UNDERSTANDING PLANT CELL WALL PHENOTYPES THAT CONTRIBUTE RECALCITRANCE TO ALKALINE -OXIDATIVE PRETREATMENTS AND ENZYMATIC HYDROLYSIS By Muyang Li A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Biosystems Engineering - Doctor of Philosophy 2015 ABSTRACT UNDERSTANDING PLANT CELL WALL PHENOTYPES THAT CONTRIBUTE RECALCITRANCE TO ALKALINE -OXIDATIVE PRETREATMENTS AND ENZYMATIC HYDROLYSIS By Muyang Li Plant cell walls represent the majority of the terrestrial plants by mass and mainly consist of polysaccharides and lignin. These constituents are considered as an abundant and renewable carbon resource providing great potential for the production of petroleum -displacing biofuels and biochemicals. However, the complex, rigid and heterogeneous plant cell wall structure typically requires a combination of biological and chemical treatments to break down the cell wall and maximize the yield of the products. Pretreatments are thermal or chemical processes that facilitate the subsequent sugar -releasing enzymatic hydrolysis. Alkaline pretreatm ents and alkaline -oxidative pretreatments are one of the main categories of pretreatments that function well on monocot grasses, which include the cereal stovers and the energy crops. These pretreatments target ester cross -links b etween lignin and hemicell ulose s within the grass cell walls formed by the hy droxycinnamates, as well as the ester linked side chains such as acetyl groups in hemicelluloses, results in significant remov al of lignin and hemicelluloses. The removal of these cell wall constituents significantly increase s the accessibility of water and enzymes to the cell walls . Structural characteristics including higher order structures within the cell wall, covalent and non -covalent cross -linking, and porosity may all impact the conversion process. Also, plant cell walls can exhibit substantial heterogeneity between plant cell type s and tissue s. The goal of this work is to employ n ovel and conventional characterization tools to identify how properties of diverse cell walls are impacted by pretreatmen ts and how this correlates to reduced cell wall ÒrecalcitranceÓ, wh ich is important for both plant design with properties suited for deconstruction and the design of deconstruction strategies. This work consists of several related studies. T he roles that glycans play in the plant cell wall recalcitrance was investigated using taxonomically and structurally diverse bi omass feedstocks. Their responses to alkaline oxidative pretreatment and how differing features of the cell wall matrix contribute to its reca lcitrance was assessed by a set of studies. Different mechanisms of improving hydrolysis yields following alk aline oxidative pretreatment were discovered for monocot s and dicots. In monocot grasses, t he ÒlooseningÓ of the glycans due to alkaline oxidative pretreatment indicating the overall weaker associations between cell wall polymers in grasses than in dicots , which results in significantly improved hydrolysis yields after the pretreatment. It was found that the hydrolysis yiel ds of alkali pretreated corn stovers are not necessarily correlated with the original contents of lign in and hydroxycinamates , but they are highly related with the removal of these compounds. Multivariate models were applied to predict the hydrolysis yields based on cell wall properties quantified or predicted by high-throughput pyrolysis molecular beam mass spectrometry (py -MBMS). It was found that the untreated cell wall composition properties including lignin and xylan content, as well as the hydrophilicity , set the initial rec alcitrance of the cell wall s. However, f or the alkaline pretreated corn stover , hydrolysis yields were only predictable by the release of lignin and hydroxycinna mates. These findings indicate that the final yield of a biomass feedstock is not necessarily c orrelated with the structural features , but is highly pretreatment -oriented , and the properties contributing to a Òreduced recalcitranceÓ phenotype following a specific pretreatment are not necessarily the same properties that contribute to recalci trance i n untreated cell wall . iv ACKNOWLEDGEMENTS The author would like to thank Dr. David Hodge, her primary academic advisor, for guiding her to conduct research in the past five years. Besides the guidance in experimental design and data analysis, Dr. David Hodge also gave her a lot of help and advices on English writing and presenting, which she appreciated the most and let her become a confident researcher and qualified doctorate candidate from a non -native English speaker with low langu age proficiency in the beginning. The author would like to thank her collaborators, Dr. Sivakumar Pattathil (University of Georgia), Robert Sykes (NREL), and Dr. Natalia de Leon (University of Wisconsin), who kindly provided their expertise in glycome prof iling, pyrolysis molecular beam mass spectrometry and maize breeding, respectively. This work was funded by Great Lake Bioenergy Research Center and National Science Foundation. v TABLE OF CONTENTS LIST OF TABLES ................................................................................................................. VII LIST OF FIGURES ............................................................................................................... VIII ! KEY TO ABBREVIATIONS ................................................................................................. XI! 1. INTRODUCTION ................................................................................................................. 1!1.1 BACKGROUND ............................................................................................................. 1!1.2 PLANT CELL WALLS ................................................................................................... 2!1.2.1 Plant cell wall organization ....................................................................................... 2!1.2.2 Plant cell wall lignins ................................................................................................ 3!1.2.3 Plant cell wall polysaccharides ................................................................................. 4!1.3 DECONSTRUCTION O F PLANT CELL WALLS ....................................................... 4!1.3.1 Cell wall deconstruction to fuels and chemicals ....................................................... 4!1.3.2 Pretreatments ............................................................................................................. 5!1.3.2 Enzymatic hydrolysis ................................................................................................ 7!1.4 PLANT CELL WALL RECALCITRANCE ................................................................... 8!1.4.1 Cell wall lignin .......................................................................................................... 8!1.4.2 Cell wall polysaccharides ......................................................................................... 9!1.4.3 Cell wall ultrastructure ............................................................................................ 10!1.4.4 Grass cell wall recalcitrance ................................................................................... 10!1.5 PLANT CELL WALL CHARACTERIZATION AND MODELING ......................... 11!1.5.1 Characterization of cell wall struc tural carbohydrates ............................................ 11!1.5.2 Characterization on cell wall lignins ....................................................................... 12!1.5.3 Novel characterization tools using specific proteins .............................................. 12!1.5.4 Data mining and models ......................................................................................... 13!1.6 OBJECTIVES ................................................................................................................ 14! 2. PLANT CELL WALL R ECALCITRANCE IN DIVE RSE BIOENERGY FEEDST OCKS ................................................................................................................................................. 16!2.1 INTRODUCTION ......................................................................................................... 16!2.2!MATERIALS AND METHOD S ............................................................................... 17!2.2.1 Pretreatment ............................................................................................................ 17!2.2.2 Composition analysis, enzymatic hydrolysis and digestibility determination ........ 18!2.2.3 Sequent ial extraction, glycome profiling ................................................................ 18!2.3 RESULTS AND DISC USSIONS ................................................................................. 19!2.3.1 Changes in composition and mass loss ................................................................... 19!2.3.2 Digestibility of poplar, goldenrod, and corn stover ................................................ 22!2.3.3 Glycome profiling of poplar, goldenrod, and corn stover ...................................... 23!2.4 CONCLUSION ............................................................................................................. 32! 3. PLANT CELL WALL R ECALCITRANCE IN DIVE RSE GRASSES ............................. 34!3.1 INTRODUCTION ......................................................................................................... 34!3.2 MATERIALS AND ME THODS .................................................................................. 35!3.2.1 Biomass pretreatment, composition analysis and enzymatic hydrolysis ................ 35!3.2.2 p-hydroxycinnamic acids determination and metal analysis .................................. 36!vi 3.2.3 Glycome profiling and data processing .................................................................. 37!3.2.4 ELISA screening of pretreatment liquors ............................................................... 37!3.2.5 Enzyme binding experiments .................................................................................. 38!3.3 RESULTS AND DISC USSIONS ................................................................................. 38!3.3.1 Correlations between grass cell wall properties and digestibility ........................... 38!3.3.2 Glycome profiling derived xylan/ !-glucan extractability of cell walls .................. 44!3.3.3 ELISA screening of the pretreatment liquor ........................................................... 50!3.3.4 Enzyme accessibility of grass cell wall s ................................................................. 52!3.4 CONCLUSION ............................................................................................................. 54! 4. CORRELATING MAIZE CELL WALL PROPERTIES TO THE RECALCITRANCE .. 56!4.1 INTRODUCTION ......................................................................................................... 56!4.2 MATERIALS AND ME THODS .................................................................................. 58!4.2.1 Maize diversity panel .............................................................................................. 58!4.2.2 Water Retention Value (WRV) measurement ........................................................ 59!4.2.3 p-hydroxycinnamic acids determination ................................................................. 59!4.2.4 Alkaline extraction, composition analysis and enzymatic hydrolysis .................... 60!4.2.5 S/G ratio prediction ................................................................................................. 60!4.3 RESULTS AND DISC USSION .................................................................................... 61!4.3.1 Cell wa ll properties and hydrolysis yields .............................................................. 61!4.3.2 Correlation between cell wall properties ................................................................ 63!4.3.3 Correlations between cell wall properties and hydrolysis yields ............................ 67!4.4 CONCLUSION ............................................................................................................. 74! 5. MODELING OF THE P LANT CELL WALL RECAL CITRANCE ................................. 76!5.1 INTRODUCTION ......................................................................................................... 76!5.2 MATERIALS AND ME THODS .................................................................................. 79!5.2.1 The determination of the maize cell wall composition properties .......................... 79!5.2.2 Py -MBMS analysis and modeling .......................................................................... 80!5.2.3 Multivariate model development ............................................................................ 80!5.3 RESULTS AND DISC USSIONS ................................................................................. 81!5.3.1 pCA/FA determination and prediction .................................................................... 81!5.3.2 PCA/PLS models for predicting hydrolysis yields ................................................. 84!5.3.3 MLR models for predicting hydrolysis yields using quantified properties ............ 85!5.4 CONCLUSION ............................................................................................................. 89! 6. CONCLUSIONS ................................................................................................................. 90! APPENDIX ............................................................................................................................. 92! BIBLIOGRAPHY ................................................................................................................. 100!vii LIST OF TABLES Table 2.1 Composition of untreated biomass on a whole sample basis rather than an extractives -free basis. Errors represent data range of duplicate measurements. ..................... 20! Table 3.1 Metal analysis of grasses ......................................................................................... 41! Table 3.2 Parameters fitting based on Langmuir adsorption model showing binding capacity and affinity of CtCBM3 on corn stovers under different treatments. ..................................... 53! Table 4.1 Variability within the data set for the 12 properties across the 27 maize lines. ...... 62! Table 4.2 Calculated proportionality constants (unscaled) between all cell wall properties and hydrolysis yields. * indicates the parameter is significant at p ! 0.05. ** indicates the param eter is significant at p ! 0.01. ........................................................................................ 68! Table 5.1 Models based on C p and B IC criteria and the modified models. ............................ 85! Table A.1 Listing of plant cell wall glycan -directed monoclonal antibodies (mAbs) used for glycome profiling analyses. The groupings of antibodies are based on a hierarchical clustering of ELISA data generated from a screen of all mAbs against a panel of plant polysaccharide preparations that groups the mAbs according to the predominant polysaccharides that they recognize. The majority of listings link to the Wall MabDB plant cell wall monoclonal antibody database ( http://www.wallmabdb.net ) that provides detailed descriptions of each mAb, including immunogen, antibody isotype, epitope structure (to the extent known), supplier information, and related literature citations. .................................... 93! Table A.2 Genotype and reference source for the 27 maize lines used in this work. ............. 99! viii LIST OF FIGURES Figure 2.1 Compositional changes (extractives -free basis) associated with AHP pretreatment with increasing H 2O2 loading showing solubilization of total cell wall mass (A), xylan (B), and lignin (C). ......................................................................................................................... 22! Figure 2.2 Enzymatic conversion of cell wall glucan to glucose in untreated and AHP pretreated biomass with increasing H2O2 loading for poplar, goldenrod, and corn stover with cellulase (Cellic CTec2) for 24 h and 72 h. ............................................................................. 22! Figure 2.3 Glycome profiling of Poplar, Goldenrod and Corn stover biomass samples before and after AHP pretreatment (12.5% H 2O2 loading). (Sequential extraction and glycome profiling were done by Sivakuma r Pattathil, University of Georgia.) .................................... 25! Figure 2.4 Comparison of the changes in the relative abundance of three glycan epitope categories due to AHP pretreatment from the oxalate (blue dots) and carbonate (red dots) extracts in glycome pro !"ling, for poplar (A, B and C, for xyloglucans, xylans and pectic polysaccharides), goldenrod (D #F for xyloglucans, xylans and pectic polysaccharides) and corn stover (G #I for xy loglucans, xylans and pectic polysaccharides). .................................. 28! Figure 2.5 Comparison of the changes in the relative abundance of three glycan epitope categories due to AHP pretreatment from each of the four harshest extracts including 1 M KOH (green triangles), 4 M KOH (blue diamonds), chlorite (red squares), and 4 M KOH PC (purple crosses) in glycome pro !"ling. ..................................................................................... 29! Figure 3.1 Glucan digestibility (hydrolysis yield) profile associated with H 2O2 loading after AHP pretreatment for the diverse grasses (0% AHP are untreated samples). ........................ 39! Figure 3.2 Composition and weight loss of switchgrass, miscanthus, bm1, bm3 and hybrid corn stover. .............................................................................................................................. 40! Figure 3.3 The c orrelation between the metal content (A. Fe content, B. Total metal content) and glucose yield for the diversely AHP pretreated grasses including corn stover, bm1 stover, bm3 stover, switchgrass and miscanthus. ................................................................................ 41! Figure 3.4 Heat map of correlation coefficients between cell wall properties and digestibility. The color blocks in the map reveal the relationship between the specific two individual cell wall properties as horizontally and vertically labeled, where the darker colors indicate positive correlations while the lighter colors indicate negative correlations. Note: this only ix includes data of 12.5% AHP pretreated and untreated corn stover, switchgrass and miscanthus. .............................................................................................................................. 43! Figure 3.5 Plots of binding quantity of individual mAbs for xylan epitopes in extracts of 1 M KOH, 4 M KOH, chlorite, and post chlorite 4 M KOH. Values on X or Y axis are respectively the binding strength of xylan specific mAbs on untreated or pretreated switchgrass (A), miscanthus (B), corn stover (C), bm1 stover (D), bm3 stover (E) and alkali -extracted corn stover (F) ......................................................................................................... 46! Figure 3.6 Plots of individual mAbs (LAMP and CCRC -M154) for !-glucan epitodes binding strength on untreated and pretreated switchgrass (SG), miscanthus (Mis), corn stover (CS), bm1 and bm3 stovers. .............................................................................................................. 48! Figure 3.7 Heat map of major cell wall non -cellulosic glycans extractability alteration by AHP pretreatment. ................................................................................................................... 50! Figure 3.8 ELISA screening of glycans solubilized in 12.5% AHP pretreatment liquors of (A) corn stover, (B) bm1 stover, (C) bm3 stover, (D) miscanthus and (E) switchgrass using 155 cell wall glycan -directed mAbs. (ELISA screening was done by Sivakumar Pattathil, University of Georgia.) ........................................................................................................... 52! Figure 3.9 Binding isotherms of CtCBM3 on untreated, 12.5% and 25% H2O2 (w/w) AHP pretreated corn stover. ............................................................................................................. 53! Figure 4.1 Range of hydrolysis yields obtained for untreated and NaOH -pretreated maize for (A) 6-hr hydrolysis yields and (B) 72 -hr hydrolysis yields. Error bars represent data range for duplicate samples. Due to missing data, some samples do not appear. .................................. 62! Figure 4.2 Correlation map of the PearsonÕs correlation coefficients for the 12 cell wall properties and 4 hydrolysis yields (red text) across the 27 maize lines as organized by hierarchical cluster analysis. Clusters of properties and yields exhibiting are highlighted. ÒInitialÓ indicates the property in the original untreated biomass sample while ÒfinalÓ indicates the property following pretreatment. ....................................................................... 64! Figure 4.3 Several between -property correlations highlighted to demonstrate relationships within the data set. Each data point represents the property for one of the 27 maize lines. PearsonÕs correlation coefficients and p-values are presented for all property correlations with p ! 0.05.Error bars on individual samples are not shown to improve clarity. ........................ 65! Figure 4.4 Summary of significant property correlations to gluco se hydrolysis yields. Open data points represent 6 -hr hydrolysis yields for untreated biomass; filled data points represent x 72-hr hydrolysis yields for NaOH -pretreated biomass. Each data point represents the value of the property and corresponding yield fo r one of the 27 maize lines. PearsonÕs correlation coefficients and p-values are presented for all property correlations with p ! 0.05. Error bars on individual samples are not shown to improve clarity. ........................................................ 70! Figure 5.1 pCA and FA solubilization in different severity of alkali treatments. ................... 82! Figure 5.2 Correlation between the mass peak intensity of 120 m/z (A) and 150 m/z (B) with pCA (A) and FA (B) content. .................................................................................................. 83! Figure 5.3 Comparison between the measured values vs. predicted values based on PLS regression coupled with PCA analysis of py -MBMS spectra for the cell wall pCA content (A) and FA content (B) of 27 maize lines. .................................................................................... 83! Figure 5.4 Comparison between the measured values vs. predicted values based on PLS regression coupled with PCA analysis of py -MBMS spectra for the 6 -hr (A) and 72 -hr (B) hydrolysis yield for 27 maize lines before pretreatment and 72 -hr hydrolysis yield after NaOH pretreatment (C). .......................................................................................................... 84! Figure 5.5 The correlation between the predicted yie lds and the measured yields for 6 -hour hydrolysis yield of untreated maize (red square), 72 -hour hydrolysis yield of untreated maize (green triangle) and 72 -hour hydrolysis yield of NaOH pretreated maize (blue diamond) using PCA/PLS model (A) and MLR model (B). ............................................................................. 88! Figure A.1 Glycome profiling results of untreated and 12.5% AHP pretreated miscanthus and switchgrass. ............................................................................................................................. 97! Figure A.2 Glycome profiling results of untreated and 12.5% AHP pretreated bm1 and bm3 stover. ...................................................................................................................................... 98! xi KEY TO ABBREVIATIONS AFEX ammonia fiber expansion AG arabinogalactan AHP alkaline hydrogen peroxide AIC Akaike information criterion BIC Bayesian information criterion CAD cinnamyl alcohol dehydrogenase CBM carbohydrate binding module CCRC complex carbohydrates research center CDTA 1,2-Cyclohexylenedinitrilotetraacetic acid CoMPP comprehensive microarray polymer profiling COMT caffeic acid O -methyl transferase Cp Cathodic potential CS corn stover Ct Clostridium thermocellum DI de-ionized ELISA enzyme -linked immuno assay FA ferulate /ferulic acid FTIR Fourier transform infrared spectroscopy G guaiacyl GAX glucuronoarabinoxylan GC gas chromatography GFP green fluorescent protein GH glycosyl hydrolase xii GHG greenhouse gas Glc glucose/glucan GM glucomannan GP glycome profiling GX glucuxylan H p-hydroxyphenyl HCA hydroxycinnamate/hydroxycinnamic acid HG/HGA homogalacturonan HPLC High -performance liquid chromatography HSQC heteronuclear single quantum coherence mAb monoclonal antibody MBMS molecular beam mass spectrometry MLG mixed -linked glycan MLR multiple linear regression MS mass spectroscopy NIR Near-infrared spectroscopy NMR nuclear magnetic resonance NREL national renewable energy laboratory PC post -chlorite pCA para -coumarate/para -coumaric acid PCA principal component analysis PLS partial least square py pyrolysis RG-I rhamnogalacturonan I S syringyl xiii SG switchgrass WRV water retention value XyG xyloglucan 1 1. INTRODUCTION 1.1 BACKGROUND The consumption of fossil fuels from ancient plants buried under the ground for approximately 300 million years is the primary contributor to GHGs emission s. Lignocellulosic biomass , as the most abundant terrestrial carbon resource , can be utilized as feedstocks for biofuels via diverse routes. These routes are beneficial to society due to the potential of reducing GHGs emission and the promotion of regional en ergy independency /diversification. These feedstocks include agricultural and forestry residues, and energy grasses [1, 2]. Structural polymers within plant cell walls ( i.e. cellulose, hemicelluloses, and lignins) offer potential for long -term sustain able production of renewable fuels, chemicals, polymers, and materials that are currently produced from petrochemicals. Many of the promising conversion pathways for these pr oducts are based on a series of biochemical and/or catalytic reactions starting wi th the sugars derived from cellulose (glucose) and hemicelluloses (pri marily xylose in angiosperms). The recalcitrance of plant cell w alls to deconstruction and conversion is considered to be the most crucial factor to overcome in order to develop success ful bioprocessing technologies for lignocellulose conversion to renewable fuels and chemical s [3]. As such, in order to generate high sugar yields from the cellulose and hemicellulose within plant cell walls, a pretreatment is required in combination with the subsequent polysaccharide hydrolysis by either enzymes or an acid catalyst [4]. The unique ultrastructure of biomass, heterogeneous and laye red plant cell wall structure, results in the difficulties in assessing the changes during the biomass conversion processes. Traditional compositional analysis can only provide limited scope of information on the cell wa ll structures such as the com position of the structural carbohydrates 2 and total content of lignin. Plant cell wall c haracte ristics including the ultrastructure of the cell walls , the covalent bonds and side -chains, the porosity and charges within the cell wall polymers and the variation in different plant parts and phenotypes of plants may all impact the deconstruction but are not receiving enough attention for detailed assessment or accurate quantification. Novel and conventional characterization tools are desired to identify how properties of diverse cell walls are impacted by type of conversion processes and how this correlates to reduced cell wall recalcitrance [3], which is important for both plant design with properties suited for deconstruction and for design deconstruction processes . 