E NVIRONMENTAL DRIVERS AND EVOLUTIONARY CONSEQUENCES O F HORIZONTAL GENE TRANSFER IN SOIL BACTERIA By Heather Kittredge A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Integrative Biology Doctor of Philosophy Ecology, Evolutionary Biology and Behavior Dual Major 202 1 ABSTRACT ENVIRONMENTAL DRIVERS AND EVOLUTIONARY CONSEQUENCES OF HORIZONTAL GENE TRANSFER IN SOIL BACTERIA By Heather Kittredge Horizontal gene transfer (HGT) is a driving force in bacterial evolution and could drive rapid adaptation in bacterial communities. Natural transformation is one mechanism of HGT that allows bacteria to pick up extracellular DNA (eDNA) fr om the environment and integrate it into their genome . B ut the rate of HGT in natural environments , and the role this process plays in facilitating rapid adaptation remains unknown. A s climate change threatens the stability of environments worldwide, understanding how quickly bacteria can adapt to novel environments is essential. My dissertation research characterizes the environmental drivers and evolutionary consequences of natural tr ansformation in a highly transformable model soil bacterium Pseudomonas stutzeri . Despite decades of research on understanding HGT at the molecular level, less is known about the ecological drivers of HGT. To understand the soil conditions relevant for tra nsformation, I first measured eDNA in the field over a short - term drying re wet ting disturbance (Ch. 2) . I found that eDNA increase d in response to the re we tting disturbance but quickly disappeared from soil , suggesting a small portion of this eDNA could be transformed by bacterial cells recover ing from the disturbance . To test the efficiency of transformation under the conditions in which eDNA disappeared , I created a novel microcosm system for quantifying tr ansformation in soil (Ch. 3) . Here, I inoculate d soil with live antibiotic - susceptible , and dead antibiotic - resistant P. stutzeri . I then tracked the evolution of antibiotic resistance over a range of soil conditions and eDNA concentrations. Transformation drove the evolution of antibiotic resistance across a wide range of soil moistures and increased in response to larger inputs of dead cells (eDNA s our ce) , with antibiotic resistance repeated ly appearing in antibiotic free soil . Despite the prevalence of transformation across bacterial species , the evolutionary origins and consequences of transformation are still largely unknown. T ransformation presumably provide s a fitness benefit in stressful or continuously changing environments, but few studies have quantified changes in transformation in response to adaptive evolution . Here, I evolve d P . stutzeri at different salinities and test ed how the growth rate and transformation efficiency change d in response to salt adaptation (Ch. 4). Overall, the growth rate increased in response to adaptation, but the transformation efficiency declined, with only ~50% of the evolved populations transforming eDNA at the end of experiment as opposed to 100% of ancestral populations transforming eDNA. Overall , my dissertation research elucidate s the factors driving transformation in soil, setting the sta ge for future experiments to scale up estimates of transformation to the whole community level. I find that transformation occurs under most soil conditions and allows genetic variants to arise at low frequencies in the absence of selection . I a l so report novel experimental evidence that transformation efficienc y can change dramatically, and in a highly variable manner, over just ~330 generations . Taken together, this body of research highligh ts a role for transformation in many natural systems of ecological significance , and points to dead cells as an important but often overlooked source of genetic diversity. iv This thesis is dedicated to Jake . T h ank you for always challenging the . --- And to the --- And to Sarah Evans whose unconditional support over the last 5 years made this dissertation possible! v ACKNOWLEDGEMENTS First and foremost, I would like to thank Sarah Evans for her incredible support in research and in life. I will forever be grateful for the intellectual freedom you granted me in my research pursuits. I would also like to thank my commi ttee for supporting my diverse research interests, your perspectives provided many unique and interdisciplinary insights. Finally, a thank you to the KBS community and especially to the members of the Evans lab , whose good energy and jovial attitudes made this experience that much better. vi TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ .... viii LIST OF FIGURES ................................ ................................ ................................ .. ix KEY TO ABBREVIATIONS ................................ ................................ ..................... xi CHAPTER 1 : INTRODUCTION ................................ ................................ .............. 1 1.1 NATURAL TRANSFORMATION ................................ ................................ .. 1 1.2 REGULATION OF COMPETENCE ................................ .............................. 2 1.3 EXTRACELLULAR DNA IN THE ENVIRONMENT ................................ ...... 3 1.4 EVOLUTIONARY ORIGINS OF TRANSFORMATION ................................ . 4 1.5 PSEUDOMONAS STUTZERI : MODEL SYSTEM ................................ ........ 5 1.6 RESEARCH QUESTION S ................................ ................................ ........... 6 CHAPTER 2 : EXTRACELLULAR DNA MASKS MICROBIAL RESPONSES TO A PULSE DISTURBANCE ................................ ................................ .......................... 7 2.1 ABSTRACT ................................ ................................ ................................ . 7 2.2 INTRODUCTION ................................ ................................ ......................... 7 2.3 MATERIALS AND METHOD S ................................ ................................ ..... 9 2.3.1 SITE AND SOIL COLLECTION ................................ ............................ 9 2.3.2 REMOVAL OF EXTRACELLULAR DNA ................................ .............. 10 2.3.3 QUA N TITATIVE PCR ................................ ................................ ........... 11 2.3.4 AMPLICON SEQUENCING AND BIOINFOR MATICS ......................... 12 2.3.5 STATISTICAL ANALYSES ................................ ................................ ... 13 2.4 RESULTS AND DISCUSSION ................................ ................................ .... 14 APPENDIX ................................ ................................ ................................ .............. 19 CHAPTER 3 : HORIZONTAL GENE TRANSFER FACILITATES THE SPREAD OF EXTRACELLULAR ANTIBIOTIC RESISTANCE GENES IN SOIL .......................... 25 3.1 ABSTRACT ................................ ................................ ................................ .. 25 3.2 INTRODUCTION ................................ ................................ .......................... 26 3.3 METHODS ................................ ................................ ................................ ... 29 3.3.1 SITE AND SOIL COLLECTION ................................ .............................. 29 3.3.2 BACTERIAL CULTURES AND EXTRACELLULAR DNA ....................... 30 3.3.3 SOIL MICROCOSMS ................................ ................................ ............. 31 3.3.4 ANTIBIOTIC RESISTANCE GENES IN LIVE VS. DEAD CELLS ........... 33 3.3.5 COUNTING TRANSFORMANTS ................................ ........................... 34 3.3.6 STATISTICAL ANALYSES ................................ ................................ ..... 35 3.4 RESULTS ................................ ................................ ................................ ..... 36 3.5 DISCUSSION ................................ ................................ ............................... 42 APPENDIX ................................ ................................ ................................ .............. 47 vii CHAPTER 4 : CHANGES IN TRANSFORMATION AFTER SALT ADAPTATION ... 50 4.1 ABSTRACT ................................ ................................ ................................ ... 50 4.2 INTRODUCTION ................................ ................................ ........................... 51 4.3 METHODS ................................ ................................ ................................ .... 54 4.3.1 SERIAL DILUTION EXPERIMENT ................................ .......................... 54 4.3.2 PREPARATION OF CELL LYSATES ................................ ...................... 55 4.3.3 GROWTH RATE DETERMINATION ................................ ....................... 56 4.3.4 TRANSFORMATION EFFICENCY AND FREQUENCY ......................... 57 4.4.5 STATISTICAL ANALYSES ................................ ................................ ...... 58 4.4 RESULTS ................................ ................................ ................................ ...... 59 4.4.1 GRO WTH RATE ................................ ................................ ..................... 59 4.4.2 LOSS OF TRANSFORMABILITY ................................ ............................ 60 4.4.3 TRANSFORMANTS AND TOTAL CELL S ................................ ............... 61 4.4.5 HIGH VARIATION IN TRANSFORMATION EFFICIENCY ...................... 62 4.4.6 TRADE - OFF BETWEEN GROWTH RATE AND TRANSFORMATION .. 64 4.5 DISCUSSION ................................ ................................ ................................ 65 APPENDIX ................................ ................................ ................................ .............. 70 CHAPTER 5 : CONCLUSIONS ................................ ................................ ................ 81 BIBLIOGRAPHY ................................ ................................ ................................ ..... 83 viii LIST OF TABLES Table 2.1 : ................................ ............................. 1 6 Table A 2.1 : Expanded table of e .......... Table A 2.2 : The top 15 taxa driving changes in response to soil rewetting ............ Table 4.1 : Review of transformation - mediated fitness effects ................................ 52 Table A 4.1 : Cell lysate sources ................................ ................................ .............. 76 Table A 4.2 : Populations that did not revive ................................ ............................ 77 Table A 4.3 : Expanded transformation efficien cy results ................................ ......... 78 Table A 4.4 : Expanded growth rate results ................................ .............................. 7 9 Table A 4.5 : Expanded population size results ................................ ........................ 80 ix LIST OF FIGURES Fig ure 2.1 : Effect of eDNA removal on b ..... 15 Figure 2.2 : ................. 17 Figure A 2.1 : ................................ .................. 22 Figure 3.1 : S oil characteristics likely to affect natural transformation ..................... 27 Figure 3.2 : General experimental design for soil microcosms ................................ 31 Figure 3.3: The relationship between eDNA availability and transformation .......... 37 Figure 3.4: The relationship between soil moisture and transformation ................. 38 Figure 3.5 : T he relationship between dispersal and transformation ....................... 39 Figure 3.6: Transformation vs. invasion of antibiotic resistant genes ..................... 41 Figure A 3.1: Transformation assays under laboratory conditions .......................... 48 Figure A 3.2: Gravimetric soil moisture from 1989 to 2019 Southwest, MI .............. 49 Figure 4.1: The serial transfer conditions for the two evolution treatments ............ 54 Figure 4.2: Graphical representation of the experimental design ........................... 55 Figure 4.3: Revival and assay conditions ................................ ............................... 56 Figure 4.4: Effects of adaptive evolution ................................ ................................ 59 Figure 4.5: Loss of t ransformability ................................ ................................ ........ 60 Figure 4.6: The relationship between transformants and total cells ........................ 61 Figure 4.7: Changes in transformation in response to experimental evolution ....... 62 Figure 4.8: Variation in transformation efficiency ................................ ................... 63 Figure 4.9: Trade - off between grow th rate and transformatio n efficiency .............. 64 Figure A 4.1: Preliminary transformation assays ................................ ..................... 71 x Figure A 4.2: Transformability in the populations that transformed eDNA ............... 72 Figure A 4.3: Effect of cell lysates on transformation ................................ .............. 73 Figure A 4.4: Variation in transformation frequency ................................ ................ 74 Figure A 4.5: Effect of cell lysates on growth rate and population size ................... 75 xi KEY TO ABBREVIATIONS ANOVA Analysis of Variance AR Antibiotic resistance ARG Antibiotic resistance genes eARG Extracellular antibiotic resistance genes eDNA Extracellular DNA Gent R Gentamicin resistance GLBRC Great Lakes Bioenergy Research Center HGT Horizontal gene transfer Kan R Kanamycin resistance LB L ysogenic Broth NaCl Sodium Chloride NMDS No nmetric multidimensional scaling OTU Operational Taxonomic Unit PERMANOVA Permutational multivariate analysis of variance PMAxx Propidium monoazide xx qPCR Quantitative PCR R2A rRNA Ribosomal ribonucleic acid 1 CHAPTER 1: INTRODUCTION Understanding how populations respond to rapid environmental change is a fundamental goal in ecology and evolutionary biology. Although bacteria are among the most ubiquitous organisms on Earth and there are many model systems for studying bacterial evolut ion in the laboratory, we still have a poor understanding of how quickly bacteria can adapt to changes in their natural environment [1], [2] . As climate change threatens to dramatically alter environments worldwide, understanding how bacterial populations will respond to these changes is essential in predicting effects on broader ecosystem processes [3], [4] . Bacteria are likely to respond faster than other organisms to environmental change due to their short generation times and their unique ability to move ge netic elements among and within species through horizontal gene transfer (HGT) [5], [6] . The exchange of genes bet we en and within bacterial species is of fundamental importance to the evolution of prokaryotic genomes, as evidenced by the rapid dissemination of antibiotic resistance genes in recent decades [7], [8] . Although th is di ssemination of foreign genes is extensively studied at the cellular level, little is known about the rate of HGT in natural environments , and the role this process plays in facilitating bacterial responses to environmental change [9], [10] . 1.1 NATURAL TRANSFORMATION Horizontal gene transfer (HGT) is the lateral exchange of genetic material bet we en potentially unrelated living cells. It is a driving force in bacterial evolution, with e stimates o f HGT - driven gene gain and loss at least comparable to the rate of point mutation s [6], [11] [17] . HGT facilitates the acquisition of foreign DNA through three 2 main mechanisms: conjugation ( pilus - mediated plasmid transfer requiring direct cell contact), transduction ( DNA transfer b y bacteriophages ) and natural transformation ( uptake of extracellula r DNA). Although HGT has been studied for decades, little is known about the relative contribution of each mechanism of HGT to bacterial evolution. My dissertation research focuses on natural transformation and its role in the soil environment. The acquisition of genes through natural transformation can accelerate genome evolution and increase a [10], [1 4], [18] . Ho we ver, to date, transformation has predominately been studied in the lab and its relevance in accelerating genome evolution in natural environments remains largely unknown. 1. 2 REGULATION OF COMPETENCE The success of transformation depends on the development of competence the physiological state of bacterial cells able to acquire free DNA from the environment [19] . Bacteria can be constitutively or conditionally competent in the laboratory . The induction of competence can be regulated by the cell cycle or induced by environmental cues [20] . However, th ese cues have been difficult to quantify experimentally since many bacteria seem unable to carry out transformation under laboratory conditions. In species tha t do undergo transformation in the lab, external factors that influence competence and transformation include temperature, pH, growth phase, nutrient concentration , and stress [21] [23] . In soil, the induction of competence may be restricted by metabolic inactivity and lack of resources , as m any soil bacter ia are starved or living in a dormant state [24], 3 [25] . In particular, water and nutrient limitations could prevent transformation, as competence often requires high cell densities [23], [26] . Transformation has been shown to occur across many different environmental microorganisms [23], [27] [29] . However , little is known about the environmental cues that regulate transformation for m ost bacterial species in situ . My dissertation research specifically addresses this knowledge gap by testing how transformation efficiency varies under different soil conditions, with a special focus on water availability which is one of the strongest controls on microbial processes in soil [30] . 