1 A HIGH-THROUGHPUT M ETHOD TO ELUCIDATE MECHANISMS OF CARBON SEQUESTRATION AND ALLOCATION IN POPLAR By JIANFEI ZHAO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2008
2 2008 Jianfei Zhao
3 To my parents, Wenbi Zhao and Qiubo Wang
4 ACKNOWLEDGMENTS I would like to thank m y adviso r, Dr. Wilfred Vermerris, w ho has helped me with great patience to deepen my knowledge in plant biol ogy and focus on the major picture of biology. He was always available providing his critical and insightful perspectiv e. Without his constant help, I would not have finished the research in this study. Also, I am very grateful to my committee members Drs. John M. Davis, Gary F. Peter and James F. Preston III for their support, valuable advice and guidance. I would like to tha nk the Forestry Genetics Laboratory for sharing their valuable data, materials and facilities. Speci al thanks to Dr. Matias Kirst, Chris Dervinis and Evandro Novaes. I am also grateful to Dr. La uren McIntyre for her efforts and support. This research has been greatly improved by their collectiv e contributions. I am also grateful to all the members in the Vermerris lab, Dr. Kenneth E. Lamb, Ana I. Saballos, Reuben Tayengwa, and Randi S. Wheeler. Special thanks to Megha n M. Brennan for stat istical consulting. I would like to thank the Graduate Program in Plant Molecular and Cell Biology, University of Florida, for providing this oppor tunity for graduate study and funding from the U.S. Department of Energy. My sincere thanks especially go to those gave me endless support: my dear parents, Wenbi Zhao, Qiubo Wang, and my fiance, Lanfei Gao.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF FIGURES .........................................................................................................................8ABSTRACT ...................................................................................................................... .............11 CHAP TER 1 BACKGROUND ....................................................................................................................13Global Carbon and Nitrogen Cycles .......................................................................................13Bioenergy and Lignocellulosic Biomass ................................................................................14Biomass and Renewable Bioenergy ................................................................................ 14Lignocellulosic Biomass ................................................................................................. 15Plant Primar y Cell Wall ....................................................................................................... ...16Cellulose ..................................................................................................................... .....16Physical characteristics of cellulose ......................................................................... 16Cellulose biosynthesis .............................................................................................. 19Cellulose synthase complex encoding genes............................................................21Hemicellulose ................................................................................................................. .24General properties of hemicellulose .........................................................................24Hemicellulose biogenesis .........................................................................................25Secondary Cell Wall ........................................................................................................... ....28Cellulose and Hemicellulose ...........................................................................................28Lignin ........................................................................................................................ ......29Lignin composition and structure ............................................................................. 29Lignin biosynthesis ..................................................................................................31Near Infrared Spectrometry and Mu ltivariate Statistics Methods .......................................... 32Utilization of a Pseudo-backcross Family for Quantitative Trait Loci Determination .......... 342 MATERIALS AND METHODS ...........................................................................................36Plant Materials ........................................................................................................................36Chemical Analysis of Root Cell Wall .................................................................................... 37Near Infrared Spectroscopy .............................................................................................37Pyrolysis-Mass Spectrometry ..........................................................................................37Lignin measurement ........................................................................................................ 38Statistical Data Analysis .........................................................................................................39Near Infrared Spectra Processing and Multivariate Data Analysis ................................. 39Quantitative trait loci analysis ......................................................................................... 39
6 3 CONSISTENCY TEST OF NEAR INFRARED SPECTRAL AND PYROLYSIS MASS SPECTROMETRY ACQUI SITION .......................................................................... 404 CHARACTERIZATION OF CELL WALL TRAITS USING NIRS AND PYROLYSI S-MASS SPECTROMETRY COMBINED WITH MULTIVARIATE STATISTICAL ANALYSIS ..................................................................................................455 QUANTATITIVE TRAIT LOCI DETERMI NATION FOR POPLAR ROOT CELL WALL TRAIT VARIATION .................................................................................................826 DISCUSSION AND FUTURE WORK ............................................................................... 112LIST OF REFERENCES .............................................................................................................118BIOGRAPHICAL SKETCH .......................................................................................................133
7 LIST OF TABLES Table page 4-1 Cell wall traits characte rized with m ass spectrum obtained after pyrolysis ...................... 505-1 Quantitative trait loci determined for ce ll wall traits under low nitrogen condition in stem and root tissues. C6, six carbon cellulose sugar ........................................................ 855-2 Quantitative trait loci determined for cell wall traits under high nitrogen condition in stem and root tissues. C6, six carbon cellulose sugar ........................................................ 86
8 LIST OF FIGURES Figure page 3-1 Consistency of NIRS settings with analysis of NIR spectra of control root genotypes. ... 423-2 Principal component analysis of NIR spectra obtained from adventitious root tissues grown in lowand highnitrogen conditions.. ...................................................................433-3 Principal component analysis of mass sp ectra obtained from two control genotypes for consistency test.. ...........................................................................................................444-1 Diagrams of NIR spectra acquired in (A ) bar, (B) landscape, (C) contour and (D) map formats .......................................................................................................................514-2 Plots showing scattering effects. E ach spectrum was plotted with the average spectrum ...................................................................................................................... .......524-3 Diagrams of NIR spectra acqui red after MSC transformation. ......................................... 534-4 Diagrams of mass spectra obtained after pyrolysis. ...........................................................544-5 Prediction of syringyl lignin m onomer content with PLS1 method.. ................................ 554-6 Prediction of syringyl lignin m onomer content with PCR method.. ..................................574-7 Prediction of guaiacyl lignin monomer content with PLS1 method .................................. 594-8 Prediction of guaiacyl lignin m onomer content with PCR method. .................................. 614-9 Prediction of syringyl-/guaiacyl lignin ratio with PLS1 method .......................................634-10 Prediction of syringyl-/guaiacyl lignin ratio with PCR method ........................................ 654-11 Prediction of total lignin content with PLS1 method ......................................................... 674-12 Prediction of total ligni n content with PCR method .......................................................... 694-13 Prediction of total carbohydr ate content with PLS1 method ............................................. 714-14 Prediction of total carbohydrate content with PCR method .............................................. 734-15 Prediction of lignin/carbohydrate ratio with PLS1 method ............................................... 754-16 Prediction of lignin/carbohydrat e content with PCR method ............................................ 774-17 Prediction of total lignin content w ith PLS1 method using the logarithmic transformed NIR spectra ....................................................................................................79