1.2 PLANT CELL WALLS 1.2.1 Plant cell wall organization Plant cell wall s comprises the majority of t he lignocellulosic biomass materials [5]. In typical plant cell walls, secondary cell wall exists around the cell membrane and lumen, surround ed by the primary cell wall. Primary cell wall is first grown and appears early than secondary cell wall. In the primary cell wall, cellulose exhibits a s the crystalline microfibrils , and hemicelluloses coalesce with the surface of cellulose microfibril s [6]. The hydrogen bonded matrix structure comprising of cellulose microfibr ils and hemicelluloses contribute to the strength of plant cell walls [7]. Lignin mostly exist in the secondary cell wall and also in the middle lamella layer between cells to resist water an d enhance rigidity [8]. Cell wall cellulose is mostly crystalline and composed of 36 chains of "-1-4 linked D -glucosyl unit [9]. In nature, cellulose crystalline structure has distinct but coexisting crystallite forms, I " and I ! [10]. Hemicelluloses are branched polysaccharides which consist of different mono meric sugars including glucose, xylose , arabin ose, mannose and galactose [11]. The composition of hemicelluloses vary by species. 3 1.2.2 Plant cell wall lignins Lignin is the most complicated and hydrophobic compound in plant cell wall materials. It is polymer composed primarily of three p-hydroxyphenyl alcohols (monolignols) inclu ding coniferyl, sinapyl, and a lesser extent p-coumaryl alcohols which form the guaicyl (G), sinapyl (S), and p-hydroxyphenyl (H) substituents in lignin [12]. Other Ònon -canonicalÓ aromatic monomers taken from other points of the monolignol bios ynthesis grid are known to be incorporated into the lignin polymer as well [13, 14]. The linkages between l ignin uni ts are primarily "-O-4 and "-O-4 ether bonds and a fraction of condensed bonds. The ÒC-CÓ bonds including "Ð5, "Ð", and 5 Ð5 and the biphenyl ether s such as 4-O-5 and 5-O-4 bonds are called condensed structure [15]. The total content and relative abundance of monolignols, their linkages, and degree of crosslinking with poly saccharides varies by plant species , plant developmental stage, and plant tissue s [16-18]. Lignin composition and cell wall structural organization in grasses is significantly different from herbaceous and woody dicots (forbs and hardwoods, respec tively) or gymnosperm lignins. One distinguishing feature of the monocot lignins is the considerable incorporation of the p-hydroxycinnamic acids including ferulic and p-coumaric acids [19-21]. Ferulate monomers and dimers are known to be ester -linked to glucuronoarabinoxylan and involved in ether and C -C linkages to the lignin polymer that act as crosslinks between polymer chains [19]. Monomers of p-coumaric acid are prop osed to be esterified to the lignin polymer at the #-carbon of the side chain region of !-O-4 linked syringyl moieties [22] and to a lesser degree esterified to glucuronoarabinoxylan [18]. Grass lignins are significantly more condensed and have h igher phenolic hydroxyl content than the lignins of di cots [21, 23]. A n important implication of this is that more than 50% of grass lignins can be solubilized by treatment with alkali [24] due to the destruction of alkali -labile ester linkages along with the high free phenolic content improv ing lignin solubility in alkali [25]. 4 1.2.3 Plant cell wall polysaccharides The ma jor classes of matrix polysaccharides include the xylans including glucuronoarabinoxylan s (GAXs) and glucuronoxylan s (GXs), glucomannans (GMs), xyloglucans (XyGs), mixed -linkage glucans (MLGs), pectins, and cell wall protein glycosylations [26-28]. The abundance, composition, and substitution patterns of these glycans vary temporally during plant growth and cell wall expansion, spatially within cell walls and between plant tissues, an d taxonomically across diver se plant s. In dicots, pectins, XyGs and GAXs are the predominant non -cellulosic glycans in the primary cell walls with GXs the predominant hemicellulose in dicot the secondary cell walls [29, 30]. In g rasses, GAXs have occupy a significant fraction of the interstitial space between cellulose microfibrils in the pr imary cell walls in addition to mixed -linkage MLGs and grass -specific XyGs and GMs [31, 32]. Pectins are complex and can comprise up to 30% of the primary cell walls of dicots and significantly less in grasses [30] where some of this function is thought to be performed by other glycans including MLGs and GAXs [32]. Structural proteins may comprise up to 10 % of the cell wall in some plant tissues and can be significantly glycosylated, for example with arabinogalactans (AGs) [26]. 1.3 DECONSTRUCTION OF PLANT CELL WALLS 1.3.1 Cell wall deconstruc tion to fuels and chemicals Plant cel l walls conversion to fuels and chemicals are generally through biochemical or thermochemical methods [33]. Thermochemical conversion including pyrolysis and gasification require high temperature to convert the biomass to bio -oils and synthesis gas, respectively. Biochemical conversion to fuels a nd chemicals via fermentation or catalytic routes usually require a low -severity thermochemical pretreatment which par tially break s down the cell wall , then an enzymatic hydrolysis step to convert the pretreated biomass to 5 fermentable sugars. Compared with thermochemical conversion , which requires low residence time but is not selective, b iochem ical conversion provide s high selectivity in converting biomass to desired end products. 1.3.2 Pretreatments Pretreatment s utilize hydrothermal or chemical trea tment to physically solubilize or chemically alter the biomass components to facilitate the enzymatic hydrolysis [34]. The best -studied pretreatment is acidic hydrothermal pretreatment , which utilize s the diluted acid in varied reaction temperature and pressure to hydrolyze and solubilize xylan, melt an d redistribute lignin and increase the enzymatic accessible surface area of the cell walls to improve digestibility [35, 36]. Steam explosion pretreatment is also a hydrothermal pretreatment to break apart the cell wall c omponents , but using high pressure steam penetrates into heated l ignocellulosic material to let the mater ial expands, which leads to size reduction and increasing porosity of the substrate [37]. The reaction variables of steam explosion pretreatment depend on the particle size and moisture content of the samples [38]. Steam explosion pretrea tment and dilute acid pretreatment both result in lignin melt ing at its glass transition point and generating small lignin globules which limit the enzyme penetration to cellulose microfibrils [39, 40]. Ammonia fiber expansion (AFEX) pretreatment is a thermoch emical pretreatment that aims to overcome biomass recalcitrance by effectively redistributi ng but without removing hemicelluloses and lignin [41-44]. During AFEX pretreatment , lignocellulosic materials are treated by liquid ammonia under high temperature and pressure for a certain time the n followed by pressure release , and the treated materials have the same composition as the untreated material but significantly increased digestibility [34, 45]. Besides the above mentioned pretreatment s, many other thermochemical pretreatm ents also have been shown to be effective for improving biomass saccharification 6 such as lime pretreatments [46], organic solvent pretreatment s [47, 48], ionic liqui d pretreatments [49] and sul fite/ SO 2 pretreatments [50], etc. Biological pretreatments usually function as biological delignification, which is another category of approaches to improve the digestibility of the lignocellulosic biomass. Compared with thermal chemical pretreatments, biological delignification requires lower energy input and is more environmental friendly in terms of the formation of less toxic byproduct s than the thermochemical pretreatments [51]. Lignin -degrading microorganisms can be found in forest leaf litter/composts , and the microorganisms applied for biological pretrea tment such as Phanerochaete chrysosporium have been reviewed recently [52].$% best known wood -decay fungi is white rot fungi, which produce s extracellular lignin -modifying enzymes, including laccase, lignin peroxidases and manganese peroxidases , and leave the lighter -colored and bleached cellulose [53]. In white -rot fung i, redox -active metals such as copper , iron and manganese act as cofactor s or are thought to participate in the reaction cycle of the active enzymes and directly participa te in the process of lignin degradation [54]. Employing the effective features of biological deligni fication, oxidative and catalytic pretreatment approaches have been developed, which offer chemical pretreatments at mild temperatures with minimal chemical input [55, 56]. Alkaline hydro gen peroxide (AHP) pretreatment is one type of alkaline oxida tive pretreatment has been studied as a chemical pretreatment [57-60] and as a delignifying post -treatment [61, 62] and is based on treatment of biomass with hydrogen peroxide at alkaline pH (optimally at pH 11.5) at ambient or near -ambient temperature s and pressure s. AHP pretreatment is originally a bleaching process widely applie d in the paper industry for brightening , which is based on hydrogen peroxide oxidatio n of lignin under alkaline condition, while alkali results in saponification reactions with hemicellulose esters and lipids . The solubiliz ation and cleav age of lignin and hemicelluloses can effectively incr ease the 7 hydrolysis conversion of residual plant cell walls [58, 59, 61]. Alkaline pretreatments are particularly effective for grasses, and it is known that AHP is less effective on herbaceous dicots [60] and woody dicots [56]. We hypothesize that the digestibility improvement resulting from AHP pretreatment ma y be attributable to the destruction of ferulate crosslinks as well as mild oxidation and solubilization of lignin , and the metals in the plant cell walls act as redox -active reactant s and participate in the lignin degradation by hydrogen peroxide . These o utcomes have the net effect of improving the overall hydrophilicity of the cell wall matrix which can allow for water and hydrolytic enzyme penetration. 1.3.2 Enzymatic hydrolysis Fungi and bacteria utilizing cellulose as a carbon source have evolved a complex array of enzymes that hydrolyze cellulose to glucose [63]. The lignocellulose -degrading enzymes are classified as families with different three -dimensional structures and catalytic mechanisms [64]. A vast natur al diversity of cellulases are belonging to GH (glycosyl hydrolases) families . In order to deconstruct the crystalline cellulose, th e enzymatic hydrolysis needs the synergy of three types of cellulolytic enzymes [65]. Cellulases i ncluding endoglucanase and cellobiohydrolase adsorb onto the substrate via binding domains. Catalytic domain of endoglucanases split s the bond in glucan chain within the cellulose microfri brils and produce the free chain ends as well as increase accessibility of the substrate by the other cellulases. Cello biohyd rolases cut the end of cellulose chain to produce cellobiose by threading the chain end into the catalytic tunnel to initiate the hydrolysis of the "-glycosidic bond and simultaneous slide forward the enzyme along the cellulose chain. The process is repeat ed until cellulases desorb from the cellulose. Meanwhile, "-glucosidase hydrolyzes cellubiose to two units of glucose. The e nzyme cocktail secreted by the filamentous fungi Trichoderma reesei represent s a most widely studied noncomplexed cellulase system [66]. Many microorganisms employ a Òcellulosome Ó, a large extracellular multiprotein complex 8 compris ing enzyme modules and scaffold protein , and bacteria Clostridium thermocellum represent s the most widely studied complexed cellulase systems [67]. In addition, accessory enzymes such as hemicellulases [68] and lytic polysaccharide monoox ygenase ( AA9) [69] can expedite the catalysi s of cell ulases . Currently , development of synthetic enzyme cocktails with feedstock -specific activities and pro tein engineering to alter the catalytic properties of enzymes are the two main routes to improve the biomass saccharification [70, 71]. 1.4 PLANT CELL WALL RECALCITRANCE 1.4.1 Cell wall lignin Lignin is a hydrophobic com ponent in the cell wall, which can prevent water penetration and provide s chemical resistance to degradation [3, 72]. Lignin impacts cell wall saccharification yield s primarily as a physical obstacle and by non -specific binding to cellulases [3], or the hydrophobic nature o f lignin retards the solub ilization of xylan associated with lignin [73]. Studies have shown strong negative correlation s between lignin content and enzym atic hydrolysis yields for different types of biomass [74, 75]. The impact of lignin content and composition on t he digestibility of biomass vary by the types of feedstock and pretreatment chemistry . Yang et al. suggested that the cellulose digestibility was strongly correlated to lignin removal for flowthrough operated pretreatment with or without acid addition [75]. Another study showed xylan content decreasing in the center of the cell wall while rema ining in the lumen and middle lamella layer during dilute a cid pretreatment, which can be explained by the hydrophobic nature of lignin preventing the solubilization of xylan associated to lignin [73]. During dilute acid hydrolysis of Populus , both lignin content and the ratio between syringyl and guaiacyl monomer (S/G ratio) in lignin have been shown to impact the sugar yield, with slightly lower S/G ratio yield ing significantly higher extent of hydrolysis [76]. Conversely, for hot water pretreatment, sugar release was higher for Populus 9 with higher S/G ratios [74]. For Arabidopsis tissue after hot water pretreatment, cell walls with higher S/G ratio also gave a higher glucose yield [77]. For alfalfa follow ing dilute acid pretreatment, significant differences in hydrolysis yields between various lines with different lignin content and S/G ratio were observed . And for those lines hav ing high H lignin content, S/G ratio alone doesn Õt impact sugar yield [78]. The correlations between the digestibility and p-coumaric acid and ferulic acid do not follo w the same mechanism and depend on their connection to other cell wall components (lignin a nd xylan). Since the majority of p-coumaric acid is esterified to lignin, the amount of p-coumaric acid esters are more of an indicator of the degree of lignification in a forage plant than a direct predictor of digestibility [79]. 1.4.2 Cell wall polysacc harides The complex matrix of non -cellulosic glycans within the cell wall is another major contributor of the cell wall recalcitrance besides lignin. The content, diversity, interactions, and distribution of polysaccharide s play role s in recalcitrance by setting the cellulose accessibility and the cell wall porosity [30, 80, 81]. Non -covalent bonds including H-bonding, van der Waals, and hydrophobic forces between and within polysaccharides are responsible for s table crystalline regions of cellulose and the ability of unsubstituted regions of glycans containing "-1,4 backbones (i.e. MLGs, xylans, GMs, and XyGs) to form junction zones with each other and to ÒsheathÓ the surface of cellulose microfibrils [82, 83]. This capacity of hemicelluloses to sheath the surface of cellulose microfibrils has been identified as one potential means through which non -cellulosic p olysaccharides contribute to cell wall recalcitrance [84]. Ionic bonds such Ca 2+ crosslinks in homogalacturonan (HGA) are important in controlling cell wall porosity (and penetration of enzymes) and potent ially aiding cell -cell adhesion [30] while electrostatic repulsion of charged uro nic acid groups in GAXs are hypothesized to play a role in ordering cellulose microfibrils, preventing aggregation, and patterning lignin deposition [85]. Covalent cross -links within the cell wall 10 include ferulate and diferulate bridges between and within xylans, lignin, and pectins [27], isodityrosine cross -links in structural proteins, and ether and aryl linkages between structural proteins and ferula tes esterified to polysaccharides [29, 86]. Taken together, this complex composite structure presents a challenge for characterization and a number of approaches have been developed recently with a focus on relating structural features of the cell wall to its recalcitrance as reviewed recently by Foston et al. [87]. 1.4.3 Cell wall ultrastructure Besides composition, many studies show ed that the enzymatic digestibility is affected by the higher order structure of biomass. A study us ing a fluorescence -labeled cellulase to probe enzyme binding with the substrates showed that the amorphous forms of the c elluloses are more accessible by enzyme than crystalline cellulose microfibrils [88]. Zhang et al. demonstrated that a n organic solvent -based pretreatment disrupt s the hydrogen bonds in cellulose to break down the biomass and the pretreated biomass with increased surface area but unchanged lignin content has dramatically higher cellulose accessibility [89]. Undoubtedly the ultrastructure of biomass including the porosity and polysaccharides accessibility are more important tha n composition on impacting the cell wall digestibility. However, previous research on characteriza tion of biomass porosity are mostly focused on hardwoods for chemical pulping and papermaking, and the nano -scale porosity properties of lignocellulosic biomass as a bioenergy feedstock still lack adequate attention. $1.4.4 Grass cell wall recalcitrance The degree of lignification and the composition of cell wall polysaccharides are diverse among grasses, woody dicots and gymnosperm biomass. Outcomes of pretrea tment are obviously dependent on the combination of pretreatment chemistry and the plant cell wall properties. A number of promising bioenergy feedstocks are commelin oid monocots, specifically the Poaceae or grasses , which include agricultural residue s [90] such as corn 11 stover, wheat straw, rice straw, and sugar cane bagasse , and dedicated perennial energy crops such as switchgrass and Miscanthus spp . Many of these commelinoid monocots such as corn stover [91], sugarcane bagasse [92], and sorghum [93] have stems that consist of high -lignin vascular bundles distributed throughout a stem consisting of primarily low -lignin pith par enchyma tissue [94]. As a consequence of the alkali -labile ferulate ester crosslinks within the hemicellulose and lignin [21, 95, 96] in graminaceous monocots , as well as high phenolic hydroxyl contents in their lignins which resulting in increased alkali solubility [25], mild alkali pretreatment of grasses such as maize has shown substantial promise as these can be employed for both fractionating biomass and generating a pretreated biomass that is highly amenable to enzymatic hydrolysis [97, 98]. For the grass cell walls, a lkaline saponification partially solubilizes the hemicelluloses by removing the side chains from xylan s such as acetate and partially removes lignin based on the degree of lignification and ferulates crosslinks, generates degradation byproducts including organic acids, aldehydes and phenols [99]. Grass cell walls swell after alkaline treatment [100] with higher glycan extractability and is more accessible to either enzyme or water, and alkaline or alkaline oxidative treated monocots yields much higher digestibility compar ed to dicots. 1.5 PLANT CELL WALL CHARACTERIZATION AND MODELING 1.5.1 Characterization of cell wall structural carbohydrates Accurate modeling and dir ect assessment of changes in plant cell wall structure is required in order to fundamentally understand the mechanisms of cell wall alteration by chemical and biological treatments. The monomer composition of biomass can be determined by either NREL two -stage sulfuric acid hydrolysis method [101], solution -state 1H NM R [102, 103], NIR/PLS (near infrared reflectance spectroscopy ) [104] and pyrolysis GC/MS [105]. The physi cal structural characteristics of biomass such as the porosity and morphology can be 12 profiled by mod ern microscopes such as scanning electron microscopy [106] and atomic force microscopy [9]. Nuclear magnetic resonance (NMR) is a powerful tool applied to determine the chemical bonds and linkages within the cell wall components such as crystall inity index of cellulose by 13C solid -state NMR [107-109] and the structural fingerprint of the polysaccharide and lignin by 2D HSQC NMR [110, 111].$1.5.2 Characterization on cell wall li gnins There are generally three categories of c haracterization methods for quantifying lignin content , including the measurements of the total mass of lignin, the total aromatics of lignin, and the total number of monolignols . To be specific, insoluble solids in two-stage NREL composition analysis is called Klason lignin , which is approximately quantifying the total mass of lignin in the sample . Acetyl bromide soluble lignin (ABSL) has been employed to determine lignin concentration based on the UV absorbance of the aromatics of the extractive -free samples [112]. Th ioaci dolysis is a method to determine the ratio of lignin monomers by cleaving the "-O-4 inter -unit linkage after lignin isolation from polysaccharides by dioxane [113]. Analytical pyroly sis-GC/MS is an other indirect method for quantify lignin content and composition by profiling the pyrolysis products in GC/MS [114-116]. The functional groups and crosslinkings of lignin have been studied by 13C, HSQC, and 31P NMR spectroscopies [117]. 1.5.3 Novel characterization tools using specific proteins Tools utilizing specific proteins such as monoclonal antibodies (mAbs) or carbohydrate binding modules (CBMs) to bind with specific site of biomass have been developed in order to characterize the cell wall structures. Cell wall directed monoclonal antibodies ( mAbs) as well as probes assembled by carbohydrate -binding modules (CBMs) from microbial cell wall glycoside hydrolases [118] have been applied to discover cell wall microstructures by loc ating cell wall epitope in situ . A technique called comprehensive microarray polymer 13 profiling (CoMPP) using microarray with probes of mAbs or CBMs for mapping cell wall composition [119] was able to detect reduction in xylan, arabinoxylans, xyloglucan, and mixed -linked glucan epitopes in hydrothermal pretreated wheat straw [120]. Pattathil et al. employed ~200 glycan -directed antibodies in a ELISA -based screen approach called glycome profiling , which can identify the abundance and extractability of 19 groups of non-cellulosic glycans within the cell walls including mixed -linked glucans, xyloglucans, xylans and pectins, etc.[121]. It has been applied to evaluate the structural change s in poplar under dilute acid pretreatments and discovered that the dilute acid pretreatment primar ily removes arabinogalactans, and the fucosyl -containing xyloglucan , then the non -fucosylated xyloglucan [122]. The CBM-GFP has been applied to quantify the accessible area o f cellulose [123] and to estimate the initiation sites of hydrolysis and saccharification yields [124]. 1.5.4 Data mining and models Data mining and model d evelopment are frequently used in the analytical field of biomass characterization . Some of the research applied multivariate approaches to manipulate the data and build up models to predict biomass component properties. Pyrolysis molecular beam mass spect rometry (py -MBMS) and Fourier transform infrared spectroscopy (FTIR) have been applied using chemometric models to predict the lignin content of switchgrass under different growth conditions [125, 126]. Near -infrared spectroscopy (NIR) has also been used to analy ze the nitrogen and carbohydrates in pine needle [127] and the weight loss in spruce wood by brown -rot fungi [128]. Principal component analysis has been utilized to s implify the spectra and correlate the principal components to easily identi fiable cell wall features [128]. Other work have been done using chemometric models to relate the cell wall properties including cellulose crystallinity [129], lignin content, p-coumaric and ferulic acids content [130], to hydrolysis yields by either neural network [131] or regression models [130]. 14 In terms of the cell wall recalcitrance, the physical properties of biomass including porosity and accessible surface area have impact on enzyme access to cellulose microfibrils . In addition, product inhibition and changes in the substrate properties affect the enzymatic hydrolysis extent and yield. There are many empir ical models that incorporate the effects of various substrate and enzyme properties on hydrolysis rates . Studies showed increasing degree of polymerization decreases the availability of chain ends, and the crystallinity index and enzymatic hydrolysis have negative correlations with cellulase accessibility [132]. Michaelis ÐMenten based models and enzyme adsorption models are frequently utilized for modeling kinetics [132, 133]. 1.6 OBJECTIVES The cell wall matrix of plants provides an effective barrier to the chemical and/or biologica l deconstruction to its monosaccharide components for use in biofuels applications. The properties of the cell wall matrix impacting its recalcitrance are set by many features of cell wall macromolecules across a wide range of length scales. Changes in the physical and chemical properties of cell wall polymers and the overall cell wall environment are import ant outcomes of pretreatments. The overall goal of th is research is to fundamentally understand the recalcitrance of plant cell walls through investigat ing the impact of pretreatment on plant cell wall s with diverse properties . To be specific, alkaline and alkaline oxidative pretreatment s are applied to alter the plant cell walls in diverse plants . The chemical and structural features will be investigated before and after the pretreatments for the diverse feedstocks . Correlations and empirical models will be developed to i n or der to understand how the cell wall recalcitrance is impacted by alkaline and alkaline -oxidative pretreatments. This series of the s tudy will focus on the following four studies : 15 1. Understand the roles that the abundance and accessibility of non-cellulosic cell wall glycan s play in the plant c ell wall recalcitrance, through quantitatively identifying how those properties in structural ly and taxonomically diverse cell walls are impacted by alkaline oxidative pretreatment s and how this correlates to reduced recalcitrance . 2. Investigate the impacts of alkaline oxidative pretreatment on diverse grass cell walls and the relationship betwe en the changes of grass cell wall properties and enzymatic hydrolysis by cellulolytic enzymes , through characteriz ing the specific grass cell wall components including hydroxycinnam ates, xylans, !-gluc ans, and cellulose accessibility associated with the al kaline oxidative pretreatment s with varied severity . 3. Identify correlations between maize cell wall properties and their relationship to the enzymatic hydrolysis yields associated with alkaline pretreatments , in order to identify the features contributing to maize cell wall recalcitrance . 4. Develop chemometric models and multivariate models to predict the cell w all composition and the hydrolysis yields, in order to examine the distribution of the cell wall recalcitrance as well as to provide fast characterization tools on the diverse plant cell walls. 16 2. PLANT CELL WALL RECAL CITRANCE IN DIVERSE BIOENERGY FEEDSTOCKS 2.1 INTRODUCTION Immunological methods using glycan directed m onoclonal antibodies (m Abs ) are widely used tools to investigate plant cell wall structure [134, 135]. Besides the cell wall composition and structure, mAbs can be used for qualitative and quantitative detection of carbohydrate epitopes in plant sequential extracts. This has been performed i n order to characterize pretreatments using one approach that plotted polysaccharides from three increasingly severe cell wall extracts (CDTA, NaOH, cadoxen) onto a microarray, which was then probed with mABs and CBMs [119] in order to identify changes in content and extractability of xylans, XyGs, and MLGs epitopes in hydrothermal pretreated wheat straw [120]. Recently, Pattathil et al. assembled a collection of glycan -specific antibodies [136] and an ELISA -based screen was used to categorize these structurally related glycans from diverse cell wall s [121, 137]. This technique called ÒGlycome profilingÓ has been applied to evaluate the structural, accessibility, and extractability changes of cell wall glycans in hybrid poplar during dilute acid pretreat ments [122] and to compare differen ces in the structural features underpinning cellulose digestibility in switchgrass and hybrid poplar [138]. In this chapter, glycome profiling was applied to understand how a mild alkaline oxidative pretreatment impacts the composition and structural organization of the cell wall. T wo distinct mechanisms by which this pretreatment overcomes cell wall recalcitrance in either herbaceous dicots or graminaceous monocots were identified . These results highlight both the taxonomic differences in cell wall organization and the differences in their response to pretreatment. 