.3.2 (Ar_16) Brown Slow Tetraspansin 8 Phvul. 003G151800 WU.3.2 (Ar_16) 6 6 6 6 6 6 6 6 6 6 72 -0.13 -0.15 -0.55 0.12 -0.27 79 22 0.72 -0.00 -0.46 0.48 0.10 16 6 0.07 0.02 -0.75 -0.21 -0.30 129 23 0.25 0.15 -0.43 0.00 -0.02 3 0.19 -0.76 -0.75 0.37 -0.03 4 0.17 0.28 -0.56 0.44 0.07 50 undef -0.25 -0.78 0.42 -0.44 5 1 1 14 1 -0.14 0.09 -0.53 0.16 -0.11 104 31 -0.05 0.19 -0.50 0.14 -0.27 829 265 0.18 -0.08 -0.40 0.19 -0.28 33 13 Table 2.3 (cont’d): Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Brown Slow Early nodulin- related Unknown function WUS-interacting protein 2 Unknown function Unknown function Unknown function Basic helix-loop- helix (Bhlh) DNA-binding superfamily protein Phosphoglycerate kinase HSP20-like chaperones superfamily protein Brown Slow Expansin-like A1 Brown Slow SKU5 similar 5 Yellow Fast Highly ABA- induced PP2C gene 3 Phvul. 003G162200 Phvul. 003G165200 Phvul. 003G167500 Phvul. 005G019201 Phvul. 005G019300 Phvul. 005G019501 Phvul. 005G036900 Phvul. 005G050800 Phvul. 006G150600 Phvul. 011G024801 Phvul. 011G102500 Phvul. 002G163900 6 6 6 6 6 6 6 6 6 6 6 6 WU.3.2 (Ar_16) WU.3.2 (Ar_16) WU.3.2 (Ar_16) WU.5.1 (Ar_16) WU.5.1 (Ar_16) WU.5.1 (Ar_16) WU.5.1 (Ar_16) WU.5.2 (Mo_16) CT.6.1 (Ar_17, Mo_17) CT.11.1 (Ar_16) CT.11.2 (Ar_16, Mo_17) CT.2.1 (Mi_C, Mi_16) 73 0.04 -0.02 -0.48 0.53 0.06 93 -0.01 -0.01 -0.69 0.60 -0.23 31 0.17 -0.33 -0.44 -0.03 -0.17 130 0.26 0.07 -0.36 0.17 0.03 201 -0.71 -0.03 -0.46 0.10 -0.17 80 -0.15 -0.28 -0.43 0.09 -0.16 112 0.07 0.30 -0.37 0.09 -0.28 24 0.00 0.15 -0.23 0.24 -0.22 81 0.48 0.16 -0.57 -0.13 0.21 10 -0.11 0.17 -0.41 -0.08 -0.28 93 0.24 0.20 -0.42 0.04 0.03 9 31 6 47 88 28 41 10 47 3 36 3 -0.06 0.00 0.17 0.05 -0.00 243 357 Table 2.3 (cont’d): Yellow Fast Yellow Fast Yellow Fast Yellow Fast HXXXD-type acyl-transferase family protein NAD(P)-linked oxidoreductase superfamily protein Major facilitator superfamily protein UDP-Glycosyl- transferase superfamily protein Phvul. 004G040200 WU.4.2 (Mi_17) Phvul. 008G091300 CT.8.1 (Mi_17) Phvul. 009G120400 WU.9.1 (Mi_16) Phvul. 009G120600 WU.9.1 (Mi_16) 6 6 6 6 0.00 0.13 0.18 -0.29 0.16 143 216 0.02 0.13 0.40 0.04 0.38 14 35 -0.14 0.07 0.21 0.17 -0.18 57 0.00 0.24 0.26 -0.27 0.32 13 92 23 Yellow Fast Unknown function Phvul. 005G019300 Yellow Slow F-box family protein Phvul. 002G222100 Yellow Slow HXXXD-type acyl-transferase family protein Phvul. 008G031900 Yellow Slow EXORDIUM like 2 Phvul. 002G210200 Yellow Slow ARM repeat superfamily protein Phvul. 008G071700 CT.5.1 (Mi_C, Mi_16) CT.2.2 (Mi_C, Mi_16) CT.8.1 (Mi_17), CT.8.2 (Mi_C) CT.2.1 (Mi_16), CT.2.2 (Mi_C, Mi_16) CT.8.1 (Mi_17), CT.8.2 (Mi_C) 12 0.02 0.23 0.17 0.38 0.02 23 34 3 3 6 0.05 -0.35 -0.05 0.10 -0.02 154 69 -0.66 -1.07 -1.32 -0.30 -0.64 23 -0.22 -0.29 -0.37 -0.24 -0.08 18 2 8 6 -0.07 -0.03 -0.44 -0.47 -0.02 11 4 74 Table 2.3 (cont’d): Yellow Slow Yellow Slow Yellow Slow Yellow Slow Yellow Slow Yellow Slow Thiamine pyrophosphate dependent pyruvate decarboxylase family protein Salt tolerance zinc finger Polynucleotidyl transferase, ribonuclease H- like superfamily protein SBP (S- ribonuclease binding protein) family protein Class I glutamine amido- transferase-like superfamily protein Yellow Slow Sugar transporter 1 Phvul. 002G221200 Yellow Slow Sphingosine kinase 1 Phvul. 002G222800 HXXXD-type acyl-transferase family protein Phvul. 008G031900 CT.8.1 (Mi_17), CT.8.2 (Mi_C) 6 6 6 6 -0.66 -1.07 -1.32 -0.30 -0.64 14 1 0.04 -0.11 -0.39 0.10 -0.23 75 31 -0.02 -0.15 -0.25 -0.14 0.11 74 41 -0.03 -0.01 -0.19 -0.19 -0.15 360 230 Phvul. 008G095400 CT.8.1 (Mi_17) Phvul. 009G070800 Phvul. 009G088600 WU.9.1 (Mi_16) WU.9.1 (Mi_16) Phvul. 009G119200 WU.9.1 (Mi_16) 6 -0.09 -0.12 -0.41 -0.07 -0.05 40 15 Phvul. 002G214100 CT.2.1 (Mi_16), CT.2.2 (Mi_C, Mi_16) CT.2.2 (Mi_C, Mi_16) CT.2.2 (Mi_C) 12 -0.21 -0.30 -0.23 -0.74 -0.12 5 1 12 -0.15 0.06 -0.05 -0.26 0.12 244 134 12 -0.21 0.11 0.06 -0.34 0.05 39 18 75 Table 2.3 (cont’d): Yellow Slow Yellow Slow Yellow Slow Yellow Slow HXXXD-type acyl-transferase family protein Thioredoxin superfamily protein Unknown function HXXXD-type acyl-transferase family protein Phvul. 004G040200 WU.4.2 (Mi_17) Phvul. 004G055600 WU.4.2 (Mi_17) 12 0.00 0.13 0.18 -0.29 0.16 191 99 12 -0.01 0.11 0.12 -0.29 0.07 38 19 Phvul. 005G025500 Phvul. 008G032200 12 12 CT.5.1 (Mi_C) CT.8.1 (Mi_17), CT.8.2 (Mi_C) -0.03 0.09 -0.20 -0.37 -0.25 328 141 -0.26 -0.20 -0.22 -0.55 0.17 25 7 76 Candidate genes identified in the brown beans Candidate genes that were more expressed in TZ-27 included genes involved in cell wall modification and expansion (xyloglucan endotransglycosylase 6/15, expansin-like A1, arabinogalactan protein 15, and leucine-rich repeat extensin-like protein 5), a gene involved in stress-induced calcium-dependent cytosolic signaling (calcium-binding EF-hand family protein), and multiple genes that participate in stress-inducible plant development pathways [ARM repeat superfamily protein, salt tolerance zinc finger, leucine-rich repeat protein kinase family protein, basic helix-loop-helix (Bhlh) DNA-binding superfamily protein, and phosphoglycerate kinase] (Mohanta et al., 2019). Stress-responsive defense genes (plant-stearoyl-acyl-carrier-protein desaturase family protein, senescence-related gene 3, ortholog of sugar beet HS1 PRO-1, 2, hypoxia-responsive family protein, and Hsp20-like chaperone superfamily protein) were also more expressed in TZ-27. Genes involved in plant development and germination were expressed in both TZ-27 and TZ-37. For example, tetraspansin 8, early-nodulin related, and WUS- interacting protein were more expressed in TZ-27, whereas responsive to abscisic acid 28 and auxin response factor 16 were more expressed in TZ-37 (Jimenez-Jimenez et al., 2019). One gene described as SKU5 similar 5 was more expressed in TZ-37 after 3 hrs of soaking but was more expressed in TZ-27 after 6 hrs of soaking. Candidate genes identified in the yellow beans Candidate genes that were more expressed in the slow-cooking yellow bean, PI527538, included genes that regulate stress-inducible development and cell expansion pathways [ARM repeat superfamily protein, F-box family protein, exordium like 2, salt tolerance zinc finger, SBP (S-ribonuclease binding protein) family protein, a protein with an unknown function containing an ARGOS-like protein domain, and sphingosine kinase 1], genes that respond to anoxia 77 (thiamine pyrophosphate dependent pyruvate decarboxylase family protein), genes that catalyze the formation of carbon-nitrogen bonds (class I glutamine amidotransferase-like superfamily protein), genes that regulate target proteins via redox reactions (thioredoxin superfamily protein), and one gene that transports monosaccharides (sugar transporter 1) (Schröder et al., 2009; Kuluev et al., 2019). By contrast, genes associated with wound and abscisic acid-induced repression of germination (highly ABA-induced PP2C gene 3), stress-inducible metabolite modification [(NAD(P)-linked oxidoreductase superfamily protein, UDP-glycosyltransferase superfamily protein], and targeted chemiosmotic-driven solute transport (major facilitator superfamily protein) were more expressed in the fast-cooking yellow bean, Ervilha (Bhaskara et al., 2012; Zhang et al., 2013). HXXXD-type acyl-transferase family proteins (a.k.a. phenolic glucoside malonyltransferase 1) were expressed in PI527538 after 6 hrs of soaking and in Ervilha after 3, 6, and 12 hrs of soaking. Candidate genes identified in both the brown and yellow beans There was some overlap in the candidate gene sets of the yellow and brown beans. Both slow-cooking beans upregulated zinc finger family proteins relative to their faster cooking counterparts. TZ-27 upregulated Phvul.001G026700 after 6 hours of soaking and PI527538 upregulated Phvul.009017800 after 6 hours of soaking. These zinc finger family proteins were both homologous with zinc finger protein ZAT10 according to SWISS-PROT. They also both contained zf-C2H2_6 domains, whereas zf-RING_2 domain-containing proteins were upregulated in both the fast- and the slow-cooking beans. In addition, TZ-27 and PI527538 expressed more ARM repeat superfamily proteins (Phvul.003G084700 and Phvul.008G071700, respectively) transcripts than the fast-cooking beans. Both gene sequences were homologous with Arm protein-encoding regions according to Pfam and with U-box domain-containing 78 protein 4 according to SWISS-PROT, meaning they could have similar molecular functions. Biochemical differences between fast- and slow-cooking beans The activity of pectin methylesterase enzymes did not change in either TZ-27 or TZ-37 due to soaking. Pectin methylesterase activity was higher overall in TZ-27 than it was in TZ-37 (Table 2.4). 79 Table 2.4: Average protein and mineral levels in TZ-27, TZ-37, PI527538, and Ervilha beans, and average pectin methylesterase (PME) activity in TZ-27 and TZ-37 beans. Beans were either unsoaked (US) or soaked (S) for 12 hours in distilled water prior to analysis. Values are percentages (%) of seed dry weight. PME activity was measured in beans grown at Montcalm, MI in 2018. All other compounds were measured in beans grown at SVERC, MI in 2019. Values with the same letter within a column are not significantly different from each other (LSD, α = 0.05). Genotype Treatment Protein Calcium Magnesium Potassium Phosphorous Sulfur Phytate PME activity (%) (%) (%) (%) (%) (%) (g/100 g) ((Δnmol/Δmin)/mg) TZ-27 TZ-37 US S US S 19.3a 0.18b 0.18a 20.9ab 0.20b 0.21bc 23.2c 0.10a 0.18ab 1.40a 1.57a 1.45a 24.9c 0.11a 0.19abc 1.48a PI527538 US 19.6ab 0.18ab 0.20abc S 20.4ab 0.15ab 0.22c Ervilha US 19.8ab 0.20b 0.18a 1.47a 1.57a 1.45a S 21.2b 0.18ab 0.19abc 1.57a 0.38a 0.44a 0.42a 0.45a 0.41a 0.45a 0.44a 0.48a 0.18a 1.01ab 0.2 0.20abc 0.982ab 0.202 0.21bc 1.15b 0.122 0.22cd 1.13b 0.137 0.19ab 0.888a 0.20abc 0.931a 0.22cd 0.944a 0.23d 1.01ab . . . . 80 TZ-27 seeds contained more calcium and protein than TZ-37 seeds. In the yellow beans, sulfur levels were higher in Ervilha compared to PI527538. Both the fast-cooking beans contained more sulfur than their slow-cooking counterparts prior to soaking (Table 2.4). The amount of copper, iron, zinc, manganese, and boron in the seeds can be found in the supplemental data (Table SI 2.9). Soaking did not significantly affect the phytate content of the beans. When the soaking treatment was excluded from the linear model, TZ-27, PI527538, and Ervilha had similar phytate concentrations, but TZ-37 had significantly more phytate than the other genotypes (Table 2.4). The biological significance of the WGCNA co-expression modules DISCUSSION In WGCNA, a correlation between a module/putative gene expression pathway and a trait indicates an association between transcript expression and trait expression. If a module was significantly positively correlated with cooking time and negatively correlated with soaking time, for instance, this could mean that 1) transcript expression tended to decrease as soaking time increased, 2) gene expression was generally higher in the slow-cooking bean, and 3) genes in the module could be or are a part of a biological pathway controlling cooking times in slow- and fast-cooking beans. In general, modules that were negatively correlated with soaking time were positively correlated with cooking time and vice versa, likely because cooking time decreased as soaking increased. Since the largest gene co-expression modules and their hub genes were positively associated with soaking time and were annotated with GO terms such as cell wall development and expansion, response to stress, and hormone-mediated signaling pathways, the hormonal control of cell wall modification in seeds during prolonged imbibition was investigated (Figs. 80 2.4-2.5; Table 2.2). Genetic mechanisms controlling cooking time in dry beans Hormonal control of cell wall modification The candidate genes identified suggest that auxin modulated gene expression in the fast- and slow-cooking beans during soaking. For example, an auxin-responsive gene identified in the candidate gene list (auxin responsive factor 16) was more expressed in the fast-cooking brown bean, TZ-37, at hour 6 of soaking (Table SI 2.7). In addition, genes involved in or targeted by hormonal signal transduction pathways, including miRNA-transport-related proteins like C2H2- domain zinc finger ZAT10, osmotic stress resistance genes, and genes with antioxidant functions were upregulated only in the slow-cooking beans (Han et al., 2020). Imbibition-induced hormone signaling pathways can lead to the differential expression of cell wall structural genes which could influence cooking time. Chickpea seeds, for instance, express auxin signaling genes in preparation for germination (Kudapa et al., 2018). In the embryonic axes of developing seeds, auxin signaling has been shown to loosen cell walls by activating expansins and XTHs, thereby enabling the cells to expand without automatically losing their integrity (Gimeno-Gilles et al., 2009; Wang and Ruan, 2013; Samalova et al., 2023). As a result of cell wall expansion, the cell wall can become more vulnerable to the activity of cell wall modification enzymes like pectin methylesterases. One study showed that the overexpression of expansin enzymes was able to increase the rate of pectin demethylesterfication in Arabidopsis root cells within three hours, leading to stiffer cell walls within six hours (Samalova et al., 2023). Changes in cell wall integrity could then lead to changes in cooking time (Chigwedere et al., 2019; Chen et al., 2022). 81 Effects of hypoxic stress on cell wall modification The results of the GO analysis show that the beans were undergoing hypoxic stress due to increasingly limited oxygen levels in the water during the later stages of soaking (Rajashekar and Baek et al., 2014; Nakayama et al., 2017; Ren et al., 2017; Butsayawarapat et al., 2019; Wang et al., 2019; Borrego-Benjumea et al., 2020; Sharmin et al., 2020). There is also evidence that genes conferring tolerance to stress were more expressed in slow-cooking beans compared to the fast-cooking beans. Firstly, genes associated with flooding tolerance in soybean, including expansin-like A1 (Phvul.011G024801), SKU5 similar 5 [putative pectin methylesterase (Phvul.011G102500)], xyloglucan endotransglycosylase 6 (Phvul.003G147400 and Phvul.003G137600), and xyloglucan endotransglycosylase/hydrolase 15 (Phvul.001G265300) were more expressed in TZ-27 during soaking relative to TZ-37 at hour 6 of soaking (Fig. 2.6; Table 2.3) (Sharmin et al., 2020). 82 Figure 2.6: Grouped boxplots of gene expression levels (in RPKM) of genes found within QTL for cooking time. Significant differences in gene expression between two genotypes at a particular soaking time are indicated with asterisk brackets, wherein * represents p≤0.05, ** represents p≤0.01**, and *** represents p≤0.001. Differentially expressed genes were identified in DESeq2 using Benjamini-Hochberg adjusted p-values and were filtered out if they did not appear within QTL for cooking time or water uptake. Gene expression levels in beans from different market classes (i.e., TZ-27 vs. Ervilha) were not compared. Data collected after 12 hours of soaking were not filtered. 83 Secondly, genes that generally increase tolerance to abiotic stress like EF-hand calcium- binding protein and exordium like 2 were upregulated in TZ-27 and PI527538, respectively, relative to their fast-cooking counterparts at hour 6 of soaking (Fig. 2.6) (Decreux and Messiaen, 2005; Schröder et al., 2009; Wang et al., 2013; Wu et al., 2018). Thirdly, out of the 109 genes that were differentially expressed between the fast- and slow-cooking genotypes, found within a WGCNA module, and found within one or more QTL for cooking time or water uptake, 101 of them (92.6%) were also found within QTL regions that overlapped with one or more QTL for flooding tolerance (Table SI 2.7) (Soltani et al., 2018). Lastly, other studies have found that seeds with pigmented/darker seed coats (which is a common characteristic of slow-cooking beans) are associated with flooding tolerance (Stanley et al., 1997; Lima et al., 2021). Genetic mechanisms controlling cooking time in the brown beans Although previous candidate genes for cooking time, including polygalacturonase, rhamnogalacturonase, pectate lyase, peroxidase, and phytase did not appear in the list of brown bean candidate genes, calcium-binding EF hand family proteins were upregulated in the slow- cooking beans (Table SI 2.2) (Stanley, 1992; Kilmer et al., 1994; Martinez-Manríque et al., 2011; Njoroge et al., 2015; Sadohara et al., 2022). EF-hand proteins can alter the concentration of cytosolic calcium in response to stress, leading divalent cations like calcium to crosslink pectin within cell walls (Wu et al., 2018; Mohanta et al., 2019). Increases in calcium concentration in the cytoplasm of TZ-27 could cause the cotyledon cell walls to break down more slowly during cooking (Cominelli et al., 2020; Temple et al., 2022). Indeed, TZ-27 contained about 2 times more calcium than TZ-37 both before and after soaking (Table 2.4). Calcium-mediated pectin dimerization in cotyledons has previously been associated with slower cooking times in dry beans and chickpea (Wood et al., 2018; Chigwedere 84 et al., 2018; Chigwedere et al., 2019). However, further studies will be needed to determine whether calcium accumulates more in the cotyledon cell walls of TZ-27 relative to those of TZ- 37 (as opposed to the seed coat, where most of the calcium in dry bean seeds is stored), as well as whether calcium-pectin crosslinks are more abundant in TZ-27 cotyledons compared to TZ-37 (Cominelli et al., 2020). Alternatively, higher levels of phytate could be preventing calcium from binding to the cotyledon cell walls of TZ-37 (Fig. 2.7). Differences in pectin methylesterase activity have been observed in TZ-27 and TZ-37 (Table 2.4). According to transcriptomic evidence, two different genes that are partially homologous with a dry bean pectin methylesterase gene were upregulated in TZ-37 (Phvul.011G025400) and TZ-27 (Phvul.011G102500) after 3 and 6 hrs of soaking, respectively (Fig. 2.6, Table 2.3). This is a potentially important finding, since pectin methylesterase has been suspected to play a role in cooking time (Martinez-Manríque et al., 2011). However, experimental evidence has shown that these proteins—called SKU5 similar 5—do not regulate pectin methylesterfication levels, but rather control cell wall expansion in roots and hypocotyls (Sedbrook et al., 2002). In all, more information will be needed to determine why TZ-27 has higher pectin methylesterase activity. It should be determined, for instance, whether pectin methylesterase and pectin methylesterease inhibitor proteins accumulate more in TZ-27 prior to seed maturity, thereby increasing pectin methylesterase activity immediately following water uptake (Tolili et al., 2022). In addition, the effects of mutations on pectin methylesterase activity should be explored. Such analyses would ideally incorporate proteomic data in addition to genomic and transcriptomic data. Enzymes controlling plant development like tetraspansin 8, early nodulin-related, and WUS-interacting protein 2 as well as cell wall modification enzymes like xyloglucan 85 endotransglycosylase/hydrolase (XTH) and expansin-like A1 protein (EXPA1) were upregulated in TZ-27 in response to abiotic stress; but the upregulation of these genes can also lead to cell wall modifications in legumes during germination (Zhu et al., 2022). This finding appears to contradict previous findings that actively expanding cell walls are more pliable and easier to break down than non-actively expanding cell walls. For example, XTH activity has been shown to accelerate post-harvest softening in strawberry fruits in combination with polygalacturonase activity (Witasari et al., 2019). The difference here could be caused by the higher activity of cell wall-strengthening enzymes like pectin methylesterase, lower phytate concentrations, and higher calcium concentrations that were measured in TZ-27 relative to TZ-37 (Table 2.4) (Witasari et al., 2019). The expression of XTH and EXPA1 could make it easier for pectin methylesterase proteins, intracellular calcium, and pectin-calcium crosslinking enzymes to access pectin that is otherwise inaccessible to enzymatic activity, leading to textural hardening in the beans during cooking (Fig. 2.7) (Peaucelle et al., 2012; Chen et al., 2022). On the other hand, an XTH variant in chickpea has been shown to limit cell wall expansion in response to biotic stress. However, lack of cell wall expansion could likewise lead to cell wall strengthening and thickening in slow- cooking beans (Niraula et al., 2021). 86 Figure 2.7: Diagram showing the water uptake pathway in dry bean. The zoomed in box shows the genetic and biological changes that take place in fast- and slow-cooking dry beans during soaking and cooking. Genetic mechanisms controlling cooking time in the yellow beans Unlike the brown beans, the yellow beans had similar calcium levels both before and after soaking (Table 2.4). Since consistent, significant differences in cooking time between the genotypes were only observed after 12 hours of soaking and calcium levels did not change after 12 hours of soaking, this suggests that another factor had a differential effect on the cooking times of the yellow beans during water uptake. While it is possible that more calcium accumulated in the cotyledon cell walls of PI527538 than in those of Ervilha during soaking, leading to different cooking times, there was no evidence in the gene expression data set that the soaking treatment affected calcium redistribution pathways in the yellow beans. Yet, exordium- like 2, an enzyme that facilitates the formation of divalent cation crosslinks between pectin chains, was more expressed in PI527538 than Ervilha. In the future, the location and abundance 87 of calcium and calcium-pectin crosslinks before and after soaking should be quantified in these genotypes to determine whether calcium crosslinks contribute to the differences in cooking time in these beans. The results of this study provide evidence that changes in gene expression and/or enzymatic activity during water uptake influenced cell wall rigidity and permeability in the yellow beans, thereby affecting their cooking times. Plant hormones and/or hypoxic stress, for example, could have increased cell wall rigidity in the yellow beans by regulating transcription factors controlling cell wall biosynthesis. In plants, secondary cell wall development is regulated by multiple tiers of transcription factors (tiers I, II, and III), wherein the tier III transcription factors (a.k.a. master regulators) control the expression of tier II transcription factors, tier II factors control tier I factors, and tier I factors regulate the expression of secondary cell wall biosynthesis genes. Importantly, a tier III transcription factor, NAC-like (Phvul.002G170200), activated by AP3/PI, was more expressed in PI527538 relative to Ervilha, but was not differentially expressed because of soaking, meaning that PI527538 expressed more NAC-like protein transcripts than Ervilha regardless of water uptake (Fig. 2.6, Table SI 2.7). Interestingly, NAC knockout Arabidopsis seedlings exhibit thinner secondary cell walls and lower anthocyanin accumulation, leading to impaired stress tolerance (Jeong et al., 2018). The abundance of NAC transcripts has also been positively associated with anthocyanin accumulation and stress tolerance in birch seedlings (Hu et al., 2019). Hence, it is possible that higher NAC-like transcript expression leads to anthocyanin accumulation and improved stress tolerance in PI527538 (Fenger, 2020; Liu et al., 2021). A gene family involved in condensed tannin biosynthesis (HXXXD-type acyl-transferase family proteins) was highly expressed in both PI527538 and Ervilha, but only PI527538 88 upregulated multiple HXXXD genes (Phvul.004G040200, Phvul.008G031900, and Phvul.008G032200) at multiple soaking times starting at hour 3 of soaking (note that Ervilha had a significantly longer cooking time than PI527538 at hour 3 of soaking) (Fig. 2.6, Table 2.3). All three of these genes were annotated as phenolic glucoside malonyltransferase 1, a member of the anthocyanin acyltransferase subfamily that acylates a range of flavonoid and naphthol glucosides. This process solubilizes and stabilizes phenols which can make them easier to deposit into cell walls/vacuoles and prevent their enzymatic degradation (Manjasetty et al., 2012). One by-product of these reactions, malonyl-CoA, can lead to the synthesis of leucoanthocyanidins and anthocyanins (Fenger, 2020). Leucoanthocyanidins can polymerize in the bean testa to form condensed tannins, an organic compound that contributes to postharvest darkening. Postharvest darkening is thought to increase the cooking time of beans by creating insoluble cell wall material, thereby increasing the mechanical strength of the cell wall and preventing water uptake (Stanley, 1992). Increased expression of HXXXD genes could help explain the higher total and insoluble fiber content, higher cotyledon cell wall thickness, and higher insoluble cotyledon cell wall isolate content that previously were observed in PI527538 relative to Ervilha (Fig. 2.7). Lower accumulation of condensed tannins was previously suggested to cause fast cooking times in Manteca-type beans like Ervilha (Sadohara et al., 2022). In dry bean, the J gene is involved in the control of postharvest seed coat darkening. Previously, a GWAS study found that a SNP within the J locus was associated with increased cooking times in beans (Erfatpour et al., 2018; Sadohara et al., 2022). In this study, QTL for water uptake and cooking time were close to but did not overlap with the J gene, meaning that this gene was excluded from the analysis. It should be noted, however, that an RNA-binding gene (Phvul.010G133101) very near 89 the J gene was more expressed in PI527538 relative to Ervilha at hours 3 and 6 of soaking. It is possible that higher basal expression levels of this gene affected the abundance of HXXXD transcripts (which are involved in postharvest darkening) during water uptake. More research will be needed to identify the regulatory mechanisms that control or are controlled by the candidate genes from this study. CONCLUSIONS Gene expression analysis was used to understand molecular changes that occur in dry beans with contrasting cooking times during soaking. The fast-cooking beans that we analyzed took 18-19 minutes to completely cook after 12 hours of soaking whereas the slow cooking beans took 23.5-30 mins. The cooking times of the beans decreased drastically from 0-12 hours of soaking and remained stable after 12 hours. Nine gene co-expression modules (i.e., pathways) were significantly associated with changes in both soaking time and cooking time in the beans. Fifty genes within these co-expression modules were also within QTL for water uptake and/or cooking time, differentially expressed in fast- and slow-cooking beans, and differentially expressed in response to water uptake. The functions of the candidate genes were related to hormone-mediated cell wall modification and response to hypoxia. The slow-cooking beans expressed more genes that increase stress tolerance than the fast-cooking beans did. Candidate genes for cooking time differed in brown and yellow beans. Genes that affect cell wall expansion and increase calcium levels in cotyledon cells were upregulated in the slow-cooking brown bean TZ-27 relative to TZ-37. By contrast, genes that form calcium-pectin crosslinks and synthesize condensed tannin precursors were upregulated in PI527538 relative to Ervilha. These candidate genes were found within predicted gene expression modules and QTL for percent water uptake during soaking and cooking time. Differentially expressed genes between the fast- and slow- 90 cooking genotypes may impact the stability of the cell wall in bean cotyledons which could lead to shorter cooking times in dry bean and less resistance to abiotic stress. 91 BIBLIOGRAPHY Alexa, A. & Rahnenführer, J. (2021). Gene set enrichment analysis with topGO. Bioconductor. Anders, S., Pyl, P. T., & Huber, W. (2015). HTSeq—a Python framework to work with high- throughput sequencing data. Bioinformatics, 31(2), 166-169. https://doi.org/10.1093/bioinformatics/btu638 Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data. Babraham Bioinformatics. AOAC International (2000) Official methods of analysis of AOAC International. Association of Official Analytical Chemists. Astudillo-Reyes, C., Fernandez, A. C., & Cichy, K. A. (2015). Transcriptome characterization of developing bean (Phaseolus vulgaris L.) pods from two genotypes with contrasting seed zinc concentrations. PLOS One, 10(9), e0137157. https://doi.org/10.1371/journal.pone.0137157 Bassett, A., Kamfwa, K., Ambachew, D., & Cichy, K. A. (2021). Genetic variability and genome-wide association analysis of flavor and texture in cooked beans (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 134, 959-978. https://doi.org/10.1007/s00122-020-03745-3 Bassett, A., Katuuramu, D. N., Song, Q., & Cichy, K. A. (2021). QTL mapping of seed quality traits including cooking time, flavor, and texture in a yellow dry bean (Phaseolus vulgaris L.) population. Frontiers in Plant Science, 12, 670284. https://doi.org/10.3389/fpls.2021.670284. Bassett, A., Hooper, S. & Cichy, K. A. (2021). Genetic variability of cooking time in dry beans (Phaseolus vulgaris L.) related to seed coat thickness and cotyledon cell wall. Food Research International, 141. https://doi.org/10.1016/j.foodres.2020.109886 Berry, M., Izquierdo, P., Jeffery, H., Shaw, S., Nchimbi-Msolla, S. & Cichy, K. A. (2020). QTL analysis of cooking time and quality traits in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 133, 2291-2305, https://doi.org/10.1007/s00122-020- 03598-w Bhaskara, G. B., Nguyen, T. T., & Verslues, P. E. (2012). Unique drought resistance functions of the highly ABA-induced clade A protein phosphatase 2Cs. Plant Physiology, 160(1), 379-395. https://doi.org/10.1104/pp.112.202408 Bi, C., Lu, N., Xu, Y., He, C., & Lu, Z. (2020). Characterization and analysis of the mitochondrial genome of common bean (Phaseolus vulgaris L.) by comparative genomic approaches. International Journal of Molecular Sciences, 21(11), 3778. 