9 4-18 Plot of klason lignin and predicted lignin obtained from mass spectra ........................ 815-1 Genetic map of 19 li nkage groups in poplar. ..................................................................... 845-2 Overview of quantitative tr ait loci analysis for 6 cell wall traits under low nitrogen condition. .................................................................................................................... .......875-3 Quantitative trait loci analysis for syringyl lignin monomer content under low nitrogen condition trait on the 19 poplar linkage groups. .................................................. 885-4 Quantitative trait loci analysis for syri ngyl lignin monomer content trait, under low nitrogen condition on lin kage group 14 (LG-14) ............................................................... 895-5 Quantitative trait loci analysis for total carbohydrate content trait under low nitrogen condition on the 19 poplar linkage groups .........................................................................905-6 Quantitative trait loci analysis for total carbohydrate content trait, under low nitrogen condition on linkage group 1 (LG-1) .................................................................................915-7 Quantitative trait loci analysis for total carbohydrate content trait, under low nitrogen condition on linkage group 18 (LG-18) .............................................................................925-8 Quantitative trait loci analysis for ligni n/carbohydrate content ratio trait under low nitrogen condition on the 19 poplar linkage groups .......................................................... 935-9 Quantitative trait loci analysis for ligni n/carbohydrate content ratio trait, under low nitrogen condition, on linka ge group 6 (LG-6) ..................................................................945-10 Quantitative trait loci analysis for guaiacyl lignin monomer content trait under low nitrogen condition on the 19 poplar linkage groups .......................................................... 955-11 Quantitative trait loci analysis for guaiacyl lignin monomer content trait under low nitrogen condition on lin kage group 1 (LG-1) ................................................................... 965-12 Quantitative trait loci analysis for total lignin monomer content trait under low nitrogen condition on the 19 poplar linkage groups .......................................................... 975-13 Quantitative trait loci analysis for total lignin monomer content trait under low nitrogen condition on lin kage group 1 (LG-1) ................................................................... 985-14 Quantitative trait loci analysis for 6 ce ll wall traits under high nitrogen condition on the 19 poplar linkage groups ..............................................................................................995-15 Quantitative trait loci analysis for syri ngyl lignin monomer content trait under high nitrogen condition on the 19 poplar linkage groups ........................................................ 1005-16 Quantitative trait loci analysis for syri ngyl lignin monomer content trait under high nitrogen condition on lin kage group-11 (LG-11) ............................................................101
10 5-17 Quantitative trait loci analysis for guaiacyl lignin monomer content trait under high nitrogen condition on the 19 poplar linkage groups ........................................................ 1025-18 Quantitative trait loci analysis for guaiacyl lignin monomer content trait under high nitrogen condition on lin kage group-15 (LG-15) ............................................................1035-19 Quantitative trait loci analysis for sy ringyl/guaiacyl lignin monomer content ratio trait under high nitrogen condition on the 19 poplar linkage groups ............................... 1045-20 Quantitative trait loci analysis for sy ringyl/guaiacyl lignin monomer content ratio trait under high nitrogen condition on linkage group-11 (LG-11) ................................... 1055-21 Quantitative trait loci analysis for total lignin content trait under high nitrogen condition on the 19 poplar linkage groups .......................................................................1065-22 Quantitative trait loci analysis for total lignin content trait under high nitrogen condition on linkage group-15 (LG-15) ...........................................................................1075-23 Quantitative trait loci analysis for total carbohydrate content trait under high nitrogen condition on the 19 poplar linkage groups .......................................................................1085-24 Quantitative trait loci analysis for total carbohydrate content trait under high nitrogen condition on linkage group-11 (LG-11) ...........................................................................1095-25 Quantitative trait loci analysis for lignin/carbohydrate content ratio trait under high nitrogen condition on the 19 poplar linkage groups ........................................................ 1105-26 Quantitative trait loci analysis for lignin/carbohydrate content ratio trait under high nitrogen condition on lin kage group-17 (LG-17) ............................................................111
11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science A HIGH-THROUGHPUT M ETHOD TO ELUCIDATE MECHANISMS OF CARBON SEQUESTRATION AND ALLOCATION IN POPLAR By Jianfei Zhao December 2008 Chair: Wilfred Vermerris Major: Plant Molecular and Cellular Biology Global warming and the pending energy crisis are presenting a serious challenge to humanity. Various approaches have been pr oposed to mitigate greenhouse emissions and our dependence on non-renewable fossil fuels. Explori ng lignocellulosic biomass as a future source of alternative energy attracts considerable attention. Tree species represent the largest lignocellulosic biomass on Earth and have an immense CO2 sequestration capacity and excellent potential as source of bioenergy fe edstocks. It is essential to el ucidate the mechanisms of carbon allocation and partitioning in tree species. The first part of this research showed that near infrared spectroscopy (NIRS) accompanied by pyrolysis-mass spectrometry can be utilized as the method to accurately quantify lignin and carbohydrate-related cell wall tr aits in poplar root tissues Using the poplar pedigree Family UMD/UF-1 which was generated by a pseudo-backcross of Populus deltoides x F1 hybrid ( P. deltoides x P. trichocarpa), NIR models were developed to estimate cell wall trait values in root tissues. Two multivariate regression methods, pa rtial least squares-1 (PLS1) and principal component regression (PCR), were used and co mpared for prediction performance. The NIR models derived in this research can be read ily used for future screening of the entire Family
12 UMD/UF-1 in large-scale. Therefore, it could aid in elucidation of genetic mechanisms of carbon sequestration and allocation in tree species. As in the second part of this research, by comparing variation in root cell wall traits of biological replicates from Family UMD/UF-1 grown in two nitrogen conditions, chromosome regions that are responsible for these variations were identifie d in each nitrogen condition. The effects of nitrogen fertilization on poplar root cell wall were examined. The high nitrogen supply decreased syringyl and guaiacyl contents, their ratio, and rati o of lignin and carbohydrate in poplar root cell walls, while it in creased carbohydrate content. A dis tinct set of genes appear to be responsible for variation in each of the cell wa ll traits at high vs. low nitrogen levels. This research also provided evidence for differential gene regulation between poplar root and stem tissues. In conclusion, this study provided a met hod, by combining NIRS and pyrolysis-mass spectrometry, to predict cell wall traits in poplar root tissues reliably and reproducibly. The NIR models derived in this study enable large-scale screening for root traits in the entire pedigree. Together with the physical marker map of the pe digree, and the QTL determined from analysis of different plant tissues, this research will contribute to the elucidation of the genetic mechanisms of carbon allocation and sequestration in Populus.