1. 3 EXTRACELLULAR DNA IN THE ENVIRONMENT In soil systems, transformation is often assumed to be in frequent compared to other mode s of gene exchange such as conjugation and transduction . This is because extracellular DNA (eDNA) degrades rapid ly in soil , reducing substrate availability for transformation [31] . Ho wev er, eDNA can readily bind to clay and sand minerals , where it is protected from degradation but still availab le for transformation [22], [32] . This protective mechanism can generate large pools of eDNA , that in rare cases, exceed the intracellular DNA pool or the live fraction of the community [33] . This eDNA adsorbs to soil particles within 90 - 120 minutes of being release d into soil and can persist for 6 months to years depending on the environmental condit ions [22], [34], [35] . L aboratory studies show that eDNA concentration is a key driver of transformation, but it is unknown whether this translates to soil [36] . In vitro studies have shown that transformation increases in spatially structured environments (solid vs. liquid media) , but even these poorly mimic the heterogenous soil environment , where cell - cell contact is likely restricted and eDNA availability could limit transformation. In this case, 4 more eDNA might be required to achieve the same transformation ef ficiencies observed in the lab. Indeed, conjugation rates are suppressed through genetic drift in spatially structured environments [37] . However, gene transfer is also more pervasive between bacteria that inhabit the same environment [38] . Because large pools of eDNA in soils could overcome spatial barriers, quant ification of soil eDNA pools and their effect on transformation are critical for understanding bacterial evolution in nature. 1.4 EVOLUTIONARY ORIGINS OF TRANSFORMATION Whether or not bacteria can survive in the absence of HGT has been debated for years [39] [41] . Several theories exist to explain the evolutionary origins and potential benefits of transformation (re viewed in Seitz and Blokesch 2013) . 1) Incoming eDNA can be used as a nutrient res our ce, 2) eDNA can serve as a template for genome repair or 3) eDNA can increase genetic diversity via recombination . A new amendment to this final theory suggests that transformation facilitates genetic recombination to rid bacterial genomes of selfish genetic elem ents [43], [44] . While t ransformation is predominately cited as a mechanism that evolved to diversify bacterial genomes, the evolutionary origins, and consequences of transformation are idiosyncratic. The presence of cellular machinery dedicated to protecting transformed e DNA from degradation inside the cell, suggests that eDNA is not acquired purely for the nutrient benefit [42], [45] . In addition, b acteria use discriminatory mechanisms that allow them to preferentially kill and transform eDNA from close relatives implying certain eDNA is preferable [46], [47] . This provides further support for the theory that transformation promot es genetic exchange within bacterial species and may in some scenarios act like meiotic sex in eukaryo tes . Ho we ver, the most convincing evidence 5 that transformation is a mechanism of genetic recombination, comes from population genetic models, [11], [48], [49] and experimental evolution studies [50], [51] , which show that bacteria capable of transformation can have higher rates of population - level adaptation than non - transformers. Despite this evidence, t ransformation also presents increased risks of acquiring deleterious DNA or unnecessarily expanding a genome [52] . Seminal [53] . Ho we ver, experimental research suggests a more complex relationship bet we en fitness and transformation, in which the benefit s of transformation are highly dependent on the fitness landscape and the environment [50], [5 4], [55] . Transformation can be beneficial in stressful or novel environments, but the fitness benefits are often small [54] . A major gap in this bo dy of work is quantification of transformation efficiency after adaptive evolution, as p revious work has disproportionately focused on differences in population - level fitness between competent and non - competent lineages without screening for differences in transformability across evolved lineages [50], [51] . Consequently, it has yet to be determined if transformation is subject to selection , and furthermore, what selective forces could act to increase or decrease transformation efficiency in natural systems. 1.5 PSEUDOMONAS STUTZERI : MODEL SYSTEM Pseudomonas stutzeri is a highly transformable soil bacterium previously used to study transformation in soil [23], [28], [56], [57] . The highly transformable lineage was originally isolated from soil and acquires eDNA at a high efficiency under a variety of conditions. P. stutzeri is broadly relevant as it exchanges genes with several 6 opportunistic pathogens, including Pseudomonas aeruginosa and enterobacteria [58] . In addition , P. stutzeri represents an environmental vector for antibiotic resistance genes to spread to new pathogenic hosts in soil [59] . My dissertation research specifically explores the potential for P. stutzeri to act as a vector in the transformation of antibiotic resistance genes in soil . 1.6 RESEARCH QUESTIONS More research is needed to understand the occurrence and importance of transformation in soil. Thus, the goal of my dissertation is to improve our understanding of the environme ntal conditions that promote transformation in natural systems . I address the following research questions: 1. How does the availability of soil eDNA change over a drying rewetting pulse disturbance? 2. What soil conditions promote transformation and do these conditions coincide with eDNA availability (determined in Q1). 3. How does the growth rate and transformation efficiency of P. stutzeri change in response to salt adaptation ? 7 CHAPTER 2 : EXTRACELLULAR DNA MASKS MICROBIAL RESPONSES TO A PULSE DISTURBANCE 2.1 ABSTRACT A major goal in microbial ecology is to predict how microbial communities will respond to global change. However, DNA - based sequencing that is intended to characterize live microbial communities includes extracellular DNA (eDNA) from dead cells. This could obscure relevant microbial responses, particularly to pulse disturbances, which kill bacteria and have disproportionate effects on ecosystems. Here, I cha racterized bacterial communities before and after a drying rewetting pulse disturbance, using an improved method for eDNA exclusion. I found that eDNA removal was important for detecting subtle yet significant changes in microbial abundance, diversity, and community composition across the disturbance. However, eDNA removal was less important for detecting differences between crop types and disturbance regimes. T he size of the eDNA pool mainly a ffected estimates of bacterial abundance, while eDNA pools enric hed in unique sequences (e.g. sequences not found in the live community), mainly a ffected estimates of community structure irrespective of size. Consequently, eDNA inclusion made bacterial communities appear as though they had the same structure before a nd after the disturbance, when the live fraction was in fact different. Overall, pulse disturbance studies have a high risk of eDNA bias and should remove eDNA to improve predictions of ecosystem responses in future climates. 2.2 INTRODUCTION Large pools of prokaryotic extracellular DNA (eDNA) can accumulate in the environment as bacteria die [23], [26] . This eDNA is included in molecular 8 abundance, di versity, and community composition. Despite this danger, we know surprisingly little about the conditions that make microbial characterizations vulnerable interpretation of th e result is altered by eDNA removal (i.e. if eDNA alters the ability to detect a change or the size of that change). Experiments testing bacterial responses to lethal disturbances could be especially susceptible to eDNA bias because bacterial death can inc rease necromass - derived eDNA . Here, I test whether eDNA alters our ability to characterize bacterial abundance, diversity, and community structure following soil drying rewetting, a stressful and likely lethal, pulse disturbance [60], [61] . To extend the generality of these findings, I also examine why eDNA bias emerges. Based on previous studies, I hypothesize that eDNA is most likely to bias results when eDNA pools are (i) large and (ii) enriched in sequences not present in the live community [33], [62] . To test this, I sampled soil bacteria that were subject to extreme drying rewetting in the field for 6 months. Rainout shelters allowed us to impose a moisture extreme outside of historical norm s for the region (28 days of drought and 80 mm rain event). I collected soil samples at 3 timepoints across a drying - rewetting event: at the end of the drought (~6hrs before rewetting, t1), 1hr after rewetting (t2) and 18hrs after rewetting (t3). Rapid rew etting after long dry periods has been shown to cause bacterial death [63] [65] , so I expected eDNA pools to increase in the first hour after rewetting, but to decline after 18hrs, as increases in soil moisture stimulate eDNA decay [66] . I repeated this sampling in contrasting soils (conventionally - tilled corn monoculture and perennial 9 switchgrass monoculture), since differences in soil moisture and bacterial communitie s could affect eDNA pool dynamics and thus sensitivity to eDNA bias [67] [69] . To isolate the effects of eDNA, I eliminated eDNA pools in one of two pair ed soil samples using the chemical propidium monoazide (PMAxx) which binds to eDNA and prevents amplification (modifying and improving the efficacy of methods in Carini et al. [33] , Fig ure A 2.1). I sequenced the 16S rRNA gene to characterize the live and extracellular fractions of the soil bacterial communit y. 2. 3 MATERIAL AND METHODS 2. 3 .1 SITE AND SOIL COLLECTION I exposed soil bacterial communities to ambient drying rewetting (6.6mm of rain every 3 days) and extreme drying rewetting (80mm of rain every 28 days) between April 3 rd and September 17 th in experimental Biofuel Cropping System plots located at the Kellogg Biological Station in Southwes a.s.l.). On September 20 th of 2017 I collected soil cores (10cm depth by 5cm diameter) from continuous corn ( Zea mays ) and continuous switchgrass ( Panicum virgatum, - in - plots at three - time poin ts; 6 - hrs before, 1 - hr after, and 18 - hrs after a rewetting event, which also coincided with the end of a 28 - day drought. The soils are well - drained mesic Typic Hapludalfs developed from glacial till and outwash consisting of co - mingled Kalamazoo (fine - loam y, mixed, semiactive) and Oshtemo (coarse - loamy, mixed, active) series (Crum and Collins 1995) with intermixed loess [70] . Bulk soil was brought back to the lab, sieved at <2mm and homogenized. Gravimetric soil moisture was determined on sieved soils dried at 60°C for 72 hrs. Soil moisture 10 increased in response to the extreme rewetting event ( Fig ure 2.1A : Corn p<0.0001; Switchgrass p=0.035), while ambient rewetting had no effect on soil moisture (data not shown). 2. 3 .2 REMOVAL OF EXTRACELLUALR DNA Methods for quantifying eDNA were modified from Carini et al. [33] . Two DNA extractions were completed for each soil sample using the DNeasy PowerSoil kit from Qiagen (previously PowerSoil DNA Isolation Kit by MO BIO). A paired sample design was used where one DNA extraction tube was treated with the chemical propidium monoazide (PMAxx Dye, 20mM in H 2 O from Biotium) which prevents amplification of eDNA. The other DNA extraction was left untreated to quantify the total DNA pool . B oth tubes wer e treated identically through the PMAxx activation steps. For each soil sample, 0.50 grams of soil was weighed out and divided evenly between the two DNA extraction tubes (0.25 g each). Prior to loading soil into the DNA extraction tubes, I removed the bea d beating beads from sample tubes . The beads were removed for the steps involving light activation of PMAxx to prevent potential cell lysis. Then 500µl of the Powerbead solution (solution that comes in the tubes with the beads) was returned to the tube. In a dark room, 3µl of PMAxx was added to the 500µl of PowerBead Solution in the transparent bead beating tubes. Samples were homogenized for 1 minute and exposed to a 1000 W halogen light source for ten minutes while undergoing frequent homogenization ( 1000 - Watt Halogen Telescoping Twin Head Tripod Work Light in a tinfoil lined cabinet). After light exposure, I returned the Powerbeads and the remaining PowerBead solution (~250µl) to the DNA extraction tube. The DNA was then extrac ted 11 handled in low light conditions minimal ambient light from windows to minimize the binding of PMAxx to DNA released during the DNA extraction. The concentration of P MAxx was selected for this experiment because it optimized eDNA removal at higher concentrations of eDNA . H owever, this also led to increased variation in estimates of eDNA in smaller eDNA pools ( Fig ure A 2.1 ). In addition, previous work has shown that agricultural soils have low eDNA concentrations compared to deciduous or coniferous forest soils (Carini et al. 2016) . Indicating future studies should optimize the PMAxx concentration to their specific study. I n general , I recommend complet ing the DNA extraction ( following PMAxx activation ) in low light conditions or a dark room, as this reduc es the probability that excess PMAxx (which is light activated) will bind to intracellular DNA that is released in the fi rst step of the DNA extraction. 2. 3 .3 QUANTITATIVE PCR I performed quantitative PCR in triplicate using 96 - well plates on an Thermofischer thermocycler. I used 3 technical replicates for each PMA - treated and un - treated soil sample ( Fig ure A 2.1C ). The total reaction volume was 20 µl with the following reaction mixture: 1 µl of each F and R primer (515f/806r at 10mM), 10 µl of iQ SYBR Green 2X Supermix (BIO - RAD), 4 µl sterile water, and 4 µl template DNA. The cycling conditions wer e: 95 °C for 15 min, followed by 40 cycles of 95°C 30 s; 50 °C 30 s; 72 °C 30 s. I generated melting curves for each run to verify product specificity by increasing the temperature from 60 - 95°C. Reactions were compared to standard curves developed using purified Pseudomon as stutzeri 28a24 genomic DNA. For all qPCR reactions the linear relationship between the log of the copy number and the threshold cycle value was reported at R 2 > 0.99. Outlier analysis was performed on qPCR 12 replicates. I removed qPCR replicates that were 2 cycles above or below the other two replicates this resulted in the removal of 1 qPCR replicate from 3 samples (samples 2_PMA, 6_Total and 42_PMA in Fig ure A 2.1C (denoted by *)). Two soil samples with negative eDNA estimates were removed from the thir d time point analysis in both corn and switchgrass soil as they fell outside of the standard curve and were biologically inaccurate (sample 34 and 44, highlighted with gray bars in Fig ure A 2.1C ). Across the entire data set, soil samples treated with PMAxx had lower variation across qPCR replicates than un treated soil samples ( Fig ure A 2.1B ). 2. 3 .4 AMPLICON SEQUENCING AND BIOINFORMATICS I characterized bacterial communities using high throughput barcoded sequencing on the Illumina MiSeq platform at the Research Technology Support Facility (RTSF) Genomics Core at Michigan State University. Sequencing was done in a 2 × 250bp paired end form at using a MiSeq v2 500 cycle reagent cartridge. The V4 hypervariable region of the 16S rRNA gene was amplified using dual - index, Illumina compatible primers 515F and 806R as described in Kozich et al. [71] . Completed libraries were normalized using Invitrogen SequalP rep DNA Normalization plates, then pooled and cleaned up using AmpureXP magnetic beads. The 16S reads were quality filtered and merged using the USEARCH pipeline ( http://drive5.come/usearch/ ) and Cutadapt was used to remove prim ers and adapter bases before reads were filtered and truncated to 250bp. OTUs were clustered at the 97% sequence identity level and classified using UPARSE. Two samples with low read number were removed, making the lowest read number 7723. Singletons, chlo roplasts, and mitochondria were removed (312 total) resulting in 15,062 bacterial OTUs and 2,259,918 reads across all samples. 13 All analyses were performed and visualized using bacterial counts that did not undergo rarefaction [72] . 2. 