92 https://doi.org/10.3390/ijms21113778 Borrego-Benjumea, A., Carter, A., Tucker, J. R., Yao, Z., Xu, W., & Badea, A. (2020). Genome- wide analysis of gene expression provides new insights into waterlogging responses in barley (Hordeum vulgare L.). MDPI Plants, 9(2), 240. https://doi.org/10.3390/plants9020240 Butsayawarapat, P., Juntawong, P., Khamsuk, O., & Somta, P. (2019). Comparative transcriptome analysis of waterlogging-sensitive and tolerant Zombi pea (Vigna unguiculata) reveals energy conservation and root plasticity controlling waterlogging tolerance. MDPI Plants, 8(8), 264. https://doi.org/10.3390/plants8080264 Cameron, R. G., Cameron, B. S., & Shaw, P. E. (1992). Adaptation of a spectrophotometric assay for pectin methylesterase to a kinetic microplate reader. Journal of Food Science, 57(4), 1006-1008. https://doi.org/10.1111/j.1365-2621.1992.tb14343.x Chen, D., Ding, A., Zhu, L., Grauwet, T., Loey, A. V., Hendrickx, M., & Kyomugasho, C. (2022). Phytate and mineral profile evolutions to explain the textural hardening of common beans (Phaseolus vulgaris L.) during postharvest storage and soaking: insights obtained through a texture-based classification approach. Food Chemistry, 404, 134531. https://doi.org/10.1016/j.foodchem.2022.134531 Chigwedere, C. M., Olaoye, T. F., Kyomugasho, C., Kermani, Z. J., Pallares, A. P., Loey, A. M., Grauwet, T., & Hendrickx, M. E. (2018). Mechanistic insight into softening of Canadian wonder common beans (Phaseolus vulgaris) during cooking. Food Research International, 106, 522-531. https://doi.org/10.1016/j.foodres.2018.01.016 Chigwedere, C. M., Nkonkola, C. M., Rai, S., Kyomugasho, C., Kermani, Z. J., Pallares, A. P., Loey, A. M., Grauwet, T., & Hendrickx, M. E. (2019). Cotyledon pectin molecular interconversions explain pectin solubilization during cooking of common beans (Phaseolus vulgaris).” Food Research International, 116, 462-470. https://doi.org/10.1016/j.foodres.2018.08.062 Cichy, K. A. (2021). Anatomy of a bean: dry bean breeding and production for consumer acceptance. Michigan Bean Commission. Cichy, K. A., Wiesinger, J. A., & Mendoza, F. A. (2015). Genetic diversity and genome-wide association analysis of cooking time in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 128, 1555-1567. https://doi.org/10.1007/s00122-015-2531-z Cominelli, E., Galimberti, M., Pongrac, P., Landoni, M., Losa, A., Paolo, D., Daminati, M. G., Bollini, R., Cichy, K. A., Vogel-Mikuṧ, K., & Sparvoli, F. Calcium redistribution contributes to the hard-to-cook phenotype and increases PHA-L lectin thermal stability in common bean low phytic acid 1 mutant seeds. Food Chemistry, 321, 126680. https://doi.org/10.1016/j.foodchem.2020.126680 93 Darvasi, A. & Soller, M. (1997). A simple method to calculate resolving power and confidence interval of QTL map location. Behavioral Genetics, 27(2), 125-132. Decreux, A. & Messiaen, J. (2005). Wall-associated kinase WAK1 interacts with cell wall pectins in a calcium-induced conformation. Plant Cell Physiology, 46(2), 268-278. https://doi.org/10.1093/pcp/pci026 Diaz, S. (2021). Genetic architecture and genomic prediction of cooking time in common bean (Phaseolus vulgaris L.). Front Plant Sci, 11. https://doi.org/10.3389/fpls.2020.622213 DOE-JGI and USDA-NIFA. (2021). Phaseolus vulgaris v2.1. Durinck, S., Spellman, P. T., Birney, E., & Huber, W. (2009). Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nature Protocols, 4, 1184–1191. https://doi.org/10.1038/nprot.2009.97 Durinck, S., Moreau, Y., Kasprzyk, A., Davis, S., De Moor, B., Brazma, A., & Huber, W. (2005). BioMart and Bioconductor: a powerful link between biological databases and microarray data analysis. Bioinformatics, 21(16), 3439–3440. https://doi.org/10.1093/bioinformatics/bti525 Erfatpour, M., Navabi, A., & Pauls, K. P. (2018). Mapping the non-darkening trait from ‘Wit- rood boontje’ in bean (Phaseolus vulgaris). Theoretical and Applied Genetics, 131, 1331- 1343. https://doi.org/10.1007/s00122-018-3081-y Fenger, J.-A. (2020). Acylated anthocyanins as natural colors: reactivity in aqueous solution, metal binding, stabilization for food applications. University of Avignon. Fu, Y., Wu, P.-H., Beane, T., Zamore, P. D., & Weng, Z. (2018). Elimination of PCR duplicates in RNA-seq and small RNA-seq using unique molecular identifiers. BMC Genomics, 19(531). https://doi.org/10.1186/s12864-018-4933-1 Garcia, R. A. V., Rangel, P. N., Bassinello, P. Z., Brondani, C., Melo, L. C., Sibov, S. T., & Vianello-Brondani, R. P. (2011). QTL mapping for the cooking time of common beans. Euphytica, 186, 779-792. https://doi.org/10.1007/s10681-011-0587-7 Gimeno-Gilles, C., Lelièvre, E., Viau, L., Malik-Ghulam, M., Ricoult, C., Niebel, A., Leduc, N., & Limami, A. M. (2009). ABA-mediated inhibition of germination is related to the inhibition of genes encoding cell-wall biosynthetic and architecture modifying enzymes and structural proteins in Medicago truncatula embryo axis. Molecular Plant, 2, 108-119. https://doi.org/10.1093/mp/ssn092 Guo, X., Castillo-Ramírez, S., González, V., Bustos, P., Fernádenz-Vázquez, J. L., Santamaría, R. I., Arellano, J., Cevallos, M. A., & Dávila, G. (2007). Rapid evolutionary change of common bean (Phaseolus vulgaris L.) plastome, and the genomic diversification of legume chloroplasts. BMC Genomics, 8(228). https://doi.org/10.1186/1471-2164-8-228 94 Han, G., Lu, C., Guo, J., Qiao, Z., Sui, N., Qiu, N., and Wang, B. (2020). C2H2 zinc finger proteins: master regulators of abiotic stress responses in plants. Frontiers in Plant Science, 11. https://doi.org/10.3389/fpls.2020.00115 Helbig, E., Oliveira, A., Queiroz, K., & Reis, S. M. P. M. (2003). Effect of soaking prior to cooking on the levels of phytate and tannin of the common bean (Phaseolus vulgaris L.) and the protein value. Journal of Nutritional Science and Vitaminology, 49(2)81-86. https://doi.org/10.3177/jnsv.49.81 Hu, P., Zhang, K., & Yang, C. (2018). BpNAC012 positively regulates abiotic stress responses and secondary wall Biosynthesis. Plant Physiology, 179, 700-717. https://doi.org/10.1104/pp.18.01167 Hu, Z., Mellor, J., Wu, J., & DeLisi, C. (2004). VisANT: an online visualization and analysis tool for biological interaction data. BMC Bioinformatics, 5(17). https://doi.org/10.1186/1471-2105-5-17 Jeong, C. Y., Lee, W. J., Truong, H. A., Trinh, C. S., Jin, J. Y., Kim, S., Hwang, K. Y., Kang, C.- S., Moon, J.-K., Hong, S.-W., & Lee, H. (2018). Dual role of SND1 facilitates efficient communication between abiotic stress signaling and normal growth in Arabidopsis. Scientific Reports, 8(10114). https://doi.org/10.1038/s41598-018-28413-x Jiminez-Jimenez, P., Santana, O., Lara-Rojas, F., Arthikala, M.-K., Armada, E., Hashimoto, K., Kuchitsu, K., Salgado, S., Aguirre, J., Quinto, C., & Cárdenas, L. (2019). Differential tetraspansin genes expression and subcellular localization during mutualistic interactions in Phaseolus vulgaris. PLOS One, 14(8). https://doi.org/10.1371/journal.pone.0219765 Kikuchi, K., Koizumi, M., Ishida, N., & Kano, H. (2006). Water uptake by dry beans observed by micro-magnetic resonance imaging. Annals of Botany, 98(3), 545-553. https://doi.org/10.1093/aob/mcl145 Kim, D., Paggi, J. M., Park, C., Bennett, C., & Salzberg, S. L. (2019). Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nature Biotechnology, 37, 907-915. https://doi.org/10.1038/s41587-019-0201-4 Kinsella, R. J., Kähäri, A., Haider, S., Zamora, J., Proctor, G., Spudich, G., Almeida-King, J., Staines, D., Derwent, P., Derhornou, A., Kersey, P., & Flicek, P. (2011). Ensembl BioMarts: a hub for data retrieval across taxonomic space. Database, 2011. https://doi.org/10.1093/database/bar030 Kilmer, O. L., Seib, P. A., & Hoseney, R. C. (1995). Effects of minerals and apparent phytase activity in the development of the hard-to-cook state of beans. Cereal Chemistry, 71(5), 476-482. Kuluev, B., Mikhaylova, E., Ermoshin, A., Veselova, S., Tugbaeva, A., Gumerova, G., 95 Gainullina, K., & Zaikina, E. (2019). The ARGOS-LIKE genes of Arabidopsis and tobacco as targets for improving plant productivity and stress tolerance. Journal of Plant Physiology, 242(153033). Kupada, H., Garg, V., Chitikineni, A., & Varshney, R. K. (2018). The RNA-Seq-based high resolution gene expression atlas of chickpea (Cicer arietinum L.) reveals dynamic spatio- temporal changes associated with growth and development. Plant, Cell, and Environment, 41, 2209-2225. https://doi.org/10.1111/pce.13210 Langfelder, P. & Horvath, S. (2008). WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics, 9(559). https://doi.org/10.1186/1471-2105-9-559 Langfelder, P. & Horvath, S. (2012). Fast R functions for robust correlations and hierarchical clustering. Journal of Statistical Software, 46(11). Law, C. W., Chen, Y., Shi, W., & Smyth, G. K. (2014). Voom: precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biology, 15. 10.1186/gb-2014- 15-2-r29 Lima, D. C. (2019). Gamma irradiation of common beans: effect on nutritional and technological properties. LWT, 116. https://doi.org/10.1016/j.lwt.2019.108539 Liu, C., Yu, H., Rao, X., & Dixon, R. A. (2021). Abscisic acid regulates secondary cell-wall formation and lignin deposition in Arabidopsis thaliana through phosphorylation of NST1. PNAS, 118(5). https://doi.org/10.1073/pnas.2010911118 Love M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology, 15(550). https://doi.org/10.1186/s13059-014-0550-8 Manjasetty, B. A., Yu, X.-H., Panjikar, S., Tajuchi, G., Chance, M. R., & Liu, C.-J. (2012). Structural basis for modification of flavonol and naphthol glucoconjugates by Nicoiana tabacum malonyltransferase (NtMaT1). Planta, 236(3), 781-793. https://doi.org/10.1007/s00425-012-1660-8 Martin, M. (2011). CutAdapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal, 17(1), 10-11. Martínez-Manrique, E., Jacinto-Hernández, C., Garza-García, R., Campos, A., Moreno, E., & Bernal-Lugo, I. (2011). Enzymatic changes in pectic polysaccharides related to the beneficial effect of soaking on bean cooking time. Journal of the Science of Food and Agriculture, 91, 2394-2398. https://doi.org/10.1002/jsfa.4474 Mohanta, T. K., Yadav, D., Khan, A. L., Hashem, A., AbdAllah, E. F., & Al-Harrasi, A. (2019). Molecular players of EF-hand containing calcium signaling event in plants. International Journal of Molecular Sciences, 20(6). https://doi.org/10.3390/ijms20061476 96 Mullins, A. P. & Arjmandi, B. H. (2020). Health benefits of plant-based nutrition: focus on beans in cardiometabolic diseases. MDPI Nutrients, 13(2). https://doi.org/10.3390/nu13020519 Nakayama, T. J., Rodrigues, F. A., Neumaier, N., Marcolino-Gomes, J., Molinari, H. B. C., Santiago, T. R., Formiheieri, E. F., Basso, M. F., Farias, J. R. B., Emygdio, B. M., de Oliveira, A. C. B., Campos, A. D., Borém, A., Harmon, F. G., Mertz-Henning, L. M., & Neponuceno, A. L. (2017). Insights into soybean transcriptome reconfiguration under hypoxic stress: functional, regulatory, structural, and compositional characterization. PLOS One, 12(11). https://doi.org/10.1371/journal.pone.0187920 Niraula, P. M., Zhang, X., Jeremic, D., Lawrence, K. S., & Klink, V. P. (2021). Xyloglucan endotransglycosylase/hydrolase increases tightly-bound xyloglucan and chain number but decreases chain length contributing to the defense response that Glycine max has to Heterodera glycines. PLOS ONE, 16(14). https://doi.org/10.1371/journal.pone.0244305 Njoroge, D. M., Kinyanjui, P. K., Christiaens, S., Shpigelman, A., Makokha, A. O., Sila, D. N., & Hendrickx, M. E. (2015). Effect of storage conditions on pectic polysaccharides in common beans (Phaseolus vulgaris) in relation to the hard-to-cook defect. Food Research International, 76, 105-113. https://doi.org/10.1016/j.foodres.2015.03.005 O‘Rourke, J. A., Iniguez, L. P., Fu, F., Bucciarelli, B., Miller, S. S., Jackson, S. A., McClean, P. E., Li, J., Dai, X., Zhao, P. X., Hernandez, G., & Vance, C. P. (2014). An RNA-seq based gene expression atlas of the common bean. BMC Genomics, 15(866). https://doi.org/10.1186/1471-2164-15-866 Parreira, J. R., Balestrazzi, A., Fevereiro, P., & Araújo, S. S. (2018). Maintaining genome integrity during seed development in Phaseolus vulgaris L.: evidence from a transcriptomic profiling study. MDPI Genes, 9(10), 463. https://doi.org/10.3390/genes9100463 Peaucelle, A., Braybrook, S., & Höfte, H. (2012). Cell wall mechanics and growth control in plants: the role of pectins revisited. Frontiers in Plant Science, 3, 1-6. https://doi.org/10.3389/fpls.2012.00121 Perez de Souza, L., Scossa, F., Proost, S., Bitocchi, E., Papa, R., Tohge, T., & Fernie, A. R. (2018). Multi-tissue integration of transcriptomic and specialized metabolite profiling provides tools for assessing the common bean (Phaseolus vulgaris) metabolome. The Plant Journal, 97(6), 1132-1153. https://doi.org/10.1111/tpj.14178 R Core Team. (2022). The R stats package. R Core Team. Rajashekar, C. B. & Baek, K.-H. (2014). Hydrogen peroxide alleviates hypoxia during imbibition and germination of bean seeds (Phaseolus vulgaris L.). American Journal of Plant Sciences, 5(24). https://doi.org/10.4236/ajps.2014.524373 97 Ren, C., Kong, C.-C., Yan, K., Zhang, H., Luo, Y.-M., & Xie, Z.-H. (2017). Elucidation of the molecular responses to waterlogging in Sesbania cannabina roots by transcriptome profiling. Scientific Reports, 7(9256). https://doi.org/10.1038/s41598-017-07740-5 Ritchie, M. E., Phipson, B., Wu, D. Hu, Y., Law, C. W., Shi, W., & Smyth, G. K. (2015). Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research, 43, e47. https://doi.org/10.1093/nar/gkv007 Rockland, L. B. & Jones, F. T. (1974). Scanning electron microscope studies on dry beans: effects of cooking on the cellular structure of cotyledons in rehydrated large lima beans. Journal of Food Science, 39(2), 342-346. https://doi.org/10.1111/j.1365- 2621.1974.tb02890.x Rstudio Team. (2020). Rstudio: Integrated development for R. Rstudio, PBC. Samalova, M., Melnikava, A., Elsayad, K., Peaucelle, A., Gahurova, E., Gumulec, J., Spyrolou, I., Zemlyanzkaya, E. V., Ubogoeva, E. V., Balkova, D., Demko, M., Blavet, N., Alexiou, P., Benes, V., Mouille, G., & Hejatko, J. (2023). Hormone-regulated expansins: Expression, localization, and cell wall biomechanics in Arabidopsis root growth.” Plant Physiology. https://doi.org/10.1093/plphys/kiad228 Sadohara, R., Izquierdo, P., Alves, F. C., Porch, T., Beaver, J., Urrea, C. A., & Cichy, K. A. (2022). The Phaseolus vulgaris L. Yellow Bean Collection: genetic diversity and characterization for cooking time. Genetic Resources and Crop Evolution, 69, 1627- 1648. https://doi.org/10.1007/s10722-021-01323-0 Schröder, F., Lisso, J., Lange, P., & Müssig, C. (2009). The extracellular EXO protein mediates cell expansion in Arabidopsis leaves. BMC Plant Biology, 9(20). https://doi.org/10.1186/1471-2229-9-20 Sedbrook, J. C., Carroll, K. L., Hung, K. F., Masson, P. H., & Somerville, C. R. (2002). The Arabidopsis SKU5 gene encodes an extracellular glycosyl phosphatidylinositol-anchored glycoprotein involved in directional root growth. Plant Cell, 14(7), 1635-1648. https://doi.org/10.1105/tpc.002360 Sharmin, R. A., Bhuiyan, M. R., Lv, W., Yu, Z., Chang, F., Kong, J., Bhat, J. A., & Zhao, T. (2020). RNA-seq based transcriptomic analysis revealed genes associated with seed- flooding tolerance in wild soybean (Glycine soja Sieb. & Zucc.). Environmental and Experimental Botany, 171. https://doi.org/10.1016/j.envexpbot.2019.103906 Siqueira, B. S., Fernandes, K. F., Brito, P. V. A., & Santos, F. C. A. (2018). Histochemical and ultrastructural characterization of easy-to-cook and hard-to-cook carioca bean genotypes. LWT, 97, 117-123. https://doi.org/10.1016/j.lwt.2018.06.048 Soltani, A., MafiMoghaddam, S., Oladzad-Abbasabadi, A., Walter, K., Kearns, P. J., Vasquez- Guzman, J., Mamidi, S., Lee, R., Shade, A. L., Jacobs, J. L., Chilivers, M. I., Lowry, D. 98 B., McClean, P., & Osorno, J. M. (2018). Genetic analysis of flooding tolerance in an Andean diversity panel of dry bean (Phaseolus vulgaris L.). Frontiers in Plant Science, 9. https://doi.org/10.3389/fpls.2018.00767 Soltani, A., Walter, K. A., Wiersma, A. T., Santiago, J. P., Quigley, M., Chitwood, D., Porch, T. G., Miklas, P., McClean, P. E., Osorno, J. M., & Lowry, D. B. (2021). The genetics and physiology of seed dormancy, a crucial trait in common bean domestication. BMC Plant Biology, 21(58). https://doi.org/10.1186/s12870-021-02837-6 Stanley, D. W. (1992). Hard beans—a problem for growers, processors, and consumers. HortTech, 2, 370-378. https://doi.org/10.21273/HORTTECH.2.3.370 Stanley, D. W. (1997). A possible role for condensed tannins in bean hardening. Food Research International, 25, 187-192. https://doi.org/10.1016/0963-9969(92)90136-S Temple, H., Phyo, P., Yang, W., Lyczakowski, J. J., Echevarría-Poza, A., Yakunin, I., Parra- Rojas, J. P., Terrett, O. M., Saez-Aguayo, S., Dupree, R., Orellana, A., Hong, M., & Dupree, P. (2022). Golgi-localized putative S-adenosyl methionine transporters required for plant cell wall polysaccharide methylation. Nature Plants, 8, 656-669. https://doi.org/10.1038/s41477-022-01156-4 Toili, M. E. M., Koning, R., Kiekens, R., Ndumba, N., Wahome, S., Anami, S., Githiri, S. M., & Angenon, G. (2022). Transcriptome-guided identification of pectin methyl-esterase- related enzymes and novel molecular processes effectuating the hard-to-cook defect in common bean (Phaseolus vulgaris L.). MDPI Foods, 11(12), 1692. https://doi.org/10.3390/foods11121692 Voorrips, R. E. (2002). MapChart: software for the graphical presentation of linkage maps and QTLs. Journal of Heredity, 93(1), 77-78. https://doi.org/10.1093/jhered/93.1.77 Wahome, S. W., Girhiri, M. S., Kinyanjui, P. K., Toili, M. E. M., & Angenon, G. (2023). “Genome-wide association study of variation in cooking time among common bean (Phaseolus vulgaris L.) accessions using Diversity Arrays Technology markers.” Legume Science, e184. https://doi.org/10.1002/leg3.184 Wang, J., Sun, H., Sheng, J., Jin, S., Zhou, F., Hu, Z., & Diao, Y. (2019). Transcriptome, physiological, and biochemical analysis of Triarrhena sacchariflora in response to flooding stress. BMC Genetics, 20(88). https://doi.org/10.1186/s12863-019-0790-4 Wang, L. & Ruan, Y.-L. (2013). Regulation of cell division and expansion by sugar and auxin signaling. Frontiers in Plant Science, 4(163). https://doi.org/10.3389/fpls.2013.00163 Wang, T.-Z., Zhang, J.-L., Tian, Q.-Y., Zhao, M.-G., & Zhang, W.-H. (2013). A Medicago truncatula EF-hand family gene, MtCaMP1, is involved in drought and salt stress tolerance. PLOS One, 8(4). https://doi.org/10.1371/journal.pone.0058952 99 Witasari, L. D., Huang, F.-C., Hoffmann, T., Rozhon, W., Fry, S. C., & Schwab, W. (2019). Higher expression of the strawberry xyloglucan endotransglucosylase/hydrolase genes FvXTH9 and FvXTH6 accelerates fruit ripening. The Plant Journal, 100(6), 1237-1253. https://doi.org/10.1111/tpj.14512 Wood, J. A., Tan, H.-T., Collins, H. M., Yap, K., Khor, S. F., Lim, W. L., Xing, X., Bulone, V., Burton, R. A., Fincher, G. B., & Tucker, M. R. (2018). Genetic and environmental factors contribute to variation in cell wall composition in mature desi chickpea (Cicer arietinum L.) cotyledons. Plant, Cell, and Environment, 41(9), 2195-2208. https://doi.org/10.1111/pce.13196 Wu, H.-C., Bulgakov, V. P., & Jinn, T.-L. (2018). Pectin methylesterases: cell wall remodeling proteins are required for plant response to heat stress. Frontiers in Plant Science, 9(1612). https://doi.org/10.3389/fpls.2018.01612 Zhang, J., Li, X., He, Z., Zhao, X., Wang, Q., Zhou, B., Yu, D., Huang, X., Yang, D., Guo, X., & Liu, X. (2013). Molecular character of a phosphatase 2C (PP2C) gene relation to stress tolerance in Arabidopsis thaliana. Molecular Biology Reports, 40(3), 2633-2644. https://doi.org/10.1007/s11033-012-2350-0 Zhu, J., Tang, G., Xu, P., Li, G., Ma, C., Li, P., Jiang, C., Shan, L., & Wan, S. (2022). Genome- wide identification of xyloglucan endotransglucosylase/hydrolase gene family members in peanut and their expression profiles during seed germination. PeerJ, 10. https://doi.org/10.7717/peerj.13428 100 CHAPTER 3: HOT, HUMID ENVIRONMENTS INDUCE THE EXPRESSION OF CELL WALL REMODELING GENES IN WHOLE MATURE DRY BEANS (PHASEOLUS VULGARIS L.) AND DISPROPORTIONATELY AFFECT THE GERMINATION RATES OF FAST-COOKING GENOTYPES 101 ABSTRACT Dry bean (Phaseolus vulgaris L.) cooking time is under genetic and environmental control. High temperature, high relative humidity (i.e., tropic-like) environments are thought to influence cooking times in dry beans by causing a phenomenon known as hard-to-cook (HTC). The goal of this work was to understand how different dry bean germplasm responds to high temperature and high humidity storage, thereby facilitating the development of varieties that are less sensitive to the effects of HTC. This study compared gene expression in four dry bean genotypes with contrasting market classes and cooking times by subjecting mature seeds to one of two soaking treatments (raw and a 12-hour soak) and one of three environmental treatments: 1) no storage, 2) storage for 3 days in a temperate (25C/50% RH) environment, or 3) storage for 3 days in a tropical (35C/75% RH) environment. The physiochemical properties of these seeds were then measured, including raw seed weight, water uptake percentage, cooking time, cell wall thickness, nutrient content, phytate content, and germination rate. RNA-seq was also performed, and reads were subjected to DESeq2, WGCNA, and topGO analysis. Genotypes that cooked faster prior to storage also cooked faster following tropical storage and 12 hours of soaking, although the cooking times of each genotype did increase on average by ~7 minutes. Cell wall thickness decreased on average in tropical beans, but this difference was only significant in the slow-cooking genotypes, suggesting that the genotypes responded differently to high heat and high humidity stress. Tropical storage induced the expression of genes related to oxidoreductase activity, plant growth/development, and/or stress tolerance across all four genotypes. A gene that participates in stress-induced cell wall remodeling and calcium signaling, wall associated kinase 5, was upregulated in all four of the genotypes studied after soaking under tropical storage conditions. Slow- and fast-cooking genotypes differed in terms of both physiochemical and gene 102 expression properties, namely the upregulation of genes sensitive to the presence of inorganic substances in the slow-cooking genotypes and impaired germination in the fast-cooking varieties under tropical storage conditions. However, impaired germination was only apparent in dry beans grown under non-irrigated conditions, suggesting that multiple stresses may be required to cause negative trade-offs in fast-cooking germplasm. INTRODUCTION Observational evidence has unequivocally demonstrated that climate change is affecting weather patterns across all seven continents on Earth (IPCC, 2007; Willett et al., 2007). In Midwestern North America alone, the frequency of stationary heat waves is projected to increase by 48% under a high-emission scenario by 2100 relative to 1976-2005 (Polasky et al., 2022). Hot, humid weather is, likewise, predicted to become increasingly common, with the most severe cases occurring mainly in the tropics and the subtropics (Matthews, 2018; Zhao et al., 2015). Increasingly common incidences of hot, humid weather have had consequences for agriculture, particularly in already vulnerable production regions. For example, a global survey published in 2016 showed that 847 localities across 78 countries reported increasingly lower crop production and reduced crop quality, with most reports originating from Africa (60%), where most crops are rain-fed (Savo et al., 2016). The increasing occurrence of high heat (≥35°C) combined with high relative humidity (≥83%) could pose a threat to staple crops in these regions such as dry beans (Phaseolus vulgaris L.) (Abay and Tolesa, 2023). Dry beans are a source of protein, carbohydrates, and key nutrients like iron, zinc, and folate for millions of people, especially in the tropical and subtropical regions of the world where meat is often scarce (Amongi et al., 2018). The co-occurrence of high heat and humidity can limit access to this nutritious food by simultaneously increasing the cooking 103 times of the seeds and by decreasing their germination percentage in the field (Aguilera et al., 1986; Cortelazzo et al., 2005). Notably, high heat and humidity can affect the quality of dry bean seeds both before and after harvest within as little as three days (Abay and Tolesa, 2023; Aguilera et al., 1986; Cichy et al., 2019). Hence, poor growth and storage conditions could place some of the most vulnerable communities around the world at risk for food insecurity. Tropical storage conditions increase the cooking times of dry beans by inducing a stress response in dry bean cotyledons known as hard-to-cook (HTC) (Njoroge et al., 2015). Dry beans with HTC appear to imbibe water like seeds without HTC but have longer than expected cooking times. This is different from conditions like hard-shell or sclerema, wherein the dry bean seeds or their cells fail to absorb any (or very little) water (Berry et al., 2020). Generally, the severity of HTC is positively associated with duration of storage, temperature, and relative humidity (Njoroge et al., 2015). While a longer than expected cooking time is possibly the most inconvenient and, thus, important trait for consumers, beans with this condition have also been observed to have stronger, thicker cell walls after cooking, lower protein content, decreased protein and pectin solubility, poor plasma membrane integrity, decreased phytic acid content, and increased polyphenol deposition in the cotyledon cell walls (Chen et al., 2023, 2022; Chigwedere et al., 2019; Coelho et al., 2007; Garcia et al., 1998; Mafuleka et al., 1993; Molina et al., 1976; Njoroge et al., 2014; Shomer et al., 1990; Wainaina et al., 2022; Yi et al., 2016). In addition, the seed coats of dry beans in the process of developing HTC tend to darken due to the accumulation of proanthocyanidin monomers, oligomers, polymers, and derivatives, more specifically flavon-3-ols [(-)-epicatechin and (+)-catechin] and procyanidins (condensed tannins) (Beninger et al., 2005; Duwadi et al., 2018; Elsadr et al., 2015; Srisuma et al., 1989; Wiesinger et al., 2021). 104 The genetic control of HTC in dry beans is less well-understood. While the severity of HTC is dependent on genotype, all dry beans appear to be at least partially susceptible to hardening in adverse growing/storage conditions. For example, pinto beans that accumulated more flavan-3-ols and procyanidins during storage also had longer cooking times compared to those that accumulated fewer of these compounds (Wiesinger et al., 2021). In another case, two yellow dry bean genotypes developed HTC at different rates when stored in a mild yet humid environment (Coelho et al., 2007). In this latter study, phytate and protein content decreased much more in the genotype that was more susceptible to HTC, suggesting that the rate of phytate and protein degradation influenced the severity of HTC in these beans (Coelho et al., 2007). It has also been hypothesized that the rapid degradation of proteins, phytate, and other compounds in the susceptible variety triggered genetic mechanisms that lead the cell to strengthen its cell walls, resulting in longer cooking times. Briefly, some of the genes that are suspected to break down compounds in adverse environmental conditions are phytases, pectinases (i.e., polygalacturonase, rhamnogalacturonase, etc.), and proteases, whereas genes that reinforce cell walls and other compounds include peroxidases and pectin methylesterases (Coelho et al., 2007; Martínez-Manrique et al., 2011; Sadohara et al., 2022). Cell wall remodeling enzymes are also thought to play a role in cooking time by making it easier for enzymes to access carbohydrates within the cell wall (Jeffery et al., 2023). The effects of post-harvest storage conditions of dry beans on gene expression remain poorly understood. Gene expression has been used to understand seed coat darkening in dry beans that accumulate fewer proanthocyanidins during storage (also known as slow- or non- darkening varieties). These studies found that the abundances of proanthocyanidin biosynthesis and transporter genes were highly correlated with the development of darker seed coats during 105 storage (Duwadi et al., 2018; Freixas Coutin et al., 2017; Islam et al., 2020). However, the only studies thus far that have associated gene expression with cooking time did not vary the environmental conditions in which the beans were grown/stored prior to sampling, so there remains a gap in our understanding of the HTC mechanism (Jeffery et al., 2023; Toili et al., 2022). This has made it difficult to identify candidate genes for the control of HTC in dry beans. Therefore, the aim of the present study was to understand how different dry bean varieties and market classes (brown and yellow) respond to hot, humid climates using RNA-sequencing as a tool to measure gene expression. Physiochemical traits including cooking time, water uptake, nutrient content, cell wall thickness, phytate content, and germination rate were measured, as well, to understand the relationship between genetic and phenotypic expression. MATERIALS AND METHODS Evaluations conducted on beans harvested from Montcalm in 2020 Germplasm Four genotypes with contrasting cooking times were analyzed, two of which were brown and two of which were yellow (Fig. 3.1). Further details about these genotypes have been previously described (Jeffery et al., 2023). The seeds used for raw seed weight, water uptake percentage, cooking time, cell wall thickness, and RNA-sequencing were grown in triplicate plots under irrigated conditions at the Michigan State University Montcalm Research Farm in Entrican, MI in 2020. Each plot consisted of four 6.1-m rows with a 0.51-m row width with a seed density of 13 seeds per meter length. The center two rows contained the experimental genotype and the outer two were a uniform kidney border. Whole plants were hand-harvested at maturity and transported back to East Lansing, MI where they were hand-threshed. The seeds were air-dried to 10-12% moisture content and stored at 4°C/75% RH for eight months prior to 106 testing. Figure 3.1: Seeds of two brown beans and two yellow beans with contrasting cooking times. The brown beans are on the top, and the yellow beans are on the bottom. TZ-27 is the slower cooking brown bean relative to TZ-37, and PI527538 is the slower cooking yellow bean relative to Ervilha. TZ-27 and TZ-37 were discovered in Tanzania in 1943 and 1994, respectively. PI527538 and Ervilha were discovered in Burundi in 1985 and Angola in 2010, respectively. The seeds in the photo are from one irrigated field replicate harvested in Montcalm, MI in 2023. Seeds that were used for RNA-sequencing, post-storage seed moisture content analysis, nutrient content analysis, and scanning electron microscopy (SEM) were surface sterilized and checked for seed coat cracks by immersing them in 2% sodium hypochlorite (w/v), 70% ethanol (w/v) solution for one minute, followed by 15 seconds of rinsing with ultrapure water. All four genotypes were subjected to one of three storage conditions: no storage (untreated), three days of storage in 27°C/50% RH (temperate), or three days of storage in 40°C/85% RH (tropical). The beans were stored in climate-controlled chambers at the Michigan State University Dept. of Packaging. At the end of the storage period, the seeds were transported back to the Plant and Soil Science Building where they were partitioned for testing (Fig. 3.2). 