17 Specifically, in this study the response of cell wall glycans of diverse plants including hybrid poplar (woody dicot), goldenrod (herbaceous dicot), and corn stover (monocot grass) to AHP pretreatment with increasing H2O2 loadings has been investigated . The cell wall response to pretreatment was characterized by compositional changes and overall mass loss of the cell wall, glucan and xylan enzymatic conversions using a commercial enzyme cocktail, and glycome profiling of the sequential glycan extracts of the untreated and AHP -pretreated cell walls at 12.5% (w/w) H 2O2 loading on biomass. Using this information, we draw conclusions about the structural changes associated with AHP pretreatment and additionally are able to gain insights into the role that differences in pla nt cell wall architecture have on cell wall recalcitrance. 2.2 MATERIALS AND METHOD S 2.2.1 Pretreatment Biomass consisted of a commercial hybrid corn stover (Pioneer Hi -Bred 36H56) provided through the Great Lakes Bioenergy Research Center (GLBRC), debarked h ybrid poplar (Populus nigra var. charkoviensis x caudina cv. NE -19) grown at the University of Wisconsin Arlington Agricultural Research Station and provided through the GLBRC, and goldenrod (Solidago canadensis) collected locally in East Lansing, MI and o btained from Dr. Jonathan Walton (MSU, Plant Biology). Biomass was initially milled with a Wiley MiniMill (Thomas Scientific) to pass a 2 mm screen and air -dried to ~5% moisture. The milled biomass was subjected to alkaline hydrogen peroxide (AHP) pretreat ment using H 2O2 loadings of either 12.5%, 25%, 50% (g H 2O2 per g biomass). These were performed in shake flasks with a 100 g total mass and 2% (w/v) solids concentration for 24 h at 30 ¼C with orbital shaking at 170 rpm and periodic pH adjustment to 11.5. After pretreatments, the liquid in the samples was removed by vacuum filtration (#113 Whatman filter paper) and the slurry was 18 washed several times with deionized water to remove solubles and air -dried at room temperature. The mass yield during pretreatmen t was quantified as the ratio between the mass (dry basis) after pretreatment and the in itial mass before pretreatment. 2.2.2 Composition analysis, enzymatic hydrolysis and digestibility determination The pretreated biomass was subjected to a two -stage aci d hydrolysis accordin g to NREL composition analysis [139] to determine neutral polysaccharide content, Klason lignin, and ash with the difference that Aminex HPX -87H (Bio -Rad, Hercules, CA) column was used to quantify the glucan, xylan+galactan+mannan, arabinan, as well as acetate content. The total uronic acids were assayed enzymatically (K -Uronic, Megazyme, Wicklow, Ireland). The extractives content was determined by a sequential 3 -step ex traction including 70% ethanol, 1:1 (v/v) methanol and ch loroform mixture, and acetone in 1 mL total volume at 5% solid load ing . Three extraction cycles for each solvent were performed and f ollowed by centrifugation at 100 00 x g for 10 minutes for each cycle. Before the enzymatic hydrolysis, the pretreated biomass was ball -milled for 3 cycles using a QIAGEN TissueLyser II equip ped with 25 mL Teflon jars and 20 mm diameter Teflon balls at 30 Hz for 2 minutes with liquid nitrogen cooling. The ball -milled samples were incubated with Cellic C -Tec2 (Novozymes, Bagsv¾rd, Denmark ) at a loading of 30 mg protein/g glucan at 50 ¼C, 10% so lid loading and 5 mL total volume in 0.05 M Na -citrate buffer pH 4.8, for 24 hours or 72 hours. The glucan and xylan yields (based on only glucan and xylan remaining after pretreatment) were determined by the HPLC analyzable glucose and xylose concentratio ns after hydrolysis divided by the original glucan and xylan cont ents in the pretreated samples. 2.2.3 Sequential extraction, glycome profiling (Sequential extraction and glycome profiling were done by Sivakumar Pattathil , University of Georgia.) 19 Sequentia l cell wall extractions and glycome profiling were carried out as described previously [121, 122, 140]. The six extractions included (in order of extraction) oxalate to remove ÒlooseÓ pectins, carbonate to remove Òmore tightlyÓ bound pectins, 1M KOH to remove ÒlooseÓ hemicelluloses along with tightly bound pectins, 4M KOH to remove ÒtightlyÓ bound hemicell uloses along with tightly bound pectins, acid chlorite to oxidize and solubilize lignin and release lignin -embedded hemicelluloses, and a 4M KOH post -chlorite treatment to remove additional lignin -bound polysaccharides. Plant glycan -directed monoclonal ant ibodies were from laboratory stocks (CCRC, JIM and MAC series) at the Complex Carbohydrate Research Center (available through CarboSource Services; http://www.carbosource.net) or were obtained from BioSupplies (Australia) (BG1, LAMP). A description of the mAbs used in this study can be found in the Appendix , Table A1, which includes links to a web database, WallMAbDB (http://www.wallmabdb.net) that provides detailed inf ormation about each antibody. 2.3 RESULTS AND DISCUSSIONS 2.3.1 Changes in composition and mass loss Three types of biomass representing three classes of plants that may offer promise as feedstocks for cellulosic biofu els were tested in this study. Corn stover represents the agricultural residue with the highest production and availabilit y for bioenergy applications in the U.S. [141], while short -rotation hybrid poplar has agronomic, logistical, and environmental advantages as a feedstock [142]. ÒLow -input high -diversityÓ bioenergy landscapes have many attractive sustainability attributes [143] and comprise mixed communities of plants on marginal or degraded lands, and in this study we use goldenrod (Solidago canadensis ) as a rep resentative herbaceous dicot that may be present in these landscapes [144]. The composition of the untreated biomass is presented in Table 2.1. 20 Notable di !!erences include the low content of pectic polysaccharides (as uronic acids) in the corn stover, which is 5 -fold lower than the goldenrod and 2 -fold lower than the hybrid poplar . The goldenrod has a substantially higher extractives content (23.5%) relative to the other two biomass types. Additionally, lignin content of the corn stover is nearly half that of the goldenrod and poplar. Hybrid Poplar (wt/wt %) Goldenrod (wt/wt %) Corn Stover (wt/wt %) Glucan 48.3±2.0 27.1±0.8 38.4±0.3 Xylan+ Mannan+ Galactan 16.9±0.6 13.6±0.8 25.2±0.2 Ara binan 1.33±0.07 2.96±0.03 3.92±0.02 Acetate 4.14±0.09 2.59±0.06 3.20±0.10 Total Uronic Acids 1.88±0.10 4.75±0.03 0.88±0.02 Lignin (Klason) 20.85±1.1 19.92±0.7 12.57±1.2 Extractives 5.79±0.22 23.5±0.23 12.0±0.74 Ash 1.95±0.48 6.08±1.02 3.03±0.29 Table 2. 1 Composition of untreated biomass on a whole sample basis rather than an extractives -free basis. Errors represent data range of duplicate measurements. AHP pretreatment was performed at increasing H 2O2 loadings (12.5%, 25%, an d 50% w/w on biomass), of which the two highest loadings would be significantly higher than would be economically practicable industrially. The reason for these high loadings was to compare and analyze how the diverse cell walls differ in their susceptibility to low temperature mild oxidative delignification and hemic ellulose extraction and as a screen for overall digestibility differences. The total cell wall mass loss (excluding extractives), xylan loss, and lignin loss for the three biomass types following pretreatment with increasing H 2O2 loadings show 21 distinct res ponses between the biomass types (Figure 2.1 A-C). For poplar, minimal material was solubilized with pretreatment (<1% by mass), while up to 20% and 25% of the mass of the cell walls of the goldenrod and corn stover, respectively, was solubilized by pretr eatment at the higher H2O2 loadings. The mass of corn stover decreased continuously with increasing H2O2 loading, while the sample mass of poplar and goldenrod decreased abruptly with the mildest treatment (12.5% H2O2). For individual cell wall components, AHP pretreatments resulted in minimal changes in glucan for all biomass types representing preservation of cellulose (data not shown) which is consistent with our previous findings [145, 146], while the xylan content decreased only for the corn stover (Figure 2.1B). For goldenrod, the Klason lignin content was only slightly reduced by AHP pretreatment and did not change significantly by increasing H2O2 loading. The simultaneous removal of xylan and Klason lignin in corn stover was increased by increasing H2O2 loading. This is a well -known property of grass cell walls and is due to the higher solubility of grass lignins [25] and alkali -only extraction of lignin and xylans in grasses is known to be significantly high er in grasses than in woody dicots [138, 147]. Besides alkali solubility, cleavage of ester and ether cross -links between xylan and lignin or lignin and ligni n mediated by ferulate [31, 148] in grasses are thought to be an important target of AHP pretreatment [146] and likely contribute to these outcomes. 22 Figure 2 .1 Compositional changes (extractives -free basis) associated with AHP pretreatment with increasing H2O2 loading showing solubili zation of total cell wall mass (A), xylan (B), and lignin (C). 2.3.2 Digestibility of poplar, goldenrod, and corn stover Figure 2 .2 Enzymatic conversion of cell wall glucan to glucose in untreated and AHP pretreated biomass with increasing H2O2 loading for poplar, goldenrod, and corn stover with cellulase (Cellic CTec2) for 24 h and 72 h. Figure 2.2 shows the enzymatic digestibility of the poplar, goldenrod, and corn stover subjected to increasing H2O2 loadings for 24 and 72 hr hydrolysis times using a commercial cellulase (Cellic CTec2) with no xylanase supplementation. These results represent screening 23 only and were not focused on optimizing enzyme cocktail or loading. As such, improved glucan (and xylan) conversions at lower enzyme loadings would be observed if x ylanase and pectinase were added. These results show that the glucan enzymatic digestibilities were increased with increasing H2O2 loading for the three types of biomass. Similar to the res ults for compositional changes (Figure 2.1), the digestibility changes with pretreatment are significantly different between the three classes of plants tested. For poplar, the digestibilities were significantly lower than goldenrod and corn stover with t he highest glucan conversions approaching 40%. For goldenrod, the glucan digestibilities ÒsaturateÓ at approximately 70% at the mildest pretreatment condition. For corn stover, the glucan digestibility approaches 100% with increased H2O2 loading . I t is kno wn that corn stover as well as many other cereal stovers are often considerably more digestible than many undomesticated grasses [146, 148]. Gould [60] determined that diverse gra minaceous monocots responded considerably better to AHP pretreatment at 100% (w/w) H2O2 loadings than herbaceous dicots/forbs (inc luding goldenrod) and that the seven grasses tested had on average more than double the improv ement in digestibility of the 11 forbs tested following AHP pretreatment. He determined that the goldenrod showed the largest improvement in digestibility of the dicots tested and was among the highest in terms of absolute glucose release per gram of biomass. 2.3.3 Glycome profiling of poplar, goldenrod, and corn stover Glycome profiling (GP) provides quantitative information on both how pretreatment impacts the strength of association between cell wall glycans and other cell wall matrix polymers and how pretreatment impacts cell wall glycan composition. For GP, the samples were subjected to increasingly severe extractions to sequentially remove the cell wall polymers, followed by quantification of recovered materials in each extraction step, and probing of the binding strength for the diverse array of antibodies covering a range of non -cellulose cell wall 24 polysaccharide epitopes [121, 122, 149]. For the sequential extractions, the stren gth of association between individual glycans and other cell wall matrix polymers may be hypothesized to be mechanistically due differences in: (1) the strength of non -covalent association or physical entanglement between polymers or ÒencrustationÓ within lignin, (2) location within the cell wall (surface vs. interior), and (3) tissue type (e.g. lignified parenchyma and sclerenchyma versus low lignin pith tissues). The six extractions included (in order of extraction) oxalate to remove ÒlooseÓ pectins, car bonate to remove ÒtightlyÓ bound pectins, 1M KOH to remove ÒlooseÓ hemicelluloses, 4M KOH to remove ÒtightlyÓ bound hemicelluloses, acid chlorite to oxidize and solubilize lignin and release lignin -embedded hemicelluloses, and a 4M KOH post -chlorite treatm ent to remove additional lignin -bound polysaccharides. The role of lignin in setting the alkali -extractability of cell wall glycans was recently demonstrated by Pattathil et al. [150], whereby it was identified that alfalfa lines with disrupted monolignol synthesis resulting in a low -lignin phenotype contained considerably more alkali -extractable glycans than the control line where much more of the cell wall glycans were extractable only a fter chlorite delignification. The GP results from this work are pre sented in Figures 2.3 for the three biomass categories for either no pretreatment or AHP pretreatment at 12.5% (w/w) H2O2 loading on the biomass. Substantial differences can be observed between biomass types and for biomass subjected to pretreatment as qu antified for both mass partitioning of extracted glycans (top panel of Figure 2.3) and differences in the abundance of glycan epitopes in these extracts (heat map in the lower part of Figure 2.3). For the extract mass partitioning of glycans, it is clear t hat the two dicots have similar profiles for the four most severe extracts, i.e. the 1 M KOH, 4 M KOH, chlorite, and 4 M KOH PC. The goldenrod shows a very high content of the oxalate - and carbonate -extractable polysaccharides relative to the other two typ es of biomass. This may be a consequence of the goldenrod having a higher fraction of pecti n-rich leaves compared to the 25 poplar which consists of only stem heartwood, and the corn stover, which as a graminaceous monocot is known to have low pectic polysacc haride content [151] Figure 2 .3 Glycome profiling of Poplar, Goldenrod and Corn stover biomass samples before and after AHP pretreatment (12.5% H2O2 loading). (Sequential extraction and glycome profiling were done by Sivakumar Pattathil , University of Georgia.) 26 A number of noteworthy differences are apparent in comparing the glycan epitope abundances within the six extracts for the three types of untreated bioma ss. One difference is that the XyG epitopes show significantly different partitioning between the three biomass types and the goldenrod is the only biomass exhibiting abundant XyG epitopes in the 1 M KOH extract. The xylan epitopes are more abundant in th e corn stover extracts and more abundantly distributed into the two most severe extracts that might correspond to Òlignin -boundÓ xylan. Pectic polysaccharides and AG domains show different partitioning behavior between the biomass types as well, with the c ell wall extracts from goldenrod exhibiting the most abundant content of these classes of epitopes. Two intense MLG epitopes are present in all the corn stover cell wall extracts corresponding to antibodies LAMP2H12H7 and BG1 and the abundance of these epi topes increase in extracts (particularly in the two mildest and the chlorite delignification) following AHP pretreatment. Weak epitope binding for both of these antibodies is present in some of the poplar extracts and goldenrod extracts. These observations are consistent with the primary cell wall models proposed by Carpita [31], where for grasses (Type II cell wall), the MLGs and GXs have a more important structural role and may act in t he capacity that XyGs and pectins in dicots (Type I cell wall). Pretreatment can conceivably alter the binding of mABs to their glycan epitopes from the same extraction condition in three ways by: (1) altering the cell wall structural integrity to shift the glycan epitope into a more easily (or more difficult) extractable category, (2) solubilizing the glycan epitope during pretreatment, and (3) structurally altering the glycan epitope, for example, through alkali -induced deacetylation or demethylesterification. These three modes of action are used to interpret the changes associated with pretreatment. T o better visualize the effects of pretreatment on glycan extractability, the GP data are replotted in Figures 2.4 and 2.5 after normalizing to epitope abundance per mass of original cell wall. In this representation, glycan epitopes that are increased in t heir abundance in individual 27 extracts after pretreatment will appear to the left of the x -y line, while epitopes that are decreased will appear to the right. It can be observed that for the poplar, pretreatment has very little effect on the total extracta ble glycans in most of the six fractions. The apparent increase in the xylan epitopes in oxalate and carbonate extracts suggest that the extractability of xylan by mild solvents may be enhanced by pretreatment (Figure 2.4, subplots A -C). However, consideri ng that the total content of carbohydrates in these extracts are unchanged and that the xylan -specific antibodies were developed for deacetylated, alkali -extracted xylans, this result likely indicates that easily extractable xylans were deacetylated by pretreatment and that the abundance of deacetylated xylan epitopes increase as a consequence. The slight differences in other epitopes in the 4 harshest extracts (Figure 2.5, subplots A -C) suggests that AHP pretreatment results in only minor alterations in th e extractability of other major cell wall glycans indicating minimal impact on the structural and compositional organization of the cell wall in agreement with the results in Figures 2.1 and 2.2. An exception is the xyloglucan epitopes in the 1M KOH extrac t (Figure 2.5A) which are slightly improved in their e xtractability by pretreatment. 28 Figure 2 .4 Comparison of the changes in the relative abundance of three glycan epitope categories due to AHP pretreatment from the oxalate (blue dots) and carbonate (red dots) extracts in glycome pro !"ling, for poplar (A, B and C, for xyloglucans, xylans and pectic polysaccharides), goldenrod (D #F for xyloglucans, xylans and pectic polysaccharides) and corn stover (G #I for xyloglucans, xylans and pectic polysaccharides). 29 Figure 2 .5 Comparison of the changes in the relative abundance o f three glycan epitope categories due to AHP pretreatment from each of the four harshest extracts including 1 M KOH (green triangles), 4 M KOH (blue diamonds), chlorite (red squares), and 4 M KOH PC (purple crosses) in glycome pro !"ling. For goldenrod, th e GP results show that xylan epitopes and XyG epitopes in the oxalate extract increased considerably (Figure 2.4, subplots D -F); potentially as a consequence of pretreatment -induced deacetylation or by pretreatment increasing the extractability in agreemen t with the increase in total glycan mass in the oxalate extract with pretreatment (Figure 2.3). Unlike the poplar, the epitopes for HG backbones, RG -I/AG, and AG are decreased in both the oxalate and carbonate extracts, indicating that these pectic polysac charides in goldenrod are likely solubilized during AHP pretreatment. In the four 30 most severe extracts, a number of important trends are apparent (Figure 2.5, subplots D -F). The first is that virtually all epitopes are decreased as a consequence of pretrea tment in the 1M KOH extract (corresponding to alkali -soluble glycans not closely associated with lignin), while slight increases in the abundance of epitopes are observed after pretreatment in the 4M KOH, chlorite and 4M KOH post -chlorite extracts (corresp onding to lignin -embedded glycans). These results indicate that the likely target of AHP pretreatment for improving digestibility in goldenrod is glycan (XyG and xylan) solubilization to improve cell wall accessibility to glycolytic enzymes and minor delig nification which slightly improves the extractability of lignin -embedded polysaccharides. The corn stover results show considerable differences in both their glycan extraction plots and antibody binding profiles relative to the hybrid poplar and goldenro d (Figure 2.3). From the glycan mass extraction profile at the top of the panel, it can be observed that, unlike the goldenrod, the amounts of extractable glycans in the three most severe extracts were significantly altered by pretreatment. One noticeable alteration is that the glycan epitopes in the 4M KOH post -chlorite extract were shifted into the chlorite extracts after AHP pretreatment. This phenomenon of the pretreatment changing the glycan extractability profile was not shown for dilute acid pretrea ted hardwood [122]. This result sup ports the identified changes in composition shown in Figure 2.1, and together with Figure 2.2, make clear that alteration of the non -cellulosic glycan extractability directly impacts the glucan digestibility for the subsequent enzymatic hydrolysis. In the two mildest extracts from corn stover, the XyG and xylan epitopes were increased, which may be a consequence of improving extractability or likely due to deacetylation of these glycans by pretreatment (Figure 2.5, subplots H and I). Unlike goldenrod and p oplar, the epitopes for pectic polysaccharides were increased in the two mildest extracts, possibly indicating differences in their structural role between monocots and dicots in pectic polysaccharides [31], Additionally, the results for 31 changes in epitope abundance as a result of pretreatment for the four harshest extracts were considerably different for the corn stover than for the goldenrod. Compared to goldenrod, where all glycan epitopes were decreased by pretreatment in the 1M KOH extract but slightly increased in the lignin -associated extracts, the cor n stover glycan epitopes showed increases across all four extracts except in the 4M KOH post -chlorite extract, in which the lignin -embedded glycan epitodes are decreased by pretreatment (Figure 2.5, subplots H -J). The se reductions of corn stover glycans in the harshest extracts are likely a consequence of these epitopes being already removed by less harsh extractions. Additionally, increases in the 2 MLG epitopes were observed with pretreatment for corn stover (Figure 2.3) for all extracts except the 4 M KOH post chlorite treatment. These differing responses to AHP pretreatment between monocots and dicots have important implications for structural features of the cell wall contributing to recalcitrance as well as the mechanism or target of pretreatment. The composition and structure of the cell wall are obviously important and many properties of the cell wall impacting recalcitrance have been described in the literature including cell wall hydrophobicity [152], porosity [153, 154], xylan content [155], lignin content, cross -linking, and higher order structure [130]. Jung et al. [156] noted that lign ified secondary cell walls were the primary obstacle hindering ruminant digestibility in dicots with stem secondary xylem (i.e. woody biomass) the most recalcitrant, while increasing lignification in grasses hinders, but does not completely inhibit digesti on. The current work identified that AHP pretreatment has a relatively minor impact on the hybrid poplar composition, digestibility, and glycan extractability profiles. Substantial work has been devoted to understanding the cell wall properties contributi ng to ruminant digestibility of grasses and properties including ferulate content, total lignin content, syringyl:guaicyl ratio of lignin monomers, and degree of arabinosylation of xylans have all been linked to differen t digestibility [130, 156-158]. Pretreatments may impac t any of these 32 cell wall properties to improve cell wall digestibility. DeMartini et al. [138] found that treatment of switchgrass with alkali alone to solubilize xylan (and lignin) from the cell wall was sufficient to result in glucan digestibilities approaching the theoretical maximum, while for hybrid poplar, chlorite delignifi cation was necessary to improve digestibility significantly past alkali -only treatment. This is consistent with models for grass cell walls that include alkali -labile ferulate ester cross -links between cell wall polymers as an important structural feature controlling lignin integration into cell walls [148]. Our previous work identified that lignin and ferulate removal by AHP pretreatment are important predictors of digestibility in diverse grasses [146]. We have previously shown that AHP pretreatment results in the destruction of "-O-4 bonds in grasses [146] and, for example, the content of free phenolics in grass lignins may enable improved alkali solubilization or potentially participate in the initiation "-O-4 scission reactions. 2.4 CONCLUSION For a woody dicot (hybrid poplar), an herbaceous dicot (goldenrod), and a graminaceous monocot (corn stover) , we found that enzymatic hydrolysis yields, cell wall biopolymer and total mass solubilization, cell wall glycan extractabilities, and glycan epitope abundances differed significantly in their response to AHP pretreatment. Using gly come profiling, we iden tified diff erent mechanisms for how AHP pretreatment overcomes cell wall recalcitrance in goldenrod &'()*)$corn stover, while it was relatively ineffective on poplar. For corn stover, mild alkaline oxidative pretreatment resulted in slight delignification and presumably disruption of cell wall polymer cross -linking. This had the consequence of disrupting the structural integrity of the cell wall , which was manifested through improved extractability of important structural glycans including xylans, MLGs, and XyGs and presumably allowed for improved accessibility for glycolytic enzymes into the cell wall during hydrolysis. Goldenrod 33 was found to respond differently in the extractability profiles where all classes of glycan epitopes exhibited considerable decre ases in the 1M KOH extracts following pretreatment rather than an increase as was observed for corn stover. Besides these differences, it was revealed that the pectic polysaccharides (HG, RG -I, and AG) were not only signi ficantly more abundant in goldenrod than in corn stover, but were solubilized by pretreatment as indicated by their decrease following pretreatment in the three mildest extracts for goldenrod. For corn stover, the pectic polysaccharides as well as the signi ficantly more abundant MLGs showed mild increases in extractability following pretreatment indicating ÒlooseningÓ from the cell wall rather than solubilization. 34 3. PLANT CELL WALL RECA LCITRANCE IN DIVERSE GRASSES 3.1 INTRODUCTION Alkaline oxidative pretreatments applied in recent years to enhance enzymatic digestibility [57, 145, 159-161] were derived from the non -delignifying bleaching in pul p and paper processes. Alkaline pre -extraction coupled to alkaline oxidative post -treatment has also been applied to compare the effects of alkaline hydrogen peroxide as a pretreatment and post -treatment [98]. The impacts of alkaline oxidative pretreatment on d iverse plant cell walls have been investigated in previous studies [56, 146, 162]. We proposed that the alkaline oxidative pretreatment was effective by removing lignin primarily through breaking !