13 CHAPTER 1 BACKGROUND Global warm ing and concerns about dwindli ng energy supplies are presenting a serious challenge to humanity. Climate change has been exacerbated by the emission of various greenhouse gas (GHG) resulting from fossil fu el consumption. Deposition of GHG into the atmosphere, such as CO2, oxides of nitrogen (NOx) and oxides of sulfur (SOx) as its major components, severely further affects the balance of carbon and nitrogen cycl es in terrestrial and ocean ecosystems, therefore resulting in ocean acidification and global climate change (Hill et al., 2006). In order to achieve a sustainable energy production, va rious approaches have been proposed as alternative energy resources to o ffset GHG emissions and to decrease humanitys dependence on non-renewable fossil fuels at the same time. Among these approaches, utilizing bioenergy serves as one of the most promising st rategies both for terrestrial ecosystems and for energy resources. Global Carbon and Nitrogen Cycles Since the 1990s, robust global population gr owth coupled with industrialization has resulted in a rapid in crease in atm ospheric CO2 concentration. This reflects the increase of the difference between the amounts of CO2 by anthropogenic carbon emission and the removal by terrestrial and ocean carbon sinks The tremendous release of CO2 into the atmosphere due to economic boom led to the highest CO2 concentration, exceeding 381 parts per million (ppm) since 2006 (Canadell et al., 2007), which is probab ly the highest concentration over the last 20 million years (Petit et al., 1999; Pearson and Pa lmer, 2000; Siegenthaler et al., 2005; de Vernal and Hillaire-Marcel, 2008). The present growth rate of CO2 concentration is the highest in the last 40 years (Canadell et al., 2007).
14 In the global carbon cycle, forests are important for several reasons. Forests currently store almost 45% of terrestrial carbon (Bonan, 2008), and take up huge amounts of carbon from the atmosphere every year. Forests contribute to the long-term carbon balance in the ecosystem through photosynthesis, respiration and bioma ss decomposition. However, carbon emissions are growing much faster than carbon sequestration by terrestrial and ocean sinks (Canadell et al., 2007). Ecological responses to the rapidly increasing CO2 concentration in the atmosphere have resulted in decreasing net carbon uptake of northern ecosystems (Piao et al., 2008) and changes in the abundance and distribution of forest plant species (Sturm et al., 2001; Lenoir et al., 2008). This human-induced climate change proposed an urge nt call for a closer examine of the role of forests in the balance of ecosystems in or der to enhance their carbon sink capacity. Bioenergy and Lignocellulosic Biomass Biomass and Renewable Bioenergy Biom ass energy is the heat energy converted from non-fossil fuels (Fie ld et al., 2008). It has long been proposed to have the advantage to offset CO2 emission compared to burning conventional fossil fuel. Bioenergy f eedstocks sequester atmospheric CO2 as they grow through photosynthesis, while fossil fuel combustion ma inly releases the stored carbon into the atmosphere. Therefore, utilization of bioene rgy has been credited as potential carbon sink to reduce GHG emissions. Biofuels currently are mainly ethanol produced from corn or sugarcane and biodiesel derived from oil crops. Limited by current conve rsion technology and land production capacity, the bioenergy industry could e xpand by planting trees on defo rested lands and grassland conversion, and at the same time, by replacing agricultural land with co rn or sugarcane for biofuel production. However, th rough these land conversion activ ities, the potential net carbon benefit for bioenergy produced from these lands could be offset if the amount of CO2 released
15 during biomass burning or microbial decomposition was taken into consideration. It would take dozens or even hundreds of years for bioener gy production to compensate the GHG emissions before any net carbon benefit (Fargione et al., 20 08; Searchinger et al., 2008). By evaluating the net carbon benefit of currently gr own major biofuel feedstocks, in cluding corn, soybean, alfalfa, reed canarygrass, switchgrass and hybrid poplar, multiple studies suggested perennial crops exceed with lowest net GHG emissions, higher elect ricity yields from biomass gasification and are most effective for compensating the CO2 emissions due to land c onversion (Adler et al., 2007; Fargione et al., 2008). Hybrid poplar is one of the most attr active biofuel stocks, especially considering its capacity grown in marg inal and abandoned agricultural lands. Lignocellulosic Biomass Biom ass from wood, straw and stover are described as lignocellulosic biomass, which is considered as the most attrac tive source for biofuel because of its high abundance and energy potential (sugars) (Farre ll et al., 2006). Tree species, like othe r plants, are autotrophic and sessile. They rely on photosynthesis to assimilate carbon into sugars and starch. Sucrose is one of the major compounds transported in the phloem from the source tissues, e.g. leaves and young bark, to non-photosynthetic sink tissues, e.g. roots and ma ture bark. A large portion of the carbon flux is allocated to cambial growth as well as secondary xylem differentiation. This carbon goes mainly towards the synthesis of cell walls, cons isting of primarily cellulose, hemicellulose, pectin and lignin. Once this carbon has been allocat ed to the primary and secondary cell wall, it cannot be converted back. Multiple approaches have been proposed to increase the carbon sink capacities of perennial plants. One of the most attractive strategies is to sequester atmospheric carbon into the soils by increasing the carbon sink strength of tree roots. Carbon stored in root tissues is ultimately decomposed by insects, f ungi and bacteria. Therefore reducing the rate of biomass decomposition is another strategy to en hance carbon sequestration in the soil. Higher
16 lignin-to-carbohydrate ratio and lo wer syringyl-to-guaiacyl lignin monomer content ratio both decrease the digestibility of root tissues. Therefore, manipulation of carbon sequestration and allocation with genetic strategies offers major potential towards mitigating climate change. Plant Primary Cell Wall In the developing plant cells, the plant cell wall is m ainly composed of a primary cell wall, composed of cellulose, hemicelluloses, pect ins, and cell wall associated proteins. In Populus tremuloides, primary cell wall contains 22% cellulo se, 47% pectin, 18% hemicellulose, 10% wall-associated protein (Melle rowicz et al., 2001).When the plant cells stops expanding, a secondary cell wall is deposited on the inside layer of the primary cell wall in certain tissues. The secondary cell wall contains high levels of lignin in certain tissu es. In wood-forming tissues of poplar, the characteristics of th e cell wall vary depend on cell type and developmental stage. Cellulose Physical characteristics of cellulose Cellulos e is a linear crystal polymer composed of -1, 4-linked-D glucose molecules (Taylor, 2008), and exists in plan t cell walls in the form of cellu lose microfibrils (Preston et al., 1948; Frey-Wyssling, 1954), which are insolubl e chains forming distinct layers. The organization of distinct layers ha s been suggested from the first direct visualization evidence of primary cell wall which came from the study of onion bulb layer (Brown and Montezinos, 1976). Multiple layers of cellulose microfibrils of distinct crystallinity regions could be clearly distinguished after unesterified pectins were removed with cycl ohexane-trans-1, 2-diaminetetraacetate (CDTA). In each layer, cellulose microfibr ils are parallel with each other and oriented nearly perpendicular to other micr ofibrils in the adjacent layers. This orientation pattern is even more obvious if subsequently removing hemicellulo ses with increasing grad ients of alkali, and the microfibrils are more tend to collapse into broader bundles with each other.