107 Figure 3.2: Data collection and analysis pipeline for untreated and aged dry beans seeds. Raw seed weight, water uptake percentage, and cooking time Thirty seeds were weighed using an analytical balance to collect raw seed weight data. The seeds were then immersed in 250 mL of distilled water for 12 hours. Seeds were removed, blotted dry with paper towels, and re-weighed. Water uptake percentage was calculated using the following formula: [(soaked weight – dry weight)/dry weight]*100. Twenty-five seeds were used to measure cooking time. Cooking time was measured using a Mattson cooker apparatus connected to a digital recording device equipped with custom recording software (Customized Machining 246 and Hydraulics Co., 247 Winnipeg, Canada). Briefly, weighted pins were placed on top of individual beans. The recording device was started after the apparatus and beans were submersed in boiling water. As the beans softened, the pins fell through the seeds and triggered the recording device. The time it took for 80% of the pins to fall completely through the seeds was recorded as the cooking time of the sample. Beans were 108 cooked in East Lansing, MI approximately 898 ft above sea level within eight months of harvest in a lab with a room temperature of 21°C. Cell wall thickness Scanning electron microscopy (SEM) was performed at the Michigan State University Center for Advanced Microscopy. Soaked samples were prepared by immersing the beans in distilled water from the Cichy lab (A367 Plant and Soil Sciences Building) for 12 hours. To prepare the beans for SEM imaging, one seed from each field replicate was cut perpendicular to the cotyledons at the hilum. Seeds were prepared by a staff member at the Center for Advanced Microscopy center as previously described (Bassett et al., 2021). Cotyledon surfaces were imaged using a JEOL JSM-6610LV scanning electron microscope with an accelerating voltage of 12 kV, a spot size of 30, and a working distance of 21 mm. Cotyledon cells were measured at x200 resolution, producing a scale bar 100 μm in length. One image was taken per field replicate, genotype, storage treatment, and soaking treatment, and cotyledon cell wall thickness was measured 40 times per image using ImageJ (Schneider et al., 2012). All images were taken perpendicular to the cotyledon surface to ensure the accuracy of measurements. Nutrient content For nutrient content analysis, soaked samples were prepared by immersing the beans in distilled water for 12 hours. Ten seeds were lyophilized, milled into a flour using a Polymix mill (PX-MFC 90D; Kinematica, Lucerne, Switzerland) (0.5-mm sieve), and sent to Great Lakes A & L (Fort Wayne, IN) for nutrient content analysis. Total nitrogen, mineral content, and moisture content were measured as previously described (AOAC International, 2000; Jeffery et al., 2023). Iron and aluminum data were not analyzed due to contamination. Sodium data was excluded because none of the samples contained a detectable amount of sodium, which is ≥ 0.01 ppm per 109 gram of material. RNA-sequencing RNA extraction and isolation Immediately after removal from storage, 10 raw/unsoaked seeds were flash frozen in 15 mL conical Falcon tubes using liquid nitrogen and stored in a -80°C freezer. For the soaked beans, 10 seeds were submersed in distilled water for 12 hours, then flash frozen and stored at - 80°C. The seeds were stored for an additional seven months prior to being processed (Fig. 3.2). Whole frozen seeds were ground in a Krups Fast Touch stainless steel coffee grinder along with a few pellets of dry ice for 10 seconds. The coffee grinder was thoroughly cleaned and dried between samples using Sigma-Aldrich RNAaseZAP, ultrapure water, 70% ethanol, pressurized air, and Kimwipes. In addition, all work surfaces and gloves were sprayed with 70% ethanol and RNAaseZAP between samples. The processed samples were left in the -80°C freezer to sublimate for 48 hours. Once the dry ice had been removed, total RNA was extracted from whole seed material using two Millipore Sigma Spectrum Plant Total RNA (STRN50) kits according to manufacturer’s instructions. The manufacturer’s recommendations regarding sterile practice were followed. Library preparation and sequencing A total of 72 samples of total RNA from plants were submitted to the Michigan State University Genomics Core for NGS library preparation and sequencing. Libraries were prepared using the Illumina Stranded mRNA Prep, Ligation kit with IDT for Illumina RNA UD Indexes, following manufacturer’s recommendations except that half volume reactions were performed. The quality and quantity of the completed libraries was determined using a Qubit dsDNA HS and an Agilent 4200 TapeStation HS DNA1000. The libraries were pooled in equimolar 110 amounts, and the pool was quantified using the Invitrogen Collibri Quantification qPCR kit. The pool was loaded onto one lane of an Illumina S4 flow cell and sequencing was performed in a 2x150bp paired end format using a NovaSeq v1.5, 300 cycle reagent kit. Base calling was done using Illumina Real Time Analysis (RTA) v3.4.4. The RTA was demultiplexed and converted to FastQ format with Illumina Bcl2fastq v2.20.0. Total read counts obtained from sequencing are provided in the supplementary information (Table SI 3.1a-b). All raw data files will be uploaded to the Sequence Read Archive upon publication of the manuscript or by August 2025. Read filtering, trimming, alignment, and quantification The quality of the samples was first checked using FastQC (v. 0.12.0) (Andrews, 2023). After passing the initial screening, the paired-end adapters were trimmed using Trimmomatic (v. 0.39) with a leading trim threshold of 20, a trailing trim threshold of 20, a sliding window threshold of 4:20, and a minimum transcript length of 30 bps (Bolger et al., 2014). The quality of the samples was then re-checked using FastQC. Transcripts were aligned to the Phaseolus vulgaris L. reference genome (v. 2.1) using HiSat2 (v. 2.1.0) and counted using HTSeq (v. 0.11.2) (Anders et al., 2015; Kim et al., 2019; Schmutz et al., 2014). Alignment and count statistics are available in the supplementary information (Table SI 3.1a). Data analysis The DESeq2 and edgeR packages were used to identify differentially expressed genes (Love et al., 2014). The similarity of the replicates was checked in RStudio using the heatmap.2 function in the gplots (v. 3.1.3) package after applying a regularized logarithmic transformation to the read counts (Warnes et al., 2022). One of the samples was concluded to be contaminated with mouse RNA, but this did not affect the similarity of the replicates (Fig. SI 3.1; Table SI 3.1b). It is unknown where the contamination originated from, as it was not investigated further. 111 The similarity of the samples was examined using multi-dimensional scaling (MDS) and principal component analysis (PCA). Un-transformed read counts were plotted with the plotMDS function, whereas DESeqTransform read counts were plotted with the plotPCA function. To identify differentially expressed genes, different storage treatments within the same genotype, different genotypes (within the same market class) exposed to the same storage treatment, and all genotypes exposed to the same storage treatment were compared. Genes were considered differentially expressed if their Benjamini-Hochman adjusted p-values were smaller than 0.05. DESeq2 outputs (i.e., lists of differentially expressed genes) were compared and Venn diagram plots were generated using Venny (Oliveros, 2015). Genes were annotated using the latest version of the Arabidopsis-derived annotation file from Phytozome (Pvulgaris_442_v2.1.annotation_info.txt) (Schmutz et al., 2014). The WGCNA package (v. 1.13) was used as previously described to identify gene co- expression modules that were significantly associated with the cooking times of raw and soaked dry beans (Jeffery et al., 2023; Langfelder and Horvath, 2012, 2008; Law et al., 2014). The topGO package (v. 3.12) was used as previously described to annotate DESeq2 and WGCNA outputs (Alexa and Rahnenführer, 2021; Durinck et al., 2009, 2005; Jeffery et al., 2023; Kinsella et al., 2011). The intersection of GO terms amongst DESeq2 outputs was visualized using the upset function in the UpSetR (v. 1.4.0) package (Conway et al., 2017). Seed moisture content percentage and phytate content Seed moisture content was tested twice. The first time, the moisture contents of raw whole untreated seeds were determined using a Dickey John benchtop moisture tester three hours before subjecting the seeds to temperate or tropical storage. The second time, remnant samples of milled raw untreated, temperate, and tropical seeds left over from RNA-sequencing were 112 analyzed after 38 months in storage at -81°C. The moisture content of the seed samples was determined within 12 hours after thawing. The moisture content of the seeds was measured using the low constant temperature oven method, wherein the material was weighed before and after being dried at 105°C for 17 hours (ISTA, 2023). The phytate content of the raw milled seed samples was also determined within 12 hours after thawing. The phytate contents (g/100 g) of the samples were measured using the Megazyme Phytic Acid kit according to the kit manufacturer’s instructions (K-PHYT). Phytate contents were then adjusted based on the moisture content of the sample using the following equation: phytate in sample/[sample weight – sample weight*(moisture content %/100)]. Evaluations conducted on beans from different seed lots Germination rate Seeds with no visible seed coat cracks or disease were selected for germination testing. The seeds were harvested in two locations (Montcalm, MI and Saginaw Valley, MI) in 2023 using a combine harvester. Seeds grown in the Montcalm location received supplemental irrigation, whereas the seeds grown in Saginaw Valley were rainfed. Seeds were harvested with a combine harvester, air-dried to 10-12% moisture content, and stored at 4°C/75% relative humidity for one month prior to testing. Seeds were transported in brown kraft paper bags to the Iowa State University Seed Testing Lab via express mail. Prior to germination testing, the moisture contents of the seeds were determined to be between 9.9-11.9% using the low constant temperature oven method (ISTA, 2023). Seeds were germinated according to the latest ISTA or AOAC protocols. For normal (warm) germination, 400 seeds were provided per field replicate. After germinating on rolled paper towel for seven days, the number of sprouted seeds was counted. Seeds stored in high temperature and relative humidity conditions are referred to as 113 ‘artificially aged’ (AA) seeds. For AA testing, 200 seeds were provided per field replicate. The test involves placing seeds on mesh screens above 40 mL of water in a sealed container. The container is then placed into a hot chamber for 3 days, then removed from the container and germinated under normal (warm) conditions. Every seed that did not germinate within seven days was determined to be viable using a tetrazolium test. Seed coat check severity The degree of seed coat checking/cracking (SCC) was determined as previously described (Wang and Cichy, 2023). Seeds were stored at 4°C/75% RH for 8 months prior to testing. Briefly, 100 seeds harvested from Montcalm, MI and Saginaw Valley, MI in 2023 were submerged in 0.33% Gram’s iodine solution for five minutes to allow for easier visualization of seed coat cracks. Seeds were categorized based on the number and the size of visible cracks in the seed coat on a scale from 1-5. The SCC severity score was calculated using the following equation [(1*# of seeds with a score of 1) + (2*# of seeds with a score of 2) + … + (5*# of seeds with a score of 5)]/100. An SCC of 1 indicates that the seed coats of all 100 seeds were completely intact, whereas a 5 indicates that all the seeds were split in half. Statistical analysis All statistical tests were conducted using RStudio software (RStudio Team, 2024). Pearson’s correlation coefficients were measured and plotted using the corrplot (v. 0.92) package (Wei and Simko, 2021). All data sets were tested for mean-shifting outliers using the outlierTest function from the car package (v. 3.1-2) (Fox and Weisberg, 2019). Following outlier removal, normality was checked using Q-Q plots, and equality of variances were checked using Levene’s test, followed by visual inspection of boxplots of the residuals. Unequal variances were corrected using either log or square root transformations. All statistical modeling was performed using the 114 lme4 (v. 1.1-35.3) package (Bates et al., 2015). Statistical models for raw seed weight, water uptake percentage, raw seed cooking time, and soaked seed cooking time included genotype and storage treatment as fixed effects and field replicate as a random effect. Models for cell wall thickness and nutrient content included genotype, storage treatment, and soaking time as fixed effects and field replicate as a random effect. The model for germination rate included genotype, growing location, and storage treatment as fixed effects and field replicate as a random effect. The significance of the interactions between fixed effects was tested using type II ANOVA (α=0.05). Since the potassium content data was unbalanced due to outlier removal, the significance of the potassium model was tested using the joint_tests function from the emmeans (v. 1.10.2 ) package (Lenth, 2024). In instances when the three-way interaction effect (genotype*soaking*storage) was not significant, the estimated means of the fixed effects or significant lower-level interaction effects were compared, instead. If the three-way interaction was significant, estimated means were calculated using the emmeans package. Back- transformation was performed if necessary using the type=”response” setting to obtain biologically meaningful estimates. Multiple comparisons testing of estimated means was conducted using the Sidak test in the multcomp (v. 1.4-25) package (Hothorn et al., 2008). RESULTS Effects of storage conditions on raw seed weight, water uptake, and cooking time The raw/unsoaked seed weight, water uptake percentage, and the cooking times of raw and soaked dry bean genotypes were determined using seeds stored in artificial storage conditions designed to replicate temperate and tropical environments. Genotype-by-storage interaction effects did not significantly affect any of these physiochemical properties. Raw seed weight varied significantly among genotypes and among storage treatments, 115 with TZ-37 having a significantly higher raw seed weight than TZ-27 regardless of storage treatment. Ervilha did not have a significantly higher average raw seed weight than PI527538, but it did have a significantly higher seed weight than TZ-27 (p=1.90e-4) (Fig. 3.3). Regardless of genotype, raw seeds stored in temperate and tropical environments weighed less on average than seeds that received no additional storage treatment (p<2.2e-16). After soaking the seeds in distilled water for 12 hours, TZ-27 and PI527538 took up 4-9% less water than TZ-37 and Ervilha, but TZ-27 did not take up significantly more water than TZ-37 (p=3.62e-06). After being subjected to temperate storage, all four genotypes took up an average of 12% more water compared to untreated and tropical beans (p=6.54e-10) (Fig. 3.4). Across all the genotypes and treatments tested, water uptake percentage was strongly negatively correlated with cooking time (r2=-0.98, p=1.22e-47) (Fig. SI 3.2). Figure 3.3: Estimated mean raw seed weights (g) of four dry bean genotypes stored in different environments. Raw weight is an average of three field plots in Montcalm, MI. Thirty seeds were used from each field plot for analysis. Treatments with different letters are not significantly different from other treatments in the same genotype (Sidak, α=0.05). 116 Figure 3.4: Estimated mean water uptake percentages of four dry bean genotypes stored in different environments. Water uptake (%) is an average of three field plots in Montcalm, MI. Thirty seeds were used from each field plot for analysis. Treatments with different letters are not significantly different from other treatments in the same genotype (Sidak, α=0.05). The cooking times of raw beans and beans pre-soaked for 12 hours in distilled water were determined (Fig. 3.5). In both cases, the interaction effect was insignificant, meaning that the storage environments had consistent effects on all four genotypes (p=0.416 and 0.156 for raw and soaked cooking times, respectively). Regarding the raw beans, untreated TZ-27 had the longest average cooking time (105.2 mins), followed by Ervilha (102.2 mins), TZ-37 (93.5 mins), and PI527538 (85.2 mins). Like previous reports, TZ-27 and Ervilha had longer cooking times than TZ-37 and PI527538 when cooked without presoaking, respectively (p=1.13e-08) (Bassett et al., 2021; Jeffery et al., 2023). The average cooking times of TZ-27, TZ-37, and PI527538 did not differ across storage treatments, but raw untreated Ervilha seeds had a significantly longer cooking time than raw temperate and tropical Ervilha seeds. 117 Figure 3.5: Estimated mean raw/unsoaked (left) and soaked (right) cooking times of four dry bean genotypes stored in different environments. Cooking times are an average of three field plots in Montcalm, MI. Twenty-five seeds were used from each field plot for analysis. Multiple comparison tests were performed separately on raw/unsoaked and soaked cooking times. Treatments with different letters are not significantly different from other treatments in the same genotype (Sidak, α=0.05). When beans were cooked following a 12 hr soak, untreated TZ-27 had the longest average cooking time (31.2 mins), followed by PI527538 (24.2 mins), TZ-37 (23.6 mins), and Ervilha (20.3 mins). The average cooking times of TZ-27 and PI527538 were higher than those of TZ-37 and Ervilha, respectively, and the cooking times of TZ-37 and Ervilha were not significantly different from each other (p=1.85e-04) (Fig. 3.5). Untreated and temperate beans took about the same amount of time to cook on average (24.5 and 25.1 mins, respectively). Three days of storage in a tropical environment, on the other hand, increased the cooking times of all four genotypes by ~7 minutes after soaking (p=1.05e-06). Effects of storage conditions on cell wall thickness The thickness of cotyledon cell walls was measured in raw and soaked dry bean seeds (Fig. 3.6). No significant differences in cell wall thickness were found between raw and soaked 118 beans when one additional factor such as genotype and storage treatment was considered. However, the three-way interaction between genotype, storage treatment, and soaking was significant (p=2.03e-06). Figure 3.6: Estimated mean cell wall thicknesses (n=3) of cotyledon cell walls in four raw/unsoaked (top) and soaked (bottom) dry bean genotypes stored in different environments. Cell wall thickness measurements are an average of three field plots in Montcalm, MI. Three seeds were used from each field plot for analysis. Means with different letters are not significantly different from each other (Sidak, α=0.05). Data was collected from scanning electron microscopy (SEM) images using ImageJ. Regardless of whether the seeds were raw or soaked, the cotyledon cell walls of the untreated slow-cooking brown bean, TZ-27, were thicker than those observed in any other treatment or genotype except for soaked tropical TZ-27. Interestingly, the cell walls of tropical TZ-27 and PI527538 were significantly thinner than the cell walls of untreated TZ-27 and PI527538, respectively, but only under certain conditions. For TZ-27, this difference was only seen before soaking, whereas for PI527538, this difference was only seen after soaking. By 119 contrast, storage treatment did not significantly affect cell wall thickness in either of the fast- cooking beans, regardless of whether the beans were soaked (Sidak, α=0.05). Effects of storage conditions on nutrient content The effect of genotype, storage, and soaking on the nutrient composition of dry beans was determined using multi-linear modeling (Table SI 3.2). The three-way interaction effect (genotype*soaking*storage) was not significant in the case of sulfur, phosphorous, potassium, and zinc content, but was significant for the other nutrients. While most of the three-way pairwise comparisons were not statistically significant, a few consistent, significant trends that were biologically relevant to the study at hand were observed. On average, calcium levels were ~36% higher in soaked tropical beans relative to soaked untreated beans and ~31% higher in soaked tropical beans relative to raw tropical beans. The brown beans, TZ-27 and TZ-37, had less boron than yellow beans stored in raw tropical conditions and yellow soaked untreated beans, but this trend was not observed across any of the other storage or soaking conditions. No significant differences were found in zinc levels across genotypes. However, zinc levels were lower in tropical beans relative to untreated beans both before and after soaking. Manganese levels were higher on average in soaked tropical beans compared to soaked untreated beans. Nitrogen, sulfur, potassium, phosphorous, magnesium, and copper levels did not change significantly in response to the tropical storage treatment either in raw or soaked beans. No significant correlations were found between cooking time and any of the nutrients tested, though cell wall thickness content was weakly negatively correlated with sulfur content (r2=-0.24, p=0.04) (Fig. SI 3.2). 120 RNA-sequencing Data quality analysis Replicates of the RNA samples were compared based on their dissimilarity scores (Fig. SI 3.1). In addition, the samples were determined to be sufficiently correlated with each other based on both Pearson’s (r2=0.96-1) and Spearman’s (r2=0.98-0.99) correlation coefficients (Table SI 3.1b). Samples with metadata are publicly available on the NCBI Sequence Read Archive (SRA) and can be found using identifiers available in the supplementary information (Table SI 3.1a). According to multidimensional scaling (MDS) and principal component analysis (PCA), samples clustered strongly by genotype and storage treatment (Fig. 3.7-3.8, Fig. SI 3.3). The first principal component of PCA clearly separates the raw from the soaked samples, whereas genotypes are more clearly distinguished by the second principal component. Within the soaked cluster, the untreated (green) and temperate (blue) samples strongly overlap. On the other hand, the tropical samples (red) cluster separately from the untreated and temperate cluster. A similar pattern can be observed within individual genotypes in both the MDS and PCA plots. Since this suggested that gene expression was similar in untreated and temperate beans, but different in the tropical beans, differential gene expression levels were explored with Venn diagrams (Fig. 3.9). These results revealed that more genes were differentially expressed between untreated and tropical or temperate and tropical samples than between untreated and temperate samples. 121 Figure 3.7: Multidimensional scaling (MDS) plot showing pairwise differences in gene expression between samples. Brown (TZ-27 and TZ-37) and yellow (PI527538 and Ervilha) bean samples are shown on the left and the right graphs, respectively. Faded colored shapes represent raw beans and solid-colored shapes represent soaked beans. TZ-27 and PI527538 take more time to cook than TZ-37 and Ervilha, respectively. 122 Figure 3.8: Principal component analysis (PCA) of the samples taken from all the dry bean samples (TZ-27, TZ-37, PI527538, and Ervilha). Lighter colors represent raw samples, and darker colors represent soaked samples. The proportion of the total variance explained by the first two principal components are displayed on the x and y axes. 123 Figure 3.9: Venn diagrams comparing gene expression in raw and soaked beans following exposure to different storage conditions. Gene expression levels in individual genotypes are shown. Only genes that were significantly differentially expressed are shown. 124 Differentially expressed genes A total of 261 genes were differentially upregulated in both slow-cooking beans relative to their fast-cooking counterparts in response to soaking and tropical storage, whereas 224 genes were upregulated in both fast-cooking beans in response to these conditions (Table SI 3.3). Analysis with topGO was performed on these sets of genes. Several GO terms were not only associated with relative increases in gene expression in individual genotypes, but also across different genotypes (Fig. SI 3.4). Notably, in raw beans, the term ‘oxidoreductase activity’ (GO:0016491) was associated with increased gene expression levels in all four genotypes under tropical conditions and in Ervilha under temperate conditions. In addition, the terms ‘UDP- glycosyltransferase activity’ and ‘sulfur compound metabolic process’ were associated with TZ- 27 and TZ-37, respectively, across all storage treatments. The term ‘response to external stimulus’ was associated with both TZ-27 and TZ-37 under temperate conditions. Furthermore, the term ‘vacuole’ was associated with both brown beans in tropical storage. Under soaked conditions, the term ‘plasma membrane’ was associated with both TZ-37 and Ervilha in temperate storage, ‘cellular homeostasis’ was associated with PI527538 under all storage conditions, ‘response to inorganic substance’ was associated with both TZ-27 and PI527538 in tropical storage, ‘phosphorelay signal transduction system’ was associated with TZ-27 and Ervilha in tropical storage. Weighted gene co-expression network analysis (WGCNA) Differences in gene co-expression patterns across genotypes, soaking treatments, and storage treatments were investigated using weighted gene co-expression network analysis (WGCNA) (Figs. SI 3.5a-b, 3.6a-b). Four gene co-expression modules were identified in raw dry beans, and three modules were identified in soaked dry beans (Figs. 3.10a-b). For clarity, 125 references to raw modules were designated ‘Raw’, and soaked modules were designated ‘Soak’. To distinguish between the different modules identified in the raw and soaked beans, modules were given numeric identifiers (i.e., Raw-1, Raw-2, etc.) (Fig. 3.10a-b). The module containing the most genes was given the lowest numeric identifier. Figure 3.10a: Z-score plots of gene co-expression modules in raw/unsoaked beans (top) and correlations between cooking time and gene co-expression modules (bottom). The blue line on each plot represents the cooking times of the samples (x-axis labels show genotype names and the storage treatments). The name of and the number of genes in each module are also shown in the table. Significance levels: >0.01 (*), >0.001 (**). 126 Figure 3.10b: Z-score plots of gene co-expression modules in soaked beans (top) and correlations between cooking time and gene co-expression modules (bottom). The dark blue line on each plot represents the cooking times of the samples (x-axis labels show genotype names and the storage treatments). The name of and the number of genes in each module are also shown in the table. Significance levels: >0.01 (*), >0.001 (**). Raw-1 and Soak-1 were the most significantly correlated with differences in cooking time among the raw and soaked samples, respectively (r2=-0.74, p >0.001; r2=0.80, p >0.001) (Figs. 3.10a-b). The Raw-2, Raw-3, Raw-4, Soak-2, and Soak-3 modules were also significantly correlated with differences in cooking times. Out of all the modules, Raw-1 contained the most genes (2817), followed by Raw-2, Raw-3, Raw-4, Soak-1, Soak-2, and Soak-3. The Raw-2, 127 Raw-3, Soak-1, and Soak-3 modules were positively correlated with cooking times, meaning that the expression levels of genes in these modules increased proportionally with cooking times across genotypes and storage treatments. By contrast, gene expression levels in the Raw-1, Raw- 4, and Soak-2 modules were negatively correlated with cooking times across samples. This could explain why many of the genes that were significantly upregulated in the faster cooking genotypes (relative to the slow-cooking beans from the same market class) belonged to the Raw- 1 and Soak-2 modules (Figs. 3.11a-b). On the other hand, genes belonging to the Raw-2 and Raw-3 were mostly upregulated in the slow-cooking varieties. The Raw-4, Soak-1, and Soak-3 modules were not as consistently associated with fast or slow cooking times. Genes from the Raw-4 module, for example, were upregulated in both TZ-27 and TZ-37. However, genes in Raw-4 were upregulated in Ervilha, the faster cooking yellow bean, relative to PI527538. Interestingly, genes in Soak-3 were upregulated in TZ-27 and Ervilha relative to TZ-37 and PI527538. 