-O-4 bonds [146], introducing carboxylic g roup on cellulose microfibirils [56] and increasing glycan extractability and enzyme accessibility, and that this pretreatme nt is especially suitable for grasses [162]. In addition, the metal complexes may be utilized as catalyst for increasing the rate and extent of hydrogen peroxide ox idation by participation in FentonÕs chemistry [55, 56]. Alkaline pretreatment s usually yield higher digestibility for grasses than the other feedstocks [163] due to the relatively abundant presence of the alkali -labile ferulate or diferulate est ers bridges in the grass lignin s or glucuronoarabinoxylans [164] and the high phenolic hydroxyl content [97]. Studies have shown that the hydrolysis yield is correlated with ligni n content of the cell walls, which is independent on the type of biomass [146], and the glucan hydrolysis yield is not necessary correlated to the xylan content after alkaline pretreatments [165] since the xylan remaining in the cel l wall is usually ÒloosenedÓ by alkaline pretreatments, which is shown the previous chapter . In addition, we proposed that the 35 metal content in the grasses may be another contributor for improving the efficiency of alkaline oxidative pretreatment. The purpose of this study was to investigate the relationship between grass cell wall properties and enzymatic hydrolysis through characterization of the specific grass cell wall components including lignin, hydroxycinnam ates, xylans and !-glucans. In addition, metal content and cellulose accessibility were analyzed and correlated with the pretreatment severity and hydrolysis yields . In this study, s witchgrass ( Panicum virgatum cv. Cave -In-Rock), two corn stovers (a commercial hybrid and inbred brown midrib lines bm1 and bm3), and Miscanthus spp. are examples of cereal stovers and energy perennials were subjected to alkaline oxidative pretreatment s at varied hydrogen peroxide loading. In addition, hybrid corn stover was subjected to an alkaline pre -extraction coupled with alkaline oxidative post -treatmen t. The differences in hydrolysis yields were compared between the grasses and correlated with their composition properties based on sugar analysis, ferulate analysis and metal analysis. ÒGlycome profilingÓ [140] was applied to identify changes in the extractability and abundance of different glycan epitopes in the individual species cell walls as functions of pretreatment severity. Glycans released during pre treatment were determined using ELISA screening with the same glycan -directed mAbs library as previously described methods [121]. Cellulose accessibility was examined through assessing the binding characteristics for pretreated and untreated corn stover using ctCBM family 3 which binds specifically to crystalline cellulose . 3.2 MATERIALS AND METHOD S 3.2.1 Biomass pretreatment, composition analysis and enzymatic hydrolysis In this study, corn stovers (Zea mays ) including Pioneer hybrid 36H56 , inbred brown midrib bm1 and bm3 , switchgrass (Panicum virgatum cv. Cave -in-Rock) and miscanthus 36 (miscanthus x giganteus) were subjected to 2.5, 5, 12.5, 25, 50% H2O2 loading (w/w biomass) Alkaline hydrogen peroxide (AHP) pretreatment s with pH adjusted to 11.5. The pretreatments were carried out in 25 0 mL Erlenmeyer flasks covered by aluminum foil, with a total reaction volume of approximately 100 mL and 2% (w/v) solid loading, 30 ¼C and 180rpm for 24 hours. The pretreated samples were washed with DI water until pH reached neutral, and then air -dried overnight. The mass loss during pretreatment was measured gravimetrically. Two -stage sulfuric acid hydrolysis was carried out according to NREL composition analysis protocol to determine the structural carbohydrates by HPLC installed with an Aminex HPX -87 H (Bio -Rad, Hercules, CA) column . P retr eated samples were ball -milled with a QIAGEN Tissuelyzer II for 3x2 minutes, which was followed b y enzymatic hydrolysis performed at pH 4.8, 50 ¼C and 180 rpm, with 30 mg/g protein loading of C -Tec 2 (Novozymes, Denmark) for 72 hours. The glucan hydrolysis yield (digestibility) was determined by the glucose released after enzymatic hydrolysis divided by the original glucan content in the pretreated samples. 3.2.2 p-hydroxycinnamic acids determination and metal analysis The hydroxycinnamic acid (HCA) content of the untreated maize and mild alkaline treated maize was determined by a severe alkaline ext raction method which represents the content of both the esterified and etherified HCA s [130, 166, 167]. In this, 0.1g biomass was treated with 5 mL of 4M NaOH in sealed Teflon tubes and held at 1 70 ¼C for 2 hour in an oven . After coo ling down in room temperature, 50 µL of 10 mg/mL methanol solution of o -coumaric acid was added as an internal standard for each sample. The mixture wa s transferred to Eppendorf tubes and centrifuged at 13,000 rpm for 10 minutes. The pH of the supernatant was adjusted to 2 with concentrated HCl, and the samples were then place d into a 4 ¼C fridge overnight. After the overnight storage, the supernatant of the samples were analyzed in a HPLC stalled with a C18 column (Discovery HS C18 HPLC Column, 5 cm # 37 2.1 mm, 5 µm). Standards containing ferulic acid, p-coumaric acid and o -cou maric acid were run in parallel to determine the concentration of the acids in t he samples. The metal analysis results of the samples was provided by A & L Great Lakes Laboratories, Inc. (Fort Wayne, IN) . 3.2.3 Glycome profiling and data processing (Sequential extraction and glycome profiling were done by Sivakumar Pattathil , Universi ty of Georgia .) The extractives -free cell wall samples were fractionated by successive extraction with six increasingly harsh solvents including oxalate, carbonate,1 M KOH, 4 M KOH, Chlorite and post -Chlorite 4 M KOH. These six extracts were collected, wei ghed and screened with the 156 glycan -directed mAbs [162]. The relative mAbs binding strength was determined from UV absorbance at OD 450-OD 680 on the basis of 20 µg glucan equivalent. The extractives content was determined gravimetrically by three cycles of organic solvent extraction performed sequentially with 70% ethanol, methanol and chloroform 1:1 (v/v) solution and pure acetone, which was followed by a 10 -minut e centrifuge at 10,000 rpm. The specific cell wall epitopes abundance in each extract was estimated by the following equation: Epitope abundance = [(OD 450-OD 680) / (g original biomass)] = N1*N2*N3*N4/20 N1 = [(OD 450-OD 680)/(20 µg glucan equiv alent )] N2 = [(µg glucan equiv alent )/(mg extracted material)] N3 = [(mg extracte d material/(g extractive -free biomass) N4 = [(g ext ractive -free biomass)/(g original biomass)] 3.2.4 ELISA screening of pretreatment liquors (ELISA screening was done by Sivakumar P attathil , University of Georgia.) 38 AHP pretreatment liquor for each grass was collected and dried at 50 ¼C in a convection oven. The ELISA screening was performed as described in previous work [98]. Plant glycan -directed monoclonal antibodies we re obtained from laboratory stocks (CCRC, JIM and MAC series) at the Complex Carbohydrate Res earch Center (available through CarboSource Services; http://www.carbosource.net) or were obtained from BioSupplies (Australia) (BG1, LAMP). Further description of the mAbs used in this study can be found in the Supporting Information Table S1, which includes lin ks to a web database, WallMAbDB (http://www.wallmabdb.net) that provides detailed information about each antibody. 3.2.5 Enzyme binding experiments Enzyme binding experiments were conducted at 1 mL total liquid volume at 4 ¼C, pH 7.0 for 4 hours with end -to-end rotation using bacterial CBM family 3 from Clostridium thermocellum (Prozomix, UK) , which bind specific to crystalline cellulose was used to test the accessible crystalline cellulose in the pretreated and untreated corn stove r. The binding isotherms were developed by a Langmuir isotherm model: !!"# !!!"# !!"# !!!!!!!"# !! The parameters of the model were estimated based on the binding data using Excel Solver. 3.3 RESULTS AND DISCUSSIONS 3.3.1 Correlations between grass cell wall properties and digestibility The digestibility profiles associated with H 2O2 loading were investigated and results are shown in Figure 3.1. Results showed that the corn stovers had higher hydrolysis yields than the two perennial grasses at comparable pretreatment conditions . This is further supported by the composition data shown in Figure 3.2, which show s that glucan content is almost constant 39 for all samples , the corn stovers have higher xylan and lignin removal than the perennial grasse s, and the three corn stovers generally have similar removal of xylan and lignin. Other researchers have found t hat corn stover generally have higher glucan hydrolysis yields than switchgrass following alkaline or neutral pretreatments such as lime a nd liquid hot water pretreatment [163]. The results in this study suggested that the lignin and xylan removal both contribute to the increased glucan hydrolysis for the diverse grasses. Figure 3 .1 Glucan digestibility (hydrolysis yield) profile associated with H 2O2 loading after AHP pretreatment for the diverse grasses (0% AHP are untreated samples). 40 Figure 3 .2 Composition and weight loss of switchgrass, miscanthus, bm1, bm3 and hybrid corn stover. Transition metals, including iron (Fe), copper (Cu) and manganese (Mn) , play important ro les in FentonÕs reactions . They can catalyze the hydrogen peroxide delignification [168] and are also essential micronutrients for growth and development of plants. Transition metal complexes have been utilized as ca talysts for improving the rate and extent of alkaline hydrogen peroxide pretreatment for poplar [55, 56]. However, the metals contained in the plant cell wall s may also contribute to FentonÕs reactions during the hydrogen peroxide pretreatment. Table 3.1 shows the concentrations of metals in the five grasses. It is worth noting that the iron content is higher than the other meta ls and the amount is diverse between different types of feedstocks. Figure 3.3A plots the relationship between the initial iron content and glucose yield of the diversely AHP pretreated grasses, and shows a clear positive trend, which implies that the cont ribution of the iron in the plants is considerable for 41 improving the peroxide oxidation of lignin . The synergistic effects of the iron and hydrogen peroxide on improving the hydrolysis yields of diverse grasses indicate that the in vivo iron participates a s a Fenton reagent in alkaline oxidation pr etreatment. Sample Name K (%) Mg (%) Ca (%) B (ppm) Zn (ppm) Mn (ppm) Fe (ppm) Cu (ppm) Al (ppm) Corn stover 1.06 0.13 0.24 3 31 44 255 5 138 Switchgrass 0.4 0.15 0.4 3 10 49 64 3 21 Miscanthus 1.06 0.11 0.18 3 28 35 43 2 27 bm1 stover 1.34 0.33 0.42 8 33 58 425 5 118 bm3 stover 1.53 0.25 0.3 7 21 27 171 5 54 Table 3. 1 Metal analysis of grasses Figure 3 .3 The correlation between the metal content (A. Fe content, B. Total metal content) and glucose yield for the diversely AHP pretreated grasses including corn stover, bm1 stover, bm3 stover, switchgrass and miscanthus. Inbred brown midrib maize lines are the result of natural mutations in corn, and are noted for their visible brown color in leave s, have been studied for over 50 years [169]. Brown midrib line s bm1 and bm 3 are well-known for their characteristics in lignin composition [170-173]. The lower amount s of cinnamyl alcohol dehydrogenas e (CAD ) in bm1 results in reduced lignin, ferulic acid and p-coumaric acid esters as well as enriched carbon -carbon monolignol linkages , and bm3 is deficient in caffeic acid O -methyl transferase (COMT) , which results in 42 decreased total lignin and syringyl monolignol incorporation [173, 174]. These characteristics are thought to result in higher digestibilit ies relative to other corn stover s [146, 169]. In this study, bm3 stover shows the highest hydrolysis yields following pretreatment but does not have a significant compositional difference compared to the three corn stovers, indicating structural differences other than lignin content and composition, for instance, the secondary cell wall thickness and deposition [175]. Monocot lignin s are rich in hydroxycinnam ates (HCAs) , including ferulate (FA) and p-coumarate ( pCA) , and it was found that the ferulic acid crosslinks grass lignins primarily with ether links, while pCA bonds with grass lignins with both esters and ethers [176, 177]. Many analytical methods for quantify HCAs in grasses apply different severities of alkaline extraction to break either ester bonds or ether bonds along with ester bonds based on the alkali -labile properties of these bonds [22, 130, 178, 179]. Previous rese arch h as found that the pyrolysis products derived from ferulate and p -coumarate both show strong inverse correlation with digestibility [146]. The HCAs concentrations in the pretreatment liquors have also been found to increase with increased H 2O2 loading during pretreatment (data not published). In this study, for the 12.5% AHP pretreated and untreated corn stover, switchgrass and miscanthus, the correlation between digestibility and major cell wall properties inclu ding lignin, xylan and ferulate content have been analyzed and the heap map of correlation coefficients generated by MATLAB is shown in Figure 3.4. The hierarchical clustering on the top of Figure 3.4 developed by MATLAB based on Euclidean geometry shows the degree of distance between values, visualizing the relationships bet ween these major grass properties. For the untreated and 12.5% AHP pretreated diverse grasses, results showed that both ferulate and lignin content are negatively correlated with glucan digestibility with correlation coefficients of -0.96 and -0.81, while xylan content and pCA content are not correlated well with glucan digestibility with correlation coefficients of 0.21 43 and -0.50. The strong correlation between ferulate and lignin revealed in the results indic ate that both lignin and FA are solubil ized in pretreatment , thus FA content would be an indicator of pretreatment effectiveness in terms of cell wall digestibility [146]. Figure 3 .4 Heat map of correlation coefficients between cell wall properties and digestibility. The color blocks in the m ap reveal the relationship between the specific two individual cell wall properties as horizontally and vertically labeled, where the darker colors indicate positive correlations while the lighter colors indicate negative correlations. Note: this only incl udes data for 12.5% AHP pretreated and untreated corn stover, switchgrass and miscanthus. However, metal content was not selected to be included in the correlation coefficient heat map. Since metals participate in the FentonÕs reaction and synergistically affect the hydrogen peroxide delignification, the hydrogen peroxide loading is critical for the correlation between the metal content and the digestibility. In addition, metal content of woody biomass has been correlated with the alkaline oxidative pretrea tment (data not published yet) , and it was shown that the poplar with the lowest metal content responses the worst with the alkaline oxidative 44 pretreatment but responses the best with Cu(bpy) catalyzed alkaline oxidative pretreatment [55, 56]. Thu s, we can conclude that lignin removal is the major contributor of the increased hydrolysis yields following alkaline oxidative pretreatment, and the lignin removal is correlated with the cleavage of HCAs in the cell walls and the improved hydrogen peroxid e oxidation catalyzed by metals. 3.3.2 Glycome profiling derived xylan/ !-glucan extractability of cell walls Glycome profiling is based on the analysis of the binding abundance of glycan -directed monoclonal antibodies in sequential cell wall extracts. T he glycome profiling results here demonstrated the alteration of extractability for non-cellulosic glycans for 12.5% AHP pretreatment (Ori ginal data shown in Appendix Figure A1 -A2). The non -cellulosic glycans were grouped into four major types of glycans i ncluding xyloglucan, xylan, !-glucan and pectins based on a previous study [162]. Among them, glucuronoarabinoxylans (GAX) are the major grass hemicellulose in the primary and secondary cell wall with ferulate substitutions [180, 181]. !-glucan s are abundant in grasses where they play a cross -linking role comparable to that of pectins in dicots [181, 182]. Thus, this part of the study is focused on comparing the changes of these two cell wall glycans between type s of pretreatment and feedstock. In order to visualize the effects of pretreatment on cell wall glycans, the epitope abundance (calculated as described in the material and methods section) before and after pretreatment have been respectively plotted on the X and Y axes of the same diagram [162]. According ly, Figure 3.5 illustrates the pretreatment effects on xylan epitope abundance of diverse grasses including switchgrass (A), miscanthus (B), corn stover (C), bm1 stover (D), bm3 stover (E) and alkali -extracted corn stover (F). Among the six sequential extracts, since the first two milder extractions using oxalate and carbonate only remove extractives and loosely bound 45 pectins, and these two extracts only contain a small fraction of the recovered materials, we only focused on the epitopes in four of the other harsher extracts including 1 M KOH, 4 M KOH, chlorite, post -chlorite 4 M KOH in these plots. In these plots, most of the trend lines are linear and relativel y close to the X=Y line. It is shown in Figure 4 that the epitope abundance in the 1 M KOH extract and the 4 M KOH extract have been unchanged or slightly increased by pretreatment for switchgrass, miscanthus and corn stover. However, the brown midribs se em to have different xylan epitopes since the mAbs in 4 M KOH for bm1 and 1 M KOH for bm3 do not follow the linear trend, which indicates the xylan extactability is different from the other grasses. This means that the xylan is more extractable in bm1 and bm3 following pretreatment relative to other grasses, which can be attributed to improved lignin removal (Figure 3.2) relative to other grasses. Vermerris et al. also showed that the reduced lignin and ferulate content in brown midribs has been compensated by increased hemicelluloses, especially GAX, which results in a thickened secondary cell wall [175]. For the corn stover, epitope abundance in PC 4 M KOH has bee n shifted to chlorite extracts, indicating that the xylan epitopes possibly have been ÒloosenedÓ by pretreatment [162]. Alkali pre -extracted corn stover is the only sample that has reduced epitope abundance after pretreatment, indicat ing xylan epitopes are removed from the cell wall instead of ÒlooseningÓ for the other untreated grasses, possibly due to the already ÒloosenedÓ xylan structure by alkali pre -extraction. Interestingly, as early as 1956 it was recognized that xylans were solubilized from hardwood cell walls according to two rate regi mes during dilute acid hydrolysis [183]. These two pools of Òfast -reactingÓ and Òslow -reactingÓ xylan have subsequently been identified in kinetic studies of dilute acid hydrolysis of corn stover and switchgrass as well [184-186], although with significantly different kinetic parameters. It is likely that these pools of xylan identified from hydrolysis kinetic studies are related to the same properties that control xylan extractability from the cell wall. In this study, that a pool 46 of xylan is easier to be extracted by alkali -pre-extraction and that another pool of xylan is only solubilized in the post -pretreatment, might be an analogous phenomenon during alkaline pretreatments to these two po ols of Òdifferent -reactingÓ xylan solubilized in acid pretreatments. Figure 3 .5 Plots of binding quantity of individual mAbs for xylan epitopes in extracts of 1 M KOH, 4 M KOH, chlorite, and post chlorite 4 M KOH . Values on X or Y axis are respectively the binding strength of xylan specific mAbs on untreated or pretreated switchgrass (A), miscanthus (B), corn stover (C), bm1 stover (D), bm3 stover (E) and alkali -extracted corn stover (F) In addition to the xyla n epitopes, another epitope in grasses that of interest are the !-glucans, since these mix ed-linkage glucans are abundant and play a role in cross -linking in the primary cell wall of cereal grasses [181]. Two mAbs, including LAMP and CCRC -M154 are known to bind to the !-glucan and the xylan epitope, were chosen to compare the extractability of !-glucan and xylan [121]. Figure 3.6 shows the binding strengths of these individual mAbs (LAMP and CCRC -M154) on untreated and AHP pretreated switchgrass, miscanthus, corn stover, bm1 and bm3 stovers, and also includes alkaline pre -extracted corn A B C D E F 47 stove r and alkaline pre -extracted coupled with AHP post -treated corn stover. These result show that xylan and !-glucan have similar binding alterations induced by AHP pretreatment. Interestingly, the AHP pretreatment increases the epitope abundance for switchgr ass, miscanthus, corn stover, bm1 and bm3, but decreases the epitope abundance for alkaline pre -extracted corn stover. This phenomenon suggests the xylan and !-glucan in the primar y cell wall have been altered to different extent s by 5% alkaline extraction and 12.5% AHP. AHP pretreatment can increase the extractability of xylan and !-glucan in grasses, while the role of AHP after alkaline p re-extraction is to oxidize or remove the xylan and !-glucan on the already alkali -ÒloosenedÓ cell wall. Another notabl e observation is the extractaibility alteration in bm3. In bm3, the increases in epitope abundance for xylan and !-glucan in the sequential extracts are the most significant among five types of grasses , indicating bm3 possibly has a ÒlooserÓ or more extrac table structure of cell wall s than the other grasses , which should be the reason why bm3 always has the highest hydrolysis yields compared to the others shown in Figure 3.1. 48 Figure 3 .6 Plots of individual mAbs (LAMP and CCRC -M154) for !-glucan epitodes binding strength on untreated and pretreated switchgrass (SG), miscanthus (Mis), corn stover (CS), bm1 and bm3 stovers. Figure 3.7 shows a summary of all of the glycome profiling results, including epitopes other than those for xylan and B glucan, (Appendix Figures A1 -A2) in terms of the epitope abundance alteration by pretreatments of diverse grasses for three major cell wall epitopes (xyloglucan, xylan and pectin). In order to quantify th e epitope abundance alteration, each series in Figure 4 has been fitted to a linear line, and the slope for each extract for each A B 49 biomass has been recorded as an absolute value. The color in the heat map was determined as the logarithm of the values of the se slopes with respect to base 10 for all of the epitopes in every extract, for all biomass. The warmer the color is, the larger the fold difference of mAbs for the specific epitope has been increased. This result illustrates three responses of AHP on glyc an alterations quantified by glycome profiling. Firstly, binding quantity shifting from milder extract to harsher extract for corn stover shows that glycans have possibly been ÒloosenedÓ by pretreatment and made more extractable. Secondly, the largest rang e of warmer colors for bm3 stover suggests bm3 has the highest increased glycan extractability after AHP pretreatment, which contributes to higher digesitibility than the other grasses under varied H 2O2 loading (Figure 3.1 ). Thirdly, the largest range of c ooler color for alkaline pre -extracted corn stover shows a reduction of glycan epitopes by AHP post -treatment, possibly due to the direct removal of glycans on the alkali -ÒloosenedÓ materials. 50 Figure 3 .7 Heat map of major ce ll wall non -cellulosic glycans extractability alteration by AHP pretreatment . 3.3.3 ELISA screening of the pretreatment liquor ELISA screening using cell wall glycan -directed mAbs could also be applied to characterize the glycans solubilized in pretreatment liquors generated from diverse grasses according to previous research [98]. Figure 3.8 shows the screening results of 12.5% AHP pretreatment liquors for five types of grasses including corn sto ver (A), bm1 stover (B), bm3 stover (C), miscanthus (D) and switchgrass (E). The brightness showing the high abundance of xylan epitopes indicates that AHP pretreatment removes xylans from grass cell walls into the liquid phase. In contrast with the signif icant amount of the xyloglucan epitopes in the glycome profiling results for all types of grasses, the mostly dark region in the top portion of Figure 8 51 shows there are much less xyloglucan epitopes in these pretreatment liquors, which indicates that AHP p retreatment either removes minimal xyloglucans to the liquid phase, or solubilizes xyloglucans to compounds that do not bind to the mAbs . Additionally, the very similar pectin epitopes abundances in the pretreatment liquors to the glycome profiling extract s indicates the AHP pretreatment also removes pectins from the grass cell walls with the similar quantity as it does for the sequential extractions. According to Figure 3.2, xylan removal of the three stovers is much more significant than the two perennial s. However, the ELISA screening results did not show large differences between the diverse species in terms of xylan epitope abundance, and only showed slightly higher xylan epitope abundance in the liquors from the bm1 and bm3 stovers. This is because of lower concentrations of the binding sites for the mAbs in the liquid phases than in the extracts [121]. The lower resolution of the quantification for the liquid phase makes the results indistinguishable. 52 Figure 3 .8 ELISA screening of glycans solubilized in 12.5% AHP pretreatment liquors of (A) corn stover, (B) bm1 stover, (C) bm3 stover, (D) miscanthus and (E) switchgrass using 155 cell wall glycan -directed mAbs . (ELISA screening was done by Sivakumar Pattathil , University of Georgia.) 3.3.4 Enzyme accessibility of grass cell walls Carbohydrate binding m odules (CBMs) from microbial cell wall polymer hydrolases [187] are another important category of molecular tools to measure the accessibility of cellulose and non-cellulosic glycans [119, 120]. In this study, enzyme binding assay was developed based on previous work [188, 189]. In order to quantify the accessibility of cellulose microfribrils, a family 3 CBM [190] from Clostridium thermocellum , which is specific to crystalline cellulose, was used to determine the celluloase accessibility of c orn stover under A B C D E 53 different severities of AHP pretreatment. The binding curve generated using CBM3 (Figure 3.9) on untreated and pretreated corn stovers could then be fit with a Langmuir model [132]: !!"# !!!"# !!"# !!!!!!!"# !! The calculated model parameters are shown in Table 3.2. The binding capacity and affinity of CBM3 increase d with increased pretreatment severity in term o f the hydrogen peroxide loading. Figure 3 .9 Binding isotherms of CtCBM3 on untreated, 12.5% and 25% H2O2 (w/w) AHP pretreated corn stover Biomass Binding capacity Emax (mg/g substrate) Binding affinity Ka ds-1 (M-1) Untreated corn stover 8.57 0.66x106 12.5% AHP corn stover 10.86 1.14x106 25% AHP corn stover 21.22 3.94x106 Table 3. 2 Parameters fitting based on Langmuir adsorption model showing binding capacity and affinity of CtCBM3 on corn stovers under different treatments . Binding characteristics of single celluases [191], cellulase mixtures [188], and CBMs [190, 192, 193] have been investigated on model cellulose such as Avicel, bacterial cellulose and 54 other regenerated cellulose [191], as well as on untreated and pretreated biomass [188, 194-196], and on isolated lignin [197, 198], to investiga te the interaction between cellulolytic enzymes and substrates . Previous study showed that the hydrogen peroxide loading increases the bound fraction of an enzyme cocktail on AHP pretreated corn stover , and the cell wall hydrophilicity quantified by WRV and enzymatic hydrolysis yield are correlated with the bound fractions [195]. In this study, a commercial CBM3 from Clostridium thermocellum specific to crystalline cellulose was shown to bind with the untreated and pretreated corn stover unde r different pretreatment severities, indicating the cellulose accessibility of corn stover was increased by the pretreatment due to the removal of lignin and hemicelluloses with increased loading of hydrogen peroxide . 3.4 CONCLUSION In this study, diverse grasses including three lines of corn stovers and two perennials commonly used as bioenergy feedstocks including switchgrass and miscanthus, were compared based on their responses towards alkaline oxidative pretreatments at increasin g hydrogen peroxide loading s. Lignin content, metal content and ferulate content were three main important factors affecting the biomass digestibility profiles associated with hydrogen peroxide loading. The glycome profiling results indicated that alkaline oxidative pretreatments impact grass cell wall glycan by ÒlooseningÓ the glycans to improve the cell wall digestibility , while impact ing the alkali pre -extracted cell wall by directly removing the glycans. The glycans released into the pretreatment liquor mainly consist of xylans and pectins, indicating the removal of these glycans but not the xyloglucans. Based on the binding isotherms generated using cellulase cocktails and cellulose specific CBMs, increased binding capacity and affinity were observed fo r corn stover under increased loading of hydrogen peroxide, which indicates the AHP pretreatment increased cellulose accessibility . Higher 55 binding for corn stover following hydrogen peroxide oxidation than sw itchgrass indicates difference s in higher order structure between the two species. 