3 .5 STATISTICAL ANALYSES I tested for differences among univariate response variables (e.g. soil moisture, Shannon diversity, and bacterial abundance or number of 16S rRNA gene copies) using Type III sum of squares ANOVA with a post - hoc Tukey HSD correction test (p < 0.05) using the lme4 and agricolae pac kages in R . For all analyses , I report average value s across the four field replicates, except for the qPCR data, where I first calculated the number of 16S rRNA gene copies for each technical replicate (n=3 qPCR replicates, Fig ure A 2.1C ). I calculated the percent eDNA by subtracting the number of 16S rRNA gene copies in the live fraction from the total number of copies, and then dividing by the total number of copies [(Total - Live)/(Total)*100]. I compared statistical differences in bact erial community composition using permutational analysis of variance (PerMANOVA) (9999 permutations) on weighted Unifrac distance matrices (Bray - Curtis distance) using the Adonis function in the vegan R package. I visualized community composition using Non - metric multidimensional scaling (NMDS) using phyloseq and ggplot2. I examined the proportion of taxa present in both the liv e community and eDNA pool using the Venn Diagram function in R. To assess shared membership in the eDNA pool I performed a Venn Dia gram analysis on each field replicate and calculated the mean proportion of taxa specific to the eDNA pool at each time point. I report the proportion of taxa in this analysis instead of the raw number of taxa to account for changes in the total number of taxa across the disturbance . To estimate the effect of eDNA removal, I calculated standardized effect 14 sizes on soil samples with and without eDNA, values [73] using the effsize package in R. 2. 4 RESULTS AND DISCUSSION Understanding when eDNA affects microbial community characterizations is important for accurately linking microbial dynamics to belowground processes [68], [74] . By comparing analyses in which eDNA was excluded to samples in which it was included (the approach of most studies), I w as able to identify the microbial responses most sensitive to eDNA bias in two different soils informing when future studies might invest in eDNA removal. Overall, I found that eDNA masked changes in bacterial community composition and abundance that occurred across drying - rewetting in corn soil ( Table 2.1, Fig ure 2.1B,D,F ). In contrast, eDNA inclus ion did not mask responses to drying - rewetting in switchgrass soil, but there were fewer responses to mask, as soil moisture was more stable and may have buffered communities from drying - rewetting stress ( Fig ure 2.1A ). T he disturbance responses most vule nrable to eDNA bias differed from expectation. I expected eDNA bias to increase after bacterial death (timepoint t1 to t2), instead it was driven by the successional recovery of the disturbance - effected community (t1 - t3 in corn soil, see Table 2.1, Fig ure 2.1A ).This post - disturbance community (t3) was larger and compositionally distinct from the drought community (t1) . However, these signatures of bacterial resilience were not detected in analyses that included eDNA, and consequently made it appear as thoug h corn communities were largely unaffected by the disturbance ( Table 2.1 ). 15 Fig ure 2.1: Effect of eDNA removal on bacterial responses to drying rewetting. (A) Soil moisture (B) a verage bacterial abundance (16S rRNA genes) and (D) average Shannon diversity including eDNA (dotted lines) or excluding eDNA (solid lines). (F) NMDS plot showing the composition of bacterial communities with eDNA (open points) or without eDNA (opaque poin ts) , with lines connecting paired soil samples. (C - G) Effect size for changes in (C) abundance, (E) diversity and (G) community composition includ ing eDNA (open points, dotted lines) and exclud ing eDNA (opaque points, solid lines) with lines showing the 95 % confidence interval. P ositive effect sizes correspond to a larger value in switchgrass soil or an increase across the disturbance. The third panel shows the magnitude of change in effect size after excluding eDNA (A - G; n=4 field replicates). 16 Treatment Measure Time Effect Size Rank ( - ) (+) Corn Abundance t2 - t3 1.939 2 0.003 0.16 Corn Diversity t1 - t2 0.76 10 0.007 0.047 Corn Diversity t1 - t3 2.63 1 0.0004 0.05 Corn Composition t1 - t2 0.14 19 0.05 0.076 Corn Composition t1 - t3 0.61 13 0.033 0.085 Overall, I did not find support for hypothesis (i ); large eDNA pools did not drive eDNA bias, but I did find support for hypothesis (ii). That is, eDNA bias was largely driven by a divergence between the live community and eDNA pool (except for changes in abundance which are inherently impacted by size). Here, eDNA bias increased when the eDNA pool was enriched in sequences not in the live community even when the eDNA pool was small (compare Fig ure 2.2A, B ). It may be that as the live community shifted in response to rewetting, the composition of the eD NA pool continued to reflect the drought community ( Fig ure 2.2A, B , see expanded results). This caused a divergence between the live community and the eDNA pool that masked a significant shift in the live fraction of the corn community post - disturbance ( Fi g ure 2.2B ) and alter ed conclusion s about which taxa were sensitive to the disturbance ( Table A2.2 ). In addition to masking signficant results, eDNA also has the potential to create false positives as was evident in switchgrass soil where eDNA inclusion inflated changes in bacterial diversity and community composition across the disturbance ( although not signficantly ) . In switchgrass communities eDNA inclusion generally inflated the effect of the d isturbance ( Fig ure 2.1E,G ) , making switchgrass communities Table 2.1 : Results impacted by eDNA bias. S ignificant results that were masked by eDNA inclusion or already significant results that became more significant after eDNA removal. The rank refers to the effect of eDNA removal with 1 being the largest change in effect size after eDNA removal. ( - ) corresponds to p - values without eDNA and (+) corresponds to p - values with eDNA. 17 appear less resilient to drying - rewetting than was true in the live - fraction of the community the opposite effect of eDNA in corn soil . Not only were trends inconsistent across c rop type , but eDNA inclusion skewed trends across the disturbance . For instance, diversity decreased in corn soil in the 18hrs after rewetting, however , eDNA inclusion created the perception that diversity increased. To further complic ate matters, the effect of eDNA removal (as measured by effect size) was often of similar magnitude in corn and switchgrass communities, despite eDNA only masking siginficant responses in corn soil . Ultimately, highlight ing how difficult it is to predict the magnitude and direction of change associated with including eDNA in microbial community chracterizations. Overall, our study suggests that eDNA removal is ne cessary for understanding microbial response s to short - term or pulse disturbances, particularl y when changes in the live community may be subtle and occur over rapid timescales. However, eDNA removal may be less important for detecting large differences across land use types, as Fig ure 2.2 : Changes to the eDNA pool in response to drying rewetting. (A) The eDNA (%) in response to soil rewetting for corn (orange circles) and switchgrass (green triangles) soil. (B) The number of eDNA - specific sequences (those found only in the eDNA pool at that sampling point) across the disturbance . (A - B) Bars and points show the average and error bars show the standard error (n=4 field replicates). 18 even across the disturbance, corn and switchgrass communities were com positionally distinct irrespective of eDNA removal. Since many global change studies are interested in the resistance and resilience of specific micorbial communities, including under more varied rainfall, pulse disutrbance studies should remove eDNA, as this will improve predictions of bacterial resilience and ecosystem function in future climates. 19 APPENDIX 20 APPENDIX 2. 5 SUPPLEMENTAL RESULTS In general, corn communit ies w ere more sensitive than switchgrass communit ies to the rewetting disturbance. This was evidenced by the significant drop in diversity from t1 - t2 in corn soil. As well as a 56% decline in the number of 16S copies during this same window in corn soil and contra sted a 6% decline in switchgrass soil. Consequently, eDNA increased 34% in corn soil from t1 - t2 ( Fig ure 2.2A ), even though the number of 16S copies in the eDNA pool actually dropped in both soils in response to rewetting ( Fig ure 2.1B ). This jump from 24 - 58 % eDNA is the largest in the dataset and masks a marginally significant shift in community composition ( Table 2.1 ). Then between t2 - t3 in corn soil, the number of 16S copies in the live fraction of the community increased significantly ( Fig ure 2.1B ), far o utweighing the decline that followed rewetting. This signature of disturbance recovery is undetectable in analyses that include eDNA though , as eDNA inflated the size of the corn community at t1 ( Fig ure 2.2B ) making it appear as though the corn community was the same size before and after the disturbance. At t3, only 5% eDNA remain ed in corn soil, yet the eDNA pool was more divergent from the live community than at the previous two sampling points (Fig ure 2.2A,B) . This suggests that the risk of eDNA bias is not solely caused by large eDNA pools. I predict that following the disturbance , only certain bacterial taxa were able to recover and increase in abundance . Thus, creating a community that look ed quite different than the t1 drought community. With only ~600 taxa remaining in the corn community at t3 in contrast to ~1400 at t1 it may be that rapid responders, likely with 21 a high 16S copy number, are in the first stages of a successional recovery event. However, this ecologically relevant and functiona lly important disturbance dynamic, is completely undetectable in analyses that include eDNA making it appear as though bacterial abundance and community composition are the same before and after the disturbance. 22 Figure A2.1: Technical aspects of eDNA removal. (A) Standard curve for the PMA - treatment in mock communities of live and dead Pseudomonas stutzeri inoculated into switchgrass soil ( Panicum virgatum, - in - eDNA (%) in each mock community regressed against the expected eDNA (%), error bars show the standard error of the mean and the dashed line corresponds to x=y (n=3 soil replicates and n=3 qPCR replicates). ( B ) Average number of 16S rRNA gene copies in the PMA - treated (red circles) and untreated soil (black triangles) samples across the time series for corn and switchgrass soil with the error bars showing the standard error (Corn n=46 and Switchgrass n=47). ( C ) The average number of 16S rRNA gene copies across the 3 technical replicates (n=3 qPCR replicates) for each PMA - treated and untreated soil sample. The corresponding point in the time series and its crop type are shown below the plot. Two samples that fel l outside the standard curve were removed and are highlighted in gray . Red asterisks indicate a PMA - treated technical replicate was removed and a black asterisk indicate an untreated technical replicate was removed. 23 Effect Size p - value Treatment Measure Time ( - ) (+) rank ( - ) (+) C vs S Abundance C vs S 2.23 1.52 0.71 11 C Abundance t1 - t2 - 0.957 - 0.81 0.147 18 S Abundance t1 - t2 - 0.488 - 0.903 0.415 16 C Abundance t2 - t3 2.93 0.991 1.939 2 0.003* 0.16 S Abundance t2 - t3 0.499 0.021 0.478 14 C Abundance t1 - t3 1.16 - 0.29 1.45 4 S Abundance t1 - t3 - 0.09 - 0.88 0.79 9 C vs S Diversity C vs S 1.47 1.55 0.08 21 C Diversity t1 - t2 - 2.45 - 1.69 0.76 10 0.007 0.047 S Diversity t1 - t2 - 0.385 - 1.18 0.795 8 C Diversity t2 - t3 - 0.39 0.436 0.826 7 S Diversity t2 - t3 0.72 0.046 0.674 12 C Diversity t1 - t3 - 4.29 - 1.66 2.63 1 0.0004 0.05 S Diversity t1 - t3 0.51 - 1 1.51 3 C vs S Composition C vs S - 0.77 - 0.35 0.42 15 C Composition t1 - t2 - 0.64 - 0.78 0.14 19 0.05 0.076 S Composition t1 - t2 0.63 - 0.49 1.12 6 C Composition t2 - t3 0.99 0.74 0.25 17 S Composition t2 - t3 - 0.8 0.58 1.38 5 C Composition t1 - t3 0.38 - 0.23 0.61 13 0.033* 0.085 S Composition t1 - t3 - 0.05 0.04 0.09 20 Table A 2.1 : Expanded table of e ffect sizes and p - values for eDNA removal. Complete list of effect sizes and p - values for changes in abundance, diversity, and community composition across the disturbance. C corresponds to corn and S to switchgrass. The rank refers to t he effect of eDNA removal with 1 being the largest change in effect size after eDNA removal. ( - ) corresponds to p - values without eDNA and (+) corresponds to p - values with eDNA. Results are listed in the same order as in Figure 2.1C, E, G. P ositive effect s izes correspond to a larger value in switchgrass soil or an increase across the disturbance (t2 higher than t1). 24 Taxa Driving Responses to Soil Rewetting ( - ) eDNA excluded (+) eDNA included Corn Bulk Soil Mizugakiibacter Mizugakiibacter Sphingomonas Thermosporotrichaceae Thermosporotrichaceae Rhodanobacter Acidobacterium Sphingomonas Xanthomonadaceae Acidobacterium Bryobacter Bryobacter Bryobacter Chitinophagaceae Acidothermus Mizugakiibacter Sphingobium Chitinophagaceae Sphingomonas Castellaniella Deltaproteobacteria (GR_WP33 - 30) Bryobacter Xanthomonadaceae Acidobacteriaceae_[Subgroup_1] Acidobacteria (Subgroup_6) Acidobacteriaceae_[Subgroup_1] DA101_soil_group Acidobacteriaceae _[Subgroup_1] Sphingomonas Bradyrhizobium Switchgrass DA101_soil_group unclassified Sphingomonas Bradyrhizobium Holophagae Luteolibacter Nitrosomonadaceae Sphingomonas Acidobacteria (subgroup_4) Flavobacterium Acidobacteria (subgroup_4) Cellvibrio Acidobacteria (subgroup_4) Acidobacteria (subgroup_12) Acidobacteria (subgroup_12) Haliangium Sphingomonas Variibacter Haliangium Pseudomonas Acidobacteria (subgroup_6) Chloroflexi Reyranella Latescibacteria DA101_soil_group Acidobacteria (subgroup_6) Acidobacteria (subgroup_12) DA101_soil_group Acidobacteriaceae_[Subgroup_1] DA101_soil_group Table A2.2: The top 15 taxa driving changes in response to soil rewetting . Dominant taxa driving shifts across the 3 sampling points when eDNA is excluded and included in analyses. Taxa or OTUs that are incorrectly identified in the presence of eDNA are italicized. 25 CHAPTER 3: HORIZONTAL GENE TRANSFER FACILITATES THE SPREAD OF EXTRACELLULAR ANTIBIOTIC RESISTANCE GENES IN SOIL 3.1 ABSTRACT It is now clear that the environment plays a major role in the dissemination of antibiotic resistance genes (ARGs), which are ubiquitous in the environment and pose a serious risk to human and veterinary health. While many studies focus on the spread of live antibiotic resistant bacteria to the environment , less is known about the contribution of extracellular ARGs to the evolution of antibiotic resistance in natural s ystems . In this study, I inoculate antibiotic - free soil with extracellular ARGs ( eARGs ) from dead Pseudeononas stutzeri cells and track the evolution of antibiotic resistance via natural transformation a mechanism of HGT involving the genomic integration of eARGs. I find that transformation facilitates the rapid evolution of antibiotic resistance even when eARGs are rare ( 0.25 µ g g - 1 soil ) . Ho wev er, when eARGs are abundant , transformation increases substantially. In general , transformation occurred under most soil conditions tested and was only inhibited at very high soil moistures (>30%) . Finally, I show that transformed eARGs are just as successful as live antibiotic resistant invaders , when challenged with a low dose of antibiotic . Overall, this work demonstrates that dead bacteria are an overlooked path to antibiotic resistance, and that disinfection alone is insufficient to stop the spread of ARGs . More generally, th e spread of ARGs in antibiotic - free soil , suggests that transformation allows genetic variants to establish at low frequenc ies in t he absence of selection. 