128 Figure 3.11a: Bar plots showing the module membership of genes that were more expressed in a specific genotype and storage treatment combination compared to a different genotype and storage treatment combination. Results are from raw beans. Differentially expressed genes (DEGs) were detected using DESeq2 (Bonferroni-Hochman adjusted p-value, α=0.05). The module membership of individual genes was determined using weighted gene co-expression network analysis (WGCNA) (α=0.05). Stacked bars represent the proportion of DEGs that belong to a specific module. Between genotype comparisons were conducted on samples from the same storage treatment and market class (i.e., TZ-27_Fr vs. TZ-37_Fr). Within genotype comparisons were conducted either between untreated and temperate samples, untreated and tropical samples, or temperate and tropical samples from the same genotype. 129 Figure 3.11b: Bar plots showing the module membership of genes that were more expressed in a specific genotype and storage treatment compared to a different genotype and storage treatment combination. Results are from soaked beans. Differentially expressed genes (DEGs) were detected using DESeq2 (Bonferroni-Hochman adjusted p-value, α=0.05). The module membership of individual genes was determined using weighted gene co-expression network analysis (WGCNA) (α=0.05). Stacked bars represent the proportion of DEGs that belong to a specific module. Between genotype comparisons were conducted on samples from the same storage treatment and market class (i.e., TZ-27_Fr vs. TZ-37_Fr). Within genotype comparisons were conducted either between untreated and temperate samples, untreated and tropical samples, or temperate and tropical samples from the same genotype. Regarding storage treatment, genes in Soak-2 and Soak-1 were down- and up-regulated, respectively, in the brown beans in tropical storage conditions relative to untreated and temperate conditions. Genes in Soak-3 were upregulated in Ervilha under untreated and temperate conditions relative to tropical conditions. However, Ervilha still expressed more Soak-3 genes than PI527538 under all storage conditions. TZ-27 and TZ-37 also upregulated genes from Soak- 3 slightly more in tropical conditions relative to untreated and temperate conditions. Very few differences in gene expression were observed in PI527538 across different storage treatments after soaking. Of note, PI527538 was the only genotype in which more differentially expressed 130 genes were identified under raw rather than soaked conditions (Fig. 3.9). It was also the only genotype wherein temperate storage appeared to affect gene expression levels in ways that differed from expectations (i.e., untreated ≤ tropical < temperate patterns were observed) (Tables SI 3.4-3.5). Before soaking, genes from the Raw-2 and Raw-3 modules were more highly expressed in PI527538 in temperate conditions than in untreated or tropical conditions. Relative to the temperate samples, the gene expression profile of PI527538 in tropical conditions more closely resembled that of TZ-37 and Ervilha. Still, PI527538 expressed genes in Raw-2 and Raw-3 at a higher level relative to Ervilha in every storage condition. The modules were annotated using topGO analysis (Figs. SI 3.6a-b). While gene co- expression does not necessarily indicate that genes in a module share similar functions, it could mean that those genes are under the control of one or several master regulators. Several of the GO terms from the Raw-1 module were associated with tRNA-mediated regulation of gene expression, energy metabolism, vitamin metabolism, and response to stress. The Raw-2 module was associated with plastids, transferase activity, cellular carbohydrate metabolic processes, and organellar lumens. The Raw-3 module was associated with insoluble dietary fiber synthesis, and the Raw-4 module was associated with chloroplasts and other non-membrane-bound organelles (i.e., ribosomes, cell walls, microtubules, etc.). The soaked modules contained relatively few GO terms, indicating that fewer genes influenced cooking time in all four genotypes following storage during soaking. The Soak-1 module was associated with protein metabolic processes, whereas the Soak-3 module was associated with protein binding. Identification of genes responsive to tropical storage conditions A small number of genes were identified that were upregulated across all four genotypes under tropical conditions relative to either untreated or temperate conditions. For this analysis, 131 the genotypes were grouped by treatment prior to comparison to find genes that were differentially expressed across all four genotypes (Tables SI 3.6a-b). Significant genes were further filtered based on whether they were significantly associated with a WGCNA module, and whether the genes were upregulated in tropical conditions relative to both untreated and temperate conditions. This produced 21 genes that could potentially play major roles in regulating transcriptomic and physiochemical responses to tropical storage in these genotypes. Seven of these genes were differentially expressed in raw beans, and 14 were differentially expressed in soaked beans (Table 3.1). Note that none of these candidate genes were differentially expressed between untreated and temperate beans. 132 Table 3.1: List of genes that were significantly differentially expressed between two storage treatments across four dry bean genotypes and which were significantly associated with a WGNCA module. Gene names and annotations are in columns 1-2. The module membership of each gene is in column 3. Values in columns 4-5 show the log-fold change in gene expression between two storage treatments. Gene Annotation Module membership Phvul.007G209332 Albino or Glassy Yellow 1 Phvul.010G093100 RAP Phvul.003G005900 Phvul.002G297700 Phvul.008G224000 Pentatricopeptide repeat (PPR) superfamily protein Leucine-rich repeat (LRR) family protein Phospholipase C 2 Phvul.008G265700 Polymerase gamma 2 Phvul.002G101900 Asparagine synthase family Phvul.003G148100 protein Protein kinase superfamily protein with octicosapeptide/Phox/Bem1p domain Raw-1 Raw-1 Raw-4 Raw-4 Raw-1 Raw-1 Raw-1 Raw-4 Untreated vs. Tropical -0.796 Temp vs. Tropical -0.608 -0.451 -0.386 -0.481 -0.321 -0.360 -0.325 -0.291 -0.272 -0.265 -0.241 -0.279 -0.220 -0.241 -0.191 Phvul.003G069300 Voltage dependent anion Raw-1 0.236 0.200 Phvul.007G218600 Phvul.003G041500 Phvul.007G247100 Phvul.003G026400 Phvul.010G134100 Phvul.002G101100 Phvul.010G033500 Phvul.005G057300 Phvul.009G136700 Phvul.002G131900 channel 1 Protein of unknown function, DUF538 SPIRAL1-like1 ATPases;nucleotide binding;ATP binding;nucleoside- triphosphatases;transcription factor binding Calmodulin-binding protein Beta-hydroxyisobutyryl-CoA hydrolase 1 Eukaryotic translation initiation factor 2 gamma subunit Ubiquitin- associated/translation elongation factor EF1B protein . Phvul.003G203200 . Phvul.002G012200 Phvul.002G131700 LOB domain-containing protein 38 Growth-regulating factor 2 Phvul.002G185700 Peptide transporter 2 Phvul.003G206600 Proline transporter 1 Raw-2 0.317 0.257 Raw-1 Raw-2 Raw-2 Soak-2 0.328 0.383 0.392 0.256 0.319 0.333 -0.704 -0.733 Soak-2 Soak-2 -0.334 -0.280 -0.275 -0.289 Soak-2 -0.236 -0.209 Soak-1 0.268 0.257 Soak-1 Soak-1 Soak-1 Soak-1 Soak-1 Soak-1 0.297 0.312 0.360 0.438 0.440 0.507 0.395 0.269 0.367 0.246 0.330 0.629 133 Table 3.1 (cont’d): Phvul.002G024200 . Phvul.003G227300 Protein of unknown function (DUF581) Soak-1 Soak-1 0.558 0.592 0.457 0.517 Phvul.001G130800 Wall associated kinase 5 Soak-1 0.856 0.805 Prior to soaking, four genes [albino or glassy yellow 1, RAP (Ras-related protein), phospholipase C 2, polymerase gamma 2, and asparganine synthase family protein] were found in the Raw-1 module, and three genes [pentatricopeptide repeat (PPR) superfamily protein, leucine-rich repeat (LRR) family protein, and protein kinase superfamily protein with octicosapeptide/Phox/Bem1p domain] were found in Raw-4. Three genes located on chromosome 3 and two genes located on chromosome 7 were upregulated in tropical beans prior to soaking. The genes found on chromosome 3 were part of the Raw-1 module. Two of these genes were annotated as voltage dependent anion channel 1 and SPIRAL1-like1, whereas the other had no known function. The two genes found on chromosome 7 were part of the Raw-2 module. Both genes had no known function. However, one gene was annotated as DUF538, whereas the other had no known annotation. In soaked beans, four genes in Soak-2 were upregulated in untreated and temperate beans, and 10 genes in Soak-1 were upregulated in tropical beans. An ATPase associated with diverse cellular activities, calmodulin-binding protein, beta-hydroxyisobutyryl-CoA hydrolase 1, and eukaryotic translation initiation factor 2 gamma subunit were more expressed in untreated and temperate beans relative to the tropical beans. By contrast, ubiquitin-associated/translation elongation factor EF1B protein, LOB domain-containing protein 38, growth-regulating factor 2, peptide transporter 2, proline transporter 1, DUF581, and wall associated kinase 5 were more expressed in the tropical beans. 134 Phytate content The phytate content in raw TZ-37, PI527538, and Ervilha did not change in response to any of the storage treatments (Fig. 3.12). However, phytate content decreased in TZ-27 in response to the tropical storage treatment. Raw, untreated TZ-27 also had significantly more phytate compared to the other genotypes (p=0.0053). Figure 3.12: Estimated mean raw/unsoaked phytate content of four dry bean genotypes stored in different environments. Phytate measurements are an average of three field plots in Montcalm, MI. One gram of pulverized mature seed was used for analysis. Multiple comparison tests were performed separately on raw/unsoaked and soaked cooking times. Treatments with different letters are not significantly different from other treatments in the same genotype (Sidak, α=0.05). Germination rates of seeds from different seed lots The effect of simulated aging (tropical) conditions on germination rate—an important agronomic trait—was tested on all four of the genotypes (Fig. 3.13). For this experiment, the seeds were harvested from two growing locations in 2023, Montcalm (MC), MI and Saginaw Valley (SV), MI. Since seed coat cracking is known to affect the rate of germination of dry beans, the seed coat cracking severity score of the seeds was tested (Palmer et al., 2022; Soltani et al., 2021) (Table SI 3.7). Three out of the four genotypes grown at the Saginaw Valley 135 location had fewer seed coat cracks compared to the same genotypes grown at Montcalm, but the severity of seed coat cracking was not significantly correlated with germination rate (r2=0.184, p=0.222). Figure 3.13: Estimated average germination rates (%) of four genotypes before and after artificial aging/tropical storage. Dry beans were grown in three plots under unstressed, irrigated conditions in Montcalm, MI and water stressed, rainfed conditions in Saginaw Valley, MI in 2023. Four hundred seeds were used from each field plot for analysis. Artificially aged/tropical seeds were stored in high temperature, high relative humidity conditions for three days prior to germination. Means with the same letter are not significantly different from each other (Sidak, α=0.05). Following exposure to post-harvest metabolic stress, the germination rates of the fast- cooking genotypes grown at SV were lower than the germination rates of the slow-cooking genotypes grown at SV (Fig. 3.13). By contrast, there were no significant differences between the germination rates of the slow- and fast-cooking genotypes grown in MC following exposure to metabolic stress. Thus, the combination of growing location, genotype, and storage treatment significantly affected the germination rates of the fast-cooking seeds. 136 Tropical storage induced cell wall modifications in dry beans DISCUSSION Possible relationships between cooking time and several physicochemical properties in dry beans stored in different environments were investigated, including water uptake, raw seed weight, cell wall thickness, nutrient content, phytate content, and germination rate. This revealed several differences between the treatments and genotypes under investigation. For instance, storage treatment increased the cooking times of all four genotypes by the same number of minutes. Hence, the faster cooking genotypes remained the faster cooking germplasm, even after developing HTC. The beans stored in a temperate environment likely took up more water on average than untreated and tropical beans due to the storage chamber having a moderate, stable relative humidity level (Fig. 3.3). Seed moisture content is known to equilibrate in response to changes in temperature and relative humidity in the environment (Hyde, 1954). Increasing environmental relative humidity typically causes environmentally equilibrated dry bean seeds to open their hilar fissures (Hyde, 1954). Variations in hilar structure have, in turn, been associated with increased water uptake in dry beans (Deshpande and Cheryan, 1986). High relative humidity during storage also had a significant effect on the seeds stored in the tropical environment, as their moisture content was about 6% higher compared to the untreated and temperate beans (Table SI 3.8). It is unclear why the tropical seeds took up less water than the temperate beans during soaking, considering that the relative humidity level of the tropical chamber was higher than that in the temperate chamber. Atmospheric water uptake combined with high temperatures likely contributed to other physicochemical and transcriptomic differences observed in seeds subjected to tropical storage. 137 Tropical storage increased the cooking times of all four genotypes by ~7 minutes (Fig. 3.5). Tropical storage also had significant effects on other physiochemical and biological traits in dry beans. For example, the thickness of the cotyledon cell walls decreased on average in dry beans both before and after soaking in response to high heat and humidity (Fig. 3.6). Yet, cell wall thickness was not equally affected by storage in all four genotypes. Whereas cell wall thickness decreased in the tropical slow-cooking beans compared to the untreated slow-cooking beans, this same trend was not observed in the fast-cooking beans. Thus, it is possible that the slow-cooking genotypes underwent more extensive cell wall remodeling than the fast-cooking genotypes while in tropical storage, leading to a noticeable decrease in cell wall thickness in the tropical slow-cooking genotypes. This same trend has been observed previously in these genotypes following water uptake after three months of storage in room temperature, low humidity (i.e., temperate) conditions (Bassett et al., 2021). Soaked tropical treated beans had higher calcium levels on average than soaked untreated and soaked temperate beans. This could have been caused by calcium binding to exposed pectin chains in the middle lamellae or storage proteins in dry bean cotyledons, as has been previously observed in pulses (Cominelli et al., 2020; Wood et al., 2018). Free calcium was originally hypothesized to have originated from the hydrolysis of phytates, as this process is known to occur more frequently in the presence of high heat and high relative humidity (Coelho et al., 2007; Galiotou-Panayotou et al., 2007; Yi et al., 2016). However, this study did not find evidence of phytate degradation in response to tropical storage except in TZ-27 (Fig. 3.12). Rather, the upregulation of gene co-expression pathways associated with cellular carbohydrate metabolic processes, macromolecule transport, and insoluble cell wall isolate biosynthesis in response to tropical storage (i.e., modules raw-2 and raw-3) suggests that free calcium could 138 have been liberated because of cell wall matrix remodeling (Table 3.1). Increasing the concentration of calcium adjacent to the exposed cell wall matrix could have then resulted in the retention of calcium in the matrix rather than being extracted during soaking and cooking, resulting in a firmer cotyledon texture during cooking (Cominelli et al., 2020). More research is needed to confirm that free calcium was released during soaking in tropical beans and subsequently disproportionately retained in the cell walls, as well as whether the degree of calcium-cell wall interactions was significantly affected by genotype. Cell wall modification under stress may have been regulated by calcium-protein interactions Regardless of genotype and soaking treatment, gene expression patterns in untreated beans more closely resembled that of the temperate beans and vice versa compared to gene expression patterns in the tropical beans (Figs. 3.7-3.8). By filtering out genes that were differentially expressed in all four genotypes in response to tropical storage relative to both untreated and temperate storage, a few candidate genes were identified. Using topGO to annotate the differential gene expression profiles and WGCNA pathway data of beans stored in different environments, several connections were identified between tropical storage and stress-induced calcium-mediated signaling pathways. For instance, an ATPase was expressed in soaked beans but was downregulated in tandem with a calmodulin-binding protein in tropical beans relative to untreated and temperate beans (Table 3.1). ATPases capable of transporting calcium have been identified (Johnson et al., 2009). Calmodulin-binding protein homologs in Arabidopsis are activated by stress-induced calcium signaling and prevent an excessive response to osmotic stress in Arabidopsis seedlings (Perruc et al., 2004). Taken together, differential expression of these genes under multiple stress 139 conditions are evidence of a calcium induced stress response in dry beans. Other stress-responsive genes were upregulated in dry beans stored in tropical conditions relative to untreated dry beans or beans stored in temperate conditions. For example, stress- responsive proteins such as wall associated kinase 5, DUF581 (also known as zf-FCS type zinc finger protein), and LOB domain-containing protein 38 were upregulated specifically in tropical beans during water uptake. Overexpression of these proteins has previously been associated with increased stress resistance in plants (Du et al., 2022; Jamsheer and Laxmi, 2015, 2014; Nietzsche et al., 2014). Wall associated kinase 5 (WAK5) may warrant further investigation because it is both a calcium and a galacturonan (pectin) receptor. Briefly, cell wall remodeling is often accompanied by the release of pectin chains into the surrounding area (Kohorn et al., 2012). Upon binding to pectin, WAKs transmit signals to the nucleus, leading to either cell expansion or a stress response depending on the type and concentration of the pectin chains (Kohorn et al., 2012; Kohorn and Kohorn, 2012; Sun et al., 2020). WAKs may also bind to calcium, facilitating the crosslinking of pectin chains in the cell wall (Decreux and Messiaen, 2005). These genes alone can confer resistance to pathogenic attacks by strengthening the cell wall (Hu et al., 2017; Kessel et al., 2023). Changes in cell wall strength across all four genotypes could explain why all four genotypes took ~7 minutes longer to cook on average following soaking (Jeffery et al., 2023). Therefore, wall associated kinase 5 may be a suitable candidate gene for future studies of HTC development in dry beans. Physiochemical evidence for a relationship between fast cooking time and germination rate is bolstered by transcriptomic evidence Tropical storage conditions decreased the germination rates of the fast-cooking varieties, but this effect was only seen in beans that were grown in a comparatively dry environment (Fig. 140 3.13). This supports previous findings that both preharvest and postharvest can influence the germination rates and, therefore, the viability of seeds (Cichy et al., 2019; Cortelazzo et al., 2005; López Herrera et al., 2001). In addition, this work demonstrates possibly for the first time that lower yields in water- and heat-stressed environments may be a trade-off to growing fast- cooking genotypes. Lower germination rates could be related to weaker (or possibly delayed) metabolic stress responses that were previously identified in these fast-cooking genotypes (Jeffery et al., 2023). Lower levels of phytate in TZ-27—a compound involved in seed germination which can decrease the cooking times of dry beans—or lower procyanidin levels— compounds that protect seeds from abiotic and biotic stress and increase the cooking times of dry beans—could also explain these observations (Cominelli et al., 2018; Galiotou-Panayotou et al., 2007; Hincks and Stanley, 1987; Raboy, 2009; Reyes-Moreno et al., 2000; Stanley, 1992; Wiesinger et al., 2021) (Fig. 3.12). Improving the protective mechanisms of dry beans to heat and drought stress could make them more adaptable to climate change. Since the germination rates of the slow-cooking genotypes were more resilient to pre- and postharvest stress, genes that were upregulated in TZ- 27 and PI527538 were explored. The GO term ‘response to inorganic substance’ was only upregulated in soaked tropical slow-cooking varieties. Considering that the slow-cooking genotypes appeared to undergo more extensive cell wall remodeling in response to tropical storage, it is possible that the ‘inorganic substance’ in this case refers to signaling molecules such as calcium (Fig. 3.6). This is in accordance with previous findings that cell wall remodeling and calcium-binding proteins were differentially upregulated in TZ-27 and PI527538 relative to TZ- 37 and Ervilha (Jeffery et al., 2023). The list of genes that were differentially expressed between fast- and slow-cooking beans 141 provides a starting point for future research on genotype-specific mechanism controlling HTC development in dry beans (Table SI 3.3). However, no basis for selecting genotype-specific candidate genes could be identified given the available information. More information will be needed to select candidate genes for breeding that improve the genetic response of dry beans to hot, humid conditions. CONCLUSIONS This work sought to understand the different responses of dry beans to hot, humid storage conditions by performing physiochemical analyses and RNA-sequencing on four dry bean genotypes with contrasting cooking times. Storage parameters that mimic conditions in temperate and tropical climates were chosen to extend the applicability of this research to growers, processors, and consumers. Relative to beans stored in untreated and temperate beans, tropical beans demonstrated symptoms of hard-to-cook (HTC). All the beans that were stored in tropical conditions took up less water than expected, took ~7 minutes longer to cook after soaking for 12 hours in distilled water, and retained more calcium and manganese on average than untreated beans after soaking. Gene expression analysis revealed that several cell wall remodeling genes were upregulated in tropical beans. Wall associated kinase 5 was identified as a potential candidate gene for the development of hard-to-cook in all four of the genotypes. The fast-cooking genotypes had superior cooking times both before and after exposure to tropical storage, emphasizing their usefulness to dry bean consumers around the world. On the other hand, the slow-cooking genotypes exhibited more tolerance to stress, as evidenced by their faster germination rates following exposure to pre-harvest drought stress and post-harvest artificial aging. The GO term ‘response to inorganic substance’ was associated with genes that were upregulated in the slow-cooking beans (but not the fast-cooking beans) after soaking under 142 tropical storage conditions, suggesting that the slow-cooking beans were more responsive to environmentally induced influxes of inorganic molecules like calcium. This work provides resources to aid future investigations of genotype-specific responses to environmental stress in the hope that this will facilitate efforts to improve the climate resiliency of dry beans. 143 BIBLIOGRAPHY Abay, M.M., Tolesa, G.N., 2023. Effects of storage temperature and relative humidity on cooking time and moisture uptake of selected common bean (Phaseolus vulgaris L.) varieties grown in Ethiopia. Cogent Food Agric. 9, 2163577. https://doi.org/10.1080/23311932.2022.2163577 Aguilera, J.M., Hau, M.I., Villablanca, W., 1986. The effect of solar drying and heating on the hardness of Phaseolus beans during storage. J. Stored Prod. Res. 22, 243–247. https://doi.org/10.1016/0022-474X(86)90017-2 Alexa, A., Rahnenführer, J., 2021. Gene set enrichment analysis with topGO. Bioconductor Improv 27, 1–26. Amongi, W., Mukankusi, C., Sebuliba, S., Mukamuhirwa, F., 2018. Iron and zinc grain concentrations diversity and agronomic performance of common bean germplasm collected from East Africa. Afr. J. Food Agric. Nutr. Dev. 18, 13717–13742. https://doi.org/10.18697/ajfand.83.17070 Anders, S., Pyl, P.T., Huber, W., 2015. HTSeq—a Python framework to work with high- throughput sequencing data. Bioinformatics 31, 166–169. https://doi.org/10.1093/bioinformatics/btu638 Andrews, S., 2023. FastQC: A quality control tool for high throughput sequence data. AOAC International, 2000. Official methods of analysis of AOAC International. Association of Official Analytical Chemists. Bassett, A., Hooper, S., Cichy, K., 2021. Genetic variability of cooking time in dry beans (Phaseolus vulgaris L.) related to seed coat thickness and the cotyledon cell wall. Food Res. Int. 141, 109886. https://doi.org/10.1016/j.foodres.2020.109886 Bates, D., Mächler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48. https://doi.org/doi:10.18637/jss.v067.i01 Beninger, C.W., Gu, L., Prior, R.L., Junk, D.C., Vandenberg, A., Bett, K.E., 2005. Changes in polyphenols of the seed coat during the after-darkening process in pinto beans (Phaseolus vulgaris L.). J. Agric. Food Chem. 53, 7777–7782. https://doi.org/10.1021/jf050051l Berry, M., Izquierdo, P., Jeffery, H., Shaw, S., Nchimbi-Msolla, S., Cichy, K.A., 2020. QTL analysis of cooking time and quality traits in dry bean (Phaseolus vulgaris L.). Theor. Appl. Genet. 133, 2291–2305. https://doi.org/10.1007/s00122-020-03598-w Bolger, A.M., Lohse, M., Usadel, B., 2014. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinforma. Oxf. Engl. 30, 2114–2120. https://doi.org/10.1093/bioinformatics/btu170 144 Chen, D., Ding, A., Zhu, L., Grauwet, T., Van Loey, A., Hendrickx, M., Kyomugasho, C., 2023. Phytate and mineral profile evolutions to explain the textural hardening of common beans (Phaseolus vulgaris L.) during postharvest storage and soaking: Insights obtained through a texture-based classification approach. Food Chem. 404, 134531. https://doi.org/10.1016/j.foodchem.2022.134531 Chen, D., Hu, K., Zhu, L., Hendrickx, M., Kyomugasho, C., 2022. Cell wall polysaccharide changes and involvement of phenolic compounds in ageing of Red haricot beans (Phaseolus vulgaris) during postharvest storage. Food Res. Int. 162, 112021. https://doi.org/10.1016/j.foodres.2022.112021 Chigwedere, C.M., Nkonkola, C.M., Rai, S., Kyomugasho, C., Kermani, Z.J., Pallares Pallares, A., Van Loey, A.M., Grauwet, T., Hendrickx, M.E., 2019. Cotyledon pectin molecular interconversions explain pectin solubilization during cooking of common beans (Phaseolus vulgaris). Food Res. Int. 116, 462–470. https://doi.org/10.1016/j.foodres.2018.08.062 Cichy, K.A., Wiesinger, J.A., Berry, M., Nchimbi-Msolla, S., Fourie, D., Porch, T.G., Ambechew, D., Miklas, P.N., 2019. The role of genotype and production environment in determining the cooking time of dry beans (Phaseolus vulgaris L.). Legume Sci. 1, e13. https://doi.org/10.1002/leg3.13 Coelho, C.M.M., de Mattos Bellato, C., Santos, J.C.P., Ortega, E.M.M., Tsai, S.M., 2007. Effect of phytate and storage conditions on the development of the ‘hard-to-cook’ phenomenon in common beans. J. Sci. Food Agric. 87, 1237–1243. https://doi.org/10.1002/jsfa.2822 Cominelli, E., Confalonieri, M., Carlessi, M., Cortinovis, G., Daminati, M.G., Porch, T.G., Losa, A., Sparvoli, F., 2018. Phytic acid transport in Phaseolus vulgaris: A new low phytic acid mutant in the PvMRP1 gene and study of the PvMRPs promoters in two different plant systems. Plant Sci. 270, 1–12. https://doi.org/10.1016/j.plantsci.2018.02.003 Cominelli, E., Galimberti, M., Pongrac, P., Landoni, M., Losa, A., Paolo, D., Daminati, M.G., Bollini, R., Cichy, K.A., Vogel-Mikuš, K., Sparvoli, F., 2020. Calcium redistribution contributes to the hard-to-cook phenotype and increases PHA-L lectin thermal stability in common bean low phytic acid 1 mutant seeds. Food Chem. 321, 126680. https://doi.org/10.1016/j.foodchem.2020.126680 Conway, J.R., Lex, A., Gehlenborg, N., 2017. UpSetR: an R package for the visualization of intersecting sets and their properties. Bioinformatics 33, 2938–2940. https://doi.org/10.1093/bioinformatics/btx364 Cortelazzo, A.L., Coutinho, J., Granjeiro, P.A., 2005. Storage and aging of French beans. J. Morphol. Sci. 22. Decreux, A., Messiaen, J., 2005. Wall-associated kinase WAK1 interacts with cell wall pectins in a calcium-induced conformation. Plant Cell Physiol. 46, 268–278. https://doi.org/10.1093/pcp/pci026 145 Deshpande, S.S., Cheryan, M., 1986. Microstructure and water uptake of Phaseolus and winged beans. J. Food Sci. 51, 1218–1223. https://doi.org/10.1111/j.1365-2621.1986.tb13089.x Du, Y., Zhao, Q., Li, W., Geng, J., Li, S., Yuan, X., Gu, Y., Zhong, J., Zhang, Y., Du, J., 2022. Genome-wide identification of the LBD transcription factor genes in common bean (Phaseolus vulgaris L.) and expression analysis under different abiotic stresses. J. Plant Interact. 17, 731–743. https://doi.org/10.1080/17429145.2022.2095449 Durinck, S., Moreau, Y., Kasprzyk, A., Davis, S., De Moor, B., Brazma, A., Huber, W., 2005. BioMart and Bioconductor: a powerful link between biological databases and microarray data analysis. Bioinformatics 21, 3439–3440. https://doi.org/10.1093/bioinformatics/bti525 Durinck, S., Spellman, P.T., Birney, E., Huber, W., 2009. Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nat. Protoc. 