56 4. CORRELATING MAIZE CE LL WALL PROPERTIES T O THE RECALCITRANCE 4.1 INTRODUCTION It was shown that c orn stover compromised 85% of 320 million dry tons of total primary residue by 2030 according to ÒThe Bil lion Ton StudyÓ , and is considered to be one of the most promising lignocelluosic bioenergy feedstocks in the US [90]. Strategies for breeding maize for its cellulosic biofuel (as well as feed or silage value) have been focused on increasing the non -starch biomass yield and the total carbohydrate content [199, 200] as well as reducing the overall recalcitrance [201, 202] for example through decreasing lignin content [203], altering the lignin monomer content [204], and redirection of carbon to non -cellulosic sugars (e.g. "-glucan, starch) [205, 206]. The cell wall composition and ruminant digestibility changes during plant development stage, and vary by parts of the plant [207], so the c ell wall recalcitrance is influenced by environmental or agronomic factors such as maturity [199, 208], harvest method [209] and genetic variation [210]. The ratio of cell types such as the epidermis, sclerenchyma, vascular bundle zone cells and pith parenchyma [93] also results in significant differences in cell wall thickness, porosity and accessibility. Many studies have been carried out previously to correlate the variability of maize cell walls with in vitro digestibility [130, 210-217], in vivo ruminant digestibility [214], and cellulolytic enzymatic digestibility with [170] or without pretreatments [218]. It was found that the untreated maize cell wall digestibility is impacted by neutral detergent fiber content, lignin content, S/G rat io, etherified FA and esterified pCA [211, 212], and the correlation between S/G ratio and esterified pCA suggests the association between pCA esters and S lignin [211]. Lignin cont ent is general ly regarded as the major factor impacting cell wall degradability 57 among grass cell walls [214]. Howeve r, regression model s correlating lignin condensation and cell wall p-coumaroylation with ruminant digestibility suggest that uncondensed lignin represented by the "-O-4 yield and S lignin linked pCA would be one of the most crucial factor impacting digesti bility for untreated maize [130]. Alkaline oxidative pretreatment cleaves !-O-4 bond to remove lignin so as to enhance the hydrolysis yield s [146], and s tudies showed that "-O-4 content is an important impediment to cell wall degradability since it has higher interactions with cellulose [130]. However, the me chanism that S/G ratio impacts hydrolysis yield s is not that clear compared with "-O-4 and the results were varied by species [22, 146]. On the other hand, the delignified grass cell wall has been regarded as a hydrogel [100] with lignin prevent the swelling of the lignified cell walls. Grass cell walls swell after alkaline treatment [100, 160] and are more easily accessible to both enzyme s and water [195], which is another reason that alkaline or alkaline oxidative treated monocots yields much higher digestibilities compared to dico ts [219]. The water retention value (WRV) can be used as an indirect measurement of a number of cell wall properties including o verall hydrophilicity, porosity, and polysaccharides accessibility, and has been used in our recent work as a predictor of enzymatic hydrolysis yields for corn stover and switchgrass subjected to alkali, alkaline -oxidative, and liquid hot water pretreatmen ts [195]. Recently, approach es for high -throughput screening of pretreatment and enzymatic hydrolysis have been developed for the purpose of screening large sets of cell wall material for the purpose of identifying potential reduced recalcitrance phenotypes both with and without a pretreatment [220, 221]. Understanding the cell wall traits associated with reduced recalcitrance phenotypes or phenotypes that exhibit specific r esponses to pretreatment are an important component of screening. However, typically only composition analysis is performed across the sample sets, limiting the understanding of the cell wall properties, traits, or phenotypes that are also associated with the cell wallÕs response to enzymatic hydrolysis. 58 Notably, recent work employing high -throughput screening of panels of promising bioenergy grasses including diverse cultivars of wheat ( Triticum aestivum L.) [222] and Miscanthus spp. [223] subjected to hydrothermal pretreatment and a sorghum ( Sorghum bicolor (L.) Moench) diversity panel subjected aqueous ammonia pretreatment [224] were able to identify substantial differences in cell wall responses to pretreatment. The goal of this work is to better understand how maize cell wall properties both impact initial recalcitrance as well as NaOH pretreatme nt. Specifically, 12 cell wall properties were selected that may influence overall cell wall recalcitrance including the WRV, the xylan and lignin contents before and after pretreatment, initial cell wall acetate content, the pCA and FA content before pret reatment, the solubilized pCA and FA during pretreatment, as well as the S/G ratio of the lignin in the untreated cell wall. These properties were quantified for a diversity panel of 26 maize lines that was previously identified as exhibiting substantial phenotypic diversity in glucose release following mild NaOH pretreatment (Muttoni et al. , 2012). These were subj ected to mild NaOH pretreatment, the glucose yields determined for all the different lines before and after the pretreatment . Subsequently, the correlations between the cell wall properties and hydrolysis yields were investigated. 4.2 MATERIALS AND ME THODS 4.2.1 Maize diversity panel (The maize diversity panel was grown and harvested by Marlies Heckwolf , University of Wisconsin.) The maize diversity set was grown in Arlington, WI in 2012 and harvested at grain physiological maturity with the Case IH¨ 2144 axial -flow combine, which allows harvesting grain and biomass in a single pass as well as measuring whole -stover and grain weight and 59 lastly chopping the whole stover. A sub -sample of approximately 1 kg per plot was obtained and dried at 50 ¼C. The dried material was ground to a particle size of 1 mm. A complete list of maize lines with accession number or resources is presented in Appendix Table A2. 4.2.2 Water Retention Value (WRV) measurement The WRV of samples pretreated with alkali were determined as follows: 2 g of biomass were added to 150 mL of 0.08% (w/w) NaOH solution. Biomass samples were soaked for three hours at room temperature, with occasional mixing. Samples were then neutralized and filtered using a fabricated 200 -mesh Buchner funnel with 500mL of deionized water. Pellet samples were created by a 200 -mesh stainless steel membrane in a ten mL plastic vial, with wet biomass placed on top of membrane. Vials were centrifuged at 900 x G for 15min. Centrifuged biomass was placed in tared aluminum trays, weighed, and dried in a convection oven at 105 ¼C for 24 hours. Dried samples were then reweighed and the WRV was determined in triplicate using the following equation [195]: !"#!!"## !!"!!"#$%&'(") !!"#$%& !"## !!"!!"#$!!!"#$%& !!! 4.2.3 p-hydroxycinnamic acids determination The hydroxycinnamic acid content of the untreated maize and mild alkaline treated maize was determined by a modified high -throughput alkaline extraction method. In this, 0.5g biomass was treated with 25 mL of 3M NaOH in sealed pressure tubes and held at 120 ¼C for 1 hour in an autoclave. After cooling down in room temperature, 250 µL of 10 mg/ mL methanol solution of o -coumaric acid was added as an internal standard for each sample. The mixture was transferred to Eppendorf tubes and centrifuged at 13,000 rpm for 10 minutes. The pH of the supernatant was adjusted to 2 with concentrated HCl, and t he samples were then place d into a 4 ¼C fridge overnight. After the overnight storage, the supernatant of the samples were analyzed in a HPLC stalled with a C18 column (Discovery HS C18 HPLC 60 Column, 5 cm # 2.1 mm, 5 µm). Standards containing ferulic acid, p-coumaric acid and o -coumaric acid were run in parallel to determine the concentration of the acids in the samples. 4.2.4 Alkaline extraction, composition analysis and enzymatic hydrolysis Milled samples of 27 maize lines were subjected to a mild alkaline pretreatment. During the alkaline extraction, 2 g biomass was added to a 20 mL 0.8% NaOH aqueous solution in 50 mL centrifuge tube and the tubes were placed in a static water bath at 80 ¼C for 1 hour. After pretreatment, the liquid was removed via filtrat ion and the residual biomass was washed by deionized water until neutral. The yield of pretreatment was determined by measuring the difference between the original and air -dried pretreated materials. Two -stage sulfuric acid hydrolysis was carried out accor ding to the NREL composition analysis protocol to determine the structural carbohydrates of the untreated and pretreated maize lines. A HPLC installed with an Aminex HPX -87 H (Bio -Rad, Hercules, CA) column, and the acid insoluble residue was regarded as th e lignin content. The enzymatic hydrolysis was performed at pH 5.0, 50 ¼C and 180 rpm shaking, with a 30mg/g protein loading of C -Tec 2 (Novozymes, Denmark) on glucan for 6 hours or 72 hours. The glucan yield was determined as the amount of glucan released after enzymatic hydrolysis divided by the original glucan content in the pretreated samples after washing. 4.2.5 S/G ratio prediction The S/G ratio was predicted by the combination of principal component analysis and partial least square (PLS) regression based on their py -MBMS spectra provided by Robert Sykes (National Renewable Energy Laboratory). The parameters of the prediction model were generated previously by correlating thioacidolysis S/G and py -MBMS spectra of a set of samples including untreated and alkaline oxidative pretreated hybrid corn stover, brown midrib stover bm1 and bm3, switchgrass and Miscanthus . This model will be further discussed in next chapter. 61 4.3 RESULTS AND DISC USSION 4.3.1 Cell wall properties and hydrolysis yields 12 types of cell wall properties, including initial and final WRV (X1 and X2), initial and final xylan content (X3 and X4), initial and final lignin content (X5 and X6), initial acetate (X7), pCA (X8) and FA (X9) contents, solubilized pCA (X10) and FA (X11), and S/G ratio (X12), were used in this study as a set of maize cell wall properties. These t welve cell wall properties or traits that may be indicators of cell wall recalcitrance were quantified across the maize diversity set in addi tion to 6 and 72 -hr hydrolysis yields prior to and following mild NaOH pretreatment. The f eatures describing the property data set presented in Table 4.1 and the yields presented in Figure 4.1. As mentioned in the Introduction, the WRV are proposed to act as a proxy variable that may be able to explain a number of phenomenon related to cell wall polysaccharide accessi bility to cellulolytic enzymes. As observed, untreated maize has WRVs ranging from approximately 1.9 to 2.4 with the WRV following alkaline tr eatment ranging from 2.3 to 3.3 (Table 4.1). This phenomenon suggests that the variety of levels of cell wall swelling in alkali for different phenotypes increases compared with cell wall swelling in water, indicating that the alkali removes lignin and oth er functional groups such as hydroxyl and carboxyl groups at different levels, which results in the variability of hydrophilicity of different maize cell walls after alkaline pretreatment [195]. Xylan concentration also exhibited substantial variability within the data set and increase d in relative abundance from an average of 0.200 g/g to 0.266 g/g as a consequence of the remo val of lignin and extractives. Wide ranges were observed for initial cell wall acetyl content, pCA content, and FA content as well as corresponding pCA and FA rel ease. The S/G ratio as determined by py -MBMS coupled to PLS models also shows a diverse range, although with in the range reported for maize [225]. 62 !!X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 Min 1.68 2.22 0.16 0.20 0.13 0.07 26.70 8.66 9.02 4.10 4.36 0.68 Avg 2.02 2.68 0.20 0.27 0.17 0.11 33.10 11.90 11.30 5.87 5.74 0.97 Max 2.39 3.32 0.26 0.31 0.21 0.14 39.20 15.90 13.90 8.49 7.43 1.52 Std 0.17 0.30 0.02 0.02 0.02 0.02 3.25 1.66 1.26 1.27 0.79 0.20 Table 4. 1 Variability within the data set for the 12 properties across the 27 maize lines. Figure 4 .1 Range of hydrolysis yields obtained for untreated and NaOH -pretreated maize for (A) 6-hr hydrolysis yields and (B) 72-hr hydrolysis yields. Error bars represent data range for duplicate samples . Due to missing data, some samples do not appear. Glucose hydr olysis yields ranged between 37.8% to 55.5% for untreated biomass and 58.6% and 82.5% for NaOH -pretreated biomass follow ing 72 hours of hydrolysis (Figure 4.1), clearly exhibiting a diverse range of digestibility phenotypes as well as diverse responses to mild NaOH pretreatment. The 6 -hr hydrolysis yields are intended to represent the initial hydrolysis rates, while the 72 -hr hydrolysis yields represent the extent of hydrolysis as only minimal additional sugar release is observed beyond 72 hours (data not s hown). It should be noted that these data are plotted for glucose yield rather than glucose release, such that results are not biased towards cell walls with higher glucan content. 63 4.3.2 Correlation between cell wall properties In order to better visuali ze the relationships between the variables, a correlation map was developed and organized using hierarchical clustering according to the Euclidian distance between sets of the Pearson correlation coefficients (R) (Fig ure 4. 2). The magnitude, scale, and sig nificance of the correlations are presented in Table 4.2. Within the correlation map (Figure 4.2), several multi -property clusters of positive correlations are observed along the top left to bottom right diagonal, while one multi -property cluster of negati ve correlations stands out in both the bottom left and top right corners. This indicates that there a number of properties that are correlated across diverse maize lines and may be responsible for both differences in the cell wallÕs response to enzymatic hydrolysis as well as the response to pretreatment. A number of strong positive correlations between related properties or yields are observable along the top left to bottom right diagonal, namely the 6 and 72 -hr hydrolysis yields, initial pCA content and pCA solubilization, initial FA content and FA solubilizat ion, and initial and final WRV. These specific results are not surprising as they may be expected to be related. A number of property correlations that highlight either causal or merely correlative r elationships between cell wall properties (but not hydrolysis yields) from within the highlighted clusters were selected and replotte d in Figure 4.3. These correlations are plotted using either initial acetate content or initial WRV as the abscissa as thes e properties appear within two of the important clusters. It should be stressed that acetate content and WRV are not necessarily the properties responsible for the variations in the other properties, but are merely correlated to them. 64 Figure 4 .2 Correlation map of the PearsonÕs correlation coefficients for the 12 cell wall properties and 4 hydrolysis yields (red text) across the 27 maize lines as organized by hierarchical cluster analysis. Clusters of properties and y ields exhibiting are highlighted. ÒInitialÓ indicates the property in the original untreated biomass sample while ÒfinalÓ indicates the property following pretreatment. 65 Figure 4 .3 Several between -property correlations highli ghted to demonstrate relationships within the data set. Each data point represents the property for one of the 27 maize lines. PearsonÕs correlation coefficients and p-values are presented for all property correlations with p ! 0.05.Error bars on individu al samples are not shown to improve clarity. Xylans (as well as other non -cellulosic cell wall polysaccharides) are known to be substantially O -acetylated and the role of these substitutions are hypothesized to be to control the strength of glycan -glycan c ross -linking by preventing H -bonding between cell wall polymers [20, 226, 227]. Of note is that while the acetate conte nt showed a diverse range (26.7-39.2 mg/g), there was no correlation with the cell wall xylan content (Fig ure 3.3A) indicating either that the xylans exhibited either a wide range of O -acetylation - calculated to be 0.25 -0.60 mol Ac per mol Xyl, which is within the reported range for maize [228] - or that 66 the acetyl groups are sub stituted on moieties other than xylan ( e.g. glucomannans, pectins, or lignins). While uncorrelated to the initial xylan content (Figure 3.3A), this diverse range of acetate contents can be observed to be exhibit correlations to a number of other cell wall properties, including a strong, statistically -significant positive correlations to the final xylan content (Figure 3.3A), the initial lignin content (Figure 3.3B), the initial content and release of pCA (Fig ure 3.3C), while showing negative correlations t o the final lignin content (Figure 3.3B), a strong positive correlation to both the final lignin content (Fi gure 3.3B) and the initial WRV (Figure 3.3D). The acetate results can be interpreted as potentially relating to the lignin release. While it may be expected that high acetate contents would consume more alkali and presumably hinder lignin removal the opposite is the case. As observed in Figure 3B, the initial cell wall acetate content is positively correlated to initial lignin contents and negatively correlated to final lignin content, indicating that high -acetate cell walls are likely to be more recalcitrant yet respond better to alkaline pretreatment with respect to lignin removal due to the cleavage of the ester bond between acetate and hemicellulos es is especially susceptible to cleavage by alkaline pretreatments. Additionally it can be observed that initial pCA content and pCA released are positively correlated to initial acetate content (Figure 3.3C). This may be a consequence of the correlation of between acetate and lignin since pCA is known to be acylated to syringyl lignins and may only represent th e correlation to lignin content [79]. The correlation to final xylan content may be a consequence o f high acetate corresponding to higher lignin removal, which would enrich the xylan content of the pretreated biomass. Overall, it is not clear whether this cluster of co -varying properties is an indication of differences in the ÒaverageÓ cell wall propert y or whether these may indicate differences in the abundance of, for example, high -lignin, high -pCA, and high -acetate tissues. Recently, work has been targeted at altering of the expression of genes associated with O -acetylation as 67 a strategy for a ltering cell wall recalcitrance [229], although itÕs currently not clear how only alteration in xylan O -acetylation will impact cell wall recalcitrance. Interestingly, the WRVs did not show a signifi cant correlation to any of the cell wall biopolymer content ( i.e. lignin and xylan) while it did exhibit significant correlations (albeit weak) to substitutions on these biopolymers ( i.e. pCA and acetate, Figure 3.3D). Specifically, the initial cell wall pCA content is inversely correlated to the initial WRV may provide evidence that highly p-coumarylated cell walls may be more hydrophobic (and higher in lignin content), potentially imparting increased recalcitrance to degradation by microbial pathogens or rumen microbiota. The (weak) negative correlation between initial cell wall acetate content and initial WRV (Figure 3.3D) could be potentially due to increasing hydrophobicity of exposed xylans ( i.e. the glycan will have an exposed ethyl group rather than a hydroxyl group). Also , the hypothesis that highly acetylated cell walls are less porous due to a decreased level of H -bonding betwee n xylans and cellulose does fit this data since increasing acetyl content corresponds to decreasing WRVÕs (Figure 3.3D). As initial lignin content is positivity correlated to the cell wall acetate content (Figure 3.3B), the decreasing initial WRV with increasing acetate could be a consequence of the increasing lignin content decreasing the cell wallÕs capacity to sorb water. 4.3.3 Correlations between cell wall properties and hydrolysis yields Besides between -property correlations, correlations between cell wall properties and hydrolysis yields are important for understanding property contributions to cell wall recalcitrance . All significant ( p ! 0.05) correlations between cell wall properties and hydrolysis yields are plotted in Figure 4. 4. Notably, this plots only correlations for untreated 6-hr yields and NaOH -pretreated 72 -hr yields. The untreated 72 -hr yields did not exhibit significant correlations to any properties other than the untreated 6 -hr yields (Figure 4.2 and Table 4.2). However, the correlations that were strongest for the untreated 6 -hr hydrolysis 68 yields were not that significant for 72 -hour digestibility, exhibiting similar correlations with p-values between 0.05 -0.15. The differences between the hydrolysis yields obtained in different time points, 6 -hr versus 72 -hr, may indicate that the initial cell wall composition impacts the hydrolysis rate more stron gly than the hydrolysis extent. The 6 -hr hydrolysis yields for NaOH -pretreated corn stover were not found to exhibit strong correlations with any other properties and furthermore, were often lower than the untrea ted 6 -hr hydrolysis yields (Figure 4.1B). Thi s contradictory result may be due to the drying of the pretreated material necessitated by the analysis that resulted in its stronger resistance to rehydration which may have introduced more variability in the data for the initial glucose release by hydrol ysis, but presumably not the extent of hydrolysis. X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 Y1 Y2 Y3 Y4 X1 1 0.54 ** -0.21 -0.33 -0.06 -0.04 -0.41 * -0.37 * -0.26 -0.16 0.19 -0.06 0.40 * 0.18 -0.40 0.06 X2 1.00 0.19 -0.40 -0.20 0.15 -0.37 -0.33 -0.36 -0.12 -0.10 0.13 0.00 0.00 -0.11 -0.12 X3 1.00 -0.09 -0.20 0.19 -0.03 0.12 0.02 0.18 -0.12 -0.18 -0.45 * -0.17 0.10 0.12 X4 1.00 0.43 * -0.39 0.63 ** 0.37 -0.02 0.52 -0.11 -0.08 -0.31 -0.26 0.27 0.24 X5 1.00 -0.16 0.50 ** 0.52 0.15 0.36 -0.18 -0.06 -0.40 * -0.30 -0.15 -0.18 X6 1.00 -0.48 ** -0.09 -0.25 -0.39 -0.27 0.05 -0.11 -0.03 -0.13 -0.50 ** X7 1.00 0.65 ** 0.24 0.73 ** -0.22 -0.14 -0.53 ** -0.34 0.13 0.20 X8 1.00 0.38 0.65 ** -0.25 -0.26 -0.58 ** -0.35 0.00 0.03 X9 1.00 0.16 0.59 ** -0.44 -0.11 -0.12 0.02 0.32 X10 1.00 -0.05 -0.32 -0.61 ** -0.30 0.02 0.40 * X11 1.00 -0.41 * 0.25 -0.11 -0.03 0.50 ** X12 1.00 0.36 ** 0.31 0.03 -0.24 Y1 1.00 0.59 ** -0.05 0.10 Y2 1.00 -0.15 0.01 Y3 1.00 0.25 Y4 1.00 Table 4. 2 Calculated p roportionality constants (unscaled) between all cell wall properties and hydrolysis yields. * indicates the parameter is significant at p ! 0.05. ** indicates the parameter is significant at p ! 0.01. 69 A number of important trends can be identified from the data in Figure 4.4. The first is the identification that the WRV is positively correlated to the untreated 6 -hr hydrolysis yield (Fig ure 4.4A). As discussed in the introduction, the WRV was included as a parameter that may offer the potential for consolidating a number of other, unquantified cell wall properties that may be able to be correlated to hydrolysis yields. Our previous work used only two types of biomass (corn stover and switchgrass) subjected to varying levels of delignification by alkaline oxidative pretreatment and found strong correlations between WRV and hydrolysis yields [195]. The current work used a wide range of (similar) biomass that were subjected to a single pretreatment condition which may explain why WRVs following pretreatment were not able to predict different hydrolysis yields following pretreatment. 70 Figure 4 .4 Summary of significant property correlations to glucose hydrolysis yields. Open data points represent 6 -hr hydrolysis yields for untreated biomass; filled data points represent 72-hr hydrolysis yields for NaOH -pretreated biomass. Each data point represents the value of the property and corresponding yield for one of the 27 maize lines. PearsonÕs correlation coefficients and p-values are presented for all property correlations with p ! 