26 3.2 INTRODUCTION Antibacterial resistance is a global threat to public health and could have a higher death toll than cancer by 2050 [75] . To reduce the impacts of antibiotic resistance on human health, we need to under stand how antibiotic resistance genes (ARGs) move through the environment [76], [77] . However, t he evolution of a ntibiotic resistance has traditionally been viewed as a clinical problem , and consequently less is known about when and how novel antibiotic resistant pathogens emerge from natural systems [78], [79] . ARGs in the environment are particul arly concerning because they pose a significant threat to food and water resources [77] and can spread to new hosts through horizontal gene transfer (HGT) [79] [82] . The spread of ARGs via HGT is a major mechanism in the rise of antibiotic resistance [7] but the environmental variables that promote the transfer of ARGs remain poorly understood [76] , despite well - documented instances of ARGs moving from the environment to the clinic [83] [85] . An important, but often overlooked source of environmental ARGs is extracellul a r DNA (eDNA) [86], [87] . Ext racellular ARGs (eARGs) enter the environment through active secretion or bacterial death, and once there can integrate into new bacterial genomes through a mechanism of HGT called natural transformation. Soil harbors one of the largest environmental res e r voirs of e ARGs [88], [89] and is home to many antib iotic producing bacteria that could select for the maintenance of e ARGs in new hosts [79] . Since eARGs can persist in soil for more t han 80 days in comparison to less than 1 day in aquatic environments the odds of a transfer even t and subsequent spread , are high in soil [34] . Overall, u nderst anding the occurrence of these transfer events will be important for the early detection of multi - drug resistance, especially since 27 11 of the 12 priority antibiotic resistant pathogens are known or predicted to be naturally transformable [90] . In soil, transformation is likely to depend on the concentration of eARGs, as transformation generally increase s with the availability of eDNA when measured in the lab ( Fig ure A 2.1 A ). However, transformation may proceed differently in complex environments like soil. For instance, soil is spatially heterogenous and spatial barriers have been shown to limit the transfer of plasmids via conjugation [37] . In addition, eDNA can persist in soil for longer than other environments , but its stability is still likely to depend on soil properties like water content [66] . These same soil properties may also impact the tendency for eDNA to undergo transformation via effects on recipient cells . In particular, t he conditions which favor eDNA stability may differ from those that favor Fig ure 3.1 : S oil characteristics likely to affect natural transformation . (A) T ransformation requires cellular competence and the presence of eARGs. But is also likely to depend on the soil moisture, soil structure, and proximity to eARGs. Arrows show possible interactions between soil characteristics and point t owards the effected variable. ( B ) In this conceptual framework, a novel A RG enters soil via a liv e invader (i) or via transformation (ii). In the transformation scenario, the blue and green cells susceptible to the antibiotic acquire the ARG via transforma tion which allows the ARG to move into multiple genetic backgrounds. This preserv es community - level genetic diversity that would have been lost if a single antibiotic resistant invader rose to fixation in the presence of an antibiotic. 28 competence the physiology state of transforming cells. For instance, wetter soils may favor active growth which often induces competence, while in contrast eDNA degradation increases with water availability and could prevent transformation . Finally , the availability of moisture could directly impact processes like biofilm formation which is known to facilitate gene transfer [91], [92] . See Fig ure 3. 1A for a conceptual diagram of soil characteristics likely to influence transformation. The conditions that promote transformation in soil could be rare, but even infrequent transformation events coul d lead to the emergence of novel drug - resistance , especially under a selective pressure [93], [94] . For this reason, antibio tics in the environment have gained increasing attention over the last decade [8] . In fact, if an antibiotic provides a strong enough selective pressure, transformed eARGs could reach high abundances in a population . While liv e antibiotic resistant cells will generally establish more successfully than an equal abundance of eARGs (DNA), due to the rarity of transformation, a strong enough selective pressure could equal the playing field such that eARGs are just as successful as live invaders. T ransfor med alleles also have the added advantage of moving into multiple genetic backgrounds [39], [40], [95] , this ultimately preserv es community - level genetic diversity and chang es population - level outcomes (see Fig ure 3. 1B ). Th e transfer of e ARGs into multiple genetic backgrounds is often facilitated by mobile genetic elements , which are a major driver in the emergence of multidrug resistance [96], [97] . These genetic parasites insert themse lves into bacterial genomes and generally appear at high abundances in soils enriched in ARGs [98], [99] . While transformation can disseminate mobile gen e tic elements into unrelated bacteria at high 29 frequencies , the prevelance of this phenomenon is poorly characterized outside of laboratory conditions. Here, I address this knowledge gap, by incoulat ing agricultural soils with eARGs carried on a mobile genetic element ( miniTn7 transposon ) , and track the evolution of antibiotic resistance into populations of Pseudomonas stutzeri a model organism for studying transformation in soil (Sikorski et al. 1998 , see Fig ure 3. 2 for the experimental design). Overall, I show that eARGs from dead bacteria are an important, but often overlooked source of antibiotic resistance in natural systems . Specifically, I find that the availability of eARGs drives the evolution of antibiotic resistance and that transformation is prevalent under a wide range of soil conditions only decreasing at very high soil moistures and in homogenized soils. More broadly though , I show that antibiotic resistant transfor mants repeatedly establish in antibiotic - free soil . Together, this provides novel in situ evidence that HGT is an evolutionary force that expands the adaptive potential of bacterial communities by facilitating the spread of non - selected antibiotic resistan ce genes. 3.3 M ETHODS 3.3.1 SITE AND SOIL COLLECTION Soil cores (10cm depth by 5cm diameter) were collected in October 2018 and April 2019 from the Great Lakes Bioenergy Research Center (GLBRC) scale - up fields located at Lux Arbor Reserve Farm in south west were established as perennial switchgrass monoculture s ( Panicum virgatum L ) in 2013, and before that were in a corn soybean rotation for more than 10 years. The soils developed on glacial outwash and are classified as we ll - drained Typic Hapludalf, fine - 30 loamy, mixed, mesic (Kalamazoo series) or coarse - loamy, mixed, mesic (Oshtemo series) or loamy sand, mixed, mesic (Boyer series) [100] . S oils were sieved at 2mm and autoclaved in two cycles (60 minutes at 121°C; gravit y cycle) separated by a 24 - hr window to target dormant and spore - forming cells resuscitated during the first autoclave cycle. 3.3.2 BACTERIAL CULTURES AND EXTRACELLULAR DNA S oil microcosms were inoculated with Pseudomonas stutzeri , strain 2 8 a2 4 [101] . Prior to inoculation, the bacterial cultures were grown at 30°C on an orbital sh aker (120 rpm) for 24 h r s in liquid luria broth (LB) media to a concentration of 10 6 CFU/mL. All LB media used throughout the experiment followed a recipe of 10% tryptone, 5% yeast extract, and 5% NaCl ( solid media contained 1.5% agarose ) . Stocks of antibiotic resistant e xtracellular DNA (eARGs) w ere made from a mutant P . stutzeri , strain encoding a gentamicin resistance gene and a LacZ gene (Tn7 transposition of pUC18 - mini - Tn7T - Gm - lacZ into strain 28a24 , see Choi and Schweizer 2006) . eDNA was also made from the wildtype P. stutzeri to act as a negative control . The gentamicin resistant P. stutzeri cells were genetically identical to the wildtype P. stutzeri cells, except for the presence of the antibiotic resistance gene . The batch cultures for eDNA stocks were prepared under the sam e conditions specified above bu t were grown for 48hrs and then resuspended in sterile nanopure water. The cells for eDNA stocks were killed via heat shock ( 90°C for 1hr ) and confirmed dead by plating. The final concentrations of eDNA ranged from 25 - 50 ng and w as appropriately diluted for each experiment (DNA concentrations were determined using Q ubit fluorometric quantification and Invitrogen 31 Quant - iT PicoGree n dsDNA Assay Kit ). transformation efficiency plateau s at ~5ng eDNA under standard laboratory conditions ( Fig ure A 3. 1 A ). 3.3.3 SOIL MICROCOSMS Soil microcosms were established in small 60 x 15mm petri dishes using 10 grams of dry, sterile , switchgrass soil. Except for the experiment in Fig ure 4, which used large 150 x 15mm petri dishes filled with 100 grams of soil. On day 0 of each experiment, the center of the microcosm was inoculated with 2mL of wildtype P. stutzeri cells suspended in liquid LB at a concentration of 10 6 CFU/mL. Immediately after adding cells, I inoculated the soil with eDNA encoding gentamicin resistance. To control for contamination or evolution of gentamycin resistance via mutation, two additional treatments were included in every experiment; 1) 5 µg of eDNA g - 1 soil made f rom gentamicin susceptible P. stutzeri cells, and 2) sterile water without eDNA . Since transformants never appeared in the control treatments the results are not shown . All microcosms were maintained at ~23°C and the soil was never mixed unless directly specified ( e.g. in Fig ure 3. 4 B ) . All microcosms were initially inoculated to ~40% soil moisture on day 0, and then dried to 20% soil moisture, where they were Fig ure 3.2 : General experimental design for soil microcosms . At time zero soil was inoculated with live antibiotic susceptible P. stutzeri and dead antibiotic resistant P. stutzeri providing a source of eARGs. Soils were then exposed to a variety of soil conditions and the number of transformant s periodically co unted by resuspending soil in a slurry and plating onto selective media. 32 maintained until the next eDNA addition ( except in the experiment s manipulating soil moisture where the soil was dried according to the treatment - level soil moisture) . In each experiment, I counted the number of transformants, and the population size e very 5 days , and then added more eDNA to simulate periodic inputs of eARGs. After eDNA additions, I gradually dried the soils over the next ~4hrs back to 20% soil moisture. To establish a baseline for transformation in th e soil system, I quantified transformation every 24hrs for 5 days after inoculation with eDNA . I ran parallel assays on petri dishes using LB and using a concentration of 5µg eDNA g - 1 soil. Each treatment had 8 replicates ( Fig ure 3. 3 A ) . To understand the relationship between the availability of eDNA and transformation, I varied the concentration of eDNA in soil bet we en 5, 2.5, 1.25, 0.25 µg of eDNA gram - 1 soil ( Fig ure 3. 3 B, C, D ) . I used these concentrations as they conservatively represent ~10%, 5%, 2.5% and 0.5% of the total eDNA pool in soil per [103] . I kept the concentration of eDNA low, as only a small percentage of eDNA will generally carry antibiotic resistance genes. The experiment lasted 15 days and eARGs wer e added on day 0, 5, and 10 , with e ach treatment consist ing of 8 replicate soil microcosms. To determine how soil moisture a ffected transformation , I maintained soil microcosms at 5, 10, 20, 30 or 40% gravimetric soil moisture over a period of 10 days ( Fig ure 3. 4 A , i.e. soil moisture = [ we ight after water addition dry wei ght] / dry wei ght ]*100 ). In this experiment, all the microcosms were inoculated with an intermediate concentration of eDNA (2.5µg g - 1 soil ) and eARGs were added on day 0 and 5. I report the number of transformants present on day 10 , us ing 8 replicate 33 microcosms per treatment. To understand if the physical structure of the soil was important for transformation, I manipulated the physical structure of the soil by mix ing the soil eve ry 2hrs, 8hrs or never , throughout a 48hr period ( Fig ure 3.4 B ) . Mixing was carried out using a sterile spatula. In each microcosm, I kept the eDNA concentration constant at 5 µg g - 1 soil and the soil moisture constant at 10% , with e ach treatment consist ing of 4 replicates. To determine if transformation was dispersal limited under different soil moisture s , I established 8 pools of eDNA in large microcosms maintained at 10, 20, 30 or 40% soil moisture ( Fig ure 3.5 ). Half of the eDNA pools contain ed e ARGs ( P. stutzeri + gentamicin resistance), and the other half did not ( P. stutzeri wildtype). The 8 pools were located 1.25 , 3.80 , 5 or 7 cm from the center of the plate. Each eDNA pool was inoculated with 2 µ g eDNA g - 1 soil (400 µl total volum e) and P. stutzeri was inoculated to the center of the microcosm (2mL total volume) . To prevent dispersal during inoculation, I dripped the eDNA and bacterial cells i nto the soil in 200µl aliquots. The experiment ran for 5 days , using 4 replicate microcosms per soil moisture . At the end of the experiment , soil was collected from the center of each eDNA pool to count the number of transformants and total cells . 3.3.4 ANTIBIOTIC RESISTANCE GENES IN LIVE VS. DEAD CELLS In a final laboratory experiment, I tested how an equal concentration of eARGs and live antibiotic resistant invaders s of P. stutzeri subjected to three different selecti ve regimes ( Fig ure 3.6 ) . Initially, I established two equal population s of kanamycin resistant P.stutzeri cells (DAB837 in [104] ) . To one of the two Kan R population s , I added 60,000 live g entamicin resistant P. stutzeri cells ( Gent R ). To 34 the other population , I added 60,000 dead Gent R cells which provided a source of eARGs . Therefore, o n day 0 of the experiment, the two treatments contained 4% Gent R cells and 0% Gent R cells, respectively. The kanamycin and gentamicin resistance genes were both carried on a miniTn7 transposon, which allowed me to track the frequency of gene replacement (just Gent R ) vs gene addition (Gent R + Kan R ) in transformed cells ( Fig ure 3.6B ). I performed parallel experiment s in liquid LB under 3 selective regimes: 0%, 10% or 25% of the lethal dose of gentamicin (equivalent to 0, 5 or 12.5 mg/ml gentamicin, respectively). Populations were founded in 1mL of media and the gentamicin added at t=0 . I ran the experiment for 10 days, providing 1 mL of fresh LB media to each population every 24hrs . On day 5 , I removed the selective pressure and transferred the populations at a 1:4 dilution to liquid LB with no gentamicin . Each day I counted the number of Kan R and Gent R genotypes using serial dilution and selective plating of 10µl dots . I counted the total number of cells on solid LB media (no antibiotic) , the number of Gent R cells on solid LB media with gentamycin (50 µ g/ml) + Xgal (20 µg/ml) , and the number of Kan R + Gent R cells on solid LB media with kanamycin ( 50 µg/ml ) + gentamycin (50 µ g/ml) + Xgal ( 4 0 µg/ml). I report the frequency of Gent R genotypes (Gent R cells/total cells) every 24hrs over the course of 10 days. In addition, on day 7 of the experiment, I counted the number of cells that were both Kan R and Gent R ( Fig ure 3.6 B ) . 3.3.5 COUNTING TRANSFORMANTS To determine the number of transformants in the soil microcosm experiments , I wei ghed out 0.2g of soil from each microcosm and placed it into a 1.5mL centrifuge 35 tube. To each tube I added 180 µ l of liquid LB and vortexed for 10 seconds (~10 - 1 dilution). After allowing the soil to settle for 10 minutes, I transferred the supernatant to a 96 - we ll plate and diluted out to 10 - 6 or 10 - 9 depending on the experiment and the expected number of cells . In the experiments that manipulated eDNA concentration and soil moisture, I plated 50µl cell suspensions . F or the remaining experiments I plated 10 µl dots. All plating was done on petri dishes with solid LB (to count the total population size) or solid LB + gentamycin (50 µ g/ml) + Xgal ( 4 0 µg/ml) (to count transformants in soil ). Plates wer e incubated at 30°C and the number of colonies counted after 48 - 72 hrs. The number of cells is reported g - 1 soil, except in Fig ure 4 where it is reported per eARG pool (0.2g soil ) and calculated according to the following equation: Cells per unit = Cells µl - 1 x [ Soil slurry volume (200µl) / Soil Mass in slurry (g)]. I report the number of transformants g - 1 soil , as it is an environmentally relevant metric, and transformation is not affected by population size in populations larger than 10 ,0 00 P. stutzeri cells ( Fig ure A 3. 1B ) . The only exceptions are Fig ure 3.3 A where the population sizes varied between soil and lab assays and in Fig ure 3 .4B where the population size fell below 1000 cells. In these instances, I instead report the Transformation Frequency = log( transformants )/log(populat ion size). 3.3.6 STATISTICAL ANALYSES Prior to analyses all data were verified to meet assumptions of normality and homogeneity of variance. Data that did not conform to assumptions of homogeneity of variance were log transformed when appropriate. Results from soil microcosm studies were analyzed by either one - way or two - test variable (i.e. soil manipulation and sampling day) as a fixed effect using the R stats 36 package (R core team 2018). E xperiments with multiple sampling days were anal y zed by two - way ANOVA, except in certian instances, when the test variables were analyzed individually by sampling day ( e.g. Fig ure 3.4 A ). Results from the soil microcosm experiment in ( Fig ure 3. 4 ) were analyzed by two - way AN variables were included in the model. Results from laboratory experiments ( Fig ure 3.6 ) were anlayzed using two - way ANOVA with the treatment ( L ive vs Dead cells ) and s election r egime ( 0, 10, 25% lethal dose gentamicin ) as fixed effects. Results were based on the frequency of gentamicin resistan t genotypes present in each population at the end of the experiment. Differences between all test variab le groups were considered 0.05. 