4, 1184– 1191. https://doi.org/10.1038/nprot.2009.97 Duwadi, K., Austin, R.S., Mainali, H.R., Bett, K., Marsolais, F., Dhaubhadel, S., 2018. Slow darkening of pinto bean seed coat is associated with significant metabolite and transcript differences related to proanthocyanidin biosynthesis. BMC Genomics 19, 260. https://doi.org/10.1186/s12864-018-4550-z Elsadr, H.T., Marles, M.A.S., Caldas, G.V., Blair, M.W., Bett, K.E., 2015. Condensed tannin accumulation during seed coat development in five common bean genotypes. Crop Sci. 55, 2826–2832. https://doi.org/10.2135/cropsci2015.01.0051 Fox, J., Weisberg, S., 2019. An R companion to applied regression. Freixas Coutin, J.A., Munholland, S., Silva, A., Subedi, S., Lukens, L., Crosby, W.L., Pauls, K.P., Bozzo, G.G., 2017. Proanthocyanidin accumulation and transcriptional responses in the seed coat of cranberry beans (Phaseolus vulgaris L.) with different susceptibility to postharvest darkening. BMC Plant Biol. 17, 89. https://doi.org/10.1186/s12870-017- 1037-z Galiotou-Panayotou, M., Kyriakidis, N.B., Margaris, I., 2007. Phytase–phytate–pectin hypothesis and quality of legumes cooked in calcium solutions. J. Sci. Food Agric. 88, 355–361. https://doi.org/10.1002/jsfa.2973 Garcia, E., Filisetti, T.M.C.C., Udaeta, J.E.M., Lajolo, F.M., 1998. Hard-to-cook beans (Phaseolus vulgaris):  Involvement of phenolic compounds and pectates. J. Agric. Food Chem. 46, 2110–2116. https://doi.org/10.1021/jf970848f Hincks, M.J., Stanley, D.W., 1987. Lignification: Evidence for a role in hard-to-cook beans. J. Food Biochem. 11, 41–58. https://doi.org/10.1111/j.1745-4514.1987.tb00111.x Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in general parametic models. Biom. J. 50, 346–363. https://doi.org/10.32614/CRAN.package.multcomp 146 Hu, K., Cao, J., Zhang, J., Xia, F., Ke, Y., Zhang, H., Xie, W., Liu, H., Cui, Y., Cao, Y., Sun, X., Xiao, J., Li, X., Zhang, Q., Wang, S., 2017. Improvement of multiple agronomic traits by a disease resistance gene via cell wall reinforcement. Nat. Plants 3, 1–9. https://doi.org/10.1038/nplants.2017.9 Hyde, E.O.C., 1954. The function of the hilum in some Papilionaceae in relation to the ripening of the deed and the permeability of the testa. Ann. Bot. 18, 241–256. https://doi.org/10.1093/oxfordjournals.aob.a083393 IPCC, 2007. Climate change 2007: Full report. Contribution of working groups I, II and III to the fourth assessment Report of the intergovernmental panel on climate change. Intergovernmental Panel on Climate Change. Islam, N.S., Bett, K.E., Pauls, K.P., Marsolais, F., Dhaubhadel, S., 2020. Postharvest seed coat darkening in pinto bean (Phaseolus vulgaris) is regulated by Psd, an allele of the basic helix-loop-helix transcription factor P. PLANTS PEOPLE PLANET 2, 663–677. https://doi.org/10.1002/ppp3.10132 ISTA, 2023. International rules for seed testing. International Seed Testing Association, Wallisellen, Switzerland. Jamsheer, M., Laxmi, A., 2015. Expression of Arabidopsis FCS-like zinc finger genes is differentially regulated by sugars, cellular energy level, and abiotic stress. Front. Plant Sci. 6. https://doi.org/10.3389/fpls.2015.00746 Jamsheer, M., Laxmi, A., 2014. DUF581 Is plant specific FCS-like zinc finger involved in protein-protein interaction. PLOS ONE 9, e99074. https://doi.org/10.1371/journal.pone.0099074 Jeffery, H.R., Mudukuti, N., Buell, C.R., Childs, K.L., Cichy, K., 2023. Gene expression profiling of soaked dry beans (Phaseolus vulgaris L.) reveals cell wall modification plays a role in cooking time. Plant Genome 16, e20364. https://doi.org/10.1002/tpg2.20364 Johnson, N.A., Liu, F., Weeks, P.D., Hentzen, A.E., Kruse, H.P., Parker, J.J., Laursen, M., Nissen, P., Costa, C.J., Gatto, C., 2009. A tomato ER-type Ca2+-ATPase, LCA1, has a low thapsigargin-sensitivity and can transport manganese. Arch. Biochem. Biophys. 481, 157–168. https://doi.org/10.1016/j.abb.2008.11.010 Kessel, B., Ouzunova, M., Scheurmann, D., Keller, B., Krattinger, S., PRAZ, C., Yang, P., ERB, M., 2023. Increased fungal resistance in plants via modulation of a wall-associated kinase or benzoxazinoids. US20230021982A1. Kim, D., Paggi, J.M., Park, C., Bennett, C., Salzberg, S.L., 2019. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37, 907– 915. https://doi.org/10.1038/s41587-019-0201-4 Kinsella, R.J., Kähäri, A., Haider, S., Zamora, J., Proctor, G., Spudich, G., Almeida-King, J., Staines, D., Derwent, P., Kerhornou, A., Kersey, P., Flicek, P., 2011. Ensembl BioMarts: 147 a hub for data retrieval across taxonomic space. Database 2011, bar030. https://doi.org/10.1093/database/bar030 Kohorn, B.D., Kohorn, S.L., 2012. The cell wall-associated kinases, WAKs, as pectin receptors. Front. Plant Sci. 3, 88. https://doi.org/10.3389/fpls.2012.00088 Kohorn, B.D., Kohorn, S.L., Todorova, T., Baptiste, G., Stansky, K., McCullough, M., 2012. A dominant allele of Arabidopsis pectin-binding wall-associated kinase induces a stress response suppressed by MPK6 but not MPK3 mutations. Mol. Plant 5, 841–851. https://doi.org/10.1093/mp/ssr096 Langfelder, P., Horvath, S., 2012. Fast R Functions for Robust Correlations and Hierarchical Clustering. J. Stat. Softw. 46, i11. Langfelder, P., Horvath, S., 2008. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559. https://doi.org/10.1186/1471-2105-9-559 Law, C.W., Chen, Y., Shi, W., Smyth, G.K., 2014. voom: precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 15, R29. https://doi.org/10.1186/gb-2014-15-2-r29 Lenth, R.V., 2024. emmeans: Estimated marginal means, a.k.a. least-squares means. López Herrera, M., Aguirre Rivera, J.R., Trejo, C., Peña-Valdivia, C.B., 2001. Differences in seed germination of wild and domesticated common bean (Phaseolus vulgaris L.) in response to storage. South Afr. J. Bot. 67, 620–628. https://doi.org/10.1016/S0254- 6299(15)31192-3 Love, M.I., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. https://doi.org/10.1186/s13059-014- 0550-8 Mafuleka, M.M., Ott, D.B., Hosfield, G.L., Uebersax, M.A., 1993. The role of phytase and lignin decorticated dry bean (Phaseolus vulgaris) hardening during storage. J. Food Process. Preserv. 17, 1–20. https://doi.org/10.1111/j.1745-4549.1993.tb00222.x Martínez-Manrique, E., Jacinto-Hernández, C., Garza-García, R., Campos, A., Moreno, E., Bernal-Lugo, I., 2011. Enzymatic changes in pectic polysaccharides related to the beneficial effect of soaking on bean cooking time. J. Sci. Food Agric. 91, 2394–2398. https://doi.org/10.1002/jsfa.4474 Matthews, T., 2018. Humid heat and climate change. Prog. Phys. Geogr. Earth Environ. 42, 391– 405. https://doi.org/10.1177/0309133318776490 Molina, M.R., Baten, M.A., Gomez-Brenes, R.A., King, K.W., Bressani, R., 1976. Heat treatment: A process to control the development of the hard-to-cook phenomenon in black beans (Phaseolus vulgaris). J. Food Sci. 41, 661–666. https://doi.org/10.1111/j.1365-2621.1976.tb00694.x 148 Nietzsche, M., Schießl, I., Börnke, F., 2014. The complex becomes more complex: protein- protein interactions of SnRK1 with DUF581 family proteins provide a framework for cell- and stimulus type-specific SnRK1 signaling in plants. Front. Plant Sci. 5. https://doi.org/10.3389/fpls.2014.00054 Njoroge, D.M., Kinyanjui, P.K., Christiaens, S., Shpigelman, A., Makokha, Anselimo.O., Sila, D.N., Hendrickx, M.E., 2015. Effect of storage conditions on pectic polysaccharides in common beans (Phaseolus vulgaris) in relation to the hard-to-cook defect. Food Res. Int., Grain legumes - Science, Technology and Impacts on Human Health 76, 105–113. https://doi.org/10.1016/j.foodres.2015.03.005 Njoroge, D.M., Kinyanjui, P.K., Makokha, A.O., Christiaens, S., Shpigelman, A., Sila, D.N., Hendrickx, M.E., 2014. Extraction and characterization of pectic polysaccharides from easy- and hard-to-cook common beans (Phaseolus vulgaris). Food Res. Int. Ott. Ont 64, 314–322. https://doi.org/10.1016/j.foodres.2014.06.044 Oliveros, J.C., 2015. Venny: An interactive tool for comparing lists with Venn’s diagrams. Palmer, J.P., Pajak, A., Robson, B., Zhang, B., Joshi, J., Diapari, M., Pauls, K.P., Marsolais, F., 2022. Pectin acetylesterase 8 influences pectin acetylation in the seed coat, seed imbibition, and dormancy in common bean (Phaseolus vulgaris L.). Legume Sci. 4, e130. https://doi.org/10.1002/leg3.130 Perruc, E., Charpenteau, M., Ramirez, B.C., Jauneau, A., Galaud, J.-P., Ranjeva, R., Ranty, B., 2004. A novel calmodulin-binding protein functions as a negative regulator of osmotic stress tolerance in Arabidopsis thaliana seedlings. Plant J. 38, 410–420. https://doi.org/10.1111/j.1365-313X.2004.02062.x Polasky, A.D., Evans, J.L., Fuentes, J.D., Hamilton, H.L., 2022. Statistical climate model downscaling for impact projections in the Midwest United States. Int. J. Climatol. 42, 3038–3055. https://doi.org/10.1002/joc.7406 Raboy, V., 2009. Approaches and challenges to engineering seed phytate and total phosphorus. Plant Sci. 177, 281–296. https://doi.org/10.1016/j.plantsci.2009.06.012 Reyes-Moreno, C., Okamura-Esparza, J., Armienta-Rodelo, E., Gómez-Garza, R.M., Milán- Carrillo, J., 2000. Hard-to-cook phenomenon in chickpeas (Cicer arietinum L): Effect of accelerated storage on quality. Plant Foods Hum. Nutr. 55, 229–241. https://doi.org/10.1023/A:1008106229189 RStudio Team, 2024. RStudio: Integrated development for R. Sadohara, R., Izquierdo, P., Couto Alves, F., Porch, T., Beaver, J., Urrea, C.A., Cichy, K., 2022. The Phaseolus vulgaris L. Yellow Bean Collection: Genetic diversity and characterization for cooking time. Genet. Resour. Crop Evol. 69, 1627–1648. https://doi.org/10.1007/s10722-021-01323-0 149 Savo, V., Lepofsky, D., Benner, J.P., Kohfeld, K.E., Bailey, J., Lertzman, K., 2016. Observations of climate change among subsistence-oriented communities around the world. Nat. Clim. Change 6, 462–473. https://doi.org/10.1038/nclimate2958 Schmutz, J., McClean, P.E., Mamidi, S., Wu, G.A., Cannon, S.B., Grimwood, J., Jenkins, J., Shu, S., Song, Q., Chavarro, C., Torres-Torres, M., Geffroy, V., Moghaddam, S.M., Gao, D., Abernathy, B., Barry, K., Blair, M., Brick, M.A., Chovatia, M., Gepts, P., Goodstein, D.M., Gonzales, M., Hellsten, U., Hyten, D.L., Jia, G., Kelly, J.D., Kudrna, D., Lee, R., Richard, M.M.S., Miklas, P.N., Osorno, J.M., Rodrigues, J., Thareau, V., Urrea, C.A., Wang, M., Yu, Y., Zhang, M., Wing, R.A., Cregan, P.B., Rokhsar, D.S., Jackson, S.A., 2014. A reference genome for common bean and genome-wide analysis of dual domestications. Nat. Genet. 46, 707–713. https://doi.org/10.1038/ng.3008 Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675. https://doi.org/10.1038/nmeth.2089 Shomer, I., Paster, N., Lindner, P., Vasiliver, R., 1990. The role of cell wall structure in the hard- to-cook phenomenon in beans (Phaseolus vulgaris L.). Food Struct. 9. Soltani, A., Walter, K.A., Wiersma, A.T., Santiago, J.P., Quiqley, M., Chitwood, D., Porch, T.G., Miklas, P., McClean, P.E., Osorno, J.M., Lowry, D.B., 2021. The genetics and physiology of seed dormancy, a crucial trait in common bean domestication. BMC Plant Biol. 21, 58. https://doi.org/10.1186/s12870-021-02837-6 Srisuma, N., Hammerschmidt, R., Uebersax, M.A., Ruengsakulrach, S., Bennink, M.R., Hosfield, G.L., 1989. Storage induced changes of phenolic acids and the development of hard-to-cook in dry beans (Phaseolus vulgaris var. Seafarer). J. Food Sci. 54, 311–314. https://doi.org/10.1111/j.1365-2621.1989.tb03069.x Stanley, D.W., 1992. A possible role for condensed tannins in bean hardening. Food Res. Int. 25, 187–192. https://doi.org/10.1016/0963-9969(92)90136-S Sun, Z., Song, Y., Chen, D., Zang, Y., Zhang, Q., Yi, Y., Qu, G., 2020. Genome-wide identification, classification, characterization, and expression znalysis of the wall- associated kinase family during fruit development and under wound stress in tomato (Solanum lycopersicum L.). Genes 11, 1186. https://doi.org/10.3390/genes11101186 Toili, M.E.M., de Koning, R., Kiekens, R., Ndumba, N., Wahome, S., Anami, S., Githiri, S.M., Angenon, G., 2022. Transcriptome-guided identification of pectin methyl-esterase-related enzymes and novel molecular processes effectuating the hard-to-cook defect in common bean (Phaseolus vulgaris L.). Foods 11, 1692. https://doi.org/10.3390/foods11121692 Wainaina, I., Wafula, E., Van Loey, A., Sila, D., Hendrickx, M., Kyomugasho, C., 2022. Kinetics of phytate hydrolysis during storage of red kidney beans and the implication in hard-to-cook development. Food Res. Int. 159, 111581. https://doi.org/10.1016/j.foodres.2022.111581 150 Warnes, G.R., Bolker, B., Bonebakker, L., Gentleman, R., Huber, W., Liaw, A., Lumley, T., Maechler, M., Magnusson, A., Moeller, S., Schwartz, M., Venables, B., 2022. gplots: Various R programming tools for plotting data. Wei, T., Simko, V., 2021. R package “corrplot”: Visualization of a correlation matrix. Wiesinger, J.A., Osorno, J.M., McClean, P.E., Hart, J.J., Glahn, R.P., 2021. Faster cooking times and improved iron bioavailability are associated with the down regulation of procyanidin synthesis in slow-darkening pinto beans (Phaseolus vulgaris L.). J. Funct. Foods 82, 104444. https://doi.org/10.1016/j.jff.2021.104444 Willett, K.M., Gillett, N.P., Jones, P.D., Thorne, P.W., 2007. Attribution of observed surface humidity changes to human influence. Nature 449, 710–712. https://doi.org/10.1038/nature06207 Wood, J.A., Tan, H.-T., Collins, H.M., Yap, K., Khor, S.F., Lim, W.L., Xing, X., Bulone, V., Burton, R.A., Fincher, G.B., Tucker, M.R., 2018. Genetic and environmental factors contribute to variation in cell wall composition in mature desi chickpea (Cicer arietinum L.) cotyledons. Plant Cell Environ. 41, 2195–2208. https://doi.org/10.1111/pce.13196 Yi, J., Njoroge, D.M., Sila, D.N., Kinyanjui, P.K., Christiaens, S., Bi, J., Hendrickx, M.E., 2016. Detailed analysis of seed coat and cotyledon reveals molecular understanding of the hard- to-cook defect of common beans (Phaseolus vulgaris L.). Food Chem. 210, 481–490. https://doi.org/10.1016/j.foodchem.2016.05.018 Zhao, Y., Ducharne, A., Sultan, B., Braconnot, P., Vautard, R., 2015. Estimating heat stress from climate-based indicators: present-day biases and future spreads in the CMIP5 global climate model ensemble. Environ. Res. Lett. 10, 084013. https://doi.org/10.1088/1748- 9326/10/8/084013 151 CHAPTER 4: NEW (BIO)MOLECULAR RESOURCES TO IMPROVE THE COOKING TIMES OF DRY BEANS (PHASEOLUS VULGARIS L.) 152 ABSTRACT Dry beans (Phaseolus vulgaris L.) are a nutritious food with highly variable and generally long cooking times that limit their use as a food. The availability of fast-cooking germplasm and reliable genetic markers for marker-assisted breeding could accelerate the development of reliably fast cooking dry bean varieties for the global market. Two within market class recombinant inbred line populations (TTRIL and YYRIL) developed by crossing a fast-by- slow cooking genotype were previously used to identify QTL for cooking time. The QTL from these studies were explored for their potential in marker assisted selection for cooking time. The fast-cooking genotypes from each study were crossed and backcrossed to the yellow parent to develop a validation population (Fast cooking backcross: FBC). In total 12 kompetitive allele- specific PCR (KASP) markers were designed based on the cooking time QTL. One phenotypic marker, UV darkening of the seed coat was also evaluated. These were tested in the backcross validation population and USDA-ARS breeding lines within the kidney, cranberry, and yellow market classes. Results indicated that cooking time was associated with the non-darkening phenotype and three single nucleotide polymorphisms (SNPs) (snpPV00038, snpPV00169, and snpPV00218). Across 331 Andean and Mesoamerican dry bean genotypes from the yellow, kidney, cranberry, brown, black, and pinto market classes, KASP markers for these SNPs failed 11, 7, and 3% of the time, respectively. Non-darkening was the most reliable predictor of fast cooking time, as it was significantly associated with fast cooking time in the YYRIL and FCBC populations. The slowest and fastest cooking varieties from the FCBC population segregated for the non-darkening trait and snpPV00218, but not the other traits. INTRODUCTION Dry beans are rich in protein and micronutrients, low in fat, broadly available, and can fix 153 nitrogen by forming symbiotic relationships with bacteria (Havemeier & Slavin, 2020; Mudryj et al., 2014; Uebersax et al., 2023). However, dry beans have long and highly variable cooking times (Cichy et al., 2015). This has the potential to limit their popularity amongst consumers who demand shorter meal preparation times, as well as consumers who rely exclusively on raw fuel like firewood for cooking (Asiimwe et al., 2024; Boy et al., 2000; Rahkovsky et al., 2018). Breeding for faster cooking time in dry beans is a viable target, considering that there is broad genetic diversity for cooking time. Breeding is also an effective and easy-to-access tool that can be used by people around the world to better meet the needs of consumers. However, resources that could speed up the process of breeding faster cooking dry bean varieties are currently scarce, expensive, or unreliable. For example, a common phenotyping device used to measure cooking time, the Mattson cooker, requires many seeds for testing, necessitates the use of specialized equipment, and is time-consuming (Bento et al., 2020). Other phenotyping methods such as the finger press method, wherein a researcher presses a bean between their fingers to determine doneness, also consume seeds and time and are less reliable since results can vary between researchers (Wood, 2017). There are a number of potential candidate QTL, genes and phenotypic markers for cooking time that can be useful in breeding for reduced cooking times (Berry et al., 2020; Cichy et al., 2019; Katuuramu et al., 2020; Sadohara, Izquierdo, et al., 2022). Procyanidins are thought to influence the cooking times of dry beans by binding to cell walls and proteins in the cotyledon and the seed coat, thereby decreasing the solubility of the seed (Stanley, 1992). Procyanidin substantially accumulate in the seed coats of some dry bean varieties, but not others, under environmental stress such as prolonged aging, storage in hot, humid conditions, or exposure to ultraviolet (UV) light, leading to the visible darkening of the seed coat (Beninger et al., 2005; 154 Wiesinger et al., 2021). Dry bean varieties that accumulate little to no procyanidins are known as slow- and non-darkening varieties, respectively. This has led to the development of a simple test to detect slow- and non-darkening dry bean varieties, wherein seeds are exposed to UV light for three days, followed by a visual evaluation to determine whether the seed coat darkened (Junk- Knievel et al., 2007). In addition, since seed coat darkening is under the control of relatively few genes, a couple of genetic markers have been developed for seed coat darkening (CGIAR, 2023; Erfatpour & Pauls, 2020) The interconversion of soluble pectin to a more tightly bound form has also been associated with the development of long cooking times in dry beans during storage (Chigwedere et al., 2019). Like seed coat darkening, both genotype and environment contribute to the development of pectin insolubility (Njoroge et al., 2015). However, the genetic explanation for the development of pectin insolubility is more complex. An abundance or a lack of divalent cations such as calcium are thought to bind to demethylesterfied pectin chains, rendering the pectin less soluble (Pelloux et al., 2007). In turn, enzymes that demethylesterfy pectin or eliminate phytate (a chemical that chelates free calcium) during environmental stress could also increase pectin insolubility (Mafuleka et al., 1993; Martínez-Manrique et al., 2011). Going further, considering that structural pectin chains are located in the interior of the cell wall matrix, it follows that the expansion of the cell wall matrix in the presence of free calcium could lead to the insolubilization of demethylesterfied pectin chains in the cell wall (Jarvis, 1992; X. Wang et al., 2020). Genes associated with cell wall expansion and pectin methylesterfication in response to stress have previously been associated with cooking time in dry beans, meaning they could be useful for marker-assisted breeding for cooking time (Jeffery et al., 2023). Some of these genes include calcium-binding EF-hand family protein, cell wall modification genes like xyloglucan 155 endotransglycosylase/hydrolases, exordium like 2, highly ABA-induced PP2C gene 3, DEX1, SKU similar 5 (a pectinesterase-type protein), HXXXD-type acyl-transferase family protein (also known as phenolic glucoside malonyltransferase 1), and expansin-like A1, and stress responsive genes like highly ABA-induced PP2C gene 3, ARM repeat superfamily protein, and Hsp20-like chaperone superfamily protein (Jeffery et al., 2023). Water uptake is also closely related to cooking time, as it is needed to solubilize proteins, starches, and pectins in the cotyledons (Berry et al., 2020). It triggers the expansion of the cell walls, as well, which may, in turn, affect pectin solubility (Jeffery et al., 2023). Since water first enters the seed via orifices (i.e., the raphe, hilum, and micropyle) in the seed coat, seed coat integrity is thought to be of utmost importance to cooking time. Cracks in the seed coat or its orifices can be caused either by environmental damage (i.e., combine harvesting, excessive drying, etc.) or by genotypic differences. One gene called pectin acetylesterase 8 (PAE8) is known to cause seed coat cracking, leading to increased water uptake (Palmer et al., 2022; Soltani et al., 2021). Thus, PAE8 is a potential marker for the genetic control of cooking time in dry beans. It has yet to be confirmed, though, whether PAE8 has a direct effect on the cooking times of dry beans. The development of reliable genetic markers for cooking time could provide breeders with a cheap, all-in-one alternative to measuring cotyledon hardness, UV light test, pectin solubility, and water uptake percentage. Numerous studies have identified genetic polymorphisms (i.e., QTL) associated with differences in cooking time (Bassett, Kamfwa, et al., 2021; Bassett, Katuuramu, et al., 2021; Berry et al., 2020a; Cichy et al., 2015; Diaz et al., 2021; Garcia et al., 2012; Sadohara, Izquierdo, et al., 2022; Wahome et al., 2023). Several kompetitive allele-specific PCR (KASP) markers have already been developed for cooking time and traits 156 associated with cooking time, but remain untested on a broader set of germplasm (CGIAR, 2023). Thus, this study aimed to develop and validate a broader range of KASP markers for cooking time using available QTL data for the purpose of advancing breeding for this important trait. Several of these KASP markers were also confirmed to be in close physical proximity to candidate genes for cooking time, meaning that the results of this study could have further implications for understanding the genetic control of cooking time in dry beans. MATERIALS AND METHODS Germplasm The full list of the genotypes used in this study is available in the supplementary information (Table SI 4.1). Recombinant inbred lines (RILs) used in validation Cooking times from the TTRIL population plus TZ-27 and TZ-37 are an average across two planting locations and two years (Berry et al., 2020a). Cooking times from the YYRIL population plus PI527538 and Ervilha analysis are also an average across two years (Bassett, Katuuramu, et al., 2021). A total of ten QTL for cooking time and eight QTL for water uptake were delineated using the TTRIL population, whereas a total of fourteen QTL for cooking time and eight QTL for water uptake were delineated using the YYRIL population (Bassett, Katuuramu, et al., 2021; Berry et al., 2020). Twelve individuals from the TTRIL population and 16 individuals from the YYRIL population were used for this study. Individuals from the TTRIL and YYRIL populations were selected if their cooking times were on extreme ends or in the middle of their respective populations. Backcross population development and usage in validation The fast-cooking backcross (FCBC) population was developed by crossing a fast-cooking 157 brown bean, TZ-37, with either: 1) an elite fast-cooking yellow bean cultivar, Ervilha; 2) one of three fast-cooking, non-darkening, high-yielding lines from the YYRIL population (YYRIL049, YYRIL211, and YYRIL264); or 3) one slow-cooking, non-darkening line from the YYRIL population (YYRIL159). All crossings were performed in the Michigan State University (MSU) Research Greenhouses using sterile practices. The F1 seeds were planted and either allowed to self-pollinate for five generations or were reciprocally backcrossed to Ervilha to create BC1 plants. The BC1 seeds were planted and either allowed to self-pollinate for four generations or were reciprocally backcrossed to Ervilha to create BC2 plants. The BC2 plants were subsequently self-pollinated for four generations. A total of 47 individuals from the FCBC population (seven F6 plants, 14 BC1 plants, and 26 14 BC2 plants) were used for the validation study. All the individuals from the FCBC population including the parental lines were grown in the MSU Research Greenhouses from January to March 2024, except for TZ-37 and YYRIL211 which were grown from April to June 2024. Breeding lines used in validation Two hundred and fifty-nine breeding lines (F4 and F6 generation) from five market classes were grown at the Montcalm Research Farm and at the Saginaw Valley Research Farm in Frankenmuth, MI in 2022. The seeds were harvested and threshed using a Hege 140 plot combine and moisture equilibrated 4°C prior to phenotyping. Since more of the breeding lines were planted at Montcalm than at Saginaw Valley, cooking times from these two locations were analyzed separately rather than as an average. Cooking time data was not available for kidney beans planted at the Saginaw Valley Research Farm in 2022. Genotyping for cooking time Molecular marker data was collected from 59 cranberry bean breeding lines including 158 one non-darkening cranberry variety, 107 kidney bean breeding lines, 78 yellow bean breeding lines, 6 black bean breeding lines, and two pinto bean breeding lines including the slow- darkening variety USDA Diamondback, 14 individuals from the TTRIL population including TZ-27 and TZ-37, 18 individuals from the YYRIL population including PI527538 and Ervilha, and 47 individuals from the FCBC population (Miklas et al., 2024). Biological replicates of TZ- 27, TZ-37, PI527538, Ervilha, YYRIL049, YYRIL159, YYRIL211, YYRIL264, and the kidney breeding line Clouseau were included as checks for kompetitive allele-specific PCR (KASP) genotyping. Cooking time data was collected from 15 of the cranberry bean breeding lines, 97 of the kidney bean breeding lines, 39 of the yellow bean breeding lines, 14 individuals from the TTRIL population including TZ-27 and TZ-37, and 18 individuals from the YYRIL population including PI527538 and Ervilha. For the FCBC population, UV response data was collected from 35 individuals, and cooking time data was collected from 27 individuals including TZ-27, TZ-37, PI527538, Ervilha, YYRIL049, YYRIL159, YYRIL211, and YYRIL264. UV light response data was collected on the same set of individuals. However, results from the brown bean and dark red kidney genotypes were not included in the analysis, as dark seed coat colors can confound the results of the UV light test. Phenotyping for cooking time Cooking time was measured as previously described (N. Wang & Daun, 2005). Briefly, 30 seeds were soaked in 250 mL of distilled water for 12 hours. Twenty-five of the seeds were immediately placed onto groves at the base of a Mattson cooker, and equally weighted pins were placed on top of the seeds. As the seeds were cooked in boiling water, the pins of the Mattson cooker fell through the seed, triggering a custom Python application to automatically record the 159 amount of time required to cook the seed. For the FCBC population, the same procedure was followed, except that only 15 seeds were soaked and cooked since fewer seeds were available from the greenhouse. The cooking time of the seeds was recorded as the amount of time needed for 80% of the pins to fall through the seeds. The cooking times of all the genotypes were measured at least in duplicate. The raw seed weight of the FCBC lines was determined using 30 seeds. Their water uptake percentage was determined after soaking the same 30 seeds in distilled water for 12 hours. The formula used to determine the moisture contents of the seeds is as follows: [(soaked weight-raw weight)/raw weight]*100. Moisture content was determined by cutting five seeds into >1 mm pieces using a flat edge razor blade. The seed fragments were weighed, then dried at 105°C for 17 hours in accordance with the latest ISTA seed testing protocols (ISTA, 2023). Moisture content percentage was calculated using the following formula: [(raw weight-dry weight)/dry weight]*100. Sample weight was always measured to the hundredth decimal place using an analytical balance. Raw seed weight, water uptake percentage, and moisture content percentage were measured once per genotype. The UVC (UV light response) test was performed as previously described at the MSU Agronomy Farm (Junk-Knievel et al., 2007). Ten seeds were placed beneath UV lamps in plastic seed trays for 72 hours. Afterwards, the seeds were visually examined to determine whether their seed coats had darkened. Seed coat darkening was defined as the development of a dark brown pigment on the exposed side of the seed. Seeds that turned a faint yellow color were categorized as slow-darkening and were classified in the same way as non-darkening (non-pigmented) seeds for data analysis. Darkening genotypes were classified as ‘1’, whereas slow-darkening and non- darkening were classified as ‘0’. UV light response testing was performed in duplicate for all 160 genotypes. All UV response data is provided in the supplementary information (Table SI 4.1). Development of molecular markers for cooking time Whole genome sequencing Six trifoliate leaves were collected for whole genome sequencing from three plants at the V1 stage of development. The trifoliate leaves were pooled and lyophilized prior to DNA extraction. Genomic DNA was purified from the pooled samples using a NucleoSpin Plant II Midi kit. dsDNA concentration was checked using a Qubit dsDNA HS assay kit. Samples were diluted accordingly in ultra-pure water and submitted to the MSU Genomics Core for library preparation and sequencing. The samples were prepared for sequencing using an Illumina TruSeq Nano Library Prep Kit and Illumina Unique Dual Index adapters. The quality and concentration of the completed libraries were checked using a Qubit dsDNA HS assay and an Agilent 4200 TapeStation HS DNA1000 assay. These four libraries were pooled in equimolar proportions and quantified using an Invitrogen Collibri Quantification qPCR kit. The samples were sequenced on an Illumina NovaSeq6000 S4 flow cell in 2x150 bp paired end format using a NovaSeq v.1.5 300 cycle reagent kit. Base calling was performed using Illumina Real Time Analysis (RTA) v.3.4.4, and the output was demultiplexed and converted to fastq format with Illumina Bclfastq v.2.20.0. All protocols used followed the recommendations of their respective manufacturers. It was estimated that 73-109X coverage of each genome was obtained from sequencing. The average Q-score of the paired-end reads ranged from 35.5-36.0, and the percentage of paired-end reads with a Q-score greater than 30 was 91.9-94.0%. Joint genotyping with GATK to identify genomic variants for molecular marker development Variants were called in the genomes of four common bean genotypes, TZ-27, TZ-37, PI527538, and Ervilha, based on GATK best practices for joint genotyping (Poplin et al., 2017). 