17 The physical dimensions of the cellulose chai n and microfibril have been determined in multiple studies. The degree of polymerization DP, glucose residues per chain) of cellulose chain varies between 8,000 in the primary cell wall (Brown, 2004), and up to over 15,000 in the secondary cell wall (Brett, 2000). The chains at the surface of microfibrils tend to be much shorter than chains buried within the crystalline microfibril (Brett, 2000). However, the exact length of individual cellulose chain is difficult to measure because of the difficulty in its extraction, purification and the fragility of cellulo se. Currently there is no clear understanding of the mechanisms controlling cellulose chain lengt h. The width of a single cellulose microfibril measured to date has been within a range of great variation. Its diameter ha s been suggested to be approximately 5-10 nm (Frey-Wyssling, 1954; Hert h, 1983). These results ar e consistent with later observations of an individual chain emerging from single rosette in Equisetum (Emons, 1985), as well as an 8-nm crystalline chain of onion parenchyma primary wall when viewed under transmission electron microscopy (TEM) (M cCann et al., 1990). However, the width of individual cellulose microfibril in quince seed mucilage has been reported as 2 nm (Franke and Ermen, 1969; Vian et al., 1994). This variation might be a result of the polysaccharide extraction method used leading to different degrees of cellulose aggregati on, distinct observation techniques employed, and various linkage degrees of other polysaccharides with cellulose, such as xyloglucans. The crystalline microfibrils are being deposited into and bent in the narrow space between the existed cell wall matrix and the plasma memb rane. Angles of the microfibril distribution in the cell wall matrix reflect the bending status of mi crofibrils and their existence status within the cellulose-hemicellulose-pectin network. Peura et al (2008) examined longitudinal sections of wood samples from Picea abies using X-ray microdiffraction for the angle distribu tion. It ranges
18 from -20 to 90 with a decreasing tendency from pith outwards from 24 into 19. This width of crystalline cellulose cross sect ion ranges 28.9 to 30.9 angstroms in mature wood samples (Peura et al, 2007). Previous solid-state 13C nuclear magnetic resonance (NMR) and X-ray diffraction studies indicated that microfibrils purifie d from almost all natural material s, such as plants or bacteria, have been shown to exist in two phases, which are designated Cellulose I and I (Atalla and Vanderhart, 1984; Sugiyama et al., 1991; Sugiya ma et al., 1991; Brown, 1996; Koyama et al., 1997; Matthews et al., 2006; Somerville, 2006). All the natural cell wall materials are a mixture of these two crystalline phases with di fferent ratios. Generally cellulose I form dominates in cell wall materials from bacteria and algae (Atall a and Vanderhart, 1984; Sugiyama et al., 1991), whereas the cellulose I from is more abundant in plant ce ll walls. Besides coexistence in the same cell wall, more interestingly, these two crys talline phases have been found to coexist in a single microfibril of cell walls of green marine al ga, and an alternative-localization of these two phases is favorable (Sugiyama et al., 1991; Imai and Sugiyama, 1998). Within the single chain, phase I exists as a triclinic unit cell with single-chain, while phase I, which has lower energy, exists as a monoclinic unit cell with two non-equivalent chains respectively (Koyama et al., 1997; Somerville, 2006). These phases have been s uggested to be interchangeable via thermal treatment (Yamamoto et al., 1989) and mechanic al bending (Jarvis, 2000 ). Considering their differences in energy state, and cellulose I is more vulnerable to hydrolysis and enzymatic degradation, a further understand ing of how cells manipulate the bi osynthesis and crystallization of these two phases into microfibrils will prov ide a better understanding of current model of cellulose biosynthesis.
19 Cellulose biosynthesis Cellulose biosynthesis m ay be as old as the emergence of the first photosynthetic organisms. It has been found widely spread among several genera of the cyanophycean (Nobles et al., 2001) which is the most ancient life on earth. Certain bact eria can synthesize cellulose (e.g., Azotobacter, Acetobacter, Rhizobium and Agrobacterium ). Cellulose has also been found in many fungi, amoebae, and green algae, and of course, in higher plants, such as mosses, ferns, angiosperms and gymnosperms. Interestingly, some animals have also been found to produce cellulose, such as the tunicate Metondrocarpa uedia (Hall and Saxl, 1960; Kimura and Itoh, 1995). Rather than a single glucose molecule, the re peat unit of cellulose is cellobiose, which consists of two reverse orientated (1->4)-linked D-glucose disaccharide. This unique successive orientation results in the relatively flat linear form of individual cellulose chains, and also enables the hydrogen bonds and Van der Waals for ces to form within the chain itself and between adjacent chains. This al so equips cellulose with all of its physical characteristics by allowing adjacent chains to stick together base d on hydrophobicity into water-insoluble, regular crystalline microfibrils. The successive glucose re sidue-inverted orientati on have led researchers to investigate the mechanisms of its biosynthesis in algae, bacterial, a nd most importantly, in higher plants. Several models have been proposed so far to describe its biosynthesis (Saxena et al., 1995; Koyama et al., 1997; Carpita and Vergara, 1998). Current data support the general model that cellulose chai ns are being synthesized at the plasma membrane possessively by cellulose sy nthase complex with a hexameric rosette structure. This structure model generally proposed that the rosette is composed six similar units which were the minimal units observed on the PF fracture side of plasma membrane. Since the analysis comparing the dimensions of single ce llulose microfibril and single cellulose chain