3.4 RESULTS Inoculating eARGs into soil facilitated gene transfer through natural transformation. Transformants appeared in soil within 24 hours after eARG addition and evolved in the presence of just 0.25 µg eDNA g - 1 soil, which, conservatively - estimated, is only a fraction (1/100) o f eDNA in field soil (reviewed by [103] ; also see [105] ) . Increasing the amount of eARGs increased t he number of transformants in soil , suggesting larger pools of eA RGs pose a greater risk to the spread of antibiotic resistance. While transformation occurred under a wide range of soil conditions, rates wer e highest at intermediate soil moistures ( 5 - 2 0%) and increased with the availability of eARGs. First, I tested how the soil environment impacted transformation by comparing the number of transformants that evolved in soil versus on low (R2A) and high - nutrient 37 (LB) agar petri dishes. I found th at transformants appeared in s oil and petri dishes at a similar frequency . Ho w e ve r, it took 4 days for the frequency of transformants in soil to equal the number of transformants on high and low nutrient agar plates ( Fig ure 3. 3 A ). This shows that soil is not a significant barrier to t ransformatio n , but that transformation Fig ure 3.3 : The relationship between e DNA availability and transformation. ( A ) Comparison of transformation in soil versus on petri dishes (High Nutrient = Luria Broth (LB); Log(Recipient Cells)]. Asteris ks indicate that the transformation frequenc y varied across the 3 treatments on that day . ( B ) The relationship between eARG concentration and the appearance of transformants in soil ( day 15 from 3.3 C). The eDNA concentration ranged from <1% to 10% of a sta ndard soil eDNA pool . ( C - D ) Time series tracking the appearance and maintenance of transformants when soil was supplemented with period inputs of eARGs at either, (C) 5 , 2.5, (D) 0.25, or 1.25 µg eDNA g - 1 soil. eARGs were added on day 0, 5, and 10 after counting transformants. The a rrows indicate the timing of the eARG additions. ( A - D ) Points/bars represent the average number of transformants g - 1 soil and error bars show the standard error of the mean (n=8 replicates). 38 may initially proceed slo wer in soil than under laboratory conditions (24 - 72hrs, p<0.001 across the 3 treatments). O ne of the strongest controls on transformation was the availability of eARGs. I found that transformation scaled linearly with the concentration of eDNA but only in soils inoculated with at least 2.5 µg of eDNA g - 1 soil ( Fig ure 3. 3 B , p<0.001). P eriodic inputs of large concentrations of eDNA (>2.5 µg), increased the number of transfor mants by an equal magnitude ( Fig ure 3. 3 C ). While periodic inputs of smal l concentrations of eDNA (<1.25 µg), did not uniformly increase the number of transformants ( Fig ure 3. 3 D ). W hen the eDNA concentration was held constant and soil microcosms were incubated at 5, 10, 20, 30 or 40% soil moisture, I found that transformation was highest at 10% soil moisture (though 10% did not significantly differ from 5% or 20%) ( Fig ure 3. 4 A , p < 0.001). However, on day 5 before the second eDNA addition there were significantly fe we r transformants at 5% soil moisture than 10% and 20% (data not Fig ure 3.4 : The relationship between soil moisture and transformation . (A ) The number of transformants in soil incubated at 5%, 10%, 20%, 30% or 40% soil moisture over a 10 - day experiment. ( B ) The relationship between the frequency of soil homogenization and the evol ution of transformants at 10% soil moisture. Homogenization was conducted every 2hrs, 8hrs or never over a 48hr window. Bars represent the average number of log 10 (transformants g - 1 soil) and error bars show the standard error of the mean (A: n=8, B: n=4 replicates ). 39 s hown). To elucidat e what factors might increase transformation at 10% soil moisture, I performed a 48 - h r transformation assay where I disturbed the soil matrix every 2h, 8h, or left the soil undisturbed ( Fig ure 3. 4 B ). Homogenizing the soil every 2h completely prevented transformation from occurring, while homogenizing every 8h r educed the frequency of transformation events compared to a non - homogenized control (p=0.03). Fig ure 3 .5 : The relationship between dispersal and transformation . ( A ) Location of eDNA poo ls in soil microcosms set - up in 150 x 15mm petri dishes. Transformable cells were added to the center of the plate; the top 4 eDNA pools have eARGs (yellow) and the bottom 4 do not (gray). ( B ) The top panel shows the average number of transformants and the b ottom panel shows the average cell count in each eDNA pool after 5 days of dispersal. The size of the dot increases as the number of cells increases . ( C ) The total transformants and ( D ) the average cells at each distance, pooled across the four soil moistu res. ( E ) The total transformants and ( F ) the average cells at each soil moisture, pooled across the four distances. ( C,E ) represent the sum of transformants across the replicates. Error bars show the standard error of the mean (n= 4 replicates) . 40 Next, I tested the effect of s eparat ing P. stutzeri cells from local eARG source s and found that this spatial separation posed a substantial barrier to transformation in soils maintained at 10, 20, 30 or 40% soil moisture ( Fig ure 3. 5 ) . Even in w et soils , where dispersal to eARG sources was high , there were no transformants in any of the eDNA pools (dispersal at 10% vs 40% soil moisture, p=0.0289, Fig ure 3.5 B - F ). Meanwhile at 10% soil moisture there were significantly more transformants than at any other soil moisture . H owever, these transformants only appeared in the closest eDNA pool, with P. stutzeri unable to disperse to the most distant eARGs located 7cm away (transformants *soil moisture, p<0.001, Fig ure 3.5 B - F ). Generally , dispersal and transformation happened at intermediate rates at 20 and 30% soil moisture . In a final experiment, I used a lab - based assay to compare the establishment of an or s of P. stutzeri ( Fig ure 3.6 A ). The recipient cells were a ll kanamycin resistant ( Kan R ), while the eARGs and live invaders encoded gentamicin resistance ( Gent R ). In almost all instances, transformed Gent R genes replaced native Kan R gene s , with gene addition ( Gent R + Kan R ) occurring at very low frequencies (<1% of transformants ) and only under the highest dose of antibiotic ( Fig ure 3.6 B ). I n the absence of a selective pressure and at 10% of the lethal dose of antibiotic , live invaders reached higher frequencies than antibiotic resistant transformants ( Fig ure 3.6 C , p<0.001). However, w hen the selective pressure increased to 25% of the lethal dose, both the live invader and t ransformed eARGs reached a high frequency in the population ( Fig ure 3.6 D , p<0.001 ). Although antibiotic resistant transformants took 24 h our s longer than live invaders to establish at high frequencies under the highest does of gentamicin ( Fig ure 3.6D ). 41 Fig ure 3.6 : Transformation vs. invasion of antibiotic resistant genes. ( A ) Frequency of kanamycin (Kan R ) and gentamicin resistance (Gent R ) at Kan R and 4% Gent R R and 0% Gent R + Gent R eDNA (right). ( B ) Gene addition (maintenance of Kan R + Gent R ) was only detected at low frequencies in populations exposed to 25% of the lethal dose of gentamicin. ( C ) The average frequency of Gent R under three selection regimes : 0%, 10%, and 25% of the lethal dose of gentamicin. ( D ) Zoomed in perspective of panel 1 and panel 3 from C. ( C - D ) Error bars show the standard error of the mea n (n=4 replicates). 42 3.5 DISCUSSION In order to reduce the impacts of antibiotic resistance on human health , we need a better understanding of the ecological dimensions that promote the tran sfer of e ARGs in natural systems [106] . In this study, I show that high concentrations of eARGs in soil increase the number of transformants , ultimately increas ing the prevalence of antibiotic resistant bacteria in antibiotic - free soil. I find that transformants appear under most conditions typical for terrestrial soils (10 - 20% moisture) , however, transforma tion efficiency decreases at high soil moisture s and with soil mixing . In addition , I find that a low dose of antibiotic allows eARGs to establish with the same success as a live antibiotic resistant invader. Overall, the sustained biological activity of e ARGs, even after bacteria death, suggests eARG removal should be incorporated into plans to combat antibiotic resistance. Several studies have now posited that the spread of antibiotic genes into diverse bacterial lineages occurs via widespread HGT [8], [82] . Here, I show that eARGs supplied by dead bacteria are readily transferred into soil bacteria, with the potential for HGT scaling linearly with the abundance of eARGs. My findings provide nove l evidence that large concentrations of environmental eARGs can drive the evolution of antibiotic resistance, and this information should be incorporated into our approach to combating antibiotic resistance. For instance, many disinfection methods focus on killing live bacteria, but may be more effective if they consider the persistence of eARG pools, which I find can be equally effective at spreading ARGs . This may explain why practices like composting manure prior to application on agricultural fields has been found to both increase and decrease the occurrence of ARGs, depending on the native bacterial 43 community [107] , and other soil conditions [99] . Interestingly, manures composted at high temperatures which promotes degradation of eDNA can be most effective in reducing ARGs [108] , supporting my findings that DNA degradation is a critical factor in reducing environmental concentrations of ARGs. Despite the dangers of low - levels of eARGs persist ing in soil for an extended time , the fate of most extracellular DNA is likely degradation [34] . While low levels of eDNA can persist in soil for 80+ days, 99% is degraded in the first ~7 days [66], [109] . Consequently, the most important role of soil conditions in regulating transformation may be the effect of moisture on the rate of eDNA de cay , and likely explain s why transformation declin ed at higher soil moistures ( Fig ure 3 . 4, 3.5 ). T ransformation was previously shown to decline at 35% soil moisture in Acinetobacter calcoacetic um [110] , however, transformation w as only measured above 18 % soil moisture . In contrast , P. stutzeri had the highest transformation efficiency between 5 and 20% soil moisture ( Fig ure 3.5 A, 3.5 ) . This finding provides novel evidence that transformation occurs at lower soil moistures than prev iously thought , but the relationship between soil moisture and transformation could var y widely across bacterial species , depending on how competence is regulated . It is also important to note that agricultural soils rarely exceed 30% soil moisture in the field , and in fact it has only happened twice since 1989 in the region where the se soil s w ere collected ( Fig ure A 3. 2 ) . H owever, wetland soils often exceed 30% soil moisture, suggesting that transformation rates may vary widely across soil habitats . T he increase in transforma bility at low er soil moistures could also be due to increased exposure to eDNA in drier conditions . Low er moisture conditions have 44 previously been shown to increase conjugation rates in P. putida , as unsaturated soils create fragmented habitats that lengthen the duration of cell contact between bacteria [111], [112] . While transformation does not require cell - cell contact, it does requir e cell - eDNA contact which increase s the duration of time near eARGs and could promote transformation. However, drier soils could also promote transformation via biofilm formation , which is a common response to drought and increases microbial survivorship at low soil moistures [113] . I mportantly , biofilm s also enhance the efficiency of gene transfer [91] and could be a critical precursor to transformation in soil as in the lab, P.stutzeri exhibits much higher transformation efficiencies in sessile o r biofilm communities than planktonic communities ( Fig ure A 3. 1C ). If biofilms are critical for transformation in soil it could explain why soil homogenization prevented transformation , as disturbing the biofilm structure every 2hrs would prevent mature biofilms from establishing and there fore could prevent transformation [114] . Future studies could use fluorescent proteins or confocal laser scanning microscopy to better quantify the relationship bet we en biofilm establishment and transformation efficiency in soil [115] . Historically, t he contribution of eARGs to antibiotic resistance was assumed to be low , p rimarily because transformed e ARGs are deleterious , and theoretically transformants should not increase in abundance in the absence of selecti on. However, the presence of antibiotics in the environment could provide a selective pressure that enables transformed eARGs to r each high frequencies in a community . In fact, my research provides unique evidence that low doses of antibiotics allow transform ed eARGs to establish in the population and produc es a population trajectory analogous to 45 an invasion by live antibiotic resistant cells. This highlights that controlling the release of antibiotics into the environm ent is critical for preventing the emergence of novel antibiotic pathogens even if there are no live antibiotic resistant bacteria in the community. A n important future research direction will be determining the antibiotic concentration s at which eARGs r epresent a major s our ce of antibiotic resistance in soil . Answering this question will require an increased understanding of antibiotic concentrations in soil and their effect on bacterial communities . T his study lays the groundwork for future studies, whi ch should examine the concentration of eARGs , and the sub - inhibitory concentrations of antibiotics which promote the rise of antibiotic resistant transform ants in soil . Taken together, my study reveals the most important variables for understanding the transmission of eARGs in soil and set s the stage for future experiments to scale up estimates of transformation to the whole community level. Here, I use d a sterile soil system , inoculate d with a single bacterium , as to prevent competitive interactions, and ensure the soil was antibiotic - free . T ransformation may be lo we r in multi - species communities, where competitive interactions would limit access to eARGs and limit the success of transformants. Regardless, this work provides novel evidence that eARGs from dead bacteria are an overlooked , but important route in the em ergence of antibiotic resistan ce . I conclude that transformation occurs under most soil conditions and show that transformed eARGs are just as successful as live antibiotic resistant invaders , when challenged with a low dose of antibiotic. Overall, this de monstrates that disinfection alone is insufficient to prevent the spread of ARGs through environmental reservoirs . 46 Furthermore, special caution should be taken in releasing antibiotics into the environment , even i f there are no live antibiotic resistant ba cteria in the community , as transformation allows DNA to maintain its biological activity past microbial death . 47 APPENDIX 48 APPENDIX Fig ure A 3.1 : Transformation assays under laboratory conditions. ( A ) The relationship between transformation and the concentration of eDNA under laboratory conditions (0.001, 0.01, 0.1, 1, 5, 10 ng/µl eDNA). ( B ) The effect of total population size on the number of transformants. ( C ) Comparison of transformation in sessile (biofilm or surface attached) communities vs. planktonic communities. 49 Fig ure A 3.2 : Gravimetric soil moisture from 1989 to 2019 Southwest, MI (regular measurements from April - November). Data from the Kellogg Biological Station Long - term ecological research center main cropping system. The red line shows 30% soil moisture and represents the soil moisture where transformation starts to decline. 50 CHAPTER 4: CHANGES IN TRANSFORMATION AFTER SALT ADAPTATION 4.1 ABSTRACT The exchange of genes between potentially unrelated bacteria is termed horizontal gene transfer (HGT) and is a driving force in bacterial evolution. Natural transformation is one mechanism of HGT where extracellular DNA (eDNA) from the environment is recombined into the host genome. The widespread conservation of transformation in bacterial lineages implies there is a fitness benefit. However, the nature of these benefits and the evolutionary origins of transformation are still unknown. Here, I examine how ~330 generations or 100 days of serial passage in either constant or increasing salinities impact s the growth rate and transformation efficiency of Pseudomonas stutzeri . While the growth rate generally improved in response to serial transfer, the transformation efficiency of the evolved lineages varied extensively , with only 39 - 6 4% o f populations undergoing transform ation at the end of adaptive evolution . I n comparison, 100% of the ancestral populations were able to undergo natural transformation. I also found that evolving P. stutzeri with different cell lysates (or populations of dead cells) minimally affected the growth rate and transformation efficiency, especially in comparison to the pervasiveness with which tr ansformation capacity was lost across the evolved populations. Taken together, I show that the efficiency of eDNA uptake changes over relatively rapid timescales, suggesting that transformation is an adaptive and selectable trait that could be lost in envi ronments where it is not beneficial. 