161 Sequence read quality was initially checked using FastQC (v. 0.11.7), and CutAdapt (v. 1.14) was used to remove poor quality base pairs from the 3’ and 5’ ends (bases with phred scores ≥ 20 were kept), adapter sequences, and truncated reads (Anders et al., 2015; Martin, 2011). After cleaning, the data quality was checked again using FastQC. BWA-MEM (v. 0.7.17) was used to align the cleaned reads to the P. vulgaris L. genome (v. 2.1) (Li, 2013; Schmutz et al., 2014). Duplicates were marked using MarkDuplicatesSpark, a tool that combines the MarkDuplicates and SortSam steps of the GATK best practices single sample pipeline into a single function (Van der Auwera & O’Connor, 2020). GATK HaplotypeCaller was used to convert input bam files into GVCF formatted files (Poplin et al., 2018). For this step, the minimum base quality score was set to 15. These files were merged into a multi-sample GVCF file using the CombineGVCFs function. The multi-sample file was then subjected to joint genotyping using the GenotypeGVCFs function. The VariantFiltering function was used to remove variants with a phred score lower than 30. This file was utilized to identify target SNPs/indels for use in marker- assisted selection. Selection of genetic variants associated with cooking time Genetic variants were selected based on a previous association with quantitative trait loci (QTL) for cooking time, water uptake, or slow darkening. The locations of the QTL had been mapped using the P. vulgaris v.2.1 reference genome (Schmutz et al., 2014). Ten of the QTL used in this study were delineated using a brown bean recombinant inbred line population consisting of 161 individuals. The population, called TTRIL, was created by crossing a fast- cooking brown bean (TZ-37) from with a slow-cooking brown bean (TZ-27) (Berry et al., 2020). Eleven of the QTL were delineated using a yellow bean recombinant inbred line population consisting of 284 individuals. This population, called YYRIL, was created by crossing a fast- 162 cooking yellow bean (Ervilha) with a slow-cooking yellow bean (PI527538) (Bassett, Katuuramu, et al., 2021). These QTL were chosen for marker-assisted selection because they had a high R2 value for cooking time. Several candidate genes within the TTRIL and YYRIL QTL were selected as targets for the validation study based on RNA-seq data confirming that they were differentially expressed in fast- and slow-cooking dry beans (Jeffery et al., 2023). These genes included a calcium-binding EF-hand family protein, cell wall modification genes like xyloglucan endotransglycosylase/hydrolases, exordium like 2, highly ABA-induced PP2C gene 3, DEX1, SKU similar 5 (a pectinesterase-type protein), HXXXD-type acyl-transferase family protein (also known as phenolic glucoside malonyltransferase 1), and expansin-like A1, and stress responsive genes like highly ABA-induced PP2C gene 3, ARM repeat superfamily protein, and Hsp20-like chaperone superfamily protein. A more detailed description of these four genotypes and the functionalities of these genes is available (Jeffery et al., 2023). A gene tightly associated with the non-darkening trait (Phvul.010G130600) and PAE8 (Phvul.003G277600) were also selected for this study based on their theoretical association with cooking time (Table SI 4.2). Genetic variants that were selected as potential KASP markers were located within 1 Mbp of the beginning or end of these QTL. An attempt was made to select variants located near the beginning, middle, and end of each of the QTL for cooking time. Candidates for KASP markers were discarded if: 1) they did not distinguish between one the pairs of fast- and slow- cooking genotypes, or 2) the SNP was heterozygous. When available, KASP markers from the Intertek catalog were used in place of custom markers (CGIAR, 2023). A full description of the KASP markers that were used in and/or developed for this study (i.e., source, sequence information, intended uses, etc.) is available in the supplementary information (Table SI 4.3). 163 KASP marker design The selected genetic variants were designed in accordance with the KASP marker design guidelines from Intertek, wherein SNPs/indels were flanked on either side by at least 50 base pairs (Alnarp, Sweden). Seven KASP markers (snpPV00255, snpPV00258, snpPV00250, snpPV00251, snpPV00257, snpPV00253, and snpPV00252) were removed from further analysis because they did not receive at least a “good” validation rating from Intertek. Four other KASP markers (snpPV00256, snpPV00256_alt1, snpPV00256_alt2, and snpPV00256_alt3) were discarded because their GC content was too low. Genotyping Genotyping was performed on all 331 genotypes over the course of three days. Tissue sampling in accordance with the general leaf sampling guidelines from Intertek (Alnarp, Sweden). Briefly, two seeds were planted into a single pot and allowed to grow to the first trifoliate stage. Leaf tissue was collected from each plant using a metal hole puncher, followed by thorough cleaning of the hole puncher using 70% ethanol and KimWipes. Two leaf disks per genotype were placed in AB-gene 96-well plates (AB056). The tissue was immediately lyophilized, sealed with a silicone cap (AB0674), and shipped to Intertek, Sweden for KASP marker validation and sequencing on an ABI QuantStudio 7K platform. A total of 12 KASP markers were used in the final genotyping experiment. snpPV00254 was not included because preliminary tests showed that it was not significantly correlated with cooking time in the YYRIL population (data not shown). Some genetic variants were scarce in some or all the populations tested. If less than two bean lines in a population had a certain polymorphism, then they were removed from the data set prior to statistical testing. For statistical analysis, one of the homozygotes was classified as “0”, the heterozygote was classified as “0.5”, 164 and the alternate homozygote was classified as “1”. The favorable homozygote was selected based on whether it was previously found to be associated with fast cooking time in genotypes with contrasting cooking times (Table SI 4.2). Hence, for snpPV00001, snpPV00038, snpPV00169, snpPV00218, and snpPV00259, the favorable homozygote was classified as “0”. All molecular marker data is provided as supplementary information (Table SI 4.1). Statistical analysis The effects of the phenotypic markers (UV light response, market class) and the molecular markers (i.e., SNPs) on cooking time were quantified using Pearson’s correlation coefficients and multiple comparisons testing in R (v. 4.0.3) (RStudio Team, 2024). In addition, correlations between cooking time, raw seed weight, water uptake percentage, and moisture content percentage in the FCBC were measured. Correlation coefficients and their significance values were calculated and plotted using the PerformanceAnalytics package (v. 2.0.4) (Peterson et al., 2020). The significance of a phenotypic or molecular marker was determined using linear modeling. Prior to multiple testing, it was determined whether the data was normally distributed and whether the residuals were equal. If the data failed to meet these assumptions, a logarithmic or an inverse transformation was applied. Since the number of genotypes from a specific phenotypic or allelic class (i.e., the number of ‘A:A’ individuals versus the number of ‘G:G’ individuals) was never exactly equal, the joint_tests function from the emmeans package (v. 1.7.4-1) was then used to test for significant differences (Lenth, 2024). Multiple comparisons testing was then performed using the ‘sidak’ test in the multcomp package (v. 1.4-25) (Hothorn et al., 2024). If the data still failed to meet the assumptions of normality and/or unequal variances 165 following a transformation, then the version of the data that best met the assumption of normality was instead subjected to non-parametric testing. Firstly, the kruskal.test was used to test for significant differences. Secondly, the wilcox.test was used to perform multiple comparisons testing on data consisting of only two groups (i.e., ‘A:A’ and ‘G:G’). However, for comparisons consisting of more than two groups (i.e., ‘A:A’, ‘A:G’, and ‘G:G’), the wilcox.test was substituted for the dunnTest from the FSA package (v. 0.9.5) (Ogle et al., 2023). All significant differences were visualized using the ggplot2 package (v. 3.4.3) (Wickham, 2016). Characteristics of the FCBC population RESULTS In contrast field grown seeds from PI527538 and Ervilha, greenhouse grown seeds from the fast-cooking yellow bean check Ervilha took the exact same amount of time to cook as the slow-cooking yellow bean check line, PI527538 (23.5 minutes). In addition, greenhouse grown YYRIL159 did not have a significantly slower cooking time than the other YYRIL check lines, contrary to reports from studies that grew the YYRILs in the field (Bassett, Katuuramu, et al., 2021). On the other hand, the second pair of check lines, TZ-27 and TZ-37, exhibited expectedly different cooking times. These differences may be the result of 1) insufficient watering, and 2) unequal temperature distributions throughout the greenhouse, as it was noted that certain pots dehydrated quickly, whereas others remained moist for extended periods of time. However, the distribution of cooking times skewed left, indicating that some of the seeds exhibited fast cooking characteristics (Fig. SI 4.1a). The fast-cooking backcross (FCBC) population consists of 40 single cross, single backcross, and double backcross lines. These individuals exhibited broad diversity for cooking time. The fastest cooking accession was the single backcross line Erv×(37×159)_4 (15.5 166 minutes) and the slowest cooking accession was the double backcross line Erv×(Erv×(37×159)_3)_3 (39.8 minutes) (Fig. 4.1). Figure 4.1: Average cooking times of dry beans. Means with the same letter are not significantly different from each other (Sidak, α=0.05). In terms of physical characteristics, the seeds exhibited a relatively normal distribution for raw seed weight and water uptake percentage (Fig. SI 4.1a). The seed coat colors of the accessions ranged from bright yellow to dark brown (data not shown). The fastest cooking accession, Erv×(37×159)_4, had a dark brown seed coat that appeared to change color from brown to a light yellow during cooking. The slowest cooking accession had a yellow seed coat both before and after cooking. While all their parental lines except PI527538 did not darken in response to UV light, about half of the lines tested darkened in response to UV light. The accessions were also polymorphic for genes and QTL related to cooking time, as 167 evidenced by snpPV00038, snpPV00170, snpPV00218, and snpPV00259. These SNPs were near or within two candidate genes for cooking time, R2R3-MYB-type transcription factor MtPAR (Phvul.010G130600) and pectin acetylesterase 8 (Phvul.003G277600), as well as several QTL associated with cooking time in the brown bean parent, TZ-37 [CT.6.1 (Ar_17, Mo_17), WU.6.3 (Mo_17), and CT.11.1 (Ar_16)]. Failure rates of the KASP markers Twelve KASP markers were tested on five dry bean market classes from Andean and Mesoamerican backgrounds a total of 344 times. In the Andean lines, the markers had an 89- 98% success rate (Table 4.1). Although only eight Mesoamerican lines were genotyped (6 blacks and 2 pintos), all the markers were 100% successful except for snpPV00201 which failed 25% of the time. The markers were more successful in the kidney breeding lines than in the cranberry and yellow breeding lines, with snpPV00038 and snpPV00169 having the lowest success rate (~84%) in the yellow and cranberry breeding lines, respectively. Failures were generally the result of unknown errors, except for snpPV00038, wherein most of the failures were classified as “uncallable”. 168 Table 4.1: Call rates of KASP markers in Andean dry bean lines. Genotyping was conducted a total of 335 times on 331 Andean dry bean genotypes from the brown, yellow, cranberry, and kidney market classes. Marker snpPV 00001 snpPV 00038 snpPV 00070 snpPV 00169 snpPV 00170 snpPV 00183 snpPV 00201 snpPV 00207 snpPV 00218 snpPV 00225 snpPV 00237 snpPV 00259 Homozygote 1 Heterozygote 4 2 109 321 33 Homozygote 2 320 155 Uncallable 0 Unknown error 9 Success rate (%) 97 28 10 89 56 0 30 19 256 276 6 17 93 1 9 97 324 317 325 277 0 0 1 10 97 1 3 0 14 96 0 1 0 9 97 48 0 0 10 97 19 2 304 1 9 315 227 1 11 0 8 7 91 1 9 97 98 97 0 0 1 13 96 Error rate (%) 3 11 4 7 3 3 4 3 3 3 2 3 167 Even though many of the individuals in this study have been inbred for several generations, snpPV00218 produced 48 heterozygous calls and 0 alternate homozygous calls. The reason for this problem is unclear and requires further investigation to understand. The presence of the G allele in the heterozygous calls will instead remain the focus of this work. Correlations with cooking time Water uptake was significantly negatively correlated with cooking time in the FCBC population, meaning that cooking time generally decreased as the amount of water taken up by the beans during soaking increased. Cooking time in the FCBC population was neither significantly correlated with raw seed weight, nor with raw seed moisture content, although raw seed weight was significantly correlated with water uptake. Significant correlations were found between cooking time and at least one genetic marker in all populations/collections tested except the yellow bean breeding line collection from Montcalm and the kidney bean breeding line collection from Saginaw Valley (Figs. SI 4.1a-h). Response to UV light was significantly correlated with cooking time in the FCBC population, the YYRIL population, and the kidney breeding lines from Montcalm (Figs. SI 4.1a-b, f). UV light response was not significantly correlated with cooking time in the kidney breeding line collection harvested in Saginaw Valley which consisted of fewer individuals. snpPV00038 and snpPV00169 were significantly correlated with cooking time in the YYRIL population and in the cranberry breeding line collections from both Montcalm and Saginaw Valley (Figs. SI 4.1d-e). The G:G variant of snpPV00038 had the opposite effect on cooking time that was predicted by PI527538 and Ervilha, in that the G:G variant appeared to be associated with the non-darkening phenotype rather than the darkening phenotype in the cranberry bean lines (Table SI 4.3). The G:G variant was, on the other hand, associated with the 168 non-darkening trait in the pinto beans. Interestingly, snpPV00038 and snpPV00169 were perfectly correlated with each other in the cranberry breeding line collections, indicating that these SNPs are tightly linked. Of note, a candidate gene for the non-darkening trait (Phvul.010G130600) is located between two snpPV00038 and snpPV00169. The relationship between these markers and response to UV light could not be determined in the cranberry lines because every cranberry genotype darkened under UV light. snpPV00218 was only correlated with cooking time in the TTRIL population, and snpPV00259 was correlated with cooking time in the TTRIL population and the yellow breeding lines from Saginaw Valley (Figs. SI 4.1c, h). The T:T variant of snpPV00259 the opposite effect that was predicted by TZ-27, TZ-37, PI527538, and Ervilha, in that the T:T variant appeared to be associated with longer rather than shorter cooking times (Table SI 4.3). By contrast, T:T was associated with shorter cooking times in the TTRIL population. snpPV00259 was not associated with water uptake percentage in the FCBC lines (r2=-0.05, p=0.73). Despite there being sufficient heterogeneity in the population for correlation analysis, cooking time in the FCBC was not correlated with snpPV00038, snpPV00170, snpPV00218, or snpPV00259. In the YYRIL population, cooking time was not significantly correlated with snpPV00259. snpPV00225 was not correlated with the cooking times of cranberry beans from either Montcalm or Saginaw Valley. The cooking times of kidney beans from Montcalm were not correlated with snpPV00038, snpPV00169, or snpPV00170. The cooking times of the yellow bean breeding lines were not significantly correlated with response to UV light or snpPV00038 in either growing environment. However, while snpPV00259 was not significantly correlated with the cooking times of the yellow bean breeding lines from Montcalm, they were correlated with the cooking times of yellow breeding lines from Saginaw Valley. 169 Predicted effects of phenotypic and molecular markers on cooking time Whereas the non-darkening trait did have a significant effect on the average cooking times of the FCBC and YYRIL populations, it did not significantly affect the average cooking times of the kidney bean breeding lines (Fig. 4.2). The non-darkening individuals in the FCBC and YYRIL populations cooked 4.4 (p=0.015) and 3.8 (p=0.038) minutes faster than the darkening individuals, respectively. Figure 4.2: Effect of UV response on the cooking times of the fast-cooking backcross population (FCBC) (left) and the yellow bean recombinant inbred line population (YYRIL) (right). Means with the same letter are not significantly different from each other (Sidak, α=0.05). snpPV00038 and snpPV00169 were significantly correlated with cooking time in the YYRIL population and the cranberry bean lines from both Montcalm and Saginaw Valley. However, these SNPs only had a significant effect on the average cooking times of the cranberry beans from Montcalm (Fig. 4.3). The two SNPs were associated with a 49.3-minute average decrease in cooking time (p=0.013). 170 Figure 4.3: Effects of snpPV00038 (top) and snpPV00169 (bottom) on cooking time in the cranberry breeding lines from Montcalm (MC) (left) and Saginaw Valley (SV) (right). Means from the YYRIL population with the same letter are not significantly different from each other (Dunn’s non-parametric test, α=0.05). Means from the cranberry bean line collection with the same letter are not significantly different from each other (Wilcoxon non-parametric test, α=0.05). snpPV00218 and snpPV00259 had significant effects on the average cooking times of the TTRIL population (p=0.025 and p=0.048, respectively) (Figs. 4.4-4.5). Individuals possessing a copy of the G allele in snpPV00218 took 10.1 fewer minutes to cook compared to individuals that contained no copies of the G allele. As for snpPV00259, individuals possessing two copies of the T allele cooked 9.8 minutes faster than individuals with two copies of the C allele. 171 Figure 4.4: Effects of snpPV00218 on the cooking times of the TTRIL population (Berry et al., 2020). Means with the same letter are not significantly different from each other (Wilcoxon non- parametric test, α=0.05). Figure 4.5: Effects of snpPV00259 on the cooking times of the TTRIL population and yellow bean breeding lines. Means with the same letter are not significantly different from each other (Sidak, α=0.05). 172 snpPV00259 also had a significant effect on the average cooking times of the yellow bean breeding lines from Saginaw Valley (p=0.045). In contrast to the TTRIL population, though, yellow breeding lines from Saginaw Valley possessing two copies of the T allele took 5 minutes longer to cook on average than individuals possessing two copies of the C allele. Yellow bean breeding lines that were heterozygous for the C and T allele had cooking times that were intermediate to those of the homozygous individuals. DISCUSSION The future potential of the FCBC population Some of the individuals in the FCBC population have good market class characteristics such as an attractive yellow seed coat. However, their cooking time data was affected by the fact that: 1) the plants were grown in a greenhouse rather than in a field environment; and 2) fewer seeds were used for the cooking time experiment compared to the normal procedure due to limited yield. For example, the two yellow bean checks, PI527538 and Ervilha, had identical cooking times, even though they have repeatedly been shown to have statistically different cooking times when grown in a field setting (Bassett, Hooper, et al., 2021; Bassett, Katuuramu, et al., 2021; Jeffery et al., 2023). On the other hand, the second pair of check lines, TZ-27 and TZ-37, had significantly different cooking times. This suggests that the fastest cooking accessions did, indeed, have significantly faster cooking times than the slowest cooking accessions in the population. More research is needed, though, to identify accessions in the FCBC population that have consistently fast cooking times across environments. Non-darkening was the most consistent predictor of fast cooking time Procyanidin accumulation in dry beans is hypothesized to be related to cooking time because these compounds are able to bind to soluble substances in the cotyledon cells and the 173 cell wall matrix, rendering the seed less soluble during cooking (Wiesinger et al., 2021). Since procyanidin accumulation is easily induced by incubating the seeds under ultraviolet (UV) light, the relationship between seed coat darkening and cooking time was tested in multiple dry bean populations with different genetic backgrounds. Indeed, cooking time was significantly correlated with response to UV light across three different populations, including the FCBC, the YYRILs, and the kidney bean breeding lines from Montcalm. However, the effect of seed coat darkening was only statistically significant in the FCBC and YYRIL populations. One possible explanation for this is that the darker seed coat colors of some kidney bean varieties make it more difficult to visually assess their responses to UV light. The use of more objective methods such as colorimetry or machine learning, which has been shown to accurately measure the seed coat colors of yellow beans, could help to ensure the accuracy of UV response data going forward, especially for seeds with darker seed coat colors (Sadohara, Long, et al., 2022). snpPV00038 and snpPV00169 also show promise as markers for non-darkening since they are tightly linked and surround a candidate gene for the non-darkening trait. These markers do not appear to be strongly associated with the response to UV light, as evidenced by the lack of a correlation between UV response and these SNPs in the YYRIL population. Hence, utilizing the response to UV light and these SNPs to predict cooking time in dry beans in a breeding program should be considered separately. On their own, they were able to distinguish fast and slow cooking cranberry breeding lines. Their KASP markers also had a decent success rate across a variety of market classes (~84% at the lowest). However, they sometimes failed to distinguish fast- from slow-cooking genotypes in the cranberry breeding line collection. While this was the result of differences in cooking times between growing locations, it is concerning that wide variation for cooking time was seen for the favorable allele, ACC:ACC, in both 174 locations, suggesting that a marker that is more tightly linked to the non-darkening trait would be needed to predict cooking time with high accuracy. Also concerning is that the ACC:ACC was deemed unfavorable to dry beans by previous researchers, as the alternative allele has been associated with resistance to common bacterial blight (CGIAR, 2023). As procyanidins are defensive compounds in dry beans, this association could be indicative of a potential trade-off between plant hardiness and cooking time (Islam et al., 2003). In the future, breeding programs will have to weigh the need to use these markers with the need for accuracy, as well as the need for resistance to disease. snpPV00218 may be associated with genes related to cell wall modification snpPV00218 was significantly associated with differences in cooking time in the TTRIL population. snpPV00218 was selected as a marker for this study in part because it is located less than 1 Mbp away from two candidate genes from cooking time, SKU5 similar 5 (Phvul.011G102500) and expansin-like A1 (Phvul.011G024801) that were differentially upregulated in the slow-cooking brown bean TZ-27 relative to the fast-cooking brown bean TZ- 37. These genes were also located within a QTL (CT.11.1 Ar_16) for cooking time that was mapped using the TTRIL population. Taken together, the association between this marker and cooking time in the TTRIL population could mean that genes in this region of the genome are associated with cooking time. Yet, given that many genes could be linked to snpPV00218, and given that a significant association was not found between snpPV00218 and cooking time in the FCBC population, more research will be needed to understand which gene(s) in the proximity of snpPV00218 may be influencing cooking time. 175 snpPV00259 alone did not explain differences in cooking time across different dry bean populations snpPV00259 is a SNP that was found within the pectin acetylesterase 8 gene in TZ-27, TZ-37, PI527538, and Ervilha (Table SI 4.3). As it was observed that the slow-cooking genotypes (TZ-27 and PI527538) possessed the C:C variant of snpPV000259, whereas the fast- cooking genotypes (TZ-37 and Ervilha) possessed the T:T variant, it was hypothesized that the T allele may contribute to faster cooking times by increasing water uptake in TZ-37 and Ervilha. This effect was observed in the TTRIL population, but the opposite effect was observed in the yellow bean breeding lines. While snpPV00259 could affect water uptake if it has a significant impact on the functionality of pectin acetylesterase 8, it alone cannot explain both an increase and a decrease in cooking time in different populations. Given that snpPV00259 was significantly associated with cooking time in both populations, though, it is possible that snpPV00259 is physically linked to another SNP in a regulatory region like a nearby promotor or a repressor in one of these populations, but not in the other. CONCLUSIONS The goal of this study was to accelerate the development of faster cooking dry bean germplasm for the benefit of consumers around the globe. The outcomes of this study are a new population of dry bean varieties developed from two fast-cooking landraces from Africa (called the ‘fast-cooking backcross’ (FCBC) population), and the identification of phenotypic and molecular markers that could be used to accurately and cheaply detect fast cooking times across a variety of market classes. Twelve KASP markers (snpPV00001, snpPV00038, snpPV00070, snpPV00169, snpPV00170, snpPV00183, snpPV00201, snpPV00207, snpPV00218, snpPV00225, snpPV00237, and snpPV00259) were tested on dry beans from both Andean and 176 Mesoamerican market classes. All twelve markers performed well, with the maximum error rate for a single marker being 11% and the average error rate for all twelve markers being 4%. Phenotypic and molecular markers were associated with cooking time in six different dry bean populations/collections, including the FCBC, two recombinant inbred line populations (TTRIL and YYRIL), and three breeding line collections. The phenotypic marker, response to UV light, was the most reliable of the tests, as it was successfully able to distinguish fast and slow cooking genotypes in the FCBC, the YYRIL population, and a kidney breeding line collection. snpPV00038 and snpPV00169 are physically linked to a candidate gene for seed coat darkening and were able to predict the cooking times of cranberry breeding lines. snpPV00038 and snpPV00169 also had the strongest effect average effect on cooking time, although the significance of the association depended on the growing location of the beans. While response to UV light, snpPV00038, and snpPV00169 are associated with seed coat darkening, only snpPV00038 and snpPV0069 were able to distinguish genotypes with faster cooking times in a cranberry bean line collection. snpPV00218 is physically linked to candidate genes for cooking time that may participate in cell wall modifications in dry beans. snpPV00259 is located within pectin acetylesterase 8, a candidate gene for cooking time that increases water uptake in dry beans. The results of this study suggest that snpPV00259 is linked to a gene that affects cooking time, but it was unclear whether snpPV00259 influenced cooking time. 177 BIBLIOGRAPHY Anders, S., Pyl, P. T., & Huber, W. (2015). HTSeq—A Python framework to work with high- throughput sequencing data. Bioinformatics, 31(2), 166–169. https://doi.org/10.1093/bioinformatics/btu638 Asiimwe, R., Katungi, E., Marimo, P., Mukankusi, C., Rubyogo, J. C., & Anthony, V. (2024). Evaluating consumer preferences for reduced cooking time, taste and colour of beans in rural and urban communities in Uganda. Agriculture & Food Security, 13(1), 19. https://doi.org/10.1186/s40066-024-00466-4 Bassett, A., Hooper, S., & Cichy, K. (2021). Genetic variability of cooking time in dry beans (Phaseolus vulgaris L.) related to seed coat thickness and the cotyledon cell wall. Food Research International, 141, 109886. https://doi.org/10.1016/j.foodres.2020.109886 Bassett, A., Kamfwa, K., Ambachew, D., & Cichy, K. (2021). Genetic variability and genome- wide association analysis of flavor and texture in cooked beans (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 134(3), 959–978. https://doi.org/10.1007/s00122-020- 03745-3 Bassett, A., Katuuramu, D. N., Song, Q., & Cichy, K. (2021). QTL mapping of seed quality traits including cooking time, flavor, and texture in a yellow dry bean (Phaseolus vulgaris L.) population. Frontiers in Plant Science, 12, 670284. https://doi.org/10.3389/fpls.2021.670284 Beninger, C. W., Gu, L., Prior, R. L., Junk, D. C., Vandenberg, A., & Bett, K. E. (2005). Changes in polyphenols of the seed coat during the after-darkening process in pinto beans (Phaseolus vulgaris L.). Journal of Agricultural and Food Chemistry, 53(20), 7777– 7782. https://doi.org/10.1021/jf050051l Bento, J. A. C., Bassinello, P. Z., Cruz, Q. A. D., Mendonça, M. A. D. S., Borba, T. C. D. O., Vanier, N. L., Souza Neto, M. A., Ferreira, K. C., Silva, G. D. L. P. E., & Oliveira, A. L. M. D. (2020). The convenience of non-conventional methods for evaluation of the culinary quality of beans. Research, Society and Development, 9(11), e44491110103. https://doi.org/10.33448/rsd-v9i11.10103 Berry, M., Izquierdo, P., Jeffery, H., Shaw, S., Nchimbi-Msolla, S., & Cichy, K. (2020a). QTL analysis of cooking time and quality traits in dry bean (Phaseolus vulgaris L.). TAG. Theoretical and Applied Genetics. Theoretische Und Angewandte Genetik, 133(7), 2291– 2305. https://doi.org/10.1007/s00122-020-03598-w Berry, M., Izquierdo, P., Jeffery, H., Shaw, S., Nchimbi-Msolla, S., & Cichy, K. A. (2020b). QTL analysis of cooking time and quality traits in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 133(7), 2291–2305. https://doi.org/10.1007/s00122- 020-03598-w Boy, E., Bruce, N., Smith, K. R., & Hernandez, R. (2000). Fuel efficiency of an improved wood- burning stove in rural Guatemala: Implications for health, environment and development. 