20 revealed that the microfibril generally consists 36 individual cellulose ch ains, therefore, it is reasonable to further propose that each unit of the six in hexameric structure may synthesize 6 individual cellulose chains. Combining these obser vations and subsequent direct visual imaging of symmetrical hexameric structur e of rosettes (Kimura et al., 1999), it is very likely that the general 8-nm cellulose microfib ril contains approximately 36 2 nm-wide individual cellulose chains which are synthesized by each cellulo se synthase catalytic subunit (Herth, 1983). However, this calculation is based on many assu mptions, such as estimation of average cellulose microfibril cross-section area, estimation of mass ratio between cellulose and xyloglucans, and assumption of equal packing dens ity within the microfibril. The sequence of CESA protein i ndicates there is only one cata lytic domain in each protein; therefore every cata lytic subunit is likely to synthesize one cellulose chain. Mutant and inhibitor analysis reveals the 36 catalytic subunits (CESA protein) could be grouped into three distinct types Type 1, 2 and 3 (Tanaka et al., 2003; Taylor et al., 2003; Taylor, 2008). The distinct three types of catalytic subunits are critical for prope r cellulose synthesis in primary and secondary cell walls. Currently there is no evidence against this model, yet no solid data to reveal how the proposed 36 CESAs make the rosette. The intact cellulose synthase complex as a hexameric structure could be observed not only on the P face of freeze-fractured plasma membrane, but also on the Golgi apparatus membrane and Golgi vesicles (Herth, 1985; Haigler and Brown, 1986). Live imaging of catalytic subunit mutant in Arabidopsis xylem cells suggested th at mutation in any one of these three types catalytic subunits will result in the rest two types of subun its retaining in the endoplasmic reticulum (ER) (Gardiner et al., 2003). The obs ervations of fluorescently-labeled catalytic subunits in the ER, and the rosette structures in Golgi vesicles, Golgi apparatus and plasma
21 membrane suggested that, although there is no so rting signal at its N-te rminal, the cellulose synthase catalytic subunits follow the secretory pathway, from ER to Golgi apparatus, through Golgi vesicles and finally reach the plasma membra ne. This sorting route is consistent with the plant Golgi organelle proteomics study (Dunkley et al., 2006), and other observations using fluorescent protein tags (Paredez et al., 2006). The assembly of the hexameric structure in the Golgi and its deposition into the plasma membrane has been suggested to be independent of the biological activity (Gardiner et al., 2003). A recent study using fluorescently labeled CESA6 s uggests that once the intact rosette being has been deposited in to the plasma membrane, in Arabidopsis hypocotyls cells it obtains motility within 10 seconds (Paredez et al., 2006), and main tains a velocity of 330 nm/min for almost 20 to 30 minutes (Paredez et al., 2006; Desprez et al., 2007), and may subseq uently dissociate or recycle-back to vesicles near the plasma membrane via endocyt osis (Jacob-Wilk et al., 2006). Cellulose synthase complex encoding genes The cloning of cellulose synthase gene s was challenging. Using the bacterial cellulos e synthase gene sequences to screen various plan ts EST databases did not provide much information. This hinted at a lack of homol ogy of overall gene sequence between bacterial and plants cellulose synthases. Assumi ng that the expression of plant cellulose synthase genes will reach its peak when cellulose synthesis achieves its maximum rate, Pear et al (1996) cloned the first plant cellulose synthase genes from the cotton ( Gossypium hirsutum ) cDNAs collected when the maximum rate of secondary wall synthesis t ook place. Protein sequences deduced from these genes suggests less than 30% hom ology between plant and bacterial cellulose synthases. There are regions along the whole sequence that shar e 50-60% homology, which may be the regions conserved for glycosyltransfer ases catalytic functions.
22 Genome analysis and expressed sequence tag (E ST) databases suggest there are at least 10 genes ( CesA cellulose synthase A ) encoding the catalytic subunit of cellulose synthase complex in Arabidopsis (Holland et al., 2000; Richmond and Somerville, 2000; Somerville, 2006). Mutant analysis in cell wall s ynthesis and whole-genome analysis reveal there are at least 12 CesA s in maize (Appenzeller et al., 2004), at least eight CesA s in barley (Burton et al., 2004), and 18 CesA s in poplar (Djerbi et al., 2005; Suzuki et al., 2006). Comparison of these CesA genes reveals the common features in higher plan ts, and also indicates specific species-unique features. In Arabidopsis, all 10 cellulose synthase genes are split with 10-13 introns and scattered across three of its five chro mosomes (Chromosome II, IV and V) (Richmond and Somerville, 2000), and the protein length ranges from 985 to 1088 amino acids. While in poplar, the cellulose synthase gene number is nearly two times that found in Arabidopsis. The genes generally share 55 to 88% homology with Arabidopsis CesA s (Suzuki et al., 2006), and are also split with 11-14 introns (Joshi et al., 2004) and could be mapped to 11 of the 19 chromosomes (Djerbi et al., 2005). Cellulose is proposed to be synthesized by the rosette complex at the plasma membrane in higher plants. Consistent with this hypothesis, all of these plant cellulo se synthase genes are predicted to encode integral pr oteins sharing a conserved stru cture of eight predicted transmembrane domains. In Arabidops is and poplar, there is no sorti ng signal at the N-terminus of typical CESA protein structure. Its N-terminus starts from a cytosolic domain containing a cysteine-rich domain with f our CxxC motifs (C=Cysteine, x=any amino acid). This motif CX2CX12FXACX2CX2PXCX2CXEX5GX3CX2C represents zinc RING finger domain, which may facilitate protein-protein in teractions, e.g. CESA dimerization, during cellulose synthase assembly. The proposed model of cellulose syntha se rosette complex depends on the interactions
23 with CESA catalytic subunit with itself, with other types of CESA subunits or with other components involved. Yeast two-hybrid experiment s and in vivo recombinant protein pull-down assays confirmed the existence of homo-dimeri zation of cotton cellulose synthase GhCESA-1and hetero-dimerization with another cellulose s ynthase GhCESA-2 in oxidized state, while GhsCESA1 exists as monomer in reduced state (Kur ek et al., 2002). This indicates the assembly of rosette complex and cellulose synthesis ma y depend on the presence of zinc RING finger motif in a redox regulated manner. Phosphorylati on in the N-terminus cytosolic domain plays roles in post-transl ational regulation. Two trans-membrane domains follow the cytoso lic region, and a larger central cytosolic domain containing the catalytic motif for processive glycosyl transferases proceeds. The processive glycosyltransferase reactions depend on this catalytic motif which has been known as the D, D, D, QXXRW motif (Saxena et al., 1995; Saxena et al., 2001). This motif was indicated through multiple-analysis and hydrophobic cluster analysis of the glycosyltransferases sequences. Together with the conserved first thre e aspartic acids (D), the QXXRW motif forms a cytosolic globule for glycosyltransfer. Site-direc ted mutation analysis c onfirmed the essential roles of the conserved aspartic residuals in su bstrate (UDP-glucose) bi nding (Pear et al., 1996; Saxena et al., 2001). In the predicted structure by Saxena et al (2001), the QXXRW motif falls in the central cavity that link two UDP-glucose molecules to form the (1, 4)-beta-linked backbone. As the cellulose chains extend out of the rosette complex during cellulose synthesis, monosaccharides are being transferred processive ly from donor UDP-glucose to the growing chain. This series of reactions in cell wall biogenesis are conser ved through bacteria, algae and higher plants. Therefore, it is reasonable that the glycosyltransferases in these species maintain a conserved structure for this preserved bioactivit y. Vergara et al (2001) proposed that in the