51 4.2 INTRODUCTION Natural transformation is a mechanism of horizontal gene transfer where by bacteria acquire extracellular DNA (eDNA) from the environment and recombine it into their genomes. Transformation plays a key role in bacterial evolution [11], [116] ; however, the fitness benefits of transformation remain unknown, despite extensive study. While it is generally accepted that transformation can facilitate adaptation through genetic recombination, the consequences of this genetic exchange can be both ben eficial and costly [52] . The t heoretical benefits of transformation are similar to meiotic sex and include speeding up adaptation, combining beneficial genes into one genome, and separating beneficial mutations from deleterious loads [11], [48], [ 95], [117], [118] . However, extracellular DNA from dead bacteria can also carry an increased mutational load or promote the spread of selfish genes [53 ], [119] . Consequently, the fitness advantages of transformation and the environmental conditions in which they are conferred have been difficult to quantify experimentally (see Table 4.1 ). There are several potential explanations for the evolutionary m aintenance of transformation [ reviewed in 42] . Several of them posit that transformation evolved as a byproduct of acquiring DNA for nutrients [120] [123] or genome repair [45], [124], [125] . However, the presence of cellular machinery dedicated to protecting extracellular DNA (eDNA) from degradation inside the cell, suggests that eDNA is not acquired purely for the nutrient benefit [42], [45] . In addition, many bacterial taxa prefer entially kill and transform eDNA from close relatives, a process somewhat analogous to the exchange of DNA in eukaryotic sex [46], [47] . 52 Transformation is also similar to eukaryotic sex in that it is primarily beneficial in stressful or continuously changing environments. Population genetic models [11], [48], [49] and experimental evolu tion studies [50], [51] have shown that transformation is beneficial in rapidly fluctuating or stochastic environments where tr ansformable cells can outcompete non - transformers [43], [129] . Theoretically, this is because transformation can increase genetic variation, thereby increasing the efficiency of natural selection [130] . Transformation does not always provide a fitness benefit in stressful environments though, as Bacher et al. [55] found that competent lineages of Citation Bacter ial Taxa Adaptation Conditions Transformation increased adaptation Exogenous DNA provided Changes in transformation Bacher et al. 2006 [ 55] Acinetobacter baylyi High Salinity & Temperature No No Decreased Baltrus et al. 2007 [50] Helicobacter pylori Novel Laboratory Conditions Yes No Not reported Perron et al. 20 12 [126] Acinetobacter baylyi Periodic Antibiotics (3 - 4x/wk) Yes, when provided resistance genes Yes Not reported Engelmoer et al. 2013 [51] Streptococcus pneumoniae Periodic Antibiotic (kanamycin 2x/wk) Yes No Not reported Utnes et al. 2015 [127] Acinetobacter baylyi Novel Laboratory Conditions Yes, but only during early stationary phase Yes Not reported Mcleman et al. 2016 [128] Acinetobacter baylyi Parasitic Phage Yes, from phage - sensitive or resistant DNA Yes Not reported Table 4.1: Review of transformation - mediated fitness effects . E xperimental evolution studies that have quantified the fitness effects of transforma tion. 53 Acinetobacter baylyi did not adapt to novel laboratory conditions faster than their non - competent competitors, and repeatedly lost the ability to transform eDNA. While several other studies have shown that transformation is beneficial in stressful environments (see Table 4.1 ), it is still unclear how the availability of beneficial genes might alter transformation - mediated fitness effects . For instance, antibiotic resistance only evolved via transformation when antibiotic resistance genes were provided, while phage resistance evolved in the presenc e of phage - sensitive or - resistant DNA [126], [128] . Since sequence similarity improves the efficiency of homologous recombination, it is generally accepted that transformation is most prevalent between closely related organisms [131] . However, sharing genes with close relatives could limit the acquisition of novel gene combinations and ultimately limit adaptation. Here, I aim to better understand the evolutionary benefits of transformation by evolving Pseudomonas stutzeri a highly transformable soil bacterium in either constant or increasing salt concentrations for 100 days, while supplying different sources of eDNA (cell lysates or dead cells). At the end of the experiment, I quantify the growth rate, population size, and transformation efficiency (transforma nts/µg eDNA) of the evolved populations and compare this to the same m easurements in the starting isolate or ancestor. I specifically address the following questions: 1) Does the transformation efficiency increase in response to evolving in a variable re lative to a constant environment (increasing salinity vs. constant low salinity)? 2) Does evolving with dead halophiles or dead Pseudomonads better facilitate adaptation to high salt concentrations? 54 4.3 METHODS 4.3.1 SERIAL DILUTION EXPERIMENT I serially transferred Pseudomonas stutzeri , strain 28a24 for ~330 generations (100 days) in 96 - well microtiter plates [ see 101 for whole genome sequence ] . Cultur es were serially transferred every 24hrs at a 1:10 dilution and maintained at 26°C. For the first 50 days (~170 generations) of the experiment, all populations were transferred as one treatment in a constant salt media (1.5% salinity; 10g/L tryptone, 5g/L yeast extract, and 15g/L NaCl). After 50 days (~170 generations), the experiment was shut down due to the global covid - 19 pandemic, and populations preserved in 40% glycerol at 20°C. Four weeks later, populations were revived and serially passed at 1.5% sa linity for 4 - 1 (~170 generations). At this point, I split the experiment into two treatments. The original treatment was maintained at a low constant salinity for the remainder of the experiment (1.5% sal t media from day 1 to 100). The new treatment, which I refer to as the increasing salinity treatment was Fig ure 4.1 : The serial transfer conditions for the two evolution treatments. The left panel shows the conditions for populations adapted to a constant low salinity environment (1.5% salinity) and the right panel shows the conditions for populations adapted to increasing salinities (1.5% to 2.5% salinity). 55 transferred to a 2% salt media (20g/L NaCl), where it was serially passed for 100 generations. Then on day 81, I increased the salt concentration to 2. 5%, were it stayed for 67 generations until Day 100 (see Fig ure 4.1 for the serial transfer conditions). The constant low and increasing salinity treatments had 96 replicates each. In addition, during each transfer (every 24hrs), populations were supplemen ted with eDNA via whole populations of dead bacteria which equated to 5ng of genomic eDNA each transfer . I refer to these as cell lysates as they contain DNA and other cellular components (see Table S4.1 for a detailed list of the cell lysate sources). T he experimental design is detailed in Fig ure 4.2. 4.3.2 PREPARATION OF CELL LYSATES Individual cell lysates were prepared in 100mL batch cultures in liquid LB on a shaker table at 120rpm and 30°C (10g/L tryptone, 5g/L yeast extract, and 5g/L NaCl). After 48hrs, each culture was plated to confirm there was no contamination (10 µl replicate dots plated 3x) . Each culture was then heat shocked at 90°C for 1hr. After heat shock , each culture was plated to confirm all the cells were dead. If bacterial strains still Fig ure 4.2 : Graphical representation of the experimental design. The first tier represents the two serial transfer conditions (n=96). Within each of these treatments, I supplemented populations with different cell lysates there were twelve total cell lysate treatments, grouped into five categories (n=8 x 12 ). The thi rd tier represents the assay conditions in which I measured the growth rate, population size, and transformation capacity at the end of the experiment. 56 had viable colonies, these cultures went through another round of heat shock at 100 - 110°C, which was sufficient to kill the remaining cells. The heat shocked cultures were then spun down and resuspended in sterile nanopore water. The lysates were filtered through a 0.22µm filter and standardized to a concentration of 1ng DNA/µl using a Qubit 2.0 fluor ometer (Life Technologies, USA). There were 12 cell lysate treatments with 8 replicates each. See Table S4.1 for expanded list of cell lysates. 4.3.3. GROWTH RATE DETERMINATION All assays were conducted on the ancestral population, and populations that evolved for 100 days (~330 generations). Strains of P. stutzeri were revived from 40% glycerol stora ge ( - 80°C) and diluted 1:10 in liquid LB media (0.5% salinity: 10g/L tryptone, 5g/L yeast extract, and 5g/L NaCl) in 250 - µl microwell plates. After revival, the populations were transferred every 24hrs at a 1:10 dilution in 0.5% salinity for 4 transfers. A fter the fourth transfer , I moved the populations to two separate salt Fig ure 4. 3: Revival and assay conditions . The ancestral and evolved lineages (day 100 populations) were revived and transferred in 0.5% salinity for 4 transfers. On the 4 th transfer, the populations were moved to the low (1.5%) and high (3%) salinity environment. I then quantified the growth rate, transformation capacity, and population size over the next 24hrs (between day 4 & 5 after the revival). 57 environments in 250 - µl microwell plates , to quantify the growth rate and transformation capacity ( Fig ure 4.3 ). In the constant low salinity treatment, 10 of the 96 populations never revived. In the increasing salinity treatment, 1 population never revived and 1 was contaminated, the se are not reported in the results (see Table S4.2 for a full list ). Each 24hr as say was conducted at low salinity (1.5% salinity: 10g/L tryptone, 5g/L yeast extract, and 15g/L NaCl) and high sal inity a novel and stressful environment for P. stutzeri (3% salinity: 10g/L tryptone, 5g/L yeast extract, and 30g/L NaCl) ( Fig ure 4.3 ). For each population, I monitored absorbance at 600nm for 24hrs using a Biotek Synergy HGT (Winooski, VT) microplate reader. The growth curve data was fit to a standard form of the logistic equation using the Growthcurver package in R studio. I used the logistic equation to describe the population size N t at time t : 4.3.4 TRANSFORMATION EFFICIENCY AND FREQUENCY Cultures were revived following the same protocol used for growth rate determination (described in 4.3.3). I quantified the transformation efficiency by tracking the acquisition of gentamicin resistance into the evolved and ancestral P. stutzeri populations which were gentamicin susceptible. The eDNA encoding gentamicin resistance was prepared from a mutant strain of P. stutzeri , strain 28a24, which carries a gentamycin resistance gene and LacZ gene fused to a miniTn7 transposon (Tn7 transposition of pUC18 - mini - Tn7T - Gm - lacZ ) . To begin the assay, I transferred 20µl from each evolved and ancestral population into 180µl of fresh LB media containing 1.5% and 3% salinity . I added genomic extracellular DNA (eDNA) resuspended in nanopore 58 water to each pop ulation and incubated at 30°C. After 24hrs I performed a serial dilution and titers were determined on selective media (LB + gentamycin [50 µg/ml] + Xgal [40 µg/ml]) and non - selective media (LB) using triplicate 10µl dots. Population level transformation e fficiency was determined by dividing the average number of transformants in a population by the µg of e DNA (0.02µg). I also report population level t ransformation frequencies by dividing the average number of transformants by the total number of cells or t he population size. 4.4.5 STATISTICAL ANALYSES Prior to analysis, I checked that data met assumptions of normality and homogeneity of variance. I corrected for increased homogeneity of variance across population sizes using a log transformation. I analyzed bacterial growth rates and populations sizes using two - factor ANOVA, with Evolution Conditions, Assay Conditions, and their interaction as factors (see Fig ure 4.2 for factors). For each evolution treatment constant salinity versus increasing salinity I used a two - factor ANOVA with the Assay Conditions, Cell Lysate treatment, and their interaction as factors. I determined differences in transformation capacity u sing a general linearized model with a negative binomial distribution to account for positive skew . T o determine statistical differences in the number of non - transforming populations (zeros) I used a two - part hurdle model from the hurdle package in R , as i t specifies one process for zero counts and one process for positive counts, and is commonly used for positively skewed data with lots of zeros [132], [133] . 59 4.4 RESULTS 4.4.1 GROWTH RATE P. stutzeri adapted to changes in salinity after ~330 generations of serial transfer. Both evolution treatments (constant vs. increasing sal inity) on average grew faster than the ancestor in the high salt environment. However, populations evolved in the increasing salinity environment grew faster in both the low and high salt environment ( Fig ure 4.4A; Salinity p < 0.001; Treatment*Salinity p = 0.052). In addition, populations exposed to the gradual increase in salt, exhibited higher growth rates than those adapted to the constant salt concentration but only when tested at the lower salinity ( Fig ure 4.4 A ; p= 0.0468). Interestingly, both of the evolved populations had larger population sizes in the high salt environment relative to the low salt environment and to the ancestor ( Fig ure 4.4B ; p <0.001). This was surprising given the evolved populations gre w significantly slower in that environment compared to the low salinity one. Figure 4.4: Effects of adaptive evolution. (A ) Growth rate and ( B ) l og transformed population size for the ancestor (red circles), constant low salinity (black closed triangles), and increasing salinity evolution treatments (black open circles) in low (1.5%) and high (3%) salinity. The points show the average across the ancestral (n=8) a nd evolved populations, and the error bars indicate the standard error ( constant low , n = 86 ; increasing n = 94) . 60 4.4.2 LOSS OF TRANSFORMABILITY Evolved populations exhibited a significant loss of transformation capacity relative to the ancestor ( Fig ure 4.5 ; constant low salinity, p=0.005; increasing salinity, p=0.02). At the end of the experiment between 39% and 6 4% of evolved populations depending on the treatment and test conditions still underwent transformation. In the remaining populations there were no transformants at a detectable level. In addition, there was a striking similarity between the two evolution treatments in terms of how many populations underwent transformation ( Fig ure 4.5 ). In both evolution treatments, there were significantly more populations undergoing transformat ion when tested at the higher salinity, despite no known difference in genotype ( Fig ure 4.5 ; p <0.001 for both treatments). Figure 4.5: Loss of transformability. The number of evolved populations with a detectable number of transformants in the low and high salinity test environment for populations evolved in constant or increasing salinities (count data). The red line corresponds to the beginning of the experiment when transformants could be detected in 100% of the ancestral populations. 61 4.4.3 TRANSFORMANTS AND TOTAL CELLS There was no relationship between the number of transformants and the number of recipient cells in independently evolving populations ( Fig ure 4.6 ). This trend was true across the treatments, as well as agreeing with preliminary work showing that transformation is not limited by population size in large r populations such as the ones in this experim ent ( Fig ure A 4.1A ; above 10,000 cells). Therefore, I report the number of transformants standardized by the amount of eDNA (transformation efficiency) . However, I also report (in Fig ure 4.7B ) the number of transformants standardized by the number of recipient cells (transformation frequency). This is done to account for the fact that, on average, there were significantly more transformants and recipient cells at higher salinities, suggesting the increase in the average number of transformants could Figure 4.6: The relationship between transformants and total cells. In the ancestral and evolved populations, the number of transformants does not increase as the population size or total number of cells increases. Each point represents an individual population - level measurement. The linear relationship between transformants and total cells is indicated by the line s with the shaded area s sh owing the 95% confidence interval (n=8 ancestor, n=86 constant low and n=94 in the increasing salinity treatment). 62 be correlated with the increase in the average number of recipient cells despite ther e being no evidence of such a correlation within the individual populations. 