178 Energy for Sustainable Development, 4(2), 23–31. https://doi.org/10.1016/S0973- 0826(08)60239-2 CGIAR. (2023, September 3). KASP low density genotyping platform. Genetic Innovation Toolbox. https://excellenceinbreeding.org/toolbox/tools/kasp-low-density-genotyping- platform Chigwedere, C. M., Nkonkola, C. M., Rai, S., Kyomugasho, C., Kermani, Z. J., Pallares Pallares, A., Van Loey, A. M., Grauwet, T., & Hendrickx, M. E. (2019). Cotyledon pectin molecular interconversions explain pectin solubilization during cooking of common beans (Phaseolus vulgaris). Food Research International, 116, 462–470. https://doi.org/10.1016/j.foodres.2018.08.062 Cichy, K. A., Wiesinger, J. A., Berry, M., Nchimbi-Msolla, S., Fourie, D., Porch, T. G., Ambechew, D., & Miklas, P. N. (2019). The role of genotype and production environment in determining the cooking time of dry beans (Phaseolus vulgaris L.). Legume Science, 1(1), e13. https://doi.org/10.1002/leg3.13 Cichy, K. A., Wiesinger, J. A., & Mendoza, F. A. (2015). Genetic diversity and genome-wide association analysis of cooking time in dry bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics, 128(8), 1555–1567. https://doi.org/10.1007/s00122-015-2531-z Diaz, S., Ariza-Suarez, D., Ramdeen, R., Aparicio, J., Arunachalam, N., Hernandez, C., Diaz, H., Ruiz, H., Piepho, H.-P., & Raatz, B. (2021). Genetic architecture and genomic prediction of cooking time in common bean (Phaseolus vulgaris L.). Frontiers in Plant Science, 11, 622213. https://doi.org/10.3389/fpls.2020.622213 Erfatpour, M., & Pauls, K. P. (2020). A R2R3-MYB gene-based marker for the non-darkening seed coat trait in pinto and cranberry beans (Phaseolus vulgaris L.) derived from ‘Wit- rood boontje.’ Theoretical and Applied Genetics, 133(6), 1977–1994. https://doi.org/10.1007/s00122-020-03571-7 Garcia, R. A. V., Rangel, P. N., Bassinello, P. Z., Brondani, C., Melo, L. C., Sibov, S. T., & Vianello-Brondani, R. P. (2012). QTL mapping for the cooking time of common beans. Euphytica, 186(3), 779–792. https://doi.org/10.1007/s10681-011-0587-7 Havemeier, S. M., & Slavin, J. (2020). Pulses and legumes: Nutritional opportunities and challenges. Cereal Foods World, 65(2). https://doi.org/10.1094/CFW-65-2-0021 Hothorn, T., Bretz, F., Westfall, P., Heiberger, R. M., Schuetzenmeister, A., & Scheibe, S. (2024). multcomp: Simultaneous inference in general parametric models (Version 1.4- 26) [Computer software]. https://cran.r-project.org/web/packages/multcomp/index.html Islam, F. M. A., Rengifo, J., Redden, R. J., Basford, K. E., & Beebe, S. E. (2003). Association between seed coat polyphenolics (tannins) and disease resistance in common bean. Plant Foods for Human Nutrition, 58(4), 285–297. https://doi.org/10.1023/B:QUAL.0000040283.51023.c2 179 ISTA. (2023). International rules for seed testing. International Seed Testing Association. Jarvis, M. C. (1992). Control of thickness of collenchyma cell walls by pectins. Planta, 187(2), 218–220. https://doi.org/10.1007/BF00201941 Jeffery, H. R., Mudukuti, N., Buell, C. R., Childs, K. L., & Cichy, K. (2023). Gene expression profiling of soaked dry beans (Phaseolus vulgaris L.) reveals cell wall modification plays a role in cooking time. The Plant Genome, 16(3), e20364. https://doi.org/10.1002/tpg2.20364 Junk-Knievel, D. C., Vandenberg, A., & Bett, K. E. (2007). An accelerated postharvest seed-coat darkening protocol for pinto beans grown across different environments. Crop Science, 47(2), 694–700. https://doi.org/10.2135/cropsci2006.05.0325 Katuuramu, D. N., Luyima, G. B., Nkalubo, S. T., Wiesinger, J. A., Kelly, J. D., & Cichy, K. A. (2020). On-farm multi-location evaluation of genotype by environment interactions for seed yield and cooking time in common bean. Scientific Reports, 10(1), 3628. https://doi.org/10.1038/s41598-020-60087-2 Lenth, R. V. (2024). Emmeans: Estimated marginal means, a.k.a. Least-squares means (Version 1.10.2) [Computer software]. University of Iowa. 10.32614/CRAN.package.emmeans Li, H. (2013). Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM (arXiv:1303.3997). arXiv. https://doi.org/10.48550/arXiv.1303.3997 Mafuleka, M. M., Ott, D. B., Hosfield, G. L., & Uebersax, M. A. (1993). The role of phytase and lignin decorticated dry bean (Phaseolus vulgaris) hardening during storage. Journal of Food Processing and Preservation, 17(1), 1–20. https://doi.org/10.1111/j.1745- 4549.1993.tb00222.x Martin, M. (2011). Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.Journal, 17(1), 10. https://doi.org/10.14806/ej.17.1.200 Martínez-Manrique, E., Jacinto-Hernández, C., Garza-García, R., Campos, A., Moreno, E., & Bernal-Lugo, I. (2011). Enzymatic changes in pectic polysaccharides related to the beneficial effect of soaking on bean cooking time. Journal of the Science of Food and Agriculture, 91(13), 2394–2398. https://doi.org/10.1002/jsfa.4474 Miklas, P. N., Soler-Garzón, A., Pastor-Corrales, M., & Cichy, K. A. (2024). Registration of ‘USDA Diamondback’ slow-darkening pinto bean. Journal of Plant Registrations, 18(1), 52–60. https://doi.org/10.1002/plr2.20334 Mudryj, A. N., Yu, N., & Aukema, H. M. (2014). Nutritional and health benefits of pulses. Applied Physiology, Nutrition, and Metabolism, 39(11), 1197–1204. https://doi.org/10.1139/apnm-2013-0557 Njoroge, D. M., Kinyanjui, P. K., Christiaens, S., Shpigelman, A., Makokha, Anselimo. O., Sila, D. N., & Hendrickx, M. E. (2015). Effect of storage conditions on pectic polysaccharides 180 in common beans (Phaseolus vulgaris) in relation to the hard-to-cook defect. Food Research International, 76, 105–113. https://doi.org/10.1016/j.foodres.2015.03.005 Ogle, D. H., Doll, J. C., Wheeler, A. P., & dunnTest()), A. D. (Provided base functionality of. (2023). FSA: Simple fisheries stock assessment methods (Version 0.9.5) [Computer software]. https://cran.r-project.org/web/packages/FSA/index.html Palmer, J. P., Pajak, A., Robson, B., Zhang, B., Joshi, J., Diapari, M., Pauls, K. P., & Marsolais, F. (2022). Pectin acetylesterase 8 influences pectin acetylation in the seed coat, seed imbibition, and dormancy in common bean (Phaseolus vulgaris L.). Legume Science, 4(3), e130. https://doi.org/10.1002/leg3.130 Pelloux, J., Rustérucci, C., & Mellerowicz, E. J. (2007). New insights into pectin methylesterase structure and function. Trends in Plant Science, 12(6), 267–277. https://doi.org/10.1016/j.tplants.2007.04.001 Peterson, B. G., Carl, P., Boudt, K., Bennett, R., Ulrich, J., Zivot, E., Cornilly, D., Hung, E., Lestel, M., Balkissoon, K., Wuertz, D., Christidis, A. A., Martin, R. D., Zhou, Z. “Zenith,” & Shea, J. M. (2020). PerformanceAnalytics: Econometric Tools for Performance and Risk Analysis (Version 2.0.4) [Computer software]. https://cran.r- project.org/web/packages/PerformanceAnalytics/index.html Poplin, R., Ruano-Rubio, V., DePristo, M. A., Fennell, T. J., Carneiro, M. O., Auwera, G. A. V. der, Kling, D. E., Gauthier, L. D., Levy-Moonshine, A., Roazen, D., Shakir, K., Thibault, J., Chandran, S., Whelan, C., Lek, M., Gabriel, S., Daly, M. J., Neale, B., MacArthur, D. G., & Banks, E. (2018). Scaling accurate genetic variant discovery to tens of thousands of samples (p. 201178). bioRxiv. https://doi.org/10.1101/201178 Rahkovsky, I., Jo, Y., & Carlson, A. (2018). Consumers balance time and money in purchasing convenience foods. USDA-ERS. https://doi.org/10.22004/ag.econ.276227 RStudio Team. (2024). RStudio: Integrated development for R [RStudio]. PBC. http://www.rstudio.com/ Sadohara, R., Izquierdo, P., Couto Alves, F., Porch, T., Beaver, J., Urrea, C. A., & Cichy, K. (2022). The Phaseolus vulgaris L. Yellow Bean Collection: Genetic diversity and characterization for cooking time. Genetic Resources and Crop Evolution, 69(4), 1627– 1648. https://doi.org/10.1007/s10722-021-01323-0 Sadohara, R., Long, Y., Izquierdo, P., Urrea, C. A., Morris, D., & Cichy, K. (2022). Seed coat color genetics and genotype × environment effects in yellow beans via machine-learning and genome-wide association. The Plant Genome, 15(1), e20173. https://doi.org/10.1002/tpg2.20173 Schmutz, J., McClean, P. E., Mamidi, S., Wu, G. A., Cannon, S. B., Grimwood, J., Jenkins, J., Shu, S., Song, Q., Chavarro, C., Torres-Torres, M., Geffroy, V., Moghaddam, S. M., Gao, D., Abernathy, B., Barry, K., Blair, M., Brick, M. A., Chovatia, M., … Jackson, S. 181 A. (2014). A reference genome for common bean and genome-wide analysis of dual domestications. Nature Genetics, 46(7), 707–713. https://doi.org/10.1038/ng.3008 Soltani, A., Walter, K. A., Wiersma, A. T., Santiago, J. P., Quiqley, M., Chitwood, D., Porch, T. G., Miklas, P., McClean, P. E., Osorno, J. M., & Lowry, D. B. (2021). The genetics and physiology of seed dormancy, a crucial trait in common bean domestication. BMC Plant Biology, 21(1), 58. https://doi.org/10.1186/s12870-021-02837-6 Stanley, D. W. (1992). A possible role for condensed tannins in bean hardening. Food Research International, 25(3), 187–192. https://doi.org/10.1016/0963-9969(92)90136-S Uebersax, M. A., Cichy, K. A., Gomez, F. E., Porch, T. G., Heitholt, J., Osorno, J. M., Kamfwa, K., Snapp, S. S., & Bales, S. (2023). Dry beans (Phaseolus vulgaris L.) as a vital component of sustainable agriculture and food security—A review. Legume Science, 5(1), e155. https://doi.org/10.1002/leg3.155 Van der Auwera, G. A., & O’Connor, B. D. (2020). Genomics in the cloud: Using Docker, GATK, and WDL in Terra (1st ed.). O’Reilly Media. Wahome, S. W., Githiri, M. S., Kinyanjui, P. K., Toili, M. E. M., & Angenon, G. (2023). Genome-wide association study of variation in cooking time among common bean (Phaseolus vulgaris L.) accessions using Diversity Arrays Technology markers. Legume Science, 5(4), e184. https://doi.org/10.1002/leg3.184 Wang, N., & Daun, J. K. (2005). Determination of cooking times of pulses using an automated Mattson cooker apparatus. Journal of the Science of Food and Agriculture, 85(10), 1631– 1635. https://doi.org/10.1002/jsfa.2134 Wang, X., Wilson, L., & Cosgrove, D. J. (2020). Pectin methylesterase selectively softens the onion epidermal wall yet reduces acid-induced creep. Journal of Experimental Botany, 71(9), 2629–2640. https://doi.org/10.1093/jxb/eraa059 Wickham, H. (2016). ggplot2: Elegant graphics for data analysis [3.4.3]. Springer-Verlag. https://cran.r-project.org/web/packages/ggplot2/citation.html Wiesinger, J. A., Osorno, J. M., McClean, P. E., Hart, J. J., & Glahn, R. P. (2021). Faster cooking times and improved iron bioavailability are associated with the down regulation of procyanidin synthesis in slow-darkening pinto beans (Phaseolus vulgaris L.). Journal of Functional Foods, 82, 104444. https://doi.org/10.1016/j.jff.2021.104444 Wood, J. A. (2017). Evaluation of cooking time in pulses: A review. Cereal Chemistry Journal, 94(1), 32–48. https://doi.org/10.1094/CCHEM-05-16-0127-FI 182 APPENDIX A: SUPPLEMENTARY INFORMATION FROM CHAPTER 1 Table SI 1.1: Regions of the genome where more than one QTL for cooking time/water uptake/other traits associated with cooking time and/or water uptake colocalize. Colocalizing QTL could be from either the same or different studies. The position of each QTL is shown in columns 1-2. The names of the QTL are in column 3. QTL names are the same as the names used in the published studies, except that the abbreviations ‘CT’ and ‘WU’ were added to the beginning of QTL for cooking time and water uptake, respectively, if the name of the QTL did not describe the trait with which it is associated. The traits associated with each cluster are listed in column 5. The source of the QTLs from each cluster is shown in column 6. Chromosome Cluster location (bps) QTL names Number of QTL 1 1 1 2 3 3 4 5 4196353 4 lpa-PvMrp1, CT.1.1, and ct1.3 4538676 4- 4627314 2 4803203 4 1088279 - 3409593 WU_Chr01p os45386764, ct1.1, ct1.3 ct1.1 CT.2.1 and CT.2.2 WU.3.1 and CT.3.1 1048427 4- 1585260 1 2554692 WU_pos2554 6920 and CT.3.1 4548944 WU_Chr04p os4548944, CT_Chr04po s4548944, and WU.4.2 WU.5.1, WU.5.2, and CT.5.1 1523099 - 2297109 3 4 3 5 3 3 3 5 Trait(s) associated with the cluster Low phytic acid, cooking time Water uptake and cooking time Cooking time Cooking time Water uptake and cooking time Water uptake and cooking time Water uptake and cooking time QTL sources Circos plot label Circos plot position C1 C2 Garcia et al., 2012 ; Panzeri et al., 2011 ; Cominelli et al., 2018 ; Berry et al., 2020 Garcia et al., 2012 ; Cichy, Wiesinger, et al., 2015 Garcia et al., 2012 Berry et al., 2020 ; Bassett, Katuuramu, et al., 2021 Berry et al., 2020 C5 C4 C3 4196353 4 4582995 3 4803203 4 2248936 1316843 8 C6 2554692 C7 4548944 Berry et al., 2020 ; Bassett, Katuuramu, et al., 2021 Bassett, Katuuramu, et al., 2021 ; Sadohara et al., 2022 Water uptake and cooking time Berry et al., 2020 ; Bassett, Katuuramu, et al., 2021 C8 1910104 183 Table SI 1.1 (cont’d): 5 6 6 6 8 8 9 10 10 3015048 - 3464525 WU.5.2, CT.5.1, and CT.5.2 2564535 9 2575082 7 2771528 3 511826 WU_ss71564 8493, CT.6.1, and WU.6.3 WU_ss71564 5753, CT.6.1, and WU.6.3 CT_Chr06po s27715283, CT.6.1, and WU.6.3 CT_Chr08po s511826, CT.8.1, CT.8.2 1429025 WU_Chr08p os1429025, CT.8.1, CT.8.2 WU.9.1 1074475 3- 1377601 0 4468484 CT_snp3377 419, CT.10.1, WU.10.1 5 3 3 3 3 4 3 7 5600172 CT_snp1000 9 96770, CT.10.1, WU.10.1 10 9440210 CT_Chr10po 3 s9440210, CT.10.1 Water uptake and cooking time Water uptake and cooking time Water uptake and cooking time Water uptake and cooking time Water uptake and cooking time Cooking time Water uptake Water uptake and cooking time Water uptake and cooking time Cooking time Berry et al., 2020 ; Bassett, Katuuramu, et al., 2021 Cichy, Wiesinger, et al., 2015 ; Berry et al., 2020 Cichy, Wiesinger, et al., 2015 ; Berry et al., 2020 Berry et al., 2020 ; Sadohara et al., 2022 Bassett, Katuuramu, et al., 2021 ; Sadohara et al., 2022 Bassett, Katuuramu, et al., 2021 ; Sadohara et al., 2022 Bassett, Katuuramu, et al., 2021 Bassett, Katuuramu, et al., 2021 ; Wahome et al., 2023 Berry et al., 2020 ; Bassett, Katuuramu, et al., 2021 ; Wahome et al., 2023 Berry et al., 2020 ; Sadohara et al., 2022 C9 3239787 C10 2564535 9 C11 2575082 7 C12 2771528 3 C13 511826 C14 1429025 C15 1226038 2 C16 4468484 C17 5600172 C18 9440210 184 Table SI 1.2a: List of 80 QTL from twelve genome-wide association/fine mapping studies (first 20 lines shown). The position of each QTL is shown in columns 1-2. The names of the QTL are in column 3. QTL names are the same as the names used in the published studies, except that the abbreviations ‘CT’ and ‘WU’ were added to the beginning of QTL for cooking time and water uptake, respectively, if the name of the QTL did not describe the trait with which it is associated. The color of each QTL on the circus plot are shown in column 4. The order in which each QTL appears on the circus plot is shown in column 5. The source of each QTL is shown in column 6. Chromosome Pv01 Pv01 Pv01 Pv01 Pv01 Pv01 Pv01 Pv02 Pv02 Pv02 Pv02 Pv02 Pv02 Pv02 Pv02 Pv03 Pv03 Pv03 Pv03 Pv03 … Position (bps) 1528388 QTL name CT_Chr01pos1528388 3691996 WU_ss715639380 37661566 WU_ss715640804 41963534 lpa-PvMrp1 Map color red blue blue green 45386764 WU_Chr01pos45386764 blue 46273142 48032034 ct1.1 ct1.1 red red 24852892 WU_Chr02pos24852892 blue 37679737 CT_ss715649687 38662923 CT_ss715647434 46670223 CKT2.1 47837868 WU_pos47837868 48072125 CT_ss715646000 481002 CT_ss715646002 red red red blue red red 49223533 WU_Chr02pos49223533 blue 832263 CT_ss715648837 863021 CT_ss715650437 983982 301942 488599 CKT3.1 CT_Chr03pos3019420 CT_pos4885990 … … red red red red red … Order on map 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 … QTL source Sadohara et al., 2022 Cichy, Wiesinger, et al., 2015 Cichy, Wiesinger, et al., 2015 Panzeri et al., 2011 ; Cominelli et al., 2018 Sadohara et al., 2022 Garcia et al., 2012 Garcia et al., 2012 Sadohara et al., 2022 Cichy, Wiesinger, et al., 2015 Cichy, Wiesinger, et al., 2015 Diaz et al., 2021 Bassett, Kamfwa, et al., 2021 Cichy, Wiesinger, et al., 2015 Cichy, Wiesinger, et al., 2015 Sadohara et al., 2022 Cichy, Wiesinger, et al., 2015 Cichy, Wiesinger, et al., 2015 Diaz et al., 2021 Sadohara et al., 2022 Bassett, Kamfwa, et al., 2021 … 185 Table SI 1.2b: List of 54 QTL from four QTL mapping studies (first 20 lines shown). See Table SI 1.2a for a complete description of this table. Chromosome Position (bps) QTL name Map color Pv01 Pv01 Pv01 Pv01 Pv01 Pv01 Pv02 Pv02 Pv02 Pv02 Pv02 Pv03 Pv03 Pv03 Pv03 Pv03 Pv04 Pv04 Pv04 Pv05 … 0-3590870 WU.1.1 blue 3168414-7474808 29379910-47273142 38804109-44666791 46084408-49511287 46555884-49071571 794387-5332427 CT.1.2 ct.1.3 CT.1.1 ct1.1 ct1.1 CT.2.1 794387-6683745 CT.2.1 1088279-3409593 1409593-12611432 CT.2.1 CT.2.2 1409593-15964804 CT.2.2 0-1418204 6990038-15852601 10484274-30424207 10484274-34398094 42174196-45170119 0-3508666 CKT3.1 WU.3.1 CT.3.1 CT.3.1 WU.3.2 WU.4.1 1893246-9013918 WU.4.2 10002648-12292681 0-2297109 … CKT4.1 WU.5.1 … red red red red red red red red red red red blue red red blue blue blue red blue … Order on map 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 … QTL source Berry et al., 2020 Berry et al., 2020 Garcia et al., 2012 Berry et al., 2020 Garcia et al., 2012 Garcia et al., 2012 Bassett, Katuuramu, et al., 2021 Bassett, Katuuramu, et al., 2021 Berry et al., 2020 Bassett, Katuuramu, et al., 2021 Bassett, Katuuramu, et al., 2021 Diaz et al., 2021 Berry et al., 2020 Berry et al., 2020 Berry et al., 2020 Berry et al., 2020 Bassett, Katuuramu, et al., 2021 Bassett, Katuuramu, et al., 2021 Diaz et al., 2021 Berry et al., 2020 … 186 APPENDIX B: SUPPLEMENTARY INFORMATION FROM CHAPTER 2 Figure SI 2.1a: Multidimensional scaling (MDS) plot showing only RNA-seq data collected from the slow-cooking beans, TZ-27 and PI527538. Each point represents a biological replicate. The shapes indicate the duration of soaking and the colors indicate the genotype. 187 Figure SI 2.1b: Heat maps showing the ontology of genes that are upregulated in the beans during soaking. Darker colors indicate that a function was strongly associated with a particular genotype during soaking. 188 Figure SI 2.2a: Multidimensional scaling (MDS) plot showing only RNA-seq data collected from the fast-cooking beans, TZ-37 and Ervilha. Each point represents a biological replicate. The shapes indicate the duration of soaking and the colors indicate the genotype. 189 Figure SI 2.2b: Heat maps showing the ontology of genes that are downregulated in the beans during soaking. Darker colors indicate that a function was strongly associated with a particular genotype during soaking. 190 Figure SI 2.3a: A bar graph of the gene ontology (GO) enrichment analysis for biological processes (BP), cellular components (CC), and molecular functions (MF) after 3, 6, 12, and 18 hours of soaking using TopGO. The top graph shows genes upregulated in TZ-27 and the bottom graph shows genes downregulated in TZ-27 (elimKS test, α=0.05). 191 Figure SI 2.3b: A bar graph of the gene ontology (GO) enrichment analysis for biological processes (BP), cellular components (CC), and molecular functions (MF) after 3, 6, 12, and 18 hours of soaking using TopGO. The top graph shows genes upregulated in TZ-37 and the bottom graph shows genes downregulated in TZ-37 (elimKS test, α=0.05). 192 Figure SI 2.3c: A bar graph of the gene ontology (GO) enrichment analysis for biological processes (BP), cellular components (CC), and molecular functions (MF) after 3, 6, 12, and 18 hours of soaking using TopGO. The top graph shows genes upregulated in PI527538 and the bottom graph shows genes downregulated in PI527538 (elimKS test, α=0.05). 193 Figure SI 2.3d: A bar graph of the gene ontology (GO) enrichment analysis for biological processes (BP), cellular components (CC), and molecular functions (MF) after 3, 6, 12, and 18 hours of soaking using TopGO. The top graph shows genes upregulated in Ervilha and the bottom graph shows genes downregulated in Ervilha (elimKS test, α=0.05). 194 Figure SI 2.4a: Results of WGCNA analysis performed on soaking brown beans. The sample dendrogram shows the clustering of brown bean RNA-seq samples and the differences in soaking time and cooking time between samples. The data did not violate the scale-free topology. A cut height of 0.95 was chosen based on the cluster dendrogram. The module-trait relationships heatmap shows the correlation between every detected module, soaking time, and cooking time. 195 Figure SI 2.4b: Results of WGCNA analysis performed on soaking yellow beans. The sample dendrogram shows the clustering of brown bean RNA-seq samples and the differences in soaking time and cooking time between samples. The data did not violate the scale-free topology. A cut height of 0.95 was chosen based on the cluster dendrogram. The module-trait relationships heatmap shows the correlation between every detected module, soaking time, and cooking time. 196 Figure SI 2.4c: Results of WGCNA analysis for soaking and cooking time on soaking brown beans. The scatterplots show the relationship between the module membership scores and the module membership significance values of each gene in the module. Only modules that were significantly associated with both soaking time and cooking time are shown. 197 Figure SI 2.4d: Results of WGCNA analysis for soaking and cooking time on soaking yellow beans. The scatterplots show the relationship between the module membership scores and the module membership significance values of each gene in the module. Only modules that were significantly associated with both soaking time and cooking time are shown. 198 Figure SI 2.5a: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the brown gene co-expression module found in brown beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the BroB module. 199 Figure SI 2.5b: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the red gene co-expression module found in brown beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the RedB module. 200 Figure SI 2.5c: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the midnightblue gene co-expression module found in brown beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the MblB module. 201 Figure SI 2.5d: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the blue gene co-expression module found in yellow beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. B) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the BluY module. 202 Figure SI 2.5e: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the brown gene co-expression module found in yellow beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. b) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the BroY module. 203 Figure SI 2.5f: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the red gene co-expression module found in yellow beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. b) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the RedY module. 204 Figure SI 2.5g: a) Z-score plot depicting individual gene expression levels (gray lines), cooking times (magenta line), and soaking time (blue line) of the white gene co-expression module found in yellow beans. These values were normalized into Z-scores so that they could be directly compared. The green line represents the average of all gene expression levels (i.e., the eigengene) in the module. Correlations between cooking time, soaking time, and eigengene values were used to determine whether a trait was significantly associated with a gene co-expression network. b) Bar plot showing the significance levels (-log10) of all the GO terms significantly associated with the WhiY module. 205 Figure SI 2.6a: VisANT map of the top 30 most co-expressed genes in the brown bean turquoise module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 206 Figure SI 2.6b: VisANT map of the top 30 most co-expressed genes in the brown bean brown module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co- expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 207 Figure SI 2.6c: VisANT map of the top 30 most co-expressed genes in the brown bean red module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 208 Figure SI 2.6d: VisANT map of the top 30 most co-expressed genes in the brown bean midnightblue module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 209 Figure SI 2.6e: VisANT map of the top 30 most co-expressed genes in the yellow bean turquoise module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 210 Figure SI 2.6f: VisANT map of the top 30 most co-expressed genes in the yellow bean brown module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are often co- expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 211 Figure SI 2.6g: VisANT map of the top 30 most co-expressed genes in the yellow bean red module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 212 Figure SI 2.6h: VisANT map of the top 30 most co-expressed genes in the yellow bean blue module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 213 Figure SI 2.6i: VisANT map of the top 30 most co-expressed genes in the yellow bean white module. Line thickness is proportional to the strength of the connection between two nodes (genes). A higher value (scale: 0-1) indicates that the genes are more co-expressed. The cutoff is the lowest connection strength displayed on the chart. Weaker connections have been removed for clarity. Nodes with squares around them are in the list of candidate genes. 214 Figure SI 2.7: RNA-seq data collection and analysis pipeline. A description of some of the bioinformatics tools used in the pipeline are provided in the table on the right. 215 Figure SI 2.8: Physical distance map of the dry bean genome with QTL for cooking time (CT) and water uptake (WU) in the TTRIL and YYRIL populations. The TTRIL QTL were discovered in either Arusha (Ar) or Morogoro (Mo) in either 2016 or 2017. The YYRIL QTL were discovered in Michigan (Mi) in either 2016, 2017, or both years (C). Scale is base pairs (1 scale bar unit = 1,000,000 bps). The locations of the candidate genes are shown to the left of the physical maps. The candidate gene markers show the name of the gene, the genotype in which the gene was more expressed, and the time (in hours) when the gene was differentially expressed in the fast- and slow-cooking beans because of soaking. 