24 central cytosolic domain there is a region may specify the substrate binding specificity and function in catalysis. This regi on has been termed Class-Specif ic Region (CSR) (Vergara and Carpita, 2001). Based on the CSR, CesA orthologs could be identif ied among different species. Six predicted transmembrane domains lie to the C-terminal of the QXXRW motif along the CesA sequence. They function in anchoring th e integral catalytic subunit in the plasma membrane. Hemicellulose General properties of hemicellulose In addition to cellu lose microfibrils, the plan t cell wall also contai ns hemicellulose, a branched matrix polysaccharide that binds via hydrogen bonds. It generally represents approximately 15-30% dry weight of wood (Plomion et al., 2001), and the contents in different developing tissues vary. In seconda ry growth tissues such as matu re wood, this value increases to ranges of 22% to 33% of dry weight (Jones et al., 1961; Northcot, 1972; Mcdougall et al., 1993; Mellerowicz et al., 2001). Hemicelluloses are render ed soluble following treatment with alkali to remove pectins They exist in the cell wall matri x, either as hetero-polymers, like glucomannan, arabinogalactan, glucuronoxylan, or as homo-polymers, like galact an and arabinan (Plomion et al., 2001). Two types of hemicellulose, xylogl ucans and arabinoxylan, are the two most abundant components in plant primary cell wall (Cosgrove, 2005). While in the secondary cell wall, arabinoxylans, glucuronoxylans and glucur onoarabinoxylans are the main hemicellulosic polysaccharides (Ebringerova and Heinze, 2000). Mannans and mixed-linkage beta-glucans are the major types of hemicellulose in multiple cell types in various plant species (Cocuron et al., 2007). Xyloglucan has a (1, 4)-linked backbone similar to cellu lose, with, xylose substitutions along the backbone (Fry et al., 1993). By partitioning its backbone into individual segments that
25 contain individual glucan residue s, xyloglucans are defined base d on the arrangement of these segments. Single letters are used for these segmen ts nomenclature. Letter G represents a linear (1, 4)-linked glycosyl residue, and X is used for G with branched (1, 6)-linked Dxylopyranose. The letter L represents -D-galactopyranose-(1, 2)-D-xylopyranose-(1, 6)D-glucopyranose residue. Subunit compositions of XXXG, XXLG, XXFG and XLFG are found in most plant species (Vincken et al., 1997; Somerville et al., 2004). Lengths of individual xyloglu cans chain vary from 30 to 500 nm (Somerville et al., 2004; Taiz, 2006) The chains that are longer than th e average distance between two adjacent cellulose microfibrils tend to be trapped and bind the mi crofibrils with hydrogen bonds and together to form a cellulose-hemicellulose network. Based on the thermodynamics of hydrogen bonds among these polymers, a physical model was proposed in which the specificity and polymerization degree of xyloglucans regulate ce ll wall enlargement (Veytsman and Cosgrove, 1998). Arabinoxylan is another abundant type of hemice llulose in plant cell walls, especially in the secondary cell wall. Arabinoxylan s share the same (1, 4)-linked -D-xylan backbone, and arabinoxylans branch from the backbone. Arabi noxylans can bind cellulose and other types of hemicellulosic polysaccharides, especially through covalent attachment via ferulic acid esters (Cosgrove, 2005). Mannans are present in moderate amounts in plant cell walls. Galactomannans and galactoglucomannans are two major mannosecontaining polysaccharides Another type of hemicellulose, the (1, 3:1, 4)-D-glucans, or termed mixed-l inkage glycans, are linear unbranched (1, 3) and (1, 4)-D-linked glycans specifically in the Poales. Hemicellulose biogenesis The cellu lose-hemicellulose network is formed to dynamically regulate multiple aspects of plant development, such as cell expansion. The plant actively regulates this network via two
26 main strategies, either by (1) adjusting polysaccha ride(s) ratio, or (2) by m odifying association of binding proteins with carbohydratebinding modules to the networ k (Taylor, 2008). Therefore, an in-depth knowledge of hemicellulo se biosynthesis is essential. Despite the central roles of hemicellulose in plant cell wall, details of its biosynthesis are not clearly understood. Owing to the conserved domain preserved for glycosyltransferases function in various sp ecies, the first plant CesA gene sequence was obtained by transcription profiling through cotton EST database collected from developing cotton fibers using the bacterial CesA gene sequence (Pear et al., 1996). Reiterative searches within the complete Arabidopsis genome revealed that besides the CesA family of 10 members that have been confirmed for cellulose synthesis (except AtCesA10 ), 29 additional genes were found to contain the domain for glycosyl-residual transferring activity (Cutle r and Somerville, 1997; Richmond and Somerville, 2000). These genes are structurally similar with the CesA genes, therefore, they were termed Cellulose Synthase-Like (Csl ) genes and grouped into six classes in Arabidopsis ( CslA -B -C D -E and -G ). Survey in the complete genome of rice ( Oryza sativa ) revealed 33 Csl genes that could be grouped into six groups (Keegstra and Walton, 2006). The letters representing Csl groups are based on sequence alignment and are for designation only. Rice and Arabidopsis share the same CslA -C -D and E groups, whereas CslF and H groups are unique to grass species including barley and rice. The release of Populus trichocarpa genome revealed 30 gene s belong to the poplar Csl gene families (denoted as PtCsl s) (Suzuki et al., 2006; Tuskan et al., 2006; Jansson and Douglas, 2007). There are 15 additional genes in poplar showing partial coding sequence that are homologous with Arabidopsis CslC -D and G families. Multiple sequence alignments among the Csl sequences of Arabidopsis, poplar and ri ce indicate that poplar share the sam e Csl groups