4.4.5 HIGH VARIATION IN TRANSFORM ATION EFFICIENCY At the end of the experiment the transformation efficiency (transformants/µg DNA) was significantly lower in the low salt environment, regardless of the evolution conditions ( Fig ure 4.7A ; p < 0.0001). However, the populations that transformed eDNA and evolved at constant salinity, did so at a higher efficienc y tha n the ancestor but only when tested in the high salt environment ( Fig ure A 4.2 ; p = 0.0216). When the number of transformants was standardized by the number of recipient cells , there w ere no statistically significant differences in transformation frequency ( Fig ure 4.7B ). Although numerically the transformation frequency was highest for populations evolved in constan t low salinities but moved to the high salinity environment for the transformation assay. Figure 4.7: Changes in transformation in response to experimental evolution . ( A ) Transformation efficiency (transformants/µg DNA) and ( B ) transformation frequency ( t ransformants/ t otal Cells) for the ancestor (red circles) and evolved populations (constant low salinity = solid black lines; increasing salinity = dashed lines). The points show the average across the ancestral (n=8) and evolved (constant low = 86, increasing =94) populations, and the error bars indicate the standar d error (averages include transforming and non - transforming populations). 63 Evolving with cells lysates or populations of dead cells did not a ffect transformation efficiency in a uniform manner ( Fig ure 4.8; Fig ure A 4.3; see Table A 4.3 for expanded results). In general, the transformation efficiency was higher in the high salt environment. However, standardizing by the population size indicated there was no difference in transformation frequency between the low and high sal t environment as the number of transformants and the number of total cells was larger in the high sal t environment ( Fig ure A 4.4 ). Moreover, there were no consistent changes in growth rate Figure 4.8: Variation in transformation efficiency. T ransformation efficiency for the a ncestral and evolved population s in low and high salinit ies . Higher transfo rmation efficiencies are denoted by larger circles with the populations evolved at a constant low salinity in the top panel and those evolved in increasing salinities in the bottom panel . T he cell lysates are shown on the x - axis and the replicate populatio ns on the y - axis corresponding to a 96 - well plate layout (n=8 replicates per cell lysate source) . 64 or population size with the addition of different cell lysates ( Fig ure A 4.5 ; see Table A 4.4 and A 4.5 for expanded results). T here was a high level of congruency between the two evolution treatmen ts in terms of which populations underwent transformation ( Fig ure 4.8 ; comparing the top and bottom panels). T hese effects may have appeared early in the experiment , since the two treatments diverged from a single set of evolving populations on day 50 of t he experiment . 4.4.6 TRADE - OFF BETWEEN GROWTH RATE AND TRANSFORMATION T here was no trade - off between the transformation efficiency and the growth rate ( Fig ure 4.9 ; the same being true for transformation frequency data not shown). While there may have been tradeoff between growth rate and transformation with in individual strains , I was unable to detect such a tradeoff in the population - level measurements conducted here . Figure 4.9: Trade - off between growth rate and transformation efficiency. Average population - level growth rate regressed against the transformation efficiency in the ( A ) low salinity and ( B ) high salinity environments. Each point represents an individual population - level measurement. The linear relationship between transformants and total cells is indicated by the line, with the shaded area showing the 95% confidence inte rval (n=8 ancestor, n=86 constant low and n=94 in the increasing salinity treatment). 65 4.5 DISCUS SION Understanding the evolutionary origins and fitness consequences of transformation can shed light on the larger question of why organisms undergo genetic recombination, and to what extent the traits governing selection are themselves selected upon. Here, I conclude that evolving P. stutzeri with different sources of eDNA or evolving them in constant versus increasing salinities did not have a large effect on growth rate or transformation efficiency. However, I did find that the transformation capacity the ability for evolved populations to transform eDNA changed dramatically over just ~330 generations or 100 days of serial transfer. By the end of the experiment, around 50% of the evolved populations did not transform any of the provided extracellular ant ibiotic resistance genes, although the exact percentage varied from 3 6 - 61% depending on the treatment. This was true, regardless of whether the number of transformants was reported as the transformation efficiency (transformants/µg eDNA) or the transformat ion frequency (transformants/total cells). I report both metrics here to account for difference s in average population size in the low versus high salinity environment , even though the total population size only limit s transformation in very small populati ons of P. stutzeri (smaller than those reported here ; Fig ure A 4.1B ). Overall , I focus the discussion on the variation in transformation capacity across the evolved lineages , as this is true irrespective of how the data is analyzed (transformation efficiency vs. transformation frequency). Several other bacterial species, in addition to P. stutzeri , undergo transformation irrespective of the population density . For instance, Vibrio parahaemolyticus and V. campbellii both undergo trans formation in the absence of quorum sensing which is the 66 ability to regulate gene expression with population size . Meanwhile, their close relative V. cholerae, and Streptococcus species both require quorum sensing for successful transformation [29], [134] . Interestingly, the genetic features that u nderpin the differences in quorum sensing across Vibrio species have yet to be identified. Similarly, different isolates of the same bacterial species often exhibit large differences in their transformation capacit y, with the genetic variation underpinning these differences often impossible to discern . For instance, isolates of P. stutzeri collected from different soil environments had highly variable transformation frequencies, with about one - third of isolates considered non - transformable [28] . Similar observations have been made in Vibrio species that inhabit different environments , and spurred the recent suggest ion that transformation may be lost in environments where it is no longer beneficial [29], [135] . Hence , it is possible that transformation was not maintained in several of the evolved lineages because it was not providing a fitness benefi t during experimental evolution . Weak selection for transforma tion could have been due to the application of only a mild stress or due to infrequent fluctuations in the environment . For instance, p revious studies that found transformation was beneficial, tended to shift the environment every 2 - 3 transfers [51], [126] , as opposed to every 20 - 30 transfers as was done in this study . Therefore, a n inter e sting follow - up study would be to compare the distribution of transformation phe notypes after evolving P. stutzeri in a constant optimal environment versus a constant but very stressful environment to better understand how stress, or fluctuations in stress, shape the evolution of transformation. 67 Another possibility is that transformation only provides a fitness benefit in response to very specific stressors. For instance, several studies that found transformation was beneficial, exposed evolving populations to periodic inputs of sub - inhibitory con centrations of antibiotics [51], [126] . Because transformation allows the reversible integration of resista nce genes, and antibiotics are usually transiently present in the environment , transformation could be a mechanism well - suited to handling antibiotic stress. In contrast, transformation may be less beneficial in response to stressors like changes in osmoti c pressure which are encoded by large and connected gene networks . In general, more work needs to be done on the specific stressors that transformation confers a benefit to , as prokaryotic genes appear to adapt to either vertical or horizontal transmission , meaning that not all processes may be well - adapted to evolve via horizontal gene transfer [136] . A final consideration is the role o f osmotic pressure in altering the efficiency of eDNA uptake. Populations adapting to high osmolarity environments generally have a large fraction of mutations in genes associated with cell wall synthesis [137] . Therefore, it could be that changes in the cell wall altered the pilus structure which captures eDNA from the environment [138] . P. stutzeri has two pili that interact to regulate transformation. The type IV pilus acquires eDNA from the environment, while the second pilus is believed to translocat e eDNA into the cytoplasm and when knocked out decreases transformation ~ 90% [139] . Therefore, changes in the cell wall in response to salt stress could have altered the interaction between these two pili, creating the gradient of transformation capacity evident in the evolved lineages. Follow - up 68 investigations which involve whole genome sequencing, will hopefully elucidate if mutatio ns in osmoregulatory genes could have altered transforma tion capacity . Despite evidence of high variation in transformation capacity in many bacterial lineages, very few studies have quantified changes in transformation during experimental evolution. To d ate, six experimental evolution studies have focused on the fitness effects of transformation but only one study has quantified transformation before and after experimental evolution. In that one study, transformation did not provide a fitness benefit and the evolved lineages repeatedly lost the capacity to undergo transformation [55] . Several other studies have identified transformation - mediated fitness benefits (primarily in stressful environments), but none of them quantified the pre valence of transformation at the end of the experiment ( Table 4.1 ). Yet, q uantifying transformation after adaptation could help disentangle the benefits of transformation from the overall benefits of competence which is the physiological state a bacteri um must enter to undergo transformation and often initiates multiple stress responses alongside transformation . F or instance, Pseudomonads also use the Type IV pilus for flagellum - independent movement or twitching motility [139] . While Bacillus subtilis , a well - studied soil - dweller, upregulates transformation as part of a general stress respons e prompted by DNA damage or antibiotics [121] . Therefore, future studies that quantify changes in transformation efficiency during and after experimental evolution, will be critical in di sentangling the specific benefits of transformation within the larger regulatory network of competence . Overall, this study provides novel experimental evidence that the ability to undergo transformation can change over relatively short timescales and may be more 69 plastic across space and time than is generally accepted. Most intersting is the substantial decrease in transformation efficiency in the low salt environment where P. stutzeri evo lved, suggesting that transformation did not provide a fitness benefit during salt adaptation . Taken together, this work suggests that transformation is an adaptive, selectable trait , that may increase or decrease rapidly in response to selection . 70 APPENDIX 71 APPENDIX Fig ure A 4. 1: Preliminary transformation assays. ( A ) The effect of total population size on the number of transformants when the eDNA concentration is held constant. ( B ) The relationship between the concentration of eDNA and the number of transformants under laboratory conditions (0.001, 0.01, 0.1, 1, 5, 10 ng/µl eDNA) in large populations (~1,000,000 recipient cells). 72 Figure A4.2: Transformability in the populations that transformed eDNA. (A ) Transformation efficiency (transformants/µg DNA) and ( B ) transformation frequency ( t ransformants/ t otal Cells) for the ancestor (red circles) and evolved populations (constant low s alinity = solid black lines; increasing salinity = dashed lines). The points show the average across the ancestral (n=8) and evolved (constant low = 86, increasing =94) populations, and the error bars indicate the standard error (averages include only popu lations that underwent transformation). 73 Fig ure A 4. 3: Effect of cell lysates on transformation. (A ) Transformation efficiency , and ( B ) t ransformation frequency for each cell lysate treatment. T he ancestor (red circles), the constant low salinity (left panel) and the increasing salinity treatment s (right panel, dashed lines) are shown at low and high salinity (n=8 replicate population s across treatments, excludin g treatments that went extinct , see table S4.2 ). 74 Figure A4.4: Variation in transformation frequency. T ransformation frequency for the a ncestral and evolved population s in low and high salinit ies . Higher transformation frequencies are denoted by larger circles with the populations evolved at a constant low salinity in the top panel and those evolved in increasing salinities in the bottom two panels . T he cell lysates are shown on the x - axis and the replicate populations on the y - axis corresponding to a 96 - well plate (n=8 replicates per cell lysate) . 75 Fig ure A 4. 5: Effect of cell lysates on growth rate and population size. ( A ) Growth rate , and ( B ) log transformed population size for each cell lysate treatment. T he ancestor (red circles), the constant low salinity (left panel) and the increasing salinity treatment s (right panel, dashed lines) are shown at low and high salinity (n=8 replicate population a cross treatments). 76 Strain Cell Lysate Category Strain Source Conductivity (mS/cm) Halomonas spp. 1 ( Halomonas taeanensis ) Halophile Namib Springs Aub Canyon 7.3 Halomonas spp. 2 ( Halomonas ) Halophile Namib Springs Kai - As 1 Halomonas spp. 3 ( Halomonas ) Halophile Namib Springs Swartmodder 190 Halomonas spp. All (1+2+3) Halophile Pseudomonas chlororaphis subsp. aureofaciens Relative DSM6698 Pseudomonas azotifigens Relative DSM17556 Pseudomonas stutzeri JM300 Relative DSM10701 Pseudomonas spp. All Relative Self - Lysate + Transposon ( P. stutzeri + miniTn7) Self Baltrus lab Self - Lysate ( P. stutzeri ) Self Baltrus lab None None All All Table A 4. 1: Cell lysate sources. The halophiles were collected from salt springs in the Namib dessert. The conductivity of the spring at the time of collection is listed for Halomonas spp . The Pseudomonas relatives were purchased from the German culture collection (accessions number list ed). Strain Cell Lysate Category Strain Source Conductivity (mS/cm) Halomonas spp. 1 ( Halomonas taeanensis ) Halophile Namib Springs Aub Canyon 7.3 Halomonas spp. 2 ( Halomonas ) Halophile Namib Springs Kai - As 1 Halomonas spp. 3 ( Halomonas ) Halophile Namib Springs Swartmodder 190 Halomonas spp. All (1+2+3) Halophile Pseudomonas chlororaphis subsp. aureofaciens Relative DSM6698 Pseudomonas azotifigens Relative DSM17556 Pseudomonas stutzeri JM300 Relative DSM10701 Pseudomonas spp. All Relative Self - Lysate + Transposon ( P. stutzeri + miniTn7) Self Baltrus lab Self - Lysate ( P. stutzeri ) Self Baltrus lab None None All All Table S 4. 1 | Cell lysate sources. The halophiles were collected from salt springs in the Namib dessert. The conductivity of the spring at the time of collection is listed for Halomonas spp . The Pseudomonas relatives were purchased from the Ger man culture collection (accessions number listed). 77 96 - Microplate Well Evolution Treatment Cell Lysate Treatment 1a Constant Low Salinity Halomonas spp. 1 ( Halomonas taeanensis ) 2a Constant Low Salinity Halomonas spp. 2 ( Halomonas ) 3a Constant Low Salinity Halomonas spp. 3 ( Halomonas ) 4a Constant Low Salinity Halomonas spp. All (1+2+3) 4b Constant Low Salinity Halomonas spp. All (1+2+3) 5c Constant Low Salinity Pseudomonas chlororaphis subsp. aureofaciens 6g Constant Low Salinity Pseudomonas azotifigens 6h Constant Low Salinity Pseudomonas azotifigens 10c Constant Low Salinity Self Lysate ( P. stutzeri ) 11b Constant Low Salinity None 1a Increasing Salinity Halomonas spp. 1 ( Halomonas taeanensis ) 3c Increasing Salinity Halomonas spp. 3 ( Halomonas ) Table A 4. 2: Populations that did not revive. List of populations that did not revive at the end of the experiment. One population was contaminated (3c). . 78 Transformation Efficiency Constant Salinity Increasing Salinity 1.50% 3.00% 1.50% 3.00% Ancestor a a b a Self Lysate b a a a No Lysate b a ab a Halophile Lysate b a ab a Relatives Lysate b a a a All spp. b a ab a p<0.0001 p=0.055 p=0.0207 p=0.0506 Table A 4. 3: Expanded transformation efficiency results. Letters for pairwise comparisons across the cell lysate treatments with p - values listed at the bottom of the column. Table S 4. 3 | Expanded growth rate results. Letters for pairwise comparisons across the cell lysate treatments with p - values listed at the bottom of the column. 