216 Table SI 2.1: Information about the RNA samples used in this study (first three lines shown). Raw transcript reads have been deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under the BioProject accession PRJNA865866. Sample name Sample description PI527538 0h PI527538 6h Ervilha 0h A04 A05 A06 … Total read count 3824113 Total cleaned read count 3665750 Alignment (%) to reference genome 89.95 Unmapped reads (%) Unique reads (%) Duplicate reads (%) SRA accession ID 10.05 81.49 8.46 SRR20823517 Alignment (%) to chloroplast DNA 3.86 Alignment (%) to mitochondrial DNA 4.21 6123740 5969411 86.07 13.93 76.48 9.59 SRR20823505 5.15 6881944 6830808 85.57 … … … … 14.43 … 76.08 … 9.48 … SRR20823493 4.69 … … 3.81 4.67 … Table SI 2.2: All genes that were differentially expressed because of soaking in TZ-27, TZ-37, PI527538, and Ervilha according to DESeq2 (first three lines shown). Transcript averages are expressed as raw read counts. Log-fold differences in gene expression were calculated as log(fast/slow). The fourth column shows when gene expression significantly changed relative to hour 0 of soaking. Genotype Gene TZ-27 TZ-27 TZ-27 … Phvul.005G089400 Phvul.010G164200 Phvul.001G251100 … Avg. Transcript 7.994193468 logDiff -3.7569 DE time 0v3 24.18813159 -2.29289 0v3 33.71385661 -1.61795 0v3 … … … 217 Table SI 2.3: Module membership scores and p-values of all the genes in the brown bean co-expression modules (first three rows and first 8 columns are shown). Membership scores and p-values are based on the correlation between gene expression levels and changes in brown bean cooking time. Gene name V7269 V7133 V2888 … Annotation Module ID GS.binary_data p.GS.binary_data MM.red p.MM.red MM.greenyellow … 007G202100 007G174100 003G035300 … black black black … 0.785463 0.782945 0.731272 … 0.007089 0.007403 0.01625 … 0.761407 0.010508 -0.79316 0.805651 0.004901 -0.86374 0.672225 0.033222 -0.58741 … … … … … … … Table SI 2.4: Module membership scores and p-values of all the genes in the brown bean co-expression modules (first three rows and first 8 columns are shown). Membership scores and p-values are based on the correlation between gene expression levels and changes in soaking time. Gene name Annotation Module ID GS.binary_data p.GS.binary_data MM.turquoise p.MM.turquoise MM.brown … V7637 V7269 008G000300 black 007G202100 black V11171 011G103500 black … … … -0.73193 -0.72411 -0.72014 … 0.016107 0.017881 0.018833 … -0.77179 -0.72224 -0.72586 … 0.008916 0.018326 0.017474 … 0.632613 0.749894 0.766014 … … … … … 218 Table SI 2.5: Module membership scores and p-values of all the genes in the yellow bean co-expression modules (first three rows and first 8 columns are shown). Membership scores and p-values are based on the correlation between gene expression levels and changes in yellow bean cooking time. Gene name Annotation Module ID GS.binary_data p.GS.binary_data MM.brown p.MM.brown MM.red … V11447 V7412 V5 … 011G201300 007G233800 001G001000 … black black black … -0.68401 -0.67938 -0.65624 … 0.029157 0.03071 0.039325 … 0.778631 0.007965 -0.11208 … 0.82463 0.003329 -0.14958 … 0.610051 0.061086 -0.28891 … … … … … Table SI 2.6: Module membership scores and p-values of all the genes in the yellow bean co-expression modules (first three rows and first 8 columns are shown). Membership scores and p-values are based on the correlation between gene expression levels and changes in soaking time. Gene name Annotation Module ID GS.binary_data p.GS.binary_data MM.blue p.MM.blue MM.turquoise … V8103 V5 V9616 … 008G119500 black 001G001000 black 009G168900 black 0.780353 0.725654 0.719986 … … … 0.007737 0.017521 0.018871 … -0.84407 0.002133 0.884461 -0.82562 0.003259 0.814723 -0.83952 0.002379 0.853248 … … … … … … … 219 Table SI 2.7: Annotated list of genes that were differentially expressed between fast and slow cooking beans in the same market class (first three rows are shown in three parts). The first column lists the QTL in which each gene was found. The second column lists any QTL that physically overlapped with the QTL in the first column. Hour of DE indicates the hour that the gene was differentially expressed between fast and slow cooking beans. Highlighted annotations are statistically significant. Gene counts are expressed in RPKM. Log-fold gene expression values were calculated as log(fast/slow). The last column shows the module membership of each gene (a gene can only belong to one module). Part 1 (columns 1-9) QTL GWAS/QTL overlap Slow DEG WU.1.1 (Mo_17) Sadohara_CT_NE, Soltani_SI_NF* Unknown function WU.1.1 (Mo_17) Sadohara_CT_NE, Soltani_SI_NF* CT.1.2 (Ar_17) Soltani_SI_NF* P-loop containing nucleoside triphosphate hydrolases superfamily protein RING/U-box superfamily protein Variety Time of DE between fast and slow cooking beans (hr) 0 Brown Brown 0 Gene Name Most TAIR description significant Arabidopsis homologs Score (bits) … Phvul. 001G005800 Phvul. 001G024200 AT2G31020 OSBP(oxysterol binding 36.5 … protein)-related protein 1A AT3G50240 Encodes a kinesin-related 129 protein Brown 0 Phvul. 001G066700 AT4G27600 Encodes a 176 … phosphofructokinase B-type carbohydrate kinase family protein called NARA5, regulates photosynthetic gene expression … … … … … … … … … … 220 Reference species SWISS- PROT Score (bits) E score Significant Pfam hits Pfam name and type Slow (RPKM) Fast (RPKM) … na na na na na na 38 A0A068FIK2 Gossypium hirsutum 5.12E+02 6.00E- 69 PF00225.23 Kinesin 6 2.00E- 44 na … … na … na … na … na … 14 3 … … … … (D) na … 253 156 … … Expression cluster GO enrichment information Table SI 2.7 (cont’d): Part 2 (columns 9-19) Most significant UniProt plant homolog na E score 0.13 2.00E- 28 … … … … … Part 3 (columns 20-21) … … … … … Kinesin- like protein KIN-4A na … LogFC -0.43 -0.30 -0.21 … BroB BroB BroB … 221 Table SI 2.8: List of genes from Table SI 2.7 combined with information from Table SI 2.2 (first three lines shown). The fifth column shows the hour(s) that a gene was differentially expressed between fast and slow cooking beans, and the sixth and seventh columns show the hour(s) that a gene was differentially expressed because of soaking. Highlighted cells indicate that a gene was both differentially expressed because of soaking and was differentially expressed between fast and slow cooking beans in the same market class. Genes with highlighted cells are listed in Table 2.3. Higher expression Gene Annotation Variety Fast Fast Fast … Phvul.001G011000 BEL1-like homeodomain 10 Phvul.003G135000 Protein of unknown function (DUF3511) Phvul.005G034600 NAD(P)-binding Rossmann-fold … superfamily protein … Time of DE between fast and slow cooking beans (hr) 0 0 Time of DE in the FAST- cooking bean (relative to unsoaked; hr) na Time of DE in the SLOW- cooking bean (relative to unsoaked; hr) na na na … 0v6 0v3 … Brown Brown Brown 0, 6 … … 222 Table SI 2.9: Average copper, iron, zinc, manganese, and boron content in freeze-dried, milled, uncooked TZ-27, TZ-37, PI527538, and Ervilha beans. Beans were either unsoaked (US) or soaked for 12 hours in distilled water (S) prior to analysis. Values are concentrations (ppm) of the total sample (water content was excluded). Values from 2018 include one field replicate from one field season in Montcalm, MI. Values from 2019 include two field replicates from one field season in Saginaw Valley, MI. Values with the same letter in the same year are not significantly different from each other (LSD, α = 0.05). Boron (ppm) Copper (ppm) Iron (ppm) Manganese (ppm) Zinc (ppm) Genotype Treatment 2018 2019 2018 2019 2018 TZ-27 US TZ-37 S US S PI527538 US Ervilha S US S 14a 12a 12a 13a 69a 65a 57a 63a 12a 12a 12a 12a 13a 13a 12a 13a 11a 13a 10a 11a 9.75abc 10.7cd 10.0bc 11.3d 8.50a 9.00ab 8.50a 9.00ab 2019 48a 55ab 61b 63b 54ab 57ab 49a 55ab 2018 2019 2018 12a 12a 11a 12a 34a 31a 32a 31a 10a 12a 10a 10a 11a 12a 10a 11a 2019 30a 33bc 33bc 33c 31abc 32abc 29ab 31abc 223 APPENDIX C: SUPPLEMENTARY INFORMATION FROM CHAPTER 3 Figure SI 3.1: Heatmap of all RNA replicates. White coloring indicates that two samples are dissimilar in terms of gene expression. 224 Figure SI 3.2: Significant correlations between physiochemical parameters. CT=cooking time; N=nitrogen; S=sulfur; P=phosphorous; Mg=magnesium; Ca=calcium; B=boron; Zn=zinc; Mn=manganese; Cu=copper; RW=raw seed weight; WU=water uptake percentage; CW=cell wall thickness. 225 Figure SI 3.3: Principal component analysis (PCA) of RNA samples taken from TZ-27 (top left), TZ-37 (top right), PI527538 (bottom left), and Ervilha (bottom right). Lighter colors represent unsoaked samples and darker colors represent soaked samples. The proportion of the total variance explained by the first two principal components are displayed on the x and y axes. 226 Figure SI 3.4: UpSet plots showing GO annotation intersections of upregulated genes. GO annotations describe the functionalities of genes obtained from DESeq2 analyses that compared two genotypes subjected to the same storage treatment and market class (i.e., TZ-27_Fr vs. TZ-37_Fr). Genes belonging to a particular genotype and treatment were upregulated in that genotype and treatment according to DESeq2. Results were obtained from unsoaked (left) and soaked (right) dry beans. 227 Figure SI 3.5a: WGCNA modules associated with changes in the cooking times of beans in response to artificial aging in unsoaked beans. 228 Figure SI 3.5b: WGCNA modules associated with changes in the cooking times of beans in response to artificial aging in soaked beans. 229 Figure SI 3.6a: Scatter plots of co-expression modules that were significantly associated with changes in cooking times of unsoaked beans due to storage. The turquoise, blue, brown, and yellow modules are referred to as raw-1, raw-2, raw-3, and raw-4 from this point forward. 230 Figure SI 3.6b: Scatter plots of co-expression modules that were significantly associated with changes in cooking times of soaked beans due to storage. The pink, cyan, and royalblue modules are referred to as soak-1, soak-2, and soak-3 from this point forward. 231 Figure SI 3.7a: Bar plots showing the significance levels (-log10) of all the GO terms significantly associated with the unsoaked gene co-expression modules (larger bars=more significant). 232 Figure SI 3.7b: Bar plots showing the significance levels (-log10) of all the GO terms significantly associated with the soaked gene co- expression modules (larger bars=more significant). 233 Table SI 3.1a: Physical and statistical descriptions of total RNA samples from whole dry beans (first three rows shown). Unique sample identifiers are in column 1, sample descriptions are in column 2, total raw read count from Illumina sequencing are in column 3, alignment and read count statistics are in columns 4-9. Sequence read archive accessions are in column 10. Sample name Sample description Total read count Total cleaned PE read count (Trimmomatic) Overall alignment rate (%) of reads to the reference genome (HiSat2) % of mate pairs that aligned concordantly exactly 1 time (HiSat2) % of mate pairs that aligned concordantly >1 time (HiSat2) Number of reads that were assigned to features (HTSeq) % of reads that were assigned to features (HTSeq) A1.bam TZ-27 R1 T0 46282486 43616487 Untreated A2.bam TZ-27 R2 T0 29941980 28239668 Untreated A3.bam TZ-27 R3 T0 36137976 34205570 … Untreated … … … 97.69 97.89 97.82 … 79.25 82.38 81.01 … 3.96 4.87 4.32 … 38012089 86.71 24477232 86.33 29740327 86.56 … … 234 Table SI 3.1b: Physical and statistical descriptions of total RNA samples from whole dry beans (first three rows shown). Pearson and Spearman correlation coefficients of read counts in biological (field) replicates. Pearson’s coefficient Spearman’s coefficient Treatment rep1˟rep2 rep2˟rep3 rep1˟rep3 rep1˟rep2 rep2˟rep3 rep1˟rep3 A B C D E F G H I J K L M N O P Q R S T U V W X 0.99 0.99 0.97 1 0.99 1 0.99 1 1 1 0.99 0.99 1 0.99 0.99 1 1 1 0.99 1 0.99 0.99 0.97 1 1 0.96 0.97 0.97 1 0.99 0.99 1 1 0.99 0.99 1 1 0.98 0.99 1 1 1 1 1 0.99 1 0.98 1 1 0.99 1 0.96 0.99 0.98 1 1 1 0.99 1 0.99 0.99 0.97 0.99 1 1 1 1 1 1 1 0.99 1 0.98 0.98 0.98 0.98 0.99 0.99 0.99 0.99 0.98 0.98 0.98 0.98 0.99 0.99 0.99 0.99 0.98 0.98 0.98 0.98 0.99 0.98 0.99 0.98 0.98 0.98 0.98 0.98 0.99 0.98 0.99 0.99 0.98 0.98 0.98 0.98 0.99 0.98 0.99 0.99 0.98 0.98 0.98 0.98 0.99 0.98 0.98 0.99 0.98 0.98 0.98 0.98 0.98 0.98 0.99 0.99 0.98 0.98 0.98 0.98 0.99 0.98 0.99 0.98 0.98 0.98 0.98 0.98 0.99 0.98 0.98 0.98 235 Table SI 3.2: Multiple comparisons of nutrient contents in raw and soaked dry beans stored in untreated, temperature, and tropical storage conditions (Sidak, α=0.05). Three-way interactions were parsed even if they were not significant. N=nitrogen; S=sulfur; P=phosphorous; K=potassium; Mg=magnesium; Ca=calcium; B=boron; Zn=zinc; Mn=manganese; Cu=copper. Unsoaked Untreated TZ-27 TZ-37 PI527538 2.90abc S† N 3.00abcd 0.197ab 0.170a 2.80b 0.186ab Ervilha Temperate TZ-27 TZ-37 PI527538 2.89abc Ervilha TZ-27 3.41abcd 0.197ab 2.94abcd 0.179ab 3.02abcd 0.186ab 0.186ab 3.22abcd 0.203ab 0.203ab 3.50ad Tropical Soaked TZ-37 0.190ab 3.39acd PI527538 3.17abcd 0.187ab 3.65d 0.213b Ervilha 3.29abcd 0.190ab Untreated TZ-27 3.21abcd 0.189ab TZ-37 0.180ab PI527538 2.90abc Ervilha 3.11abcd 0.200ab Temperate TZ-27 2.83bc 0.180ab 2.96abcd 0.193ab TZ-37 PI527538 2.98abcd 0.183ab 3.58ad Ervilha 0.200ab 3.15abcd 0.190ab TZ-27 Tropical TZ-37 3.29abcd 0.203ab PI527538 3.06abcd 0.200ab 3.22abcd 0.210ab Ervilha P† 0.510ab 0.460ab 0.493ab 0.493ab 0.420a 0.453ab 0.437ab 0.480ab 0.487ab 0.480ab 0.463ab 0.533b 0.463ab 0.467ab 0.433ab 0.493ab 0.483ab 0.517ab 0.513ab 0.487ab 0.477ab 0.483ab 0.473ab 0.490ab †Genotype x treatment x soaking interaction effect was not significant. Mg 0.230f 0.210abcf 0.216cf 0.180abde 0.190abcde 0.176ade 0.163e 0.169de 0.170de B Ca 0.137abcd 11.3abef 0.130abcd 10.7abdef 33.7ab 11.0abdef 34.9ab 0.163bd Zn† 36.3a 34.3ab 9.32cde 0.0800a 0.110abcd 11.3abef 30.2ab 0.113abcd 10.6abdef 29.0b 0.133abcd 9.97bcdef 29.0b 0.140abcd 11.0abdef 32.8ab 32.2ab 8.65c 0.0867ac 0.180abde 30.3ab 0.0900abc 8.65c 0.193abcdef 0.107abcd 10.7abdef 30.0ab 35.3ab 0.207abcf 32.7ab 0.177ade 33.3ab 0.180abde 32.0ab 0.190abcde 0.133abcd 11.7abf 0.0833a 9.65cdef 0.0900abc 9.65cdef 0.107abcd 12.0ab 0.193abcdef 0.107abcd 12.7a 0.213bcf 0.223cf 0.213bcf 0.176ade 0.200abcdf 35.3ab 0.127abcd 10.0bcdef 30.3ab 31.7ab 0.137abcd 9.65cdef 31.9ab 0.147abcd 9.65cdef 0.0933abc 8.65c 31.3ab 0.127abcd 11.0abdef 30.7ab 0.180abde 0.177ade 0.187abcde 29.7ab 0.143abcd 9.00cd 9.32cde 0.173d 29.7ab 9.97bcdef 29.6ab 0.160bcd Mn 15.3ab 15.3ab 16.3ab 13.7ab 14.7ab 14.0ab 14.3ab 15.3ab 13.3ab 13.3ab 15.7ab 17.3b 13.0ab 12.6a 14.0ab 15.3ab 16.3ab 16.0ab 16.6ab 13.3ab 16.7ab 17.0b 17.3b 16.6ab Cu 12.7bc 11.3abc 12.7c 11.0abc 9.64a 10.7abc 9.98ab 10.6abc 10.7abc 9.95a 10.0ab 11.0abc 11.0abc 10.7abc 10.3abc 11.3abc 11.7abc 11.3abc 11.3abc 10.3abc 9.66a 10.3abc 10.0ab 10.7abc K† 1.48a 1.41a 1.47a 1.57a 1.47a 1.53a 1.50a 1.54a 1.56a 1.51a 1.44a 1.57a 1.57a 1.49a 1.45a 1.54a 1.54a 1.52a 1.51a 1.59a 1.48a 1.49a 1.50a 1.42a 236 Table SI 3.3: Genes that were differentially expressed between both slow- or both fast-cooking genotypes (first three rows shown in two parts). The truncated name of the gene is in column 1. The name of the transcript is in column 2. Gene annotation information is provided in columns 3-10. Data from brown and yellow beans are in columns 11-13 and 14-16, respectively. Average raw read counts are in columns 11 and 14. Log fold differences in gene expression between fast- and slow-cooking genotypes are in columns 12 and 15. Negative and positive log fold values indicate the gene was more expressed in the slow- and fast-cooking variety, respectively. Bonferroni-Hochman adjusted p-values are in columns 13 and 16 (α=0.05). Part 1 (columns 1-8) Gene Transcript Pfam Panther KO KEGG GO 008G118300 008G118300.1 PF03171 PTHR10209,PTHR10209:SF184 005G164400 005G164400.2 PF00560, PF07714, PF12819 002G163000 002G163000.1 PF07765 PTHR27003,PTHR27003:SF105 KOG1187 PTHR18937,PTHR18937:SF216 … … … … Part 2 (columns 9-16) TAIR … AT3G11180.2 … AT1G51800.1 … GO:0055114, GO:0016491 GO:0005515, GO:0006468, GO:0004672 AT5G41790.1 … … … … … … … … … Domain Annotation baseMean_brown l2fc_brown padj_brown baseMean_yellow l2fc_yellow padj_yellow 2-oxoglutarate (2OG) and Fe(II)- dependent oxygenase superfamily protein Leucine-rich repeat protein kinase family protein COP1-interactive protein 1 … CIP1 … 19.18308 -7.46923 1.14E-07 113.6904 -10.356 1.05E-15 7.167511 -6.04911 0.000193 5.307467 -5.93232 0.000987 661.5076 -5.4739 1.11E-138 88.95846 -1.52504 9.00E-06 … … … … … … 237 Table SI 3.4: All genes within raw genotypes that were differentially expressed between two storage treatments (first three rows shown). The comparison that was conducted in DESeq2 is in column 1. The name of the gene is in column 2. The log fold change in expression between the two storage treatments is in column 3. Page 1: TZ-27; page 2: TZ-37; page 3: PI527538; page 4: Ervilha. Genotype Comparison Gene TZ-27 TZ-27 TZ-27 … Untrt vs. Trop Phvul.001G002000 Temp vs. Trop Phvul.001G002000 Temp vs. Trop Phvul.001G007700 … … Log Fold Change 0.587654176 0.577458974 0.223524632 … Table SI 3.5: All genes within soaked genotypes that were differentially expressed between two storage treatments (first three rows shown). The comparison that was conducted in DESeq2 is in column 1. The name of the gene is in column 2. The log fold change in expression between the two storage treatments is in column 3. Page 1: TZ-27; page 2: TZ-37; page 3: PI527538; page 4: Ervilha. Genotype Comparison Gene Log Fold Change TZ-27 TZ-27 TZ-27 … Untrt vs. Trop Untrt vs. Trop Phvul.001G000800 Phvul.001G001500 Temp vs. Trop Phvul.001G001500 … … -0.416475782 1.015383538 0.810447322 … 238 Table SI 3.6a: Genes that were differentially expressed between two storage treatments across all four genotypes (a combination of TZ-27, TZ-37, PI527538, and Ervilha) before soaking and which were found in a WGCNA module (first three rows shown). The comparison that was conducted in DESeq2 is in column 1. The module membership of the gene is in column 2. The name of the gene is in column 3. The log fold change in expression between the two storage treatments is in column 4. Gene annotation information from Phytozome is in column 5. The RPKM values of genotypes in different storage conditions are in columns 6-8. Comparison Module Gene name Log Fold Change Annotation Avg. Untreated (RPKM) Avg. Temperate (RPKM) Avg. Tropical (RPKM) Untrt vs. Trop Untrt vs. Trop Untrt vs. Trop … Raw-1 Raw-1 Raw-1 … Phvul. 011G125800 Phvul. 007G209332 Phvul. 005G006700 … -0.95 -0.80 -0.55 … Plant mitochondrial ATPase, F0 complex, subunit 8 protein Albino or Glassy Yellow 1 39.2 17.2 Zincin-like metalloproteases family protein 6.5 … … 33.0 17.9 5.1 … 19.4 15.6 4.4 … Table SI 3.6b: Genes that were differentially expressed between two storage treatments across all four genotypes (a combination of TZ-27, TZ-37, PI527538, and Ervilha) after soaking and which were found in a WGCNA module (first three rows shown). The comparison that was conducted in DESeq2 is in column 1. The module membership of the gene is in column 2. The name of the gene is in column 3. The log fold change in expression between the two storage treatments is in column 4. Gene annotation information from Phytozome is in column 5. The RPKM values of genotypes in different storage conditions are in columns 6-8. Comparison Module Gene name Log Fold Change Annotation Avg. Untreated (RPKM) Avg. Temperate (RPKM) Avg. Tropical (RPKM) Untrt Temp Untrt Trop Untrt Trop … vs. Soak-2 vs. Soak-1 vs. Soak-1 … Phvul. 001G241100 Phvul. 009G194700 Phvul. 008G025000 … -0.13 eukaryotic translation initiation factor 2 336.3 308.4 316.0 0.19 0.20 … signal recognition particle receptor alpha subunit family protein Galactose oxidase/kelch protein … superfamily repeat 40.0 17.2 … 42.9 18.2 … 46.0 19.9 … 239 Table SI 3.7: Average seed coat check severity scores of dry beans grown in Montcalm, MI (MC) and Saginaw Valley, MI (SV) in 2023. Standard deviations of the values are shown. Genotype Location Seed coat check severity score 1.69±0.0819 1.27±0.0300 1.15±0.0503 1.05±0.0141 1.02±0.0200 TZ-27 TZ-27 TZ-37 TZ-37 MC SV MC SV PI527538 MC PI527538 SV 1.00±0.00 Ervilha MC Ervilha SV 1.32±0.0551 1.11±0.0586 Table SI 3.8: Average seed moisture contents of whole and milled dry bean samples. Seeds were harvested from Montcalm, MI in 2020. The data collected in 2021 were obtained from whole seeds using a Dickey John benchtop moisture tester. The data collected in 2024 were obtained from raw milled samples that had been frozen at -81°C for 38 months prior to testing. Moisture testing of the milled samples was conducted using the low constant oven temperature method. Standard deviations of the values are shown. TZ-27 TZ-37 Genotype Storage Measurement Untreated date 5/27/2021 Moisture content 11.50 ± 0.3033 Untreated 5/27/2021 12.03 ± 0.8756 PI527538 Untreated 5/27/2021 11.43 ± 0.6713 Ervilha Untreated 5/27/2021 11.53 ± 0.8091 TZ-27 TZ-37 Untreated 7/26/2024 12.47 ± 0.1327 Untreated 7/26/2024 12.27 ± 0.5971 PI527538 Untreated 7/26/2024 Ervilha Untreated 7/26/2024 11.47 ± 0.07420 12.17 ± 0.4856 TZ-27 TZ-37 Temperate 7/26/2024 11.97 ± 0.2106 Temperate 7/26/2024 12.09 ± 0.5022 PI527538 Temperate 7/26/2024 12.19 ± 1.397 Ervilha Temperate 7/26/2024 11.77 ± 0.1747 TZ-27 Tropical 7/26/2024 TZ-37 Tropical 7/26/2024 18.46 ± 0.08884 18.38 ± 0.2409 PI527538 Tropical 7/26/2024 17.88 ± 0.2364 Ervilha Tropical 7/26/2024 17.46 ± 0.2915 240 APPENDIX D: SUPPLEMENTARY INFORMATION FROM CHAPTER 4 Figure SI 4.1a: Correlation plot of highly polymorphic SNPs with cooking time in the FCBC (fast-cooking backcross) population. A histogram of the values along with a distribution curve are shown on the diagonal. Values are plotted as empty circles with scale bars at the base of the graph. Fitted lines are displayed over the plotted values to show goodness-of-fit. P-values: ≥0.001***, ≥0.01**, ≥0.05*, 0.1•. Abbreviations: CT: cooking time; MC%: raw seed moisture content percentage; RW%: raw seed weight percentage; WU%: water uptake percentage; UV: ultraviolet light; SNP: single nucleotide polymorphism; PV: Phaseolus vulgaris. 241 Figure SI 4.1b: Correlation plot of highly polymorphic SNPs with cooking time in the TTRIL population (Berry et al., 2020). See Fig. SI 4.1a for more details about the layout of the plot. 242 Figure SI 4.1c: Correlation plot of highly polymorphic SNPs with cooking time in the YYRIL population (Bassett et al., 2021). See Fig. SI 4.1a for more details about the layout of the plot. 243 Figure SI 4.1d: Correlation plot of highly polymorphic SNPs with cooking time in cranberry bean breeding lines grown in Montcalm Research Farm (Lakeview, MI) in 2022. See Fig. SI 4.1a for more details about the layout of the plot. 244 Figure SI 4.1e: Correlation plot of highly polymorphic SNPs with cooking time in cranberry bean breeding lines grown in Saginaw Valley Research Farm (Frankenmuth, MI) in 2022. See Fig. SI 4.1a for more details about the layout of the plot. 245 Figure SI 4.1f: Correlation plot of highly polymorphic SNPs with cooking time in kidney bean breeding lines grown in Montcalm Research Farm (Lakeview, MI) in 2022. See Fig. SI 4.1a for more details about the layout of the plot. 246 Figure SI 4.1g: Correlation plot of highly polymorphic SNPs with cooking time in yellow bean breeding lines grown in Montcalm Research Farm (Lakeview, MI) in 2022. See Fig. SI 4.1a for more details about the layout of the plot. 247 Figure SI 4.1h: Correlation plot of highly polymorphic SNPs with cooking time in yellow bean breeding lines grown in Saginaw Valley Research Farm (Frankenmuth, MI) in 2022. See Fig. SI 4.1a for more details about the layout of the plot. 248 Table SI 4.1: Complete description of the dry bean lines used in this study (first three lines shown in two parts). Descriptions include the market class, molecular and phenotypic marker data, and cooking times of each genotype (if available). The cooking times of seeds planted at the Montcalm Research Farm in Lakeview, MI and the Saginaw Valley Research Farm in Frankenmuth, MI in 2022 are in the columns labeled “CT_MC” and “CT_SV”, respectively. The cooking times of the TTRIL and YYRIL populations and their parent lines are in the column labeled “CT_RIL”. The cooking times of the fast-cooking backcross (FCBC) population and their parent lines are in the column labeled "CT_BC”. Part 1 (columns 1-9) Genotype Market class snp PV00001 snp PV00038 snp PV00070 snp PV00169 snp PV00170 snp PV00183 snp PV00201 … Adams B1904-3-1 B1904-3-2 … Black Black Black … CC CC CC … CC CC CC … GG TT TT … ACCACC ACCACC ACCACC … CC CC CC … CC AA AA … TT TT TT … … … … … Part 2 (columns 10-19) … snp PV00207 snp PV00218 snp PV00225 snp PV00237 snp PV00259 CT_MC CT_SV CT_RIL CT_BC UV response (0=non- darkening; 0.5=slow- darkening; 1=regular darkening) … AA … AA … AA … … AG AG AG … TT TT TT … GG GG GG … CC CC CC … … … … … … … … … … … … … … … … … … … … … 249 Table SI 4.2: Information about the KASP markers used in this study (first three columns shown in two parts). The Intertek ID of the markers are in column 1. QTL for cooking time or water uptake that are within 1 Mbp of the KASP marker are in column 3. Candidate genes for cooking time that are within 1 Mbp of the KASP marker are in column 4. The physical position of each KASP marker on the P. vulgaris v.2.1 reference genome is described in columns 5-6 (Schmutz et al., 2014). Other details about the design of the markers are found in columns 7-10. Part 1 (columns 1-8) Intertek SNP ID Customer ID Associated QTL Associated candidate gene (if applicable) Position Chromosome SNP Marker source snpPV00255 CaEF_C01 snpPV00256 XTH15_C01 snpPV00256_al t1 XTH15_C01_a lt1 … … CT.1.2 (Ar_17) (Berry et al., 2020) CT.1.1 (Mo_17) (Berry et al., 2020) CT.1.1 (Mo_17) (Berry et al., 2020) … Calcium-binding EF- hand family protein (Phvul.001G067400) Xyloglucan endotransglycosylase / hydrolase 15 (Phvul.001G265300) Xyloglucan endotransglycosylase / hydrolase 15 (Phvul.001G265300) … 8616158 1 G/A 51132997 1 A/C 51132947 1 A/T … … … Karen Cichy (karen.cichy@usda.gov); Hannah Jeffery (jeffery96hr@gmail.com) Karen Cichy (karen.cichy@usda.gov); Hannah Jeffery (jeffery96hr@gmail.com) Karen Cichy (karen.cichy@usda.gov); Hannah Jeffery (jeffery96hr@gmail.com) … … … … … … 250 Table SI 4.2 (cont’d): Part 2 (columns 9-10) … … … … Notes Sequence Inconclusive ATCTAGAATTTACCCCAATCACAATATTACTACATGTATACATGTGTTAT[R] CCCCCCCCATACTTTCCAGCAACTCCTCCATCTCTACTAAACTTCCTTGTA Failed (GC content too low) Failed (GC content too low) ATTATTCCAGTATTGCAAATATTAATCAAATTTATTTATTTTCTTCAAAA[M] AAATTTAATACTCTACCAATCTAGTCTTACTTATCATTTTACACTTAAAAA TATGTATATAAGTATTAAATAAATAAATGAAAAATAAAGTATAATTAATAA[W] AACATAATATACTCATATTGGAATTGTATTATTATTGCATAAACATGATGG … … … 251 Table SI 4.3: Molecular and phenotypic marker data of selected genotypes with contrasting cooking times and/or seed coat darkening traits (first four rows shown in two parts). Traits that are polymorphic in pairs of genotypes with contrasting cooking times are in bold font. Genotypes highlighted with darker colors are slower cooking relative to genotypes from the same market class that are highlighted with lighter colors. Part 1 (columns 1-6) Replicate Genotype Description Market class Avg. cooking time (mins) Avg. UV response (0=non-darkening; 0.5=slow-darkening; 1=regular darkening) 1 2 1 2 … TZ-27 TZ-27 TZ-37 TZ-37 … Part 2 (columns 7-18) TTRIL parent TTRIL parent TTRIL parent TTRIL parent … Brown 60.30 Brown . Brown 33.00 Brown … . … . . . . … … … … … … … … snp PV00001 snp PV00038 snp PV00070 snp PV00169 snp PV00170 snp PV00183 snp PV00201 snp PV00207 snp PV00218 snp PV00225 snp PV00237 snp PV00259 … C:C … C:C … C:C … C:C … … G:G G:G G:G G:G … T:T T:T T:T T:T … ACC:ACC C:C ACC:ACC C:C ACC:ACC C:C ACC:ACC C:C … … A:A A:A A:A A:A … C:C C:C C:T C:C … A:A A:A A:A A:A … A:A A:A A:G A:G … T:T T:T T:T T:T … A:A A:A G:A A:A … C:C C:C C:T T:T … 252 APPENDIX E: MISCELLAENOUS SUPPLEMENTARY INFORMATION (NOT IN ANY CHAPTERS) TZ-27 TZ-37 Figure SI E.1: Relative distribution of the pectic and hemicellulosic polysaccharides of cell wall isolates from raw TZ-27 and TZ-37 seeds. Values were obtained by the strategy outlined in Table B.3 from the data in Table B.2. HG(XHG) = (xylo)homogalacturonan, RG-I = rhamnogalacturonan-I, AG-I = type I (arabino)galactan, AG-II = type II arabinogalactan(protein), GM = glucomannan, GAX = glucuronoarabinoxylan, 4-Glc = unaccounted for 4-linked glucan, XyG = (fucogalacto)xyloglucan. Seeds were grown at the Montcalm Research Farm in Lakeview, MI in 2022. 253 Figure SI E.2: MAS-CP solid state nuclear magnetic resonance (ssNMR) spectral data of raw dry bean flours. The beans were grown at the Montcalm Research Farm in Lakeview, MI in 2023. A 35-40 mg sample was processed in a 3.2 mm rotor. Bean Flor 1: TZ-27; Bean Flor 4: TZ-37; Bean Flor 7: PI527538. Results are from a single sample. 254 Table SI E.1. Monosaccharide profiles of raw TZ-27 and raw TZ-37 cell wall isolates. Seeds were grown at the Montcalm Research Farm in Lakeview, MI in 2022. Values are the means of duplicates of three biological replicates ± standard deviation. Proportions of uronic acids were determined by carboxyl reduction with NaBD4, as described by Carpita and McCann (1997). Data was collected by Dr. Nicholas Carpita. Sample Rha Fuc Ara Xyl Man Gal GalA Glc GlcA mol % TZ-27 3.9±0.5 2.9±0.6 46.2±5.0 11.8±1.1 0.6±0.2 7.2±0.7 9.3±2.0 17.7±7.5 0.4±0.1 TZ-37 3.4±0.3 2.7±1.0 51.9±2.2 11.9±1.2 0.4±0.1 7.4±0.9 9.9±1.2 12.0±1.9 0.4±0.2 255 Table SI E.2. Linkage analysis of bean cell walls. Seeds were grown at the Montcalm Research Farm in Lakeview, MI in 2022. Values are the mean ± standard deviation of three biological replicates. tr=trace amounts less than 0.05 mol %. Data was collected by Dr. Nicholas Carpita. Linkage group TZ-27 TZ-37 mol % t-Rhap 2-Rhap 2,4- Rhap 0.1±0.0 1.8±0.4 0.1±0.1 1.7±0.3 2.0±0.1 1.6±0.2 t-Fucp 2.9±0.6 2.7±1.1 t-Araf t-Arap 2-Araf 3-Araf 10.1±1.9 13.7±0.8 0.1±0.0 0.1±0.0 0.1±0.0 0.1±0.0 0.4±0.1 0.3±0.1 4-Arap 0.1±0.1 0.1±0.0 5-Araf 16.9±1.4 18.2±0.2 2,5-Araf 7.2±0.4 6.1±2.0 3,5-Araf 4.0±0.4 5.1±1.7 2,3,5- Araf t-Xylp 2-Xylp 4-Xylp 7.3±0.4 8.2±0.4 6.1±0.7 6.7±0.5 4.5±0.6 4.0±0.7 0.8±0.2 0.7±0.1 2,3-Xylp 0.3±0.2 0.4±0.1 2,4-Xylp 0.1±0.1 0.1±0.0 t-Manp 0.1±0.0 tr 4-Manp 0.4±0.1 0.3±0.0 4,6- Manp t-Galp 2-Galp 3-Galp 4-Galp 6-Galp 0.1±0.0 0.1±0.0 3.6±0.2 3.8±0.3 0.6±0.6 1.6±0.4 0.6±0.2 0.4±0.2 0.9±0.1 0.6±0.3 1.1±0.2 0.8±0.4 3,4-Galp 0.0±0.0 0.0±0.0 3,6-Galp 0.4±0.1 0.2±-0.1 256 Table SI E.2 (cont’d) : t-GalAp 0.2±0.1 0.2±0.1 4-GalAp 3.4±0.6 4.6±0.7 3,4- GalAp t-Glcp 3-Glcp 4-Glcp 5.7±1.5 5.1±1.8 0.4±0.2 0.1±0.1 0.4±0.3 0.1±0.0 12.3±5.0 7.5±1.4 4,6-Glcp 4.6±2.2 4.3±1.0 t-GlcAp 0.4±0.1 0.4±0.2 257 Table SI E.3: Diagnostic linkage groups determined by gas-liquid chromatography-electron impact mass spectrometry of partly methylated alditol acetate derivatives of polysaccharides. Uronic acids were carboxyl-reduced with NaBD4 to generate 6,6-dideutero-isomers of galactose (gal) and glucose (glc) resolved from their respective neutral sugar by mass spectroscopy. Polysaccharide abundances were calculated from the sum of the mol % of these diagnostic linkages presented in Table B.2. Data was collected by Dr. Nicholas Carpita. Polysaccharide Diagnostic Linkage Groups Pectins Rhamnogalacturonan I t-Rha, 2-Rha, 2,4-Rha, plus 4-GalA equal to Rha residues (Xylo)homogalacturonan Remaining 4-GalA, 3,4-GalA, plus t-Xyl equal to 3,4- Arabinan GalA 2-Araf, 3-Araf, 5-Araf, 2,5-Araf, 3,5-Araf, 2,3,5-Araf and t-Araf equal to branch-point residues Type I (Arabino)galactan 4-Gal, 3,4-Gal, and t-Araf equal to branch-point Type II Arabinogalactan(- protein) Hemicelluloses (Fucogalacto)xyloglucan residues 3-Gal, 6-Gal, 3,6-Gal, plus t-Araf equal to 3,6-Gal 4,6-Glc and 4-Glc equal to 1/3 the 4,6-Glc, t-Xyl and 2-Xyl equal to 4,6-Glc, t-Gal and 2-Gal equal to 2- Xyl, and t-Fuc equal to 2-Gal (Glucuronoarabino)xylan 4-Xyl, 2,3-Xyl, 2,4-Xyl, plus t-GlcA and t-Araf equal to branch-point residues 4-Man and 4,6-Man residues, plus an equal amount of 4-Glc, plus t-Gal equal to the 4,6-Man residues Glucomannan 4-Glucan Any remaining 4-Glc not associated with xyloglucan or glucomannan 258