27 with Arabidopsis, but not with monocots such as rice (Keegstra and Walton, 2006). Therefore although some families of cellulose synthase-like ge nes are found in all land plants, there are still specific groups that are conserved in dicotyledonous angiosperms, such as CslA through E and G shared by Arabidopsis and poplar, while CslF -H and J only exist in monocots (Burton et al., 2006; Farrokhi et al., 2006). In the eight Csl families shared in dicots, the gene numbers of each family are not equal in Populus trichocarpa and Arabidopsis thaliana genomes (Richmond and Somerville, 2000; Tuskan et al., 2006). For CslA family, the number of genes nearly doubles in Arabidopsis than poplar. This is more evident for CslB family; Arabidopsis contains six genes comparing with the two genes in Populus trichocarpa Intriguingly, black cottonwood contains 10 PtCslD s in its genome and Arabidopsis has five AtCslD s. Since gene sequence divergence often reflects functional divergen ce, it is reasonable to speculate that the differences between AtCsl and PtCsl families are due to the specific re quirements of unique polysaccharides biogenesis in their cell wall matrix. Biochemi cal examination of global polysaccharides composition from different tissues at various developmental stages, accompanying with determination of expression levels of gene me mbers by transcription a bundance profiling, may reveal clues to explain such a divergence. Ho wever, such attempts offered limited success. Although traditional biochemical methods are help ful to elucidate cell wall composition, during extractions loose bonds and weak association within the polysaccharides are always diminished, thus the results obtained could reflect partial information. Also, extrac tion methods are always developed for characterizing specific components. The complexity and spatial heterogeneity of cell wall matrix are hard to be directly de termined through these assays. Recently, several methods have been developed to systematically map cell wall polymers, either using monoclonal antibodies against specific plant cell wall componen ts (Willats et al., 1998; Willats et al., 1999;
28 Willats and Knox, 1999; Pettolino et al., 2001; Willats et al., 2002), or based on binding specificity of carbohydrate-bindi ng modules (McCartney et al ., 2004; McCartney et al., 2006; Moller et al., 2007). These methods are useful to screen cell wall materials and provide semiquantitative information. However, they are limited by the variety, sensitiviti es and specificity of monoclonal antibodies developed, and also aff ected by methods breaking down cell walls and materials used to immob ilize diverse glycans. Cellulose synthase-like genes have been identified in plants based on their homology with cellulose synthase genes (Cutler and Somerv ille, 1997; Richmond and Somerville, 2001). All proteins encoded by Csl genes have the same D, D, D, QXXRW motif as CesA genes in Arabidopsis and bacteria. Theref ore they are presumably associ ated with this motif of UDPglycosyltransferase catalytic functions. Recent protein organelle proteomics studies confirmed their presence in the Golgi, which is the comp artments where branched polysaccharides are synthesized (Dunkley et al., 2006). Secondary Cell Wall When the plant cell stops expanding, the sec ondary cell wall is deposited to the inside layer of the primary cell wall. Besides cellulose, hemicelluloses, the secondary cell wall is also composed of lignin. Cellulose and Hemicellulose Cellulose extent varies between primary walled and secondary walled cells, as well as am ong their spatial differences. This suggests that these differences may be the consequences of the different sets of cellulose synthase genes f unctional for primary and secondary walls. This is consistent with the hypothesis propo sed for distinct roles of cellu lose synthases in priming and elongation of cellulose bi ogenesis in plants
29 Secondary growth is essential in wood form ation in tree species, as it contributes substantially to carbon allo cation and participation in perennia l tree species. In the tree model poplar, xylem fibers are the majo r portion of wood, and cellulose re presents 50% of its total biomass (Balatinecz et al., 2001). (Peng et al., 20 02; Read and Bacic, 2002). In the secondary walled tissues of Populus nigra (L.), cellulose represen ts 48% of dry weight. Mutagenesis analysis also revealed specific cellulose genes involved in the secondary cell wall deposition. In Arabidopsis, among its total 10 cellulose synthase genes, AtCesA4 -7 and -8 are associated with secondary wall biogenesis (T urner and Somerville, 1997; Taylor et al., 1999; Taylor et al., 2000; Ha et al., 2002; Taylor et al., 2003). They may associate with each other to form the intact rosette structure specifically in tissues of seco ndary growth of Arabidopsis. EST libraries and microarray analysis indicated certain poplar cellulose synthase genes are particularly expressed during xylogenesis (A speborg et al., 2005; Suzuki et al., 2006). Transcripts of PtCesA4 -5 -7 -8 and -17 are abundant in xylem, and PtCesA13 and -18 showed almost 104 times higher in xylem than other tissues examined. Besides dicot plant, in monocot species, members of multiple cellulose synthase gene family are specifically functionally active in secondary growth. In Oryza sativa a set of mutations that affect three genes, OsCesA4 -7 and -9 which are orthologs of corresponding number of AtCesA (Tanaka et al., 2003). The null mutants showed similar phenotypes, such as decr eased cellulose content in secondary growth tissues, and for, also showed same expression pattern in secondary growth tissues. Lignin Lignin composition and structure Lignin is a m ajor component of the seconda ry wall. This polymer of heterogeneous hydroxycinnamyl alcohols (monolignols) gaps in be tween cellulose -hemi cellulose and pectins (Boerjan et al., 2003). Li gnin gives structural support to the plant to allow upward growth,
30 provide impermeability for water and nutrient tran sport (Jones et al., 2001), and contribute to plant defense system against biotic and abiotic st resses (Lange et al., 1995; Franke et al., 2002). Structural analysis by microa utoradiography and UV-microsp ectrometry reveals lignin is primarily polymerized by three types of hydroxycinnamyl alcohols (monolignols), p-coumaryl alcohol, coniferyl alcohol and si napyl alcohol. These three different precursors differ from each only by phenyl ring substitution/methoxylation degr ee. Once they have been produced from phenylalanine through the monolignol biosynthetic pa thway, they polymerize, giving rise to phydroxyphenyl (H), guaiacyl (G) an d syringyl(S) subunits. Besides the basic three monolignol structures, many structural components, either produced from monolignol precursors or other phenolic compounds have been revealed (Hu et al., 1999; Lu and Ral ph, 1999; Ralph et al., 2001; Lu and Ralph, 2002; Boerjan et al., 2003; Morreel et al., 2004; Lepl et al., 2007). The most common linkage found is the -O-4 linkage, and other linkages, such as -5, 5-5, -1 and 5-O-4, has also been found between two units. The cross-linkages are affected by the ratio of syringyl and guaiacyl units, as th e extra OMe group on the aromatic ring of syringyl unit allows only two linkages formed between it and other un its, whereas guaiacyl units can be covalently linked with a maximum of three other units. Poplar lignin contains primarily G and S units, and only a trace amount of H units (Baucher et al., 1998; Coleman et al., 2008). The ra tio of S and G monomers (S/G) in poplar is a major cell wall characteristic. It wa s reported to range from 0.725 in Populus trichocarpa (Boerjan et al., 2003), 1.8 in hybrid poplar (Coleman et al., 2008), 1.72-1.93 in Populus trichocarpa x P.deltoides (Pitre et al., 2007), to 2.3 in Populus tremuloides (Hu et al., 1999). The monolignol subunit ratio indicate s the carbon flow through branches of monolignol biosynthetic pathway. This cell wall trait (S/G) was also r ecently proposed to be important for paper pulping