0.05. Error bars on individual samples are no t shown to improve clarity. The initial cell wall acetate content demonstrated a negative correlation to the untreated 6 -hr hydrolysis yield although not to the pretreated 72 -hr yield (Figure 4.4B). As discussed previously, the acetate content is strongly correlated to many other properties and potentially this cluster of related properties may be able to explain differences in hydrolysis yields rather than a single property. For example, initial acetate content is strongly correlated to initial 71 lignin cont ent (Figure 4.3B), which may be the property most strongly responsible for the response in the hydrolysis yields. Previously, nonlinear models have been developing relating decreases in acetate and lignin to increases in hydrolysis yields for lime -pretreat ed corn stover [230] and hybrid poplar [131], although these correlate simultaneous acetate and lignin removal by alkali to improved hydrolysis yiel ds rather than initial variability in these properties as is done in the present work. Lignin is thought to impact saccharification yields by physical occlusion of polysaccharides [73], providing resistance t o swelling [231], and by non -specific binding to cellulases [232] and studies have generally shown strong negative correlation between lignin content and hydrolysis yields for a wide range of untreated grasses [130, 214, 233] with strong correlations following a delignifying pretreatment [146, 230, 234]. As expected, the untreated lignin content is n egatively correlated to the untre ated 6 -hr hydrolysis yield (Figure 4.4C ) and the lignin content following pretreatment is negatively correlated to the pretreated 72 -hr hydrolysis yield (Fig ure 4.4D). However, the lignin contents prior to and following pretreatment are no t correlated to each other (Figure 4.2 and Table 4.2) indicating that lignin removal for mild NaOH pretreatment is not necessarily dependent on the initial lignin content, but is dependent on m any other cell wall properties. This is signifi cant in that many strategies for reduced cell wall recalcitrance have targeted lignin levels [235], although it may be sufficient that the lignin is easily removable by the pretrea tment to decrease the final cell wall recalcitrance rather than the initial cell wall recalcitrance. As weÕve already identified in this work, the lignin removal is more important than the initial lignin content for mild NaOH pretreatment and consequently , lignin properties that may contribute to improved lignin removal are important. Another interesting lignin -related finding is that the initial S/G ratio is positively correlated to the untreated 6 -hr h ydrolysis yield (Figure 4.4E), while the trend appare ntly reverses for the pretreated hydrolysis yield 72 (although this is not a statistically significant correlation). The correlation between S/G ratio in untreated biomass and in vitro ruminant digestibility for grasses has been somewhat contradictory in the literature [22, 130, 211] although a general trend is that increasing S/G ratio may be linked to increasing digestibility. The initia l xylan content is found to be negatively correlated to the untreated 6 -hr hydrolysis yield (Fig ure 4.4F). Considering that both initial lignin and xylan are negatively correlated to the untreated hydrolysis yields, yet, these two propertie s are uncorrelat ed to each (Figure 4.2) may indicate that both of these components play an important role in the limited access of cellulolytic enzymes to cellulose. The enzyme cocktail utilized for this study was not supplemented with xylanase, which may be a factor cont ributing to this result. Ferulate and diferulate esters are known to be attached to the primary hydroxyl at the C5 position of $-L-arabinofuranosyl residues in xylans and be involved in cross -linking between xylans to lignin by incorporation of the aroma tic moiety through ether linkages into growing lignin polymers or by cross -coupling [21]. This cross -linking of cell wall polymers by ferulates is generally accepted to play an important role in cell wall recalcitrance [236]. Negative correlations have been found in the literature for total cell wall ferulate content and in vitro ruminant digestibility in maize [237] as well as other diverse grasses [96] and decreasing cell wall ferulate content has been investigated as a strategy for i mprov ing the forage quality of maize [238]. However, while strong, statistically significant negative correlations were identified between etherified ferulates and in vitro digestibility, esterified ferulates content was found to be positively correlated to in vitro digestibility in smooth bromegrass ( Bromus inermis Leyss subsp. inermis), cocksfoot ( Dactylis glomerata L.), and reed canary grass ( Phalaris arundinacea L.) [239, 240], and other recent work has validated a strategy for increasing esterified diferulate co ntent in maize to improve its digestibility [241]. The present work did not distinguish between etherified and esterified ferulates although the 73 method for ferulate content was performed under harsh enough conditions that this term likely comprises both etherified and este rifed as increasing the alkali loading and temperature during saponification did not result in additional ferula te release (unpublished data). The ferulate release during pretreatment may be more representative of esterified ferulate due to mild conditions utilized for pretreatment. Nonetheless, ferulate content and ferulate released were found to be strongly inversel y correlated to each other ( Figure 4.2 and Table 4.1), and importantly, the ferulate released showed a strong correlation to the 72 -hr hydrolysis yields following NaOH pretreatment (Fig ure 4.4G), although not any of the other hydrolysis conditions. This can be understood as cell wall s that are highly cross -linked are more susceptible to alkaline pretreatment. Recently, incorporation of ferulate esters into hybrid poplar has been validated as a strategy to improve the enzymatic hydrolysis following an alkaline pretreatment [242]. Lignin content, pCA content, and S/G ratio in maize stem internodes and rinds have been shown to increase with increasing maturity while f erulate content does not change [225]. The simultaneous increase in these properties has been implicated in the decrease in maize digestibility with increasing maturity [18, 91]. Bot h the initial pCA (Figure 4.4H) and solubilized pCA (Figure 4.4I) exhibit statistically significant correlations to the hydrolysis yields. These results are notable in that the correlation is of the opposite for untreated biomass compared to NaOH -pretreate d biomass, whereby increasing release or content of pCA corresponds to decreasing hydrolysis yields for untreated cell walls and increasing hydrolysis yields for pretreated cell walls. High pCA content in untreated maize has been correlated to low in vitro digestibilities [130, 211], although the finding that high initial p-coumarylation of lignin may be related to high hydrolysis yields following pretreatment is novel. As discussed previously, the pCA content is positively correlated to the initial lignin content so this could be an indirect measure of the impact of lignin content. As another 74 alternative, pCA-containing lignins may be responsible for sorbing and inactivating cellulolytic enzymes as is known for polyphenolic compounds such as tannins and lignins and phenolic acid monomers [243]. 4.4 CONCLUSION A diversity panel of 26 maize lines were subjected to mild -NaOH pretreatment and a broad set of properties relating to cell wall recalcitrance were characterized. The hydrolysis yields of un -pretreated cell walls were found to be positively correlated to the WRV and the S/G rati o and negatively correlated to the xylan and acetate content. While the initial cell wall xylan and acetate content were uncorrelated to each other, the acetate content was found to exhibit a number of strong correlations to other cell wall properties. Th e pretreated cell wall hydrolysis yields were positively correlated to the ferulate released by pretreatment, indicating that breaking of ferulate cross -links between cell wall polymers is an important outcome of pretreatment. As expected, statistically si gnificant negative correlations were identified between the cell wall lignin content and the hydrolysis yields for both untreated and NaOH -pretreated maize . It has long been known that cell wall lignin content can be negatively correlated to enzymatic hydr olysis yields, in vitro digestibility, as well as in vivo digestibility in ruminants. However, the data demonstrated that the initial cell wall lignin content and the pretreated cell wall content were not correlated to each other, and the cell wall lignin content following pretreatment was not correlated to untreated hydrolysis yields. This important finding indicates that while enzymatic hydrolysis yields may be set by the cell wall lignin content, the cell wallÕs response to a delignifying pretreatment s uch as mild NaOH pretreatment is not necessarily set by the initial lignin content. The pCA that is saponifiable by mild NaOH pretreatment showed a negative correlation to the hydrolysis yields of untreated maize and the inverse response for pretreated mai ze indicating that cell 75 walls with a high content of saponifiable pCA are more recalcitrant without treatment, yet respond better to pretreatment than cell walls with a low content of saponifiable pCA. 76 5. MODELING OF THE PLANT CELL WALL RECA LCITRANCE 5.1 INTRODUCTION Plant cell wall modification is a consequence of thermochemical or biological pretreatment and is a prerequisite for the enhancement of enzymatic hydrolysis to generate high yields of fermentable sugars. Detailed and accurate quantitative a nalysis of the quantity or structures of macromolecular cell wall components are essential for selecting the feedstock, monitoring the deconstruction processes, and calculating the details of the final sugar yields. However, compared with other biological macromolecules such as proteins or DNAs, plant cell wall macromolecules are relative difficult to be analyzed in a high -throughput manner due to their complexity. Nondestructive quantitative characteriza tion methods of the cell wall are desired for fundame ntal research on the cell wall recalcitrance. Data mining and modeling are frequently used in the analytical field of biomass characterization. One high -throughput approach is to apply spectroscopic techniques using multivariate appro aches to build up mod els to convert spectra to compositional properties for biomass. Principal component analysis (PCA) was usually utilized to simplify the spectra, and partial least square regression (PLS) models were applied to correlate the principal components to easily i dentifiable cell wall features. Prediction of the lignin content in grasses [125, 244] is one application of these models based on pyr olysis -molecular beam mass spectrometry (py -MBMS). Also, PCA/PLS models is capable of predicting other cell wall composition properties including the content of glucan, xylan, mannan, arabinan and galactan, by incorporation of composition analysis with eit her py -MBMS spectra [245] or NMR spectra [246]. Near -infrared spectroscopy (NIR) has been also frequently used to analysis the composition properties [127, 128] and to predict the digestibility [247] by multivariate 77 approach for diverse types of biomass including softwoods [127, 128], hardwoods or grasses [247]. Besides those spectroscopic techniques, other types of models have been developed to relate the cell wall properties include cellulose crystallinity [129], ligni n content, p-coumaric and ferulic acids content [130], to hydrolysis yields by either neural network [131] or regression models [130]. Li et al. reviewed multivariate data analysis as a high -throughput characterization tool for biomass using py -MBMS and NIR [248]. Pyrolysis is a process rapidly breaks biomass down to small fragments derived from the cell wall components in the condition of high temperature in the absence of oxygen [249-251]. Pyrolysis has been used as an analytical tool since the 1980s for characterizing plant cell wall composition [252, 253] by profiling the major pyrolysis derived cell wall fragments by mass spectrometry in a high -throughput manner. Although lignin macromolecules are relatively difficult to be depolymerized by traditional wet chemistry methods, pyrolysis -GC/MS can provide estimates of lignin properties such as S/G ratio and hydroxycinnamic acid content by quantifying these pyrolyzable lignin -derived fragments [146]. However, since the lignin -derived fragments maybe re -condensed during transferring to GC column due to the low temperature in the transfer line, the mass spectrometer only quantifies a small fraction of the lignin -derived fragments [146]. Pyrol ysis molecular beam mass spectrometry (py -MBMS) is another high -throughput method which profiles all of the mass peaks without GC separation, and PCA/PLS models are frequently applied to convert the unseparated peaks to identifiable cell wall properties. Researchers have done a lot of studies trying to find the relationship with the composition of lignin, such as the S/G ratio and the hydroxycinnamic content, and the digestibility of the lignocellulosic feedstock [254]. A well -developed method to quantify the S/G ratio of lignin is thioacidolysis, an acid Ðcatalyzed reaction to depolymerize lignin [255], which was first 78 utilized to estimate the S/G ratio by Lapierre et al. at 1980s [256, 257]. However, thioacidolysis yield is based on the con tent of b -O-4 linkage in lignin [146] because it proceeds mainly by cleavage of the ether linkages in lignin by boron trifluoride etherate in ethanethiol -dioxane [255]. Since b -O-4 is the most preferable linkage in syringyl -rich lignin, thioacidolysis yields lower amount of lignin monomers in grasses for its lower S lignin content than dicots [18, 258]. Besides the roughly estimation of the content of lignin and carbohydrates by the varie d types of spectroscopies, py - MBMS and py -GCMS allow to analyze more details of the cell wall lignin properties, such as S/G ratio and hydroxycinnamic acids content. Gerber et al. developed multivariate models to estimate the proportion of lignin as S/G r atio in the wood samples by py -GCMS [259]. Py -MBMS is frequently used to estimate the S/G ratio by summing the syringyl peaks intensity devided by guaiacyl peaks intensity for a large group of samples [260, 261]. However, this method usually overest imate S/G in grasses because the peak derived from ferulates cannot be distinguished from the peaks derived from G lignin, and grasses was shown to have higher ferulates content [21]. Many previous studies have found that p -coumarate (pCA) and ferulates (FA) can exert a strong influence on cell wall recalcitrance in terms of the ruminant digestibility [237] or the recalcitrance associated with enzymatic hydrolysis [236], and recently scientists are trying to increase the content of ferulate esters into model plants to increase the hydrolysis yields after alkaline pretreatments [242]. The content of esterified or etherified of pCA and FA can be quantitatively determined by alkaline extractions in different severity [167], and they were shown to impact differently on the cell wall degradability [130]. Maize ( Zea mays L.) stover has the largest production area in the United States as a bioenergy feedstock. The environmental and agronomic factors such as maturity, nutrients and genetics may impact the maize cell wall properties. Maize breeders for bioenergy production p roposes have focused on increasing the total biomass, increasing carbohydrate 79 content, or reducing recalcitrance [199]. High -throughput screening has been applied to select maize lines with higher enzymatic hydrolysis yields before or after hydrothermal pretreatments [222]. In this work, a diversity panel of 30 maize lines was utilized to develop chemometric models to predict cell wall properties based on pyrolysis -MBMS spectra through principal component analysis coupled with partial least square regression (PCA/PLS) model. A mild alk aline pretreatment, which removes recalcitrant components in monocot grasses in previous study [254], was utilized to prov ide the maize diversity panel with different responses under pretreatment. Py -MBMS was applied to develop the chemometric models to predict cell wall hydrolysis yields from the MS spectra through PCA/PLS models for the maize diversity panel before and after alkaline pretreatment. Also, multiple line ar regression (MLR) models were developed to predic t the hydrolysis yields for maize before and after alkaline pretreatment based on a series of measured cell wall properties including the water retention value (WRV), representing porosity and hydrophilicity, and the other compositional properties includin g lignin, xylan and p-hydroxycinnamic acids content. In addition, the S/G ratio predicted by py -MBMS and quantified by thioacidolysis have been compared and both utilized to incorporate to the prediction model for enzymatic hydrolysis. 5.2 MATERIALS AND ME THODS 5.2.1 The determination of the maize c ell wall comp osition properties Structural carbohydrates in maize cell walls were determined by the NREL two -stage composition analysis , and the acid insoluble solids after the two -stage acid composition analysis were regarded as Klason lignin content [139]. The contents o f ferulic acid ( FA) and p-coumaric acid ( pCA) were determined by a modified alkaline saponification method introduced in 4.2.3. However, the development and significance of this method will be 80 discussed in the following section. In general, Chapter 4 and Chapter 5 shared the same data set for quantifiable maize cell wall properties and hydrolysis yields . 5.2.2 Py-MBMS analysis and modeling The py -MBMS analy sis was performed using a pyrolysis furnace coupled to a free -jet molecular beam mass spectrometer without gas chromatographic column separation by Robert Sykes (NREL) according to previous procedure [125]. In py -MBMS spectra, peaks across the span of 51 to 450 m/z at interval of 1.0 m/z were subjected to a princip al component analysis [128]. For the predict ion model Y= !X, X was t he matrix of the fir st 10 principle components of the principal component analysis, and Y was one of the cell wall properties . The parameter s in the matrix ! of the regression model were calculated by MATLAB using partial least squar e regression method . 5.2.3 Multivariate m odel development 12 types of cell wall properties, including initial and final WRV (X1 and X2), initial and final xylan content (X3 and X4), initial and final lignin content (X5 and X6), initial acetate (X7), pCA (X8) and FA (X9) contents, solubilized pCA (X10) and FA (X11), and S/G ratio (X12), were used in this study as a set of maize cell wall properties. The correlation coefficients of those properties were calculated by MATLAB . 4 types of hydrolysis yields, including 6 -hr and 72 -hr glucose yields for untreated maize (Y1 and Y2) as well as 6-hr and 72 -hr glucose yields for pretreated maize (Y3 and Y4), were used as the response variables . To normalize the unit s and distributio n of measurement s, the Z -score tr ansformation was applied to manipulate the raw data by the formula: !"#!!"#$%!!"# !!"#$%!!"#$ !!"#$!%# !"#$%#&% !!"#$%&$'( 81 Partial least square regressions were done by MATLAB between the property data set and individual hydrolysis yields. Multiple variable se lection models were compared and t he top 3 models were selected by Cp and BIC criteria by R (free software environment for statistical computing and graphics), and they were compared by the correlation between the predicted values and measured values. Modi fied models for the hydrolysis yields were developed based on the above gathered information. 5.3 RESULTS AND DISCUSSI ONS 5.3.1 pCA/FA determination and prediction Monocot lignin s are rich in hydroxycinnam ates (HCAs) including ferulate (FA) and p-coumarate ( pCA) , which can be identifiable markers of the grass cell walls in the pyrolysis products [146]. Analytical methods to quantify these compounds in grasses typically involve the application alka line extraction to break ester and/or ether bonds [22, 130, 178, 179]. The relationship between digestibility, lignin content and ferulate content of the residue cell wall determined semi-quantitatively by py -GC/MS were correlated in previou s work , and it was found that the percentage of the pyrolysis pro ducts derived from ferulate was correlated with Klason lignin content, which had a strong negative correlation with the cell wall digestibility [146]. The HCA concentrations in the liquid phase a fter pretreatment also have been found to be increased by the increasing H 2O2 loading applied in pret reatment (data not published ). In order to quantify the HCAs in the residual cell walls, the severe alkaline extraction using 4M NaOH under 170 ¼C for 2 h ours extract both ester and ether linked FA and pCA [22, 130, 178]. Figure 5.3 shows the amount of these acids released by alkaline treatment at different severities for corn stover. It shows that the 3M NaOH condition yields the maximum amount of HCAs among other alkaline conditions, and the HCA yields at 170 ¼C is similar as the 82 HCA yields at 120 ¼C with the same NaOH loadings. This result suggest s that degradation or oxidation of HCAs possibly happens in high severity conditions . Figure 5. 1 pCA and FA solubilization in different severity of alkali treatments. The HCA content determined by the above mentioned method was found to be strongly correlated with the specific peaks 120 m/z and 150 m/z in the py -MBMS spectr a with R -squared of 0.47 and 0.33 of the linear correlation (Figure 5.4) [262]. The first ten principal components of the py -MBMS spectra of the maize diversity set were correlated to the quantified HCAs content to obtain a series of parameters !. And the dot product of ! and the principal components of the py -MBMS spectra were regarded as the predicted values of HCA content. Figure 5. 5 plots the predicted values vesus the measured values f or HCAs contents, indicating py -MBMS spectra is capable of making an accu rate prediction of HCAs content in the maize cell wall . 83 Figure 5 .2 Correlation between the mass peak intensity of 120 m/z (A) and 150 m/z (B) with pCA (A) and FA (B) content. Figure 5.3 Comparison between the measured values vs. predicted values based on PLS regression coupled with PCA analysis of py -MBMS spectra for the cell wall pCA content (A) and FA content (B) of 27 maize lines. A B A B 84 5.3.2 PCA/PLS model s for predicting hydrolysis yields PCA/PLS model s for predicting hydrolysis yields were also developed based on correlating the first ten principal components to the 6 -hr and 72 -hr hydrolysis yields o f untreated and pretreated maize. Figure 5.6 A -C shows that the PCA/PLS model provides an accurate prediction of the 6-hr hydrolysis yield s for untreated biomass (Figure 5.6A), relative weaker prediction for 72-hr hydrolysis yield s for untreated biomass (Figure 5. 6B), and the poorest prediction for 72-hr hydrolysis yield s after alkaline pretreatment (Figure 5. 6C). The improved enzymatic hydrolysis yields in plant cell walls after alkaline pretreatments has been explained by the increase s in the cell wall po rosity and hydrophilicity due to the removal o f lignin, hydroxycinnamic acids and hemicelluloses [59, 263]. The previous c hapter has shown that initial and final lignin content and HCAs content are negatively correlated with the initial and final hydrolysis yield, respectively, while the removal of these components is related with the increasing hydrolysis yield. Clearly, aft er alkaline pretreatment, the mass peaks of components including lignin and HCAs as Òmarkers Ó are eliminated due to removal of these compounds, which resulting in poor prediction for hydrolysis yield of pretreated maize based on the model develope d using t he py -MBMS spectra of the untreated maize . Figure 5 .4 Comparison between the measured values vs. predicted values based on PLS regression coupled with PCA analysis of py -MBMS spectra for the 6 -hr (A) and 72 -hr (B) C A B 85 hydrolysis yield for 27 maize lines before pretreatment and 72 -hr hydrolysis yield after NaOH pretreatment (C). 5.3.3 MLR models for predicting hydrolysis yields using quantified properties In order to further understand the relationship between the cell wall properties and the response to hydrolysis , multiple linear regression (MLR) models were developed in this study. To select the properties, multiple variable selection criteria have b een compared including Akaike information criterion (AIC) [264], R 2, adjusted R 2, MallowÕs C p [265], and Bayesian information criterion ( BIC) [266]. To prevent overfitting, MallowÕs Cp and BIC were chosen for model selection, which are more restrictive and limit additional variables to ones that impact the dependent variable [267]. The top three models based on these criteria were selected using the Leaps function in R, which performs an exhaustive search for the best subsets of the variables in Xs for predicting Ys in linear regression, and the top 3 models with the smallest C p and BIC values were sele cted for each response variab le ( Table 5.1). The results showed that Y 3 has a limited number of variables in all of the selected models. As stated in previous chapter , hornification [268] of the treated materials may result in difficulties in comparing the 6 -hour hydrolysis yield among diverse maize lines, therefore the model was not able to predic t 6 -hour hydrolysis yield of pretreated maize. Y1 Y2 Y3 Y4 Top 3 models based on C p and BIC criteria X3+X6+X7+X9+X11+X12 X5+X12 X1 X5+X6+X10+X11+X12 X3+X5+X6+X7+X9+X11+X12 X8 X4 X1+X5+X6+X10+X11+X12 X3+X5+X10 X3+X5+X12 X1+X5 X1+X5+X10+X11+X12 Modified model X1+X3+X5+X10 X5+X12 N/A X5+X10+X11+X12 Table 5. 1 Models based on C p and B IC criteria and the modified models. In order to normalize the vari ables with varied units and ranges , Z -score transformation s [269] were applied to transform the raw data to a new data set with similar range for the dif ferent 86 variables. The top three models were regressed in MATLAB and the relative parameter magnitudes of the variables were compared. For the prediction model of Y 1, X 6 and X 7 (definitions of the variables are in 5.2.3) were shown to have a small impact on the hydrolysis yield compared to X 3, X 5 and X 10. The variable X1, which was shown to have a p value less than 0.05 in the individual linear correlations with Y 1, has also been added to the modified model. Thus , the variable s selected for prediction model Y1 are X 1, X 3, X 5 and X 10, with parameter s for the Z -transformed model as: !!!!!!"!!!!!!"!!!!!!!!!!!!!"!!" This model indicates that, besides WRV (X 1) that positively impacts hydrolysis yield, xylan content (X 3), lignin content (X 5) and solubilized pCA (X 10) have similar magnitude of negative impacts on the initial rate of the hydrolysis of untreated maize. Previous results have shown that WRV, lignin and xylan content are negatively correlated to hydrolysis yields for alkaline hydrogen peroxide (AHP) pretreated grasses [146, 195]. This model also includes solubilized pCA content and the solubilized pCA was shown to correlate with lign in content in the Table 1. X5 and X 12 was shown to be able to fit Y 2 as using the least number of variables, with parameter s for the Z -transformed model as : !!!!!!!"!!!!!!"!!" X1 and X 6 were shown to have minimal impact on Y 4. Thus Y 4 would be fitted using X 5, X 10, X11, and X 12, would be: !!!!!!!"!!!!!!"!!"!!!!"!!!!!!!"!!" These two models demonstrat e the predictions of 72 -hour hydrolysis yield s for the pretreated (Y1) and untreated (Y 4) maize include both initial lignin content (X 5) and S/G ratio (X 12), with the only difference that the solubilized pCA (X 10) and FA (X 11) have been also utilized 87 in the prediction m odel of the pretreated maize. In the previous research demonstrating that alkaline pretreatments enhanc e digestibility of maize cell wall, it was shown that the enzymatic degradability is highly correlated with the solubilized amount of pCA and FA after alkaline pretreatment [170], which is consistent as the findings in this study. Although the multivariate model depends on the total number of variables, the relative parameter magnitude is not comparable between different treatments, and the model was only able to compare the contribution of different cell wall properties for one of the hydrolysis yields. Figure 5.5 compares the predictions of the three hydrolysis yie lds based on PCA/PLS model s and MLR model s, and shows that the prediction s based on MLR model s are with comparable accuracy while the predictions based on PCA/PLS models are with incomparable accuracy . Unlike the MLR models, which are developed based on th e three sets of maize properties as the variables, the PCA/PLS model s are all based on initial composition of the cell walls, which is profiled by py -MBMS spectra. However, the cell wall changes that occur during the pretreatment and enzymatic hydrolysis i mpact the hydrolysis yields , but are not predictable by the initial composition. A B 88 Figure 5 .5 The correlation between the predicted yields and the measured yield s for 6 -hour hydrolysis yield of untreated maize (red square), 72-hour hydrolysis yield of untreated maize (green triangle) and 72 -hour hydrolysis yield of NaOH pretreated maize (blue diamond) using PCA/PLS model (A) and MLR model (B). These models suggest that, the higher cell wall hydrophilicity represented by WRV, initial xylan and lignin content can set the initial cell wall recalcitrance and impact the cell wall degradability. However, only lignin has strong impacts on the hydrolysis extent and pretreatment effectiveness. Initial acetate and pCA content have a st ronger negative impact on the original cell wall hydrolysis extent (72 -hour yield) instead of the initial hydrolysis rate (6-hour yield), which have been eliminated in the model since both of them are strongly correlated with lignin content. Additionally, the solubility of pCA and FA quantified as solubilized pCA and FA in the liquid phase after pretreatment positively impacts the final hydrolysis yield. The correlation between S/G ratio and in vitro ruminant digestibility for monocot grasses has been shown in many of the previous research with contradictory findings [22, 130, 211]. We have found that S/G ratio did not change during the alkaline oxidative pretreatments since the thioacidolysi s method can only quantify the !-O-4 linked monolignols thus the thioacidolysis yields decreased significantly after alkaline oxidative pretreatment without changing the S/G ratio [146]. Characterization of lignin based on pyrolysis techniques have problems with that the ferulates and G lignin derived compounds are not distinguishable, which usually results in underestimation of S/G. In this study, we predicted S/G based on a PCA/ PLS model developed previously by correlating pyrolysis -MBMS spectra and thioacidolysis S/G results of another grass set, which avoids the overestimation of G lignin and estimates S/G in a high -throughp ut manner. Our results showed that S/G ratio is one of t he variables in the models to predict the hydrolysis yields for alkaline pretreated maize . 