79 Growth Rate (r) Constant Salinity Increasing Salinity 1.50% 3.00% 1.50% 3.00% Ancestor a a a a Self Lysate a ab ab ab No Lysate a b a abc Halophile Lysate a b ab bc Relatives Lysate a b b c All spp. a b ab c p=0.697 p<0.001 p=0.08 p<0.001 Table A 4. 4: Expanded growth rate results. Letters for pairwise comparisons across the cell lysate treatments with p - values listed at the bottom of the column. 80 Log 10 (Populations size) Constant Salinity Increasing Salinity 1.50% 3.00% 1.50% 3.00% Ancestor abc ab a ab Self Lysate bc ab a ab No Lysate ab a b c Halophile Lysate abc b ab c Relatives Lysate a ab a bc All spp c ab a a p=0.008 p=0.02 p<0.001 p<0.001 Table A 4. 5: Expanded population size results. Letters for pairwise comparisons across the cell lysate treatments with p - values listed at the bottom of the column. 81 CHAPTER 5: CONCLUSIONS The goal of my dissertation was to understand if natural transformation promotes rapid adaptation in soil bacteria. I first studied eDNA dynamics after a drying - rewetting disturbance as increases in eDNA from bacterial death could promote transformation. Overall, I found that eDNA cycled rapidly through soil in response to drying rewetting , ultimately disappearing from soil as bacterial communities recover ed from the disturbance (Ch. 2 ). Interestingly , most sequences present in the eDNA po ol were also present in the live community, suggesting that eDNA even if slightly deleterious could promote genetic admixing within bacterial species recovering from a disturbance . I also used this dataset to test if including eDNA in microbial communi ty characterizations led to false conclusions about live bacterial communities. I showed that including eDNA in analyses minimally affected most conclusions about bacterial sensitivity but did mask subtle signatures of bacterial resilience and recovery pos t - disturbance that could ultimately skew predictions of ecosystem stability. In a second study , I quantified transformation in soil microcosms to determine the environmental drivers of eDNA acquisition (Ch. 3 ). I found that P. stutzeri could transform eDNA under most soil conditions including those seen across the drying rewetting disturbance. I show that eDNA encoding antibiotic resistance genes can be transformed at high rates in soil. These findings suggest that widespread efforts to reduce the spread of antibiotic resistan ce genes in the environment should incorporate methods that eliminate both live and extracellular sources of drug resistance. 82 In a final laboratory experiment, I tested the fitness effects of transformation by adapting P. stutzeri to high salt concentrations . After ~330 generations of adaptation, I quantified the growth rate and transformation efficiency of the evolved populations . I found that P. stutzeri grew fa ster than the ancestor at high salt concentrations, but that the transformation efficiency was greatly diminished (Ch. 4 ). Overall, ~50% of the evolved populations did not undergo transformation, despite the original or ancestral populations all undergoing transformation. This suggests that transformation was not responsible for salt adaptation and may have been selected against during experimental evolution. In general, this work contributes to the larger question of why bacteria undergo transformation, an d sets the stage for future experiment s to investigate the question: under what conditions do transformable cells outcompete their non - transformable counterparts? Taken together, I find that eDNA cycles through soil rapidly after a pulse disturbance and is readily transformed under these soil conditions. I also find that the transformation efficiency can change dramatically over just ~330 generations , suggest ing that transformation capacity may be an adaptive trait that diminishes in the absence of selection . Overall, this body of research supports the hypothesis that transformation is periodically adaptive but often maladaptive and is likely important in spatially structured environments where new niches regularly open for invasion [11] . It also suggests that we need to expand our view of the community metagenome to include extracellular DNA which is biologically active after bacterial death through the process of natural transformation. 83 BIBLIOGRAPHY 84 BIBLIOGRAPHY [1] E. Low - Décarie et al. Proc. Natl. Acad. Sci. , no. 10, p. 201513125, 2015. [2] N. D. Youngblut, A. Shade, J. S. Read, K. D. Mcmahon, and R. J. Whitaker, - Appl. Environ. Microbiol. , vol. 79, no. 1, pp. 39 47, 2013. [3] sistance, resilience, and redundancy in Proc. Natl. Acad. Sci. , vol. 105, no. Supplement 1, pp. 11512 11519, 2008. [4] Front. Microbiol. , vol. 5, no. SEP, pp. 2012 2014, 2014. [5] Mol. Biol. Evol. , vol. 20, no. 10, pp. 1598 1602, 2003. [6] H. Ochman, J. G. L Nature , vol. 405, no. 6784, pp. 299 304, 2000. [7] C. J. H. Von Wintersdorff et al. microbial ecosystems through hor Front. Microbiol. , vol. 7, no. FEB, pp. 1 10, 2016. [8] Science (80 - . ). , vol. 321, 2008. [9] Curr. Biol. , vol. 26, no. 3, pp. R112 R115, 2016. [10] Philos. Trans. B , 2019. [11] E. V. F1000Research , vol. 5, no. 0, p. 1805, 2016. [12] f Genome Res. , vol. 16, no. 5, pp. 636 643, 2006. [13] 85 Mol. Biol. Evol. , vol. 23, no. 12, pp. 2379 2391, 2006. [14 ] BMC Evol. Biol. , vol. 7, no. SUPPL. 1, pp. 27 29, 2007. [15] and its Genome Biol. Evol. , vol. 6, no. 6, pp. 1514 1529, 2014. [16] ISME J. , vol. 393, pp. 199 208, 2009. [17] R. Theor. Popul. Biol. , vol. 61, no. 4, pp. 489 495, 2002. [18] mutati Proc. Natl. Acad. Sci. U. S. A. , vol. 116, no. 36, pp. 17906 17915, 2019. [19] 2007. [20] Curr. Biol. , vol. 26, no. 21, pp. R1126 R1130, 2016. [21] K. L. Meibom, M. Blokesch, N. A. Dolganov, C. - Y. Wu, and G. K. Schoolnik, 5. [22] Appl. Environ. Microbiol. , vol. 58, no. 6, pp. 1930 1939, 1992. [23] M. G. Lorenz and W. Wackernag Microbiol. Rev. , vol. 58, no. 3, pp. 563 602, 1994. [24] Proc. Natl. Acad. Sci. , vol. 107, no. 13, pp. 5881 5886, 2010. [25] Philos. Trans. R. Soc. B Biol. Sci. , vol. 375, no. 1798, 2020. [26] D. J. Levy - Booth et al. Soil Biol. Biochem. , vol. 39, no. 12, pp. 2977 2991, 2007. 86 [27] transformation of Acinetobacter sp. by homology - facilitated illegitim ate PNAS , vol. 99, no. 4, pp. 2094 2099, 2001. [28] transformation are associated with genomic subgroups within a local population of Pseudomonas stutzeri fro Appl. Environ. Microbiol. , vol. 68, no. 2, pp. 865 873, 2002. [29] C. A. Simpson, R. Podicheti, D. B. Rusch, A. B. Dalia, and C. Van Kessel, pp. 1 16, 2019. [30] Spec. Publ. Georg. Agric. Exp. Station. , 1983. [31] K. M. Nielsen, P. J. Johns Environ. Biosafety Res. , vol. 6, no. 1 2, pp. 37 53, 2007. [32] E. Gallori et al. - ster FEMS Microbiol. Ecol. , vol. 15, no. 1 2, pp. 119 126, 1994. [33] P. Carini, P. J. Marsden, J. W. Leff, E. E. Morgan, M. S. Strickland, and N. Fierer, Nat. Mic , v ol. 53, no. 9, p. 680840, 2016. [34] Appl. Environ. Microbiol. , vol. 53, no. 12, pp. 2948 2952, 1987. [35] L. S. England, L. Hung, and J. T. Trevo Soil Biol. Biochem. , vol. 29, no. 9 10, pp. 1521 1527, 1997. [36] transformation of Pseudomonas stutzeri in a non - Microbiology , vol. 144, no. 2, pp. 569 576, 1998. [37] Biophys. J. , vol. 106, no. February, pp. 944 9 54, 2014. [38] C. S. Smillie, M. B. Smith, J. Friedman, O. X. Cordero, L. A. David, and E. J. Alm, Nature , vol. 480, no. 7376, pp. 241 244, 2011. 87 [39] J. Felsenstein and S. Genetics , vol. 83, no. 4, pp. 845 859, 1976. [40] Mutat. Res. - Fundam. Mol. Mech. Mu tagen. , vol. 1, no. 1, pp. 2 9, 1964. [41] Ninth Int. Conf. Artif. Life , pp. 340 345, 2004. [42] competence and transformation in pathogenic and environmental Gram - negative FEMS Microbiol. Rev. , vol. 37, no. 3, pp. 336 363, 2013. [43] G. Carvalho et al. transformation buffers environmental fluctuations 12, 2019. [44] contribution of natural transformation to the shuffling Curr. Opin. Microbiol. , vol. 38, pp. 22 29, 2017. [45] Nature Reviews Microbiology , vol. 12, no. 3. 2014. [46] S. Gui - programmed predation of noncompetent cells in the human pathogen Streptococcus PNAS , 2005. [47] trategy for DNA Nat. Rev. Microbiol. , vol. 15, no. 10, pp. 621 629, 2017. [48] om dead cells and population G3 Genes, Genomes, Genet. , vol. 4, no. 2, pp. 325 339, 2014. [49] J.Hered. , vol. 84, no. February, pp. 339 344, 1993. [50] Evolution (N. Y). , vol. 62, no. 1, pp. 39 49, 2007. [51] D. J. P. Engelmoer, I. Donaldson, and D. E. PLoS Pathog. , vol. 9, no. 11, pp. 1 7, 2013. [52] Trends Ecol. Evol. , 88 vol. 28, no. 8, pp. 489 495, 2 013. [53] [54] Evolution ( N. Y). , no. 1, 2011. J. Bacteriol. , vol. 188, no. 24, pp. 8534 8542, 2006. [56] S. Graupner, V. Frey, R. Hashemi, and M. G. and pilC of Pseudomonas stutzeri Are Required for Natural Genetic Transformation , and pilA Can Be Replaced by Corresponding Genes from 2190, 2000. [57] M. G. Lorenz and transformation of Pseudomonas stutzeri in soil extract supplemented with a Appl. Environ. Microbiol. , vol. 57, no. 4, pp. 1246 1251, 1991. [58] S. Domingues, K. Harms , W. F. Fricke, P. J. Johnsen, G. J. da Silva, and K. M. PLoS Pathog. , vol. 8, no. 8, 2012. [59] M. Camiade et al. esistance patterns of Pseudomonas spp. isolated FEMS Microbiol. Ecol. , vol. 96, no. 2, pp. 1 15, 2020. [60] al ISME J. , vol. 7, no. 11, pp. 2229 2241, 2013. [61] S. J. Blazewicz, E. Schwartz, M. K. Firestone, S. J. Blazewicz, E. Schwartz, and derlie rainfall - induced 1162 1172, 2017. [6 2] 14, 2018. [63] E. K. Bünemann, B. Keller, D. Hoop, K. Jud, P. Boivin, and E. Frossard, ailability of phosphorus after drying and rewetting of a grassland Plant Soil , vol. 370, no. 1 2, pp. 511 526, 2013. 89 [64] floors: dynamics of soluble phosphorus, microbial biomass - phosphorus, and the Biol. Fertil. Soils , vol. 54, no. 6, pp. 761 768, 2018. [ 65] Responses to Soil Drying - Front. Environ. Sci. , vol. 7, no. September, pp. 1 9, 2019. [66] rning extracellular DNA degradation Environ. Microbiol. Rep. , vol. 11, no. 2, pp. 173 184, 2019. [67] S. Karlowsky, A. Augusti, J. Ingrisch, M. K. U. Akanda, M. Bahn, and G. Gleixner, - induced accumulation of root exudates suppor ts post - drought recovery Front. Plant Sci. , vol. 871, no. November, pp. 1 16, 2018. [68] D. Naylor and D. Coleman - - associated bacterial Front. Plant Sci. , vol. 8, no. January, pp. 1 16, 2018. [69] moisture dependences and responses to drying rewetting: The legacy of 18 Glob. Chang. Biol. , vol. 25, no. 3, pp. 1005 1015, 2019. [70] M . D. Luehmann et al. - storied soils on the outwash plains of Ann. Am. Assoc. Geogr. , vol. 106, no. 3, pp. 551 571, 2016. [71] J. J. Kozich, S. L. Westcott, N. T. Bax ter, S. K. Highlander, and P. D. Schloss, - index sequencing strategy and curation pipeline for Appl. Environ. Microbiol. , vol. 79, no. 17, pp. 5112 5120, 201 3. [72] PLoS Comput. Biol. , vol. 10, no. 4, 2014. [73] - Psychol. Bull. , vol. 96, no. 3, pp. 573 580, 1984. [74] Agric. Ecosyst. Environ. , vol. 266, no. April, pp. 122 132, 2018. [75] nitmicrobial Resistance. Tackling drug - resistant infections [76] resistance in the environment and its relevance to environmental regulators: Fi nal 90 Front. Microbiol. , vol. 7, no. NOV, pp. 1 22, 2016. [77] R. L. Finley et al. Clin. Infect. Dis. , vol. 57, no. 5, pp. 704 710, 2013. [78] B. Ber Infect. Ecol. Epidemiol. , vol. 5, no. 1, p. 28564, 2015. [79] bacteria: Relationships between resistance determinants of antibiotic producers, Front. Microbiol. , vol. 9, no. NOV, pp. 1 21, 2018. [80] X. Jiang et al. antibiotic Nat. Commun. , vol. 8, pp. 1 7, 2017. [81] J. Bengtsson - FEMS Microbiol. Rev. , vol. 42, no. 1, pp. 68 80, 2018. [82] Curr. Opin. Microbiol. , vol. 53, pp. 35 43, 2020. [83] L. Poirel, J. M. Rodriguez - Martinez, H. Mammeri, A. Liard, and P. Nordmann, - Antimicrob. Agents Chemother. , vol. 49, no. 8, pp. 3523 3525, 2005. [84] L. Poirel, P. Kämpfer, and P. Nordmann, - encoded ambler class a - lactamase of Kluyvera georgiana, a probable progenitor of a subgroup of CTX - M extended - - Antimicrob. Agents Chemother. , vol. 46, no. 12, pp. 4038 4040, 2002. [85] Curr. Opin. Microbiol. , vol. 13, no. 5, pp. 589 594, 2010. [86] antibiotic resistance genes (eARGs) in typical environmental samples and the Environ. Int. , vol. 125, no. October 2018, pp. 90 96, 2019. [87] D. Mao et al. Environ. Sci. Technol. , vol. 48 , no. 1, pp. 71 78, 2014. [88] 91 Environ. Microbiol. , vol. 9, no. 3, pp. 657 666, 2007. [89] C. T. T. Binh, H. Heuer FEMS Microbiol. Ecol. , vol. 66, no. 1, pp. 25 37, 2008. [90] ransfer of antibiotic Can. J. Microbiol. , vol. 65, no. 1, pp. 34 44, 2019. [91] S. Molin and T. Tolker - biofilms and induces enhanced stabilisation of the Curr. Opin. Biotechnol. , vol. 14, no. 3, pp. 255 261, 2003. [92] FEMS Immunol. Med. Microbiol. , vol. 65, no. 2, pp. 183 195, 2012. [93] Appl. Microbiol. Biotechnol. , vol. 87, no. 3, pp. 925 941, 2010. [94] Q. Chang, W. Wang, G. Regev - Yochay, M. Lipsitch, and W. P. Hanage, Evol. Appl. , vol. 8, no. 3, pp. 240 247, 2015. [95] between beneficial mutations in popula PLoS Biol. , vol. 5, no. 9, pp. 1899 1905, 2007. [96] - Resistant Enterococci Lack CRISPR - MBio , vol. 1, no. 4, pp. 1 10, 2010. [97] antibiotic resistance in J. Basic Microbiol. , vol. 58, no. 11, pp. 905 917, 2018. [98] Y. G. Zhu et al. Proc. Natl. Acad. Sci. U. S. A. , vol. 110, no. 9, pp. 3435 3440, 2013. [99] abundance, and persistence of antibiotic resistance genes in various types of J. Hazard. Mater. , vol. 344, pp. 716 722, 2 018. [100] management of intensively cropped agro - ecosystems improves soil quality with Agric. Ecosyst. Environ. , vol. 140, no. 3 4, pp. 419 429, 92 2011. [ 101] Genome Announc. , vol. 2, no. 3, p. 2014, 2014. [102] K. - - Tn7 insertion in bacteria with single attTn7 Nat. Protoc. , vol. 1, no. 1, pp. 153 161, 2006. [103] E. M. Morrissey et al. bacte Soil Biol. Biochem. , vol. 86, pp. 42 49, 2015. [104] A. Romanchuk et al. Plasmid , vol. 73, pp. 16 25, 2 014. [105] G. Pietramellara, J. Ascher, F. Borgogni, M. Ceccherini, G. Guerri, and P. 235, 2009. [106] T. Zhang et al. FEMS Microbiol. Ecol. , vol. 96, no. 8, pp. 1 2, 2020. [107] A. Ezzariai et al. or manure: Global perspectives on persistence, degradation, and resistance J. Haza rd. Mater. , vol. 359, no. July, pp. 465 481, 2018. [108] H. Li et al. Ecotoxicol. Environ. Saf. , vol. 140, no. February, pp. 1 6, 2017. [109] C. W. McKinney and extracellular and intracellular antibiotic resistance genes in manure - amended soil: Soil Sci. Soc. Am. J. , vol. 84, no. 3, pp. 747 759, 2020. [110] K. M. Nielsen, M. D. M. van Weerelt, T. N. Berg, A. M. Bones, A. N. Hagler, and J. Acinetobacter calcoaceticus Appl. Environ. Microbiol. , vol. 63, no. 5, pp. 1945 1952, 1997. [111] - to - cell bacterial Proc. Natl. Acad. Sci. U. S. A. , vol. 115, no. 39, pp. 9791 9796, 2018. [112] A. Dechesne, - controlled Proc. Natl. Acad. Sci. U. S. A. , vol. 93 107, no. 32, pp. 14369 14372, 2010. [113] - based approach to bac terial biofilms in Environ. Microbiol. , vol. 18, no. 8, pp. 2732 2742, 2016. [114] polymeric substances in biofilm formation by Pseudomonas stutzeri strain XL - Appl. Micro biol. Biotechnol. , vol. 103, no. 21 22, pp. 9169 9180, 2019. [115] M. Wilson, R. Mcnab, and B. Henderson, Chapter 14 - Bacterial Cell Biology and Development . 2017. [116] O. H. Ambur, J. Engelstädter, P. J. Johnsen, E. L. Miller, and D. E. Rozen, t the wheel: Conservative sex and the benefits of bacterial Philos. Trans. R. Soc. B Biol. Sci. , vol. 371, no. 1706, 2016. [117] the Fisher - G enetics , vol. 171, no. 3, pp. 1377 1386, 2005. [118] 1483, 2014. [119] Evolution of Bacterial [120] [121] J. as a General Response to Stress in Gram - [122] - your - cake and - eat - lt - too of J. Hered. , vol. 84, no. 5, pp. 400 404, 1993. [123] vol. 88, no. May, pp. 1106 1119, 2013. [124] R. Genetics , vol. 118, no. 1, pp. 31 39, 1988. [125] for genetic Curr. Opin. Microbiol. , vol. 15, no. 5, pp. 570 576, 2012. [126] G. G. Perron, A. E. G. Lee, Y. Wang, W. E. Huang, and T. G. B arraclough, 94 - drug - resistance in Proc. R. Soc. B Biol. Sci. , vol. 279, no. 1733, pp. 1477 1484, 2012. [127] A. L. G. Utnes et al. - specific evolutionary benefits of natural 2231, 2015. [128] natural transformation on the evolution of resistance to bacteriophages in the Acinetob Nat. Sci. Reports , no. November, pp. 6 11, 2016. [129] Competence for Transformation Due to Host - 22, 2018. [130] N. H. Barton an September, pp. 1986 1991, 1998. [131] Nat. Rev. Genet. , vol. 16, no. 8, pp. 472 482, 2015. [132 ] U. Gonzales - distributions and their zero - modified equivalents as a framework for modelling Int. J. Food Microbiol. , vol. 136, no . 3, pp. 268 277, 2010. [133] Caries Res. , vol. 50, no. 6, pp. 517 526, 2016. [134] sensing regulation of competence and Genes (Basel). , vol. 8, no. 1, 2017. [135] - dependent regulation of natural competence and type VI s Nucleic Acids Res. , vol. 46, no. 20, pp. 10619 10634, 2018. [136] Biol. Philos. , vol. 35, no. 1, pp. 1 22, 2020. [137] S. Cesar et al. , - 4, pp. 1 17, 2020. [138] S. Graupner, V. Frey, R. Hashemi, M. G. Lorenz, G. Brandes, and W. required for natural genetic transformation, and pilA can be replaced by 95 J. Bacteriol. , vol. 182, no. 8, pp. 2184 2190, 2000. [139] Competence Genes pilT and pilU of Pseudomonas stutzeri for Natural Transformation and Suppression of pilT Deficiency by a Hexahistidine Tag on the 4701, 2001.