89 5.4 CONCLUSION In this study, chemometric models and MLR models were dev eloped and compared for predicting cell wall properties and hydrolysis yields of diverse maize and grasses. A high throughput method for quantifying the cell wall HCAs content is developed in this study. The PCA/PLS model gave a more accurate prediction of hydrolysis rates than hydrolysis yields, and a very poor prediction for the hydrolysis yields after alkaline pretreatment. T his indicate s that the initial cell wall composition profiled in py -MBMS impact s the hydrolysis rate s more than the yields, and the removal of the alkali -labile markers such as HCAs leads to less predictable hydrolysis yields after pretreatment by py -MBMS spectra . The MLR model of the prediction on hydro lysis yields also suggested similar findings. The re lative parame ter magnitudes of the MLR model showed that the hydrolysis rate was set by the composition properties and WRV, and the hydrolysis yields of the maize cell wall before pretreatment are high ly dependent on the lignin c ontent and S/G ratio. For t he hydrolysis yields after pretreatment, lignin content, S/G ratio and the releasing amounts of pCA and FA in pretreatment liquor are the most influential factors . These findings provide insights for maize breeding for bioenergy production based on stovers -suitable alkaline pretreatments. 90 6. CONCLUSIONS Untreated and alkaline oxidative pretreated biomass from phylogenetically diverse plants were compared in this study to understand fundamental features impacting cell wall recalcitrance. The roles that gl ycans play in the plant cell wall recalcitrance was investigated using taxonomically and structurally diverse bi omass feedstocks. Their response to alkaline oxidative pretreatment and how differing features of the cell wall matrix contribute to its recalci trance wa s assessed by glycome profiling, and we identified different mechanisms for how AHP pretreatment overcomes cell wall recalcitrance in goldenrod versus corn stover, while it was relatively ineffective on poplar. $Furthermore, grass cell wall features were characterized to understand how those features may impact recalcitrance reduced by alkaline oxidative pretreatment . Lignin removal in grasses induced by cleavage of hydroxycinnamic esters and internal metal-expedited peroxide oxidation in the alkaline oxidative pretreatment is the major contributor to impr ove the following hydrolysis yi elds. The results call attention to the important role that differences in cell wall structure and organization play in establishing cell wall recalcitrance to deconstruction by pretreatment and enzymatic hydrolysis. $Based on a maize diversity panel , a number of both expected as well as non -intuitive correlations between cell wall properties and recalcitrance were identified for untreated and alkaline pretreated maize. It was found that the properties contributing to a Òreduced recalcitranceÓ phenotype following a specific pretreatment are not necessarily the same properties that contribute to recalcitrance in untreated cell wall s. In order to compare and validat e the findings, multivariate models including PCA/PLS models and MLR models were applied to predict the hydrolysis yields based on cell wall properties whic h were predicted using chemome tric models or directly quantified . It was found that the initial composition sets the cell wall recalcitrance which impacts the initial hydrolysis rate, but the hydrolysis 91 extent before and after pretreatment is highly impacted by the content, composition and removal of lignin. Pretreatment conditions that are feedstock -speci fic are likely to be more effective than general approaches, and future work in breeding and engineering plants with cell walls designed for speci fic deconstruction approaches should make use of positive synergistic interactions bet ween spec ific pret reatments and particular cell wall features. 92 APPENDIX 93 Glycan Group Recognized mAb Names Non -Fucosylated Xyloglucan -1 CCRC-M95 CCRC-M101 Non -Fucosylated Xyloglucan -2 CCRC-M104 CCRC-M89 CCRC-M93 CCRC-M87 CCRC-M88 Non -Fucosylated Xyloglucan -3 CCRC-M100 CCRC-M103 Non -Fucosylated Xyloglucan -4 CCRC-M58 CCRC-M86 CCRC-M55 CCRC-M52 CCRC-M99 Non -Fucosylated Xyloglucan -5 CCRC-M54 CCRC-M48 CCRC-M49 CCRC-M96 CCRC-M50 CCRC-M51 CCRC-M53 Non -Fucosylated Xyloglucan -6 CCRC-M57 Fucosylated Xyloglucan CCRC-M102 CCRC-M39 CCRC-M106 CCRC-M84 CCRC-M1 Xylan -1/XG CCRC-M111 CCRC-M108 CCRC-M109 Table A. 1 Listing of plant cell wall glycan -directed monoclonal antibodies (mAbs) used for glyc ome profiling analy ses. The groupings of antibodies are based on a hierarchical clustering of ELISA data generated from a screen of all mAbs against a panel of plant polysaccharide preparations that groups the mAbs according to the predominant polysaccharides that they recog nize. The majority of listings link to the Wall MabDB plant cell wall monoclonal antibody database ( http://www.wallmabdb.net ) that provides detailed descriptions of each mAb, including immunogen, antibody isotype, epitope structure (to the extent known), supplier information, and related literature citations. 94 Table A.1 (contÕd) Xylan -2 CCRC-M119 CCRC-M115 CCRC-M110 CCRC-M105 Xylan -3 CCRC-M117 CCRC-M113 CCRC-M120 CCRC-M118 CCRC-M116 CCRC-M114 Xylan -4 CCRC-M154 CCRC-M150 Xylan -5 CCRC-M144 CCRC-M146 CCRC-M145 CCRC-M155 Xylan -6 CCRC-M153 CCRC-M151 CCRC-M148 CCRC-M140 CCRC-M139 CCRC-M138 Xylan -7 CCRC-M160 CCRC-M137 CCRC-M152 CCRC-M149 Galactomannan -1 CCRC-M75 CCRC-M70 CCRC-M74 Galactomannan -2 CCRC-M166 CCRC-M168 CCRC-M174 CCRC-M175 Acetylated Mannan CCRC-M169 CCRC-M170 "-Glucan LAMP BG1 HG Backbone -1 CCRC-M131 CCRC-M38 JIM5 HG Backbone -2 JIM136 JIM7 RG-I Backbone CCRC-M69 CCRC-M35 CCRC-M36 95 Table A.1 (contÕd) CCRC-M14 CCRC-M129 CCRC-M72 Linseed Mucilage RG-I JIM3 CCRC-M40 CCRC-M161 CCRC-M164 Physcomitrella Pectin CCRC-M98 CCRC-M94 RG-Ia CCRC-M5 CCRC-M2 RG-Ib JIM137 JIM101 CCRC-M61 CCRC-M30 RG-Ic CCRC-M23 CCRC-M17 CCRC-M19 CCRC-M18 CCRC-M56 CCRC-M16 RG-I/Arabinogalactan CCRC-M60 CCRC-M41 CCRC-M80 CCRC-M79 CCRC-M44 CCRC-M33 CCRC-M32 CCRC-M13 CCRC-M42 CCRC-M24 CCRC-M12 CCRC-M7 CCRC-M77 CCRC-M25 CCRC-M9 CCRC-M128 CCRC-M126 CCRC-M134 CCRC-M125 CCRC-M123 CCRC-M122 CCRC-M121 CCRC-M112 CCRC-M21 96 Table A.1 (contÕd) JIM131 CCRC-M22 JIM132 JIM1 CCRC-M15 CCRC-M8 JIM16 Arabinogalactan -1 JIM93 JIM94 JIM11 MAC204 JIM20 Arabinogalactan -2 JIM14 JIM19 JIM12 CCRC-M133 CCRC-M107 Arabinogalactan -3 JIM4 CCRC-M31 JIM17 CCRC-M26 JIM15 JIM8 CCRC-M85 CCRC-M81 MAC266 PN 1 6.4B4 Arabinogalactan -4 MAC207 JIM133 JIM13 CCRC-M92 CCRC-M91 CCRC-M78 Unidentified MAC265 CCRC-M97 97 Figure A. 1 Glycome profiling results of untreated and 12.5% AHP pretreated miscanthus and switchgrass. Switchgrass (Monocot) Miscanthus (Monocot) 98 Figure A. 2 Glycome profiling results of untreated and 12.5% AHP pretreated bm1 and bm3 stover. Brown Midribs (Monocot) Untreated 12.5% AHP Untreated 12.5% AHP 99 Genotype Accession Number or Source 52220 Ames 2336 A659 NSL 81599 B14A PI 550461 B47 PI 601009 B66 PI 550464 B7 NSL 65863 CH157 Ames 22441 F431 PI 257515 HUANYAO Ames 2337 INB 101LFY/LFY (A632 X M16 S5) NSL 174429 Ky226 Ames 27131 LH143 (Maintainer) Ames 27450 LP1NR HAT PI 600729 Mo5 PI 558523 N215 PI 595367 NC290A Ames 27141 NC368 Ames 27180 NK807 PI 601430 NO. 380 PI 303924 Oh7B Ames 19323 PHG86 PI 601442 W611S a W64A NSL 30058 W64A[7] bm4 S8 a W64A[8]bm2 S6 a Wf9 Ames 19293 YE 4 Ames 2335 Table A. 2 Genotype and reference source for the 27 maize lines used in this work. (The maize diversity panel was grown and harvested by Marlies Heckwolf , University of Wisconsin.) a Dr. Natalia de Leon - University of Wisconsin (ndeleongatti@wisc.edu ) 100 BIBLIOGRAPHY 101 BIBLIOGRAPHY 1. Hill, J., et al., Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. Proceedings of the National Academy of Sciences, 2006. 103(30): p. 11206 -11210. 2. Wyman, C.E., Ethanol from lignocellulosic biomass: Technology, economics, and opportunities. Bioresource Technology, 1994. 50(1): p. 3 -15. 3. Himmel, M.E ., et al., Biomass Recalcitrance: Engineering Plants and Enzymes for Biofuels Production. Science, 2007. 315(5813): p. 804 -807. 4. Wyman, C., et al., Comparative data on effects of leading pretreatments and enzyme loadings and formulations on sugar yields from different switchgrass sources. Biores Technol, 2011. 102: p. 11052 - 11062. 5. McNeil, M., et al., Structure and Function of the Primary Cell Walls of Plants. Annual Review of Biochemistry, 1984. 53(1): p. 625 -663. 6. Hayashi, T., M.P.F. Marsden, a nd D.P. Delmer, Pea Xyloglucan and Cellulose: V. Xyloglucan-Cellulose Interactions in vitro and In vivo. Plant Physiology, 1987. 83(2): p. 384-389. 7. Whitney, S.E.C., et al., Roles of cellulose and xyloglucan in determining the mechanical properties of primary plant cell walls. Plant Physiology, 1999. 121(2): p. 657-663. 8. Terashima, N., Formation of macromolecular lignin in ginkgo xylem cell walls as observed by field emission scanning electron microscopy. Comptes Rendus Biologies, 2004. 327(9-10): p. 903-910. 9. Ding, S. -Y., et al., How Does Plant Cell Wall Nanoscale Architecture Correlate with Enzymatic Digestibility? Science, 2012. 338(6110): p. 1055 -1060. 10. Atalla, R.H. and D.L. Vanderhart, NATIVE CELLULOSE - A COMPOSITE OF 2 DISTINCT CRYSTALLINE FORMS. Science, 1984. 223(4633): p. 283 -285. 11. Souza, A., et al., Composition and Structure of Sugarcane Cell Wall Polysaccharides: Implications for Second -Generation Bioethanol Production. BioEnergy Research, 2013. 6(2): p. 564 -579. 12. Adler, E., Lignin chemistryÑpast, present and future. Wood Science and Technology, 1977. 11(3): p. 169 -218. 13. Bonawitz, N. and C. Chapple, The Genetics of Lignin Biosynthesis: Connecting Genotype to Phenotype. Annu Rev Gen, 2010. 44: p. 337 - 363. 14. Ralph, J., Hydroxycinnamates in lignification. Phytochemistry Reviews, 2010. 9: p. 65 - 83. 102 15. Grabber, J.H., et al., Genetic and molecular basis of grass cell -wall degradability. I. Lignin -cell wall matrix interactions. Comptes Rendus Biologies, 2004. 327(5): p. 455-465. 16. Chen, L., et al., Lignin Deposition and Associated Changes in Anatomy, Enzyme Activity, Gene Expression, and Ruminal Degradability in Stems of Tall Fescue at Different Developmental Stages. J Agric Food Chem, 2002. 50: p. 5558 - 5565. 17. Terashima, N., J. Nakashima, and K. Takabe, Proposed Structure for Protolignin in Plant Cell Walls. Lignin and Lignan Biosynthesis, American Chemical Society, 1998. 697: p. 180 - 193. 18. Grabber, J.H., et al., Genetic and molecular basis of grass cell-wall degradability. I. Lignin Ðcell wall matrix interactions. Comptes Rendus Biologies, 2004. 327(5): p. 455-465. 19. Iiyama, K., T.B.T. Lam, and B.A. Stone, Phenolic acid bridges between polysaccharides and lignin in wheat internodes. Phytochemistry, 1990. 29(3): p. 733 -737. 20. Carpita, N., Structure and biogenesis of the cell walls of grasses. Annu Rev Plant Physiol Plant Molec Biol, 1996. 47: p. 445 - 476. 21. Hatfield, R., J. Ralph, and J. Grabber, Cell wall cross -linking by ferulates and diferu lates in grasses. J Sci Food Agric, 1999. 79: p. 403 - 407. 22. Mechin, V., et al., In Search of a Maize Ideotype for Cell Wall Enzymatic Degradability Using Histological and Biochemical Lignin Characterization. Journal of Agricultural and Food Chemistry, 2005. 53(15): p. 5872 -5881. 23. Higuchi, T., Y. Ito, and I. Kawamura, p-hydroxyphenylpropane component of grass lignin and role of tyrosine -ammonia lyase in its formation. Phytochemistry, 1967. 6(6): p. 875 -881. 24. Hartley, R., Degradation of cell walls of forages by sequential treatment with sodium hydroxide and a commercial cellulase preparation. J Sci Food Agric, 1983. 34: p. 29 - 36. 25. Lapierre, C., D. Jouin, and B. Monties, On the molecular origin of the alkali solubility of Gramineae lignin s. Phytochemistry, 1989. 28(5): p. 1401 -1403. 26. Jos‘-Estanyol, M. and P. Puigdom‘nech, Plant cell wall glycoproteins and their genes. Plant Physiology and Biochemistry, 2000. 38(1Ð2): p. 97 -108. 27. Voragen, A., et al., Pectin, a versatile polysaccharide present in plant cell walls. Structural Chemistry, 2009. 20(2): p. 263 -275. 28. Ebringerov⁄, A., Structural Diversity and Application Potential of Hemicelluloses. Macromolecular Symposia, 2005. 232(1): p. 1 -12. 103 29. Harris, P.J. and B.A. St one, Chemistry and Molecular Organization of Plant Cell Walls , in Biomass Recalcitrance . 2009, Blackwell Publishing Ltd. p. 61 -93. 30. Willats, W.G.T., et al., Pectin: cell biology and prospects for functional analysis. Plant Molecular Biology, 2001. 47(1): p. 9 -27. 31. Carpita, N.C., Structure and biogenesis of the cell walls of grasses. Annual Review of Plant Physiology and Plant Molecular Biology, 1996. 47: p. 445 -476. 32. McCann, M.C. and N.C. Carpita, Designing the deconstruction of plant cell walls . Current Opinion in Plant Biology, 2008. 11(3): p. 314 -320. 33. Chundawat, S.P.S., et al., Deconstruction of Lignocellulosic Biomass to Fuels and Chemicals. Annual Review of Chemical and Biomolecular Engineering, 2011. 2(1): p. 121-145. 34. McMillan, J. D., Pretreatment of Lignocellulosic Biomass, in Enzymatic Conversion of Biomass for Fuels Production , M.E. Himmel, J.O. Baker, and R.P. Overend, Editors. 1994. p. 292-324. 35. Foston, M. and A.J. Ragauskas, Changes in lignocellulosic supramolecular and ultrastructure during dilute acid pretreatment of Populus and switchgrass. Biomass and Bioenergy, 2010. 34(12): p. 1885 -1895. 36. Jung, S., et al., Surface Characterization of Dilute Acid Pretreated Populus deltoides by ToF -SIMS. Energy & Fuels, 2010. 24(2): p. 1347 -1357. 37. Grous, W.R., A.O. Converse, and H.E. Grethlein, EFFECT OF STEAM EXPLOSION PRETREATMENT ON PORE -SIZE AND ENZYMATIC -HYDROLYSIS OF POPLAR. Enzyme and Microbial Technology, 1986. 8(5): p. 274 -280. 38. Brownell, H.H., E.K.C. Yu, and J.N . Saddler, STEAM -EXPLOSION PRETREATMENT OF WOOD - EFFECT OF CHIP SIZE, ACID, MOISTURE -CONTENT AND PRESSURE -DROP. Biotechnology and Bioengineering, 1986. 28(6): p. 792-801. 39. Donohoe, B.S., et al., Visualizing lignin coalescence and migration through mai ze cell walls following thermochemical pretreatment. Biotechnology and Bioengineering, 2008. 101(5): p. 913 -925. 40. Kristensen, J.B., et al., Cell -wall structural changes in wheat straw pretreated for bioethanol production. Biotechnology for Biofuels, 20 08. 1. 41. Chundawat, S.P.S., B. Venkatesh, and B.E. Dale, Effect of particle size based separation of milled corn stover on AFEX pretreatment and enzymatic digestibility. Biotechnology and Bioengineering, 2007. 96(2): p. 219 -231. 104 42. Chundawat, S.P.S., et al., Multifaceted characterization of cell wall decomposition products formed during ammonia fiber expansion (AFEX) and dilute acid based pretreatments. Bioresource Technology, 2010. 101(21): p. 8429 -8438. 43. Chundawat, S.P.S., et al., Multi -scale vis ualization and characterization of lignocellulosic plant cell wall deconstruction during thermochemical pretreatment. Energy & Environmental Science, 2011. 4(3): p. 973 -984. 44. Li, C., et al., Influence of physico-chemical changes on enzymatic digestibility of ionic liquid and AFEX pretreated corn stover. Bioresource Technology, 2011. 102(13): p. 6928-6936. 45. Sun, Y. and J. Cheng, Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresource Technology, 2002. 83(1): p. 1 -11. 46. Sierra, R., C. Granda, and M. Holtzapple, Lime Pretreatment , in Biofuels , J.R. Mielenz, Editor. 2009, Humana Press. p. 115 -124. 47. Sannigrahi, P., S.J. Miller, and A.J. Ragauskas, Effects of organosolv pretreatment and enzymatic hydrolysis on cellulo se structure and crystallinity in Loblolly pine. Carbohydrate Research, 2010. 345(7): p. 965 -970. 48. Kumar, P., et al., Methods for Pretreatment of Lignocellulosic Biomass for Efficient Hydrolysis and Biofuel Production. Industrial & Engineering Chemistr y Research, 2009. 48(8): p. 3713 -3729. 49. Cheng, G., et al., Transition of Cellulose Crystalline Structure and Surface Morphology of Biomass as a Function of Ionic Liquid Pretreatment and Its Relation to Enzymatic Hydrolysis. Biomacromolecules, 2011. 12(4): p. 933 -941. 50. Wang, Z., T. Lan, and J. Zhu, Lignosulfonate and elevated pH can enhance enzymatic saccharification of lignocelluloses. Biotechnology for Biofuels, 2013. 6(1): p. 9. 51. Narayanaswamy, N., et al., Biological Pretreatment of Lignocellulosic Biomass for Enzymatic Saccharification , in Pretreatment Techniques for Biofuels and Biorefineries , Z. Fang, Editor. 2013, Springer Berlin Heidelberg. p. 3 -34. 52. Saritha, M., A. Arora, and Lata, Biological Pretreatment of Lignocellulosic Substrates for Enhanced Delignification and Enzymatic Digestibility. Indian Journal of Microbiology, 2012. 52(2): p. 122 -130. 53. Hatakka, A., Lignin -modifying enzymes from selected white -rot fungi: production and role from in lignin degradation. FEMS Mic robiology Reviews, 1994. 13(2Ð3): p. 125 -135. 54. Baldrian, P., Interactions of heavy metals with white-rot fungi. Enzyme and Microbial Technology, 2003. 32(1): p. 78 -91. 105 55. Li, Z., et al., Rapid and effective oxidative pretreatment of woody biomass at mild reaction conditions and low oxidant loadings. Biotechnology for Biofuels, 2013. 6(1): p. 119. 56. Li, Z., et al., Catalysis with CuII(bpy) improves alkaline hydrogen peroxide pretreatment. Biotechnology and Bioengineering, 2013. 110(4): p. 1078 -1086. 57. Banerjee, G., et al., Alkaline peroxide pretreatment of corn stover: effects of biomass, peroxide, and enzyme loading and composition on yields of glucose and xylose. Biotechnology for Biofuels, 2011. 4(1): p. 16. 58. Saha, B.C. and M.A. Cotta, Etha nol production from alkaline peroxide pretreated enzymatically saccharified wheat straw. Biotechnology Progress, 2006. 22(2): p. 449 -453. 59. Sun, R., J.M. Lawther, and W.B. Banks, Influence of alkaline pre -treatments on the cell wall components of wheat straw. Industrial Crops and Products, 1995. 4(2): p. 127-145. 60. Gould, J.M., Enhanced polysaccharide recovery from agricultural residues and perennial grasses treated with alkaline hydrogen peroxide. Biotechnology and Bioengineering, 1985. 27(6): p. 893 -896. 61. Yang, B., et al., Fast and efficient alkaline peroxide treatment to enhance the enzymatic digestibility of steam -exploded softwood substrates. Biotechnology and Bioengineering, 2002. 77(6): p. 678 -684. 62. Kumar, L., R. Chandra, and J. Saddler, Influence of steam pretreatment severity on post -treatments used to enhance the enzymatic hydrolysis of pretreated softwoods at low enzyme loadings. Biotechnology and Bioengineering, 2011. 108(10): p. 2300 -2311. 63. Bayer, E.A., et al., Cellulose, cellul ases and cellulosomes. Current Opinion in Structural Biology, 1998. 8(5): p. 548 -557. 64. Cantarel, B.L., et al., The Carbohydrate -Active EnZymes database (CAZy): an expert resource for Glycogenomics. Nucleic Acids Research, 2009. 37(suppl 1): p. D233 -D238. 65. Mosier, N., et al., Reaction Kinetics, Molecular Action, and Mechanisms of Cellulolytic Proteins , in Recent Progress in Bioconversion of Lignocellulosics . 1999, Springer Berlin / Heidelberg. p. 23 -40. 66. Zhang, Y. -H.P. and L.R. Lynd, Towa rd an aggregated understanding of enzymatic hydrolysis of cellulose: Noncomplexed cellulase systems. Biotechnology and Bioengineering, 2004. 88(7): p. 797 -824. 67. Felix, C.R. and L.G. Ljungdahl, The Cellulosome: The Exocellular Organelle of Clostridium. Annual Review of Microbiology, 1993. 47(1): p. 791 -819. 106 68. Hu, J., V. Arantes, and J. Saddler, The enhancement of enzymatic hydrolysis of lignocellulosic substrates by the addition of accessory enzymes such as xylanase: is it an additive or synergistic e ffect? Biotechnology for Biofuels, 2011. 4(1): p. 36. 69. Hemsworth, G.R., G.J. Davies, and P.H. Walton, Recent insights into copper -containing lytic polysaccharide mono -oxygenases. Current Opinion in Structural Biology, 2013. 23(5): p. 660 -668. 70. Bane rjee, G., et al., Synthetic multi -component enzyme mixtures for deconstruction of lignocellulosic biomass. Bioresource Technology, 2010. 101(23): p. 9097 -9105. 71. Ward, R.J., Cellulase Engineering for Biomass Saccharification Routes to Cellulosic Ethanol , M.S. Buckeridge and G.H. Goldman, Editors. 2011, Springer New York. p. 135-151. 72. Zeng, Y., et al., Lignin plays a negative role in the biochemical process for producing lignocellulosic biofuels. Current Opinion in Biotechnology, 2014. 27(0): p. 38-45. 73. Brunecky, R., et al., Redistribution of xylan in maize cell walls during dilute acid pretreatment. Biotechnology and Bioengineering, 2009. 102(6): p. 1537 -1543. 74. Studer, M.H., et al., Lignin content in natural Populus variants affects sugar release. Proceedings of the National Academy of Sciences, 2011. 108(15): p. 6300 -6305. 75. Yang, B. and C.E. Wyman, Effect of xylan and lignin removal by batch and flowthrough pretreatment on the enzymatic digestibility of corn stover cellulose. Biotechn ology and Bioengineering, 2004. 86(1): p. 88 -98. 76. Davison, B.H., et al., Variation of S/G ratio and lignin content in a Populus family influences the release of xylose by dilute acid hydrolysis. Applied Biochemistry and Biotechnology, 2006. 130(1-3): p . 427-435. 77. Li, X., et al., Lignin monomer composition affects Arabidopsis cell -wall degradability after liquid hot water pretreatment. Biotechnology for Biofuels, 2010. 3(1): p. 1 -7. 78. Chen, F. and R.A. Dixon, Lignin modification improves fermentable sugar yields for biofuel production. Nature Biotechnology, 2007. 25(7): p. 759 -761. 79. Jung, H.G. and M.S. Allen, Characteristics of plant cell walls affecting intake and digestibility of forages by ruminants. Journal of Animal Science, 1995. 73(9): p. 2774-90. 80. Somerville, C., et al., Toward a Systems Approach to Understanding Plant Cell Walls. Science, 2004. 306(5705): p. 2206 -2211. 81. Cosgrove, D.J., Growth of the plant cell wall. Nat Rev Mol Cell Biol, 2005. 6(11): p. 850-861. 107 82. Kıhnke, T., †. –stlund, and H. Brelid, Adsorption of Arabinoxylan on Cellulosic Surfaces: Influence of Degree of Substitution and Substitution Pattern on Adsorption Characteristics. Biomacromolecules, 2011. 12(7): p. 2633 -2641. 83. Linder, †., et al., Mechanism of Assembly of Xylan onto Cellulose Surfaces. Langmuir, 2003. 19(12): p. 5072 -5077. 84. Mosier, N., et al., Optimization of pH controlled liquid hot water pretreatment of corn stover. Bioresource Technology, 2005. 96(18): p. 1986 -1993. 85. Reis, D., B. Vian, and J. -C. Roland, Cellulose -glucuronoxylans and plant cell wallstructure. Micron, 1994. 25(2): p. 171 -187. 86. Fry, S.C. and J.G. Miller, Toward a Working Model of the Growing Plant Cell Wall , in Plant Cell Wall Polymers . 1989, American Chem ical Society. p. 33 -46. 87. Foston, M. and A. Ragauskas, Biomass Characterization: Recent Progress in Understanding Biomass Recalcitrance Ind Biotechnol, 2012. 8(4): p. 191 -208. 88. Jeoh, T., et al., Cellulase digestibility of pretreated biomass is limit ed by cellulose accessibility. Biotechnology and Bioengineering, 2007. 98(1): p. 112 -122. 89. Rollin, J.A., et al., Increasing cellulose accessibility is more important than removing lignin: A comparison of cellulose solvent -based lignocellulose fractiona tion and soaking in aqueous ammonia. Biotechnology and Bioengineering, 2011. 108(1): p. 22-30. 90. Sun, R., Cereal straw as a resource for sustainable biomaterials and biofuels chemistry, extractives, lignins, hemicelluloses and cellulose . 2010, Amsterdam ; Boston: Elsevier. 91. Jung, H.G. and M.D. Casler, Maize stem tissues: Impact of development on cell wall degradability. Crop Science, 2006. 46(4): p. 1801 -1809. 92. Sanju⁄n, R., et al., Morphological and Chemical Composition of Pith and Fibers from Mexican Sugarcane Bagasse. Holz als Roh - und Werkstoff, 2001. 59(6): p. 447 -450. 93. Wilson, J.R., D.R. Mertens, and R.D. Hatfield, Isolates of cell types from sorghum stems: Digestion, cell wall and anatomical characteristics. Journal of the Science of Food and Agriculture, 1993. 63(4): p. 407 -417. 94. Wilson, J.R. and R.D. Hatfield, Structural and chemical changes of cell wall types during stem development: consequences for fibre degradation by rumen microflora. Australian Journal of Agricultural Resear ch, 1997. 48(2): p. 165 -180. 95. Scalbert, A., et al., Ether linkage between phenolic acids and lignin fractions from wheat straw. Phytochemistry, 1985. 24(6): p. 1359 -1362. 96. Iiyama, K. and T. Lam, Structural characteristics of cell walls of forage grasses: their nutritional evaluation for ruminants. Asian-Aust J Anim Sci, 2001. 14: p. 862 - 879. 108 97. Chen, Y., et al., Understanding of alkaline pretreatment parameters for corn stover enzymatic saccharification. Biotechnology for Biofuels, 2013. 6(1): p . 8. 98. Liu, T., et al., Coupling alkaline pre -extraction with alkaline -oxidative post -treatment of corn stover to enhance enzymatic hydrolysis and fermentability. Biotechnology for Biofuels, 2014. 7(1): p. 48. 99. Klinke, H.B., et al., Characterization of degradation products from alkaline wet oxidation of wheat straw. Bioresource Technology, 2002. 82(1): p. 15 -26. 100. Westman, L. and T. Lindstroem, Swelling and mechanical properties of cellulose hydrogels. V. Swelling in concentrated alkaline solutions . 1983. Medium: X; Size: Pages: 363 -376. 101. Sluiter, J.B., et al., Compositional Analysis of Lignocellulosic Feedstocks. 1. Review and Description of Methods. Journal of Agricultural and Food Chemistry, 2010. 58(16): p. 9043 -9053. 102. Kiemle, D.J., A.J. Stipanovic, and K.E. Mayo, Proton NMR methods in the compositional characterization of polysaccharides , in Hemicelluloses: Science and Technology , P. Gatenholm and M. Tenhanen, Editors. 2004, Amer Chemical Soc: Washington. p. 122 -139. 103. Mittal, A., et al., Quantitative analysis of sugars in wood hydrolyzates with 1H NMR during the autohydrolysis of hardwoods. Bioresource Technology, 2009. 100(24): p. 6398-6406. 104. Hames, B.R., et al., Rapid biomass analysis - New tools for compositional a nalysis of corn stover feedstocks and process intermediates from ethanol production. Applied Biochemistry and Biotechnology, 2003. 105: p. 5 -16. 105. Liu, Q., et al., Pyrolysis of wood species based on the compositional analysis. Korean Journal of Chemical Engineering, 2009. 26(2): p. 548 -553. 106. Zhao, H., et al., Studying cellulose fiber structure by SEM, XRD, NMR and acid hydrolysis. Carbohydrate Polymers, 2007. 68(2): p. 235 -241. 107. Sannigrahi, P., A.J. Ragauskas, and S.J . Miller, Effects of Two -Stage Dilute Acid Pretreatment on the Structure and Composition of Lignin and Cellulose in Loblolly Pine. BioEnergy Research, 2008. 1(3-4): p. 205 -214. 108. Larsson, P.T., K. Wickholm, and T. Iversen, A CP/MAS13C NMR investigation of molecular ordering in celluloses. Carbohydrate Research, 1997. 302(1-2): p. 19 -25. 109. Larsson, P.T. and P. -O. Westlund, Line shapes in CP/MAS 13C NMR spectra of cellulose I. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2005. 62(1-3): p. 539 -546. 109 110. Kim, H., J. Ralph, and T. Akiyama, Solution -state 2D NMR of Ball -milled Plant Cell Wall Gels in DMSO -d 6. BioEnergy Research, 2008. 1(1): p. 56 -66. 111. Kim, H. and J. Ralph, Solution -state 2D NMR of ball -milled plant cell wall gels in DMSO -d6/pyridine -d5. Organic & Biomolecular Chemistry, 2010. 8(3): p. 576. 112. Fukushima, R.S. and R.D. Hatfield, Comparison of the Acetyl Bromide Spectrophotometric Method with Other Analytical Lignin Methods for Determining Lignin Concentration in Forage Samples. Journal of Agricultural and Food Chemistry, 2004. 52(12): p. 3713 -3720. 113. Grabber, J.H., S. Quideau, and J. Ralph, p-coumaroylated syringyl units in maize lignin: Implications for [beta] -ether cleavage by thioacidolysis. Phytochemistry, 1996. 43(6): p. 1189 -1194. 114. Alves, A., et al., Analytical pyrolysis as a direct method to determine the lignin content in wood. Journal of Analytical and Applied Pyrolysis, 2006. 76(1-2): p. 209 -213. 115. Alves, A., et al., Analytical pyrolysis as a direct method to determine the lignin content in woodPart 2: Evaluation of the common model and the influence of compression wood. Journal of Analytical and Applied Pyrolysis, 2007. 116. Alves, A., et al., Analytical pyrolysis as a direct method to d etermine the lignin content in woodPart 3. Evaluation of species -specific and tissue -specific differences in softwood lignin composition using principal component analysis. Journal of Analytical and Applied Pyrolysis, 2009. 85(1-2): p. 30 -37. 117. Samuel, R., et al., Structural Characterization and Comparison of Switchgrass Ball -milled Lignin Before and After Dilute Acid Pretreatment. Applied Biochemistry and Biotechnology, 2009. 162(1): p. 62 -74. 118. J Paul, K., Revealing the structural and functional d iversity of plant cell walls. Current Opinion in Plant Biology, 2008. 11(3): p. 308 -313. 119. Moller, I., et al., High -throughput mapping of cell -wall polymers within and between plants using novel microarrays. The Plant Journal, 2007. 50(6): p. 1118-1128. 120. Alonso -SimŠn, A., et al., High -throughput microarray profiling of cell wall polymers during hydrothermal pre -treatment of wheat straw. Biotechnology and Bioengineering, 2010. 105(3): p. 509 -514. 121. Pattathil, S., et al., A Comprehensive Toolkit of Plant Cell Wall Glycan -Directed Monoclonal Antibodies. Plant Physiology, 2010. 153(2): p. 514 -525. 122. DeMartini, J.D., et al., Application of monoclonal antibodies to investigate plant cell wall deconstruction for biofuels production. Energy & Environmental Science, 2011. 4(10): p. 4332 -4339. 110 123. Hong, J., X. Ye, and Y.H.P. Zhang, Quantitative Determination of Cellulose Accessibility to Cellulase Based on Adsorption of a Nonhydrolytic Fusion Protein Containing CBM and GFP with Its Applicati ons. Langmuir, 2007. 23(25): p. 12535 -12540. 124. Kawakubo, T., et al., Analysis of exposed cellulose surfaces in pretreated wood biomass using carbohydrate -binding module (CBM) Ðcyan fluorescent protein (CFP). Biotechnology and Bioengineering, 2010. 105(3): p. 499 -508. 125. Mann, D.G.J., et al., Rapid Assessment of Lignin Content and Structure in Switchgrass (Panicum virgatum L.) Grown Under Different Environmental Conditions. BioEnergy Research, 2009. 2(4): p. 246 -256. 126. Smith-Moritz, A., et al., Combining multivariate analysis and monosaccharide composition modeling to identify plant cell wall variations by Fourier Transform Near Infrared spectroscopy. Plant Methods, 2011. 7(1): p. 26. 127. Hiukka, R., A multivariate approach to the analysis of pine needle samples using NIR. Chemometrics and Intelligent Laboratory Systems, 1998. 44(1Ð2): p. 395 -401. 128. Kelley, S.S., J. Jellison, and B. Goodell, Use of NIR and pyrolysis -MBMS coupled with multivariate analysis for detecting the chemical changes asso ciated with brown -rot biodegradation of spruce wood. FEMS Microbiol Lett, 2002. 209(1): p. 107 -11. 129. Zhu, L., et al., Structural features affecting biomass enzymatic digestibility. Bioresource Technology, 2008. 99(9): p. 3817 -3828. 130. Zhang, Y., et al., Impact of Lignin Structure and Cell Wall Reticulation on Maize Cell Wall Degradability. Journal of Agricultural and Food Chemistry, 2011. 59(18): p. 10129-10135. 131. P. O'Dwyer, J., et al., Neural Network Prediction of Biomass Digestib ility Based on Structural Features. Biotechnology Progress, 2008. 24(2): p. 283 -292. 132. Bansal, P., et al., Modeling cellulase kinetics on lignocellulosic substrates. Biotechnology Advances, 2009. 27(6): p. 833 -848. 133. Andri % , P., et al., Reactor design for minimizing product inhibition during enzymatic lignocellulose hydrolysis: I. Significance and mechanism of cellobiose and glucose inhibition on cellulolytic enzymes. Biotechnology Advances, 2010. 28(3): p. 308 -324. 134. Willats, W.G.T., et al., Making and using antibody probes to study plant cell walls. Plant Physiology and Biochemistry, 2000. 38(1Ð2): p. 27 -36. 135. Knox, J.P., The use of Antibodies to Study the Architecture and Developmental Regulation of Plant Cell Walls , in International Re view of Cytology , W.J. Kwang, Editor. 1997, Academic Press. p. 79 -120. 111 136. Banerjee, G., et al., Scale -up and integration of alkaline hydrogen peroxide pretreatment, enzymatic hydrolysis, and ethanolic fermentation. Biotechnology and Bioengineering, 2011 : p. n/a -n/a. 137. Pattathil, S., et al., Immunological Approaches to Plant Cell Wall and Biomass Characterization: Glycome Profiling # , in T Biomass Conversion . 2012. p. 61-72. 138. DeMarini, J., et al., Investigating plant cell wall components that aff ect biomass recalcitrance in poplar and switchgrass. Energy & Environmental Science, 2013. 139. Sluiter, A., et al., Determination of Structural Carbohydrates and Lignin in Biomass. NREL Analytical Procedure, 2004. National Renewable Energy Laboratory, Golden, CO. 140. Pattathil, S., et al., Immunological Approaches to Plant Cell Wall and Biomass Characterization: Glycome Profiling , in Biomass Conversion , M.E. Himmel, Editor. 2012, Humana Press. p. 61 -72. 141. Energy., U.S.D.o., U.S. Billion -Ton Update : Biomass Supply for a Bioenergy adn Bioproducts Industry. , in ORNL/TM -2011/224. Oak Ridge National Laboratory, Oak Ridge, TN. , R.D. Perlack and B.J. Stokes, Editors. 2011. p. 227. 142. Dale, V.H., et al., Interactions among bioenergy feedstock choices, l andscape dynamics, and land use. Ecological Applications, 2011. 21(4): p. 1039 -1054. 143. Tilman, D., J. Hill, and C. Lehman, Carbon -Negative Biofuels from Low -Input High -Diversity Grassland Biomass. Science, 2006. 314(5805): p. 1598 -1600. 144. Gelfand, I., et al., Sustainable bioenergy production from marginal lands in the US Midwest. Nature, 2013. 493(7433): p. 514 -517. 145. Banerjee, G., et al., Scale -up and integration of alkaline hydrogen peroxide pretreatment, enzymatic hydrolysis, and ethanolic fe rmentation. Biotechnology and Bioengineering, 2012. 109(4): p. 922 -931. 146. Li, M., et al., Structural characterization of alkaline hydrogen peroxide pretreated grasses exhibiting diverse lignin phenotypes. Biotechnology for Biofuels, 2012. 5(1): p. 38. 147. Stoklosa, R.J. and D.B. Hodge, Extraction, Recovery, and Characterization of Hardwood and Grass Hemicelluloses for Integration into Biorefining Processes. Industrial & Engineering Chemistry Research, 2012. 51(34): p. 11045 -11053. 148. Ralph, J., et al., Genetic and molecular basis of grass cell -wall biosynthesis and degradability. III. Towards a forage grass ideotype. Comptes Rendus Biologies, 2004. 327(5): p. 467 -479. 112 149. Pattathil, S., et al., Immunological Approaches to Plant Cell Wall and Biomass Characterization: Glycome Profiling Biomass Conversion , in Biomass Conversion: Methods and Protocols , M.E. Himmel, Editor. 2012, Humana Press. p. 61 -72. 150. Pattathil, S., et al., Changes in Cell Wall Carbohydrate Extractability Are Corr elated with Reduced Recalcitrance of HCT Downregulated Alfalfa Biomass Ind Biotechnol, 2012. 8(4): p. 217 -221. 151. Jarvis, M.C., W. Forsyth, and H.J. Duncan, A Survey of the Pectic Content of Nonlignified Monocot Cell Walls. Plant Physiology, 1988. 88(2): p. 309 -314. 152. Nakagame, S., et al., Enhancing the enzymatic hydrolysis of lignocellulosic biomass by increasing the carboxylic acid content of the associated lignin. Biotechnology and Bioengineering, 2011. 108(3): p. 538 -548. 153. Grous, W.R., A.O. Converse, and H.E. Grethlein, Effect of steam explosion pretreatment on pore size and enzymatic hydrolysis of poplar. Enzyme and Microbial Technology, 1986. 8(5): p. 274 -280. 154. Thompson, D.N., H. -C. Chen, and H.E. Grethlein, Comparison of pretreatment methods on the basis of available surface area. Bioresource Technology, 1992. 39(2): p. 155-163. 155. McMillan, J.D., Pretreatment of Lignocellulosic Biomass, in Enzymatic Conversion of Biomass for Fuels Production . 1994, American Chemical Society. p. 292 -324. 156. Jung, H. -J.G., D.A. Samac, and G. Sarath, Modifying crops to increase cell wall digestibility. Plant Science, 2012. 185Ð186(0): p. 65 -77. 157. Bernard Vailh”, M.A., et al., Effect of Phenolic Structures on the Degradability of Cell Walls Isolated from Newly Extended Apical Internode of Tall Fescue (Festuca arundinacea Schreb.). Journal of Agricultural and Food Chemistry, 2000. 48(3): p. 618-623. 158. Mohagheghi, A., et al., Cofermentation of glucose, xylose, and arabinose by genomic DNA -integrated xylose/arabinose fermenting strain of Zymomonas mobilis AX101. Appl Biochem Biotechnol, 2002. 98-100(1): p. 885 -898. 159. Gould, J.M., ALKALINE PEROXIDE DELIGNIFICATION OF AGRICULTURAL RESIDUES TO ENHANCE ENZYMATIC SACCHARIFICATION. Biotechnolog y and Bioengineering, 1984. 26(1): p. 46 -52. 160. Gould, J.M., Studies on the mechanism of alkaline peroxide delignification of agricultural residues. Biotechnology and Bioengineering, 1985. 27(3): p. 225 -231. 161. Gupta, R. and Y.Y. Lee, Pretreatment of corn stover and hybrid poplar by sodium hydroxide and hydrogen peroxide. Biotechnology Progress, 2010. 26(4): p. 1180 -1186. 113 162. Li, M., et al., Identification of features associated with plant cell wall recalcitrance to pretreatment by alkaline hydrogen peroxide in diverse bioenergy feedstocks using glycome profiling. RSC Advances, 2014. 4(33): p. 17282 -17292. 163. Wyman, C.E., et al., Comparative Performance of Leading Pretreatment Technologies for Biological Conversion of Corn Stover, Poplar Wood, and Switchgrass to Sugars , in Aqueous Pretreatment of Plant Biomass for Biological and Chemical Conversion to Fuels and Chemicals. 2013, John Wiley & Sons, Ltd. p. 239 -259. 164. Grabber, J.H., et al., Coniferyl Ferulate Incorporation into Lignin Enhances the Alkaline Delignification and Enzymatic Degradation of Cell Walls. Biomacromolecules, 2008. 9(9): p. 2510 -2516. 165. Qing, Q. and C.E. Wyman, Hydrolysis of different chain length xylooliogmers by cellulase and hemicellulase. Bioresource Technology, 2011. 102(2): p. 1359 -1366. 166. Lam, T.B.T., K. Kadoya, and K. Iiyama, Bonding of hydroxycinnamic acids to lignin: ferulic and p -coumaric acids are predominantly linked at the benzyl position of lignin, not the !-position, in grass cell walls. Phytochemistry, 2001. 57(6): p. 987 -992. 167. Sun, R. -C., X. -F. Sun, and S. -H. Zhang, Quantitative Determination of Hydroxycinnamic Acids in Wheat, Rice, Rye, and Barley Straws, Maize Stems, Oil Palm Frond Fiber, and Fast-Growing Poplar Wood. Journal of Agricultural and Food Chemistry, 2001. 49(11): p. 5122 -5129. 168. Smith, P.K. and T.J. McDonough, TRANSITION -METAL ION CATALYSIS OF THE REACTION OF A RESIDUAL LIGNIN -RELATED COMPOUND WITH ALKALINE HYDROGEN-PEROXIDE. Svensk Papperstidning -Nordisk Cellulosa , 1985. 88(12): p. R106-R112. 169. Barri‘re, Y. and O. Argillier, Brown -midrib genes of maize: a review. Agronomie, 1993. 13(10): p. 865 -876. 170. Hartley, R. and E. Jones, Phenolic components and degradability of the cell walls of the brown midrib mutan t, bm3, of Zea mays. J Sci Food Agric, 1978. 29: p. 777 - 782. 171. Chabbert, B., et al., Biological variability in lignification of maize: Expression of the brown midrib bm3 mutation in three maize cultivars. J Sci Food Agric, 1994. 64: p. 349 - 355. 172. Lam, T., K. Iiyama, and B. Stone, Lignin and Hydroxycinnamic Acids in Walls of Brown Midrib Mutants of Sorghum, Pearl Millet and Maize Stems. J Sci Food Agric, 1996. 71: p. 174 - 178. 173. Marita, J.M., et al., Variations in the Cell Wall Composition of Maize brown midrib Mutants. Journal of Agricultural and Food Chemistry, 2003. 51(5): p. 1313 -1321. 174. Barriere, Y., et al., Genetic variation for organic matter and cell wall digestibility in silage maize. Lessons from a 34 -year long experiment with sh eep in digestibility crates. MYDCAH 2004. 49: p. 115 -126. 114 175. Vermerris, W., D.M. Sherman, and L.M. McIntyre, Phenotypic plasticity in cell walls of maize brown midrib mutants is limited by lignin composition. Journal of Experimental Botany, 2010. 176. Lam, T.B.T., K. Iiyama, and B.A. Stone, Changes in phenolic acids from internode walls of wheat and phalaris during maturation. Phytochemistry, 1992. 31(8): p. 2655 -2658. 177. Tuyet Lam, T.B., K. Iiyama, and B.A. Stone, An approach to the estimation of ferulic acid bridges in unfractionated cell walls of wheat internodes. Phytochemistry, 1994. 37(2): p. 327 -333. 178. Pan, G.X., J.L. Bolton, and G.J. Leary, Determination of Ferulic and p -Coumaric Acids in Wheat Straw and the Amounts Released by Mild Acid and Alkaline Peroxide Treatment. Journal of Agricultural and Food Chemistry, 1998. 46(12): p. 5283 -5288. 179. Anderson, W.F. and D.E. Akin, Structural and chemical properties of grass lignocelluloses related to conversion for biofuels. J Ind Microbiol Bi otechnol, 2008. 35(5): p. 355 -66. 180. Ebringerov⁄, A. and T. Heinze, Xylan and xylan derivatives Ð biopolymers with valuable properties, 1. Naturally occurring xylans structures, isolation procedures and properties. Macromolecular Rapid Communications, 2 000. 21(9): p. 542 -556. 181. Carpita, N.C., STRUCTURE AND BIOGENESIS OF THE CELL WALLS OF GRASSES. Annual Review of Plant Physiology and Plant Molecular Biology, 1996. 47(1): p. 445 -476. 182. Silva, G.B., et al., Cell wall polysaccharides from fern leave s: Evidence for a mannan-rich Type III cell wall in Adiantum raddianum. Phytochemistry, 2011. 72(18): p. 2352-2360. 183. Kobayashi, T. and Y. Sakai, Hydrolysis Rate of Pentosan of Hardwood in Dilute Sulfuric Acid. Bull. Agr. Chem. Soc. Japan, 1955. 20(1): p. 1-7. 184. Schell, D., et al., Dilute -sulfuric acid pretreatment of corn stover in pilot -scale reactor. Applied Biochemistry and Biotechnology, 2003. 105(1): p. 69 -85. 185. Chen, R., Y. Lee, and R. Torget, Kinetic and modeling investigation on two -stage reverse-flow reactor as applied to dilute -acid pretreatment of agricultural residues. Applied Biochemistry and Biotechnology, 1996. 57-58(1): p. 133 -146. 186. Estaghlalian, A., et al., Modeling and Optimization of the Dilute -Sulfuric -Acid Pretreatment of Corn Stover, Poplar, and Switchgrass. Bioresource Technology, 1997. 59: p. 129 -136. 187. Shoseyov, O., Z. Shani, and I. Levy, Carbohydrate Binding Modules: Biochemical Properties and Novel Applications. Microbiology and Molecular Biology Reviews, 2006. 70(2): p. 283 -295. 115 188. Gao, D., et al., Binding characteristics of Trichoderma reesei cellulases on untreated, ammonia fiber expansion (AFEX), and dilute -acid pretreated lignocellulosic biomass. Biotechnology and Bioengineering, 2011. 108(8): p. 1788 -1800. 189. Abbott, D.W. and A.B. Boraston, Quantitative approaches to the analysis of carbohydrate -binding module function. Methods Enzymol, 2012. 510: p. 211 -31. 190. Lehtiı, J., et al., The binding specificity and affinity determinants of family 1 and family 3 cellulose binding modules. Proceedings of the National Academy of Sciences, 2003. 100(2): p. 484 -489. 191. Jung, H., D.B. Wilson, and L.P. Walker, Binding and reversibility of Thermobifida fusca Cel5A, Cel6B, and Cel48A and their respective catal ytic domains to bacterial microcrystalline cellulose. Biotechnology and Bioengineering, 2003. 84(2): p. 151 -159. 192. McLean, B.W., et al., Analysis of binding of the family 2a carbohydrate -binding module from Cellulomonas fimi xylanase 10A to cellulose: specificity and identification of functionally important amino acid residues. Protein Eng, 2000. 13(11): p. 801 -9. 193. Boraston, A.B., et al., Co-operative binding of triplicate carbohydrate -binding modules from a thermophilic xylanase. Mol Microbiol, 20 02. 43(1): p. 187 -94. 194. Piccolo, C., et al., Enzyme adsorption on SO2 catalyzed steam -pretreated wheat and spruce material. Enzyme and Microbial Technology, 2010. 46(3Ð4): p. 159 -169. 195. Williams, D. and D. Hodge, Impacts of delignification and hot water pretreatment on the water induced cell wall swelling behavior of grasses and its relation to cellulolytic enzyme hydrolysis and binding. Cellulose, 2014. 21(1): p. 221 -235. 196. Hansen, M., et al., Enzyme affinity to cell types in wheat straw (Triticum aestivum L.) before and after hydrothermal pretreatment. Biotechnology for Biofuels, 2013. 6(1): p. 54. 197. Tu, M., X. Pan, and J.N. Saddler, Adsorption of Cellulase on Cellulolytic Enzyme Lignin from Lodgepole Pine. Journal of Agricultural and Food Chemistry, 2009. 57(17): p. 7771 -7778. 198. Rahikainen, J., et al., Inhibition of enzymatic hydrolysis by residual lignins from softwood Ñstudy of enzyme binding and inactivation on lignin -rich surface. Biotechnology and Bioengineering, 2011. 108(12): p. 2823 -2834. 199. de Leon, N., S.M. Kaeppler, and J.G. Lauer, Breeding Maize for Lignocellulosic Biofuel Production , in Bioenergy Feedstocks . 2013, John Wiley & Sons, Inc. p. 151 -171. 200. Vermerris, W., et al., Molecular Breeding to Enhance Ethanol P roduction from Corn and Sorghum Stover. Crop Science, 2007. 47(Supplement_3): p. S -142-S-153. 116 201. Barri‘re, Y., et al., Genetic variation and breeding strategies for improved cell wall digestibility in annual forage crops. A review. Anim. Res., 2003. 52(3): p. 193 -228. 202. Hisano, H., R. Nandakumar, and Z. -Y. Wang, Genetic modification of lignin biosynthesis for improved biofuel production. In Vitro Cellular & Developmental Biology - Plant, 2009. 45(3): p. 306 -313. 203. He, X., et al., Improvement of Forage Quality by Downregulation of Maize O -Methyltransferase. Crop Science, 2003. 43(6): p. 2240 -2251. 204. Piquemal, J., et al., Down -Regulation of Caffeic Acid O -Methyltransferase in Maize Revisited Using a Transgenic Approach. Plant Physiology, 2002. 130(4): p. 1675 -1685. 205. Chuck, G., A high -carb diet for biofuels? Biofuels, 2013. 4(3): p. 243 -245. 206. Chuck, G.S., et al., Overexpression of the maize Corngrass1 microRNA prevents flowering, improves digestibility, and increases starch content of switchgrass. Proceedings of the National Academy of Sciences, 2011. 108(42): p. 17550 -17555. 207. Hansey, C., A. Lorenz, and N. Leon, Cell Wall Composition and Ruminant Digestibility of Various Maize Tissues Across Development. BioEnergy Research, 2010. 3(1): p. 28 -37. 208. Salvi, S., et al., Conserved noncoding genomic sequences associated with a flowering-time quantitative trait locus in maize. Proceedings of the National Academy of Sciences, 2007. 104(27): p. 11376 -11381. 209. Garlock, R., et al., Opt imizing harvest of corn stover fractions based on overall sugar yields following ammonia fiber expansion pretreatment and enzymatic hydrolysis. Biotechnology for Biofuels, 2009. 2(1): p. 29. 210. Lorenz, A.J., et al., Genetic Analysis of Cell Wall Traits Relevant to Cellulosic Ethanol Production in Maize (Zea mays L.). Crop Science, 2010. 50(3): p. 842 -852. 211. M”chin, V., et al., Relationship of cell wall composition to in vitro cell wall digestibility of maize inbred line stems. Journal of the Science of Food and Agriculture, 2000. 80(5): p. 574 -580. 212. Jung, H. -J.G. and D.R. Buxtono, Forage quality variation among maize inbreds: Relationships of cell -wall composition and in -vitro degradability for stem internodes. Journal of the Science of Food and Agriculture, 1994. 66(3): p. 313 -322. 213. Lundvall, J.P., et al., Forage Quality Variation among Maize Inbreds: In Vitro Digestibility and Cell -Wall Components. Crop Science, 1994. 34(6): p. 1672 -1678. 214. Jung, H.G., D.R. Mertens, and A.J. Payne, Corr elation of Acid Detergent Lignin and Klason Lignin with Digestibility of Forage Dry Matter and Neutral Detergent Fiber. Journal of Dairy Science, 1997. 80(8): p. 1622 -1628. 117 215. Fontaine, A. -S., et al., Variation in Cell Wall Composition among Forage Maize (Zea mays L.) Inbred Lines and Its Impact on Digestibility: " Analysis of Neutral Detergent Fiber Composition by Pyrolysis -Gas Chromatography -Mass Spectrometry. Journal of Agricultural and Food Chemistry, 2003. 51(27): p. 8080 -8087. 216. Jung, H. -J.G ., et al., Impact of accessibility and chemical composition on cell wall polysaccharide degradability of maize and lucerne stems. Journal of the Science of Food and Agriculture, 2000. 80(3): p. 419 -427. 217. Argillier, O., V. M”chin, and Y. Barri e ` re, Inbred Line Evaluation and Breeding for Digestibility -Related Traits in Forage Maize. Crop Science, 2000. 40(6): p. 1596 -1600. 218. Oakey, H., et al., Identification of crop cultivars with consistently high lignocellulosic sugar release requires the use o f appropriate statistical design and modelling. Biotechnology for Biofuels, 2013. 6(1): p. 185. 219. Li, M., et al., Identification of Features Associated with Plant Cell Wall Recalcitrance to Pretreatment by Alkaline Hydrogen Peroxide in Diverse Bioenerg y Feedstocks Using Glycome Profiling. RSC Advances, 2014. 220. Decker, S., et al., High -Throughput Screening Techniques for Biomass Conversion. BioEnergy Research, 2009. 2(4): p. 179 -192. 221. Santoro, N., et al., A High -Throughput Platform for Screening Milligram Quantities of Plant Biomass for Lignocellulose Digestibility. BioEnergy Research, 2010. 3(1): p. 93-102. 222. Lindedam, J., et al., Evaluation of high throughput screening methods in picking up differences between cultivars of lignocellulosic b iomass for ethanol production. Biomass and Bioenergy, 2014. 66(0): p. 261 -267. 223. Zhang, T., et al., Rapid selection and identification of Miscanthus genotypes with enhanced glucan and xylan yields from hydrothermal pretreatment followed by enzymatic hy drolysis. Biotechnology for Biofuels, 2012. 5(1): p. 56. 224. Vandenbrink, J.P., et al., A sorghum diversity panel biofuel feedstock screen for genotypes with high hydrolysis yield potential. Industrial Crops and Products, 2010. 31(3): p. 444 -448. 225. Morrison, T.A., et al., Cell -Wall Composition of Maize Internodes of Varying Maturity. Crop Sci., 1998. 38(2): p. 455 -460. 226. Gille, S. and M. Pauly, O-acetylation of Plant Cell Wall Polysaccharides. Frontiers in Plant Science, 2012. 3. 227. Pawar, P.M. -A., et al., Acetylation of woody lignocellulose: significance and regulation. Frontiers in Plant Science, 2013. 4. 118 228. Selig, M., et al., The impact of cell wall acetylation on corn stover hydrolysis by cellulolytic and xylanolytic enzymes. Cellulose, 2009. 16(4): p. 711 -722. 229. Xiong, G., K. Cheng, and M. Pauly, Xylan O-Acetylation Impacts Xylem Development and Enzymatic Recalcitrance as Indicated by the Arabidopsis Mutant tbl29. Molecular Plant, 2013. 230. Kim, S. and M. Holtzapple, Effect of structural features on enzyme digestibility of corn stover. Bioresource Technology, 2006. 97(4): p. 583 -591. 231. OÕNeill, M.A. and W.S. York, The composition and structure of plant primary cell walls. The plant cell wall, 2003: p. 1 -54. 232. Palonen, H., et al., Adsorption of Trichoderma reesei CBH I and EG II and their catalytic domains on steam pretreated softwood and isolated lignin. Journal of Biotechnology, 2004. 107(1): p. 65 -72. 233. Jung, H. and K. Vogel, Influence of Lignin on Digest ibility of Forage Cell Wall Material. J Anim Sci, 1986. 62: p. 1703 - 1712. 234. Sewalt, V.J.H., et al., Lignin impact on fiber degradation. IV. Enzymatic saccharification and in vitro digestibility of alfalfa and grasses following selective solvent delig nification. Bioresource Technology, 1997. 61(3): p. 199 -206. 235. Simmons, B.A., D. Loqu”, and J. Ralph, Advances in modifying lignin for enhanced biofuel production. Current Opinion in Plant Biology, 2010. 13(3): p. 312 -319. 236. Grabber, J.H., R.D. Hatfield, and J. Ralph, Diferulate cross -links impede the enzymatic degradation of non -lignified maize walls. Journal of the Science of Food and Agriculture, 1998. 77(2): p. 193 -200. 237. Jung, H.G., D.R. Mertens, and R.L. Phillips, Effect of reduced ferulate -mediated lignin/arabinoxylan cross -linking in corn silage on feed intake, digestibility, and milk production. Journal of Dairy Science, 2011. 94(10): p. 5124 -5137. 238. Jung, H.G. and R.L. Phillips, Putative Seedling Ferulate Este r (sfe) Maize Mutant: Morphology, Biomass Yield, and Stover Cell Wall Composition and Rumen Degradability All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher. Crop Sci., 2010. 50(1): p. 403-418. 239. Casler, M.D. and H. -J.G. Jung, Relationships of fibre, lignin, and phenolics to in vitro fibre digestibility in three perennial grasses. Animal Feed Science and Technology, 2006. 125(1Ð2): p. 151 -161. 119 240. Casler, M.D., H.G. Jung, and W.K. Coblentz, Clonal Selection for Lignin and Etherified Ferulates in Three Perennial Grasses All rights reserved. . Crop Sci., 2008. 48(2): p. 424 -433. 241. Barros -Rios, J., et al., Divergent selection for ester -linked diferulates in maize pith stalk tissues . Effects on cell wall composition and degradability. Phytochemistry, 2012. 83(0): p. 43 -50. 242. Wilkerson, C.G., et al., Monolignol Ferulate Transferase Introduces Chemically Labile Linkages into the Lignin Backbone. Science, 2014. 344(6179): p. 90 -93. 243. Ximenes, E., et al., Deactivation of cellulases by phenols. Enzyme and Microbial Technology, 2011. 48(1): p. 54 -60. 244. Penning, B., et al., Validation of PyMBMS as a High -throughput Screen for Lignin Abundance in Lignocellulosic Biomass of Grasses. BioEnergy Research, 2014. 7(3): p. 899-908. 245. Sykes, R., et al., High -Throughput Method for Determining the Sugar Content in Biomass with Pyrolysis Molecular Beam Mass Spectrometry. BioEnergy Research, 2015: p. 1 -9. 246. Gjersing, E., et al., Rapid determination of sugar content in biomass hydrolysates using nuclear magnetic resonance spectroscopy. Biotechnology and Bioengineering, 2013. 110(3): p. 721 -728. 247. Nousiainen, J., et al., Prediction of indigestible cell wall fraction of grass silage by near infrared reflectance spectroscopy. Animal Feed Science and Technology, 2004. 115(3Ð4): p. 295 -311. 248. Xiao, L., et al., NIR and Py -mbms coupled with multivariate data analysis as a high -throughput biomass characterization technique: a review. Fron tiers in Plant Science, 2014. 5: p. 388. 249. Shafizadeh, F., Introduction to pyrolysis of biomass. Journal of Analytical and Applied Pyrolysis, 1982. 3(4): p. 283 -305. 250. Saiz-Jimenez, C. and J.W. De Leeuw, Lignin pyrolysis products: Their structures and their significance as biomarkers. Organic Geochemistry, 1986. 10(4-6): p. 869 -876. 251. Evans, R.J. and T.A. Milne, Molecular characterization of the pyrolysis of biomass. Energy & Fuels, 1987. 1(2): p. 123 -137. 252. Obst, J.R., Analytical pyrolysis of hardwood and softwood lignins and its use in lignin-type determination of hardwood vessel elements. Journal Name: J. Wood Chem. Technol.; (United States); Journal Volume: 3:4, 1983: p. Medium: X; Size: Pages: 377-397. 253. Ralph, J. and R.D. Hatfield, Pyrolysis -GC-MS characterization of forage materials. Journal of Agricultural and Food Chemistry, 1991. 39(8): p. 1426 -1437. 120 254. Li, M., et al., Cell -wall properties contributing to improved deconstruction by alkaline pre -treatment and enzymatic hydrolys is in diverse maize (Zea mays L.) lines. Journal of Experimental Botany, 2015. 255. Rolando, C., B. Monties, and C. Lapierre, Thioacidolysis , in Methods in Lignin Chemistry , S. Lin and C. Dence, Editors. 1992, Springer Berlin Heidelberg. p. 334 -349. 256. Lapierre, C., B. Monties, and C. Rolando, Thioacidolysis of Poplar Lignins: Identification of Monomeric Syringyl Products and Characterization of Guaiacyl -Syringyl Lignin Fractions. Holzforschung, 1986. 40(2): p. 113 -118. 257. Lapierre, C. and C. Rolando , Thioacidolyses of Pre -Methylated Lignin Samples from Pine Compression and Poplar Woods , in Holzforschung - International Journal of the Biology, Chemistry, Physics and Technology of Wood . 1988. p. 1. 258. Lu, F. and J. Ralph, Chapter 6 - Lignin , in Cere al Straw as a Resource for Sustainable Biomaterials and Biofuels, R. -C. Sun, Editor. 2010, Elsevier: Amsterdam. p. 169-207. 259. Gerber, L., et al., Multivariate curve resolution provides a high -throughput data processing pipeline for pyrolysis -gas chromatography/mass spectrometry. Journal of Analytical and Applied Pyrolysis, 2012. 95(0): p. 95 -100. 260. Hu, Z., et al., Chemical profiles of switchgrass. Bioresource Technology, 2010. 101(9): p. 3253 -3257. 261. Sykes, R., et al., High -Throughput Scre ening of Plant Cell -Wall Composition Using Pyrolysis Molecular Beam Mass Spectroscopy, in Biofuels, J.R. Mielenz, Editor. 2009, Humana Press. p. 169 -183. 262. Sykes, R., et al., Within tree variability of lignin composition in Populus. Wood Science and Te chnology, 2008. 42(8): p. 649 -661. 263. Tarkow, H. and C. Feist William, A Mechanism for Improving the Digestibility of Lignocellulosic Materials with Dilute Alkali and Liquid Ammonia , in Cellulases and Their Applications . 1969, AMERICAN CHEMICAL SOCIETY. p. 197-218. 264. Bozdogan, H., Model selection and Akaike's Information Criterion (AIC): The general theory and its analytical extensions. Psychometrika, 1987. 52(3): p. 345 -370. 265. Lance, C.E., Mallows' Cp Statistic , in Encyclopedia of Statistics in B ehavioral Science . 2005, John Wiley & Sons, Ltd. 266. Burnham, K.P. and D.R. Anderson, Multimodel Inference: Understanding AIC and BIC in Model Selection. Sociological Methods & Research, 2004. 33(2): p. 261 -304. 267. Sauerbrei, W., P. Royston, and H. Binder, Selection of important variables and determination of functional form for continuous predictors in multivariable model building. Statistics in Medicine, 2007. 26(30): p. 5512 -5528. 121 268. Luo, X., et al., Effects of wet -pressing -induced fiber hornification on enzymatic saccharification of lignocelluloses. Cellulose, 2011. 18(4): p. 1055 -1062. 269. Cheadle, C., et al., Analysis of Microarray Data Using Z Score Transformation. The Journal of molecular diagnostics : JM D, 2003. 5(2): p. 73 -81.