31 industry. This is due to the roles of different aromatic structures of syringyl and guaiacyl units on lignin structure. Various major lignin structural linkages and th e monomer abundance have been revealed by chemical degradation analysis (Lundquist and Kirk, 1971; Lu and Ralph, 1997; Holtman et al., 2003) and by noninvasive methods, such as nuclear magnetic resonance (NMR) spectroscope analysis (Hu et al., 1999; Ralph et al., 2001; Lepl et al., 2007), and Fouriertransform infrared spectroscopy (Schultz et al., 1985; Faix, 1986; McCann et al., 2007). Lignin biosynthesis The lignin structure formed by polymerization of monolignol renders lignin relatively resistant to biotic and abiotic degradation. Extens ive efforts have led to the identification of genes participating in m onolignol biosynthetic pathway in plants. The release of complete genomes of Arabidopsis thaliana Oryza sativa and Populus trichocarpa has enabled the systematic characterization and comparison of phenylpropanoid-related genes, and other components in the lignification toolbox (Whe tten and Sederoff, 1992; Osakabe et al., 1995; Bell-Lelong et al., 1997; Betz et al ., 2001; Kumar and Ellis, 2001; Ro et al., 2001; Harding et al., 2002; Kao et al., 2002; Nair et al ., 2002; Costa et al., 2003; Kirst et al., 2003; Raes et al., 2003; Ehlting et al., 2005; Ehlting et al., 2006; Li et al., 2006; Tsai et al., 2006 ; Tuskan et al., 2006; Hamberger et al., 2007). There are detailed reviews focusing on monolignol biosynthetic pathway in many species recently, such as Arabidopsis thaliana (Raes et al., 2003), Oryza sativa (Jumanova et al., 2006; Bao et al ., 2007; Hamberger et al., 2007), Zea mays L. (Andersen et al., 2008), Picea abies (Koutaniemi et al., 2007), and Populus trichocarpa (Tsai et al., 2006; Tuskan et al., 2006; Hamberge r et al., 2007) Total lignin content in poplar has been re ported to range from 19.0-23.8% of total dry weight in poplar stem samples (Hu et al., 1999; Mellerowicz et al., 2001; Le pl et al., 2007; Pitre et al., 2007; Coleman et al., 2008). However, curr ent reported values are mostly based on wood
32 samples collected from poplar stems, either fr om stem xylem with or without barks or from extract-free wood. Lignin conten t is not equally distributed among poplar tissues. Measurement of lignin content in stems, leaves a nd roots of wild type and transgenic Populus tremuloides in which the 4CL or coniferaldehyde 5-hydroxylase ( CAld5H ) gene had been downregulated showed statistically significant differences am ong each other (Hancock et al., 2007). This may indicate the tissue-specific lignin deposition mechanisms, or this may depend on different efficiencies of lignin extraction. Currently, there is no single prot ocol to extract and determine total lignin in a quantitative manner. Different extraction methods also result in different values from the same sample (Fukushima and Hatfie ld, 2004; Hatfield and Fukushima, 2005). For example, the gravimetric Klason method, acid de tergent lignin method, permanganate lignin, and the spectrophotometric acetyl bromide (AcBr) proce dure all result in differe nt concentrations of lignin in the sample (The measured lignin c ontents are called after the procedure name, respectively, such as Klason lignin). Therefore, it is essential to deve lop an accurate, efficient determination method to quantify lignin as well as other cell wall components in multiple lignocellulosic biomass tissues. Near Infrared Spectrometry and Mu ltiva riate Statistics Methods Developing high-throughput/high-information as says is essential for characterizing cell wall traits on a large scale, such as determin ing the digestibility of biomass feedstocks and screening for certain traits in a large population. Combined with the availabili ty of whole genome sequences of Arabidopsis thaliana (Arabidopsis-Genome-Initiative, 2000), Populus trichocarpa (Tuskan et al., 2006) and Oryza sativa (Goff et al., 2002; Yu et al., 2002), as well as tissue-specific expressed sequence tag (EST) databases in various species, these integrative highthroughput strategies are showing great promise to identify chromosome regions that are responsible for variation in many cell wall traits and comparative genomics.
33 Traditional methods to annotate cell wall gene s rely on mutagenesis analysis with an emphasis on visible phenotypes, such as e ffects on plant height, pigmentation, xylem development and root hair growth. However, even after extensive mutagenesis screening, still not every cell wall related gene ha s been identified. This is probably in part due to the fact that cell wall biogenesis is so critical to plant deve lopment that multigene families with overlapping functions exist to regulate as gene networks. Interruptions to the key regulatory genes at the centers of network may yield visible effects. Theref ore in order to elucidat e the genetic control of cell wall biogenesis, besides trad itional research methods, focusi ng on methods of invisible cell wall phenotypes are being employed (McCann a nd Carpita, 2005) These methods rely on a combination of spectroscopy and chemical composition analysis. Near infrared spectroscopy (NIRS) has been wi dely used to determine various biological traits, especially those associated with plant ce ll wall properties (Bai et al., 2004; Zenoni et al., 2004; McCann and Carpita, 2005; Yeh et al., 2005; Petisco et al., 2006; Philippe et al., 2006; Roumet et al., 2006; Lepl et al., 2007; McCann et al., 2007; Philippe et al., 2007; Vermerris et al., 2007). NIRS is based on the sample absorbance of near infrared light energy which is within the 800-2500 nm range. Combined with multivariate statistical analysis methods, such as principal component analysis (PCA), discriminate analysis (DA), cl ustering analysis (CA) and partial least squares (PLS) analysis, NIR spectra taken from various samples can be projected into hyperspace with reduced dimensions, normalized, and distinguished from each other. Therefore, according to the spatial distribution subtle differences between th e spectra can be identified and samples can be clustered into distinct qualitative groups.
34 Besides generating qualitative data, NIRS has also been used to predict the values of cell wall traits. Considering the intensity at each wave length of the spectrum as independent variables and a given cell wall trait as the dependent variable, multivariate equations can be established that define the relationship betw een trait and spectrum. NIRS ha s been used to successfully predict Klason lignin content, S/ G unit ratio, cellulose and xylan content from stem wood and leaves of loblolly pine and aspen (Yeh et al ., 2005; Petisco et al., 2006; Yamada et al., 2006), predict H/G unit ratio in maritime pine (Alves et al., 2006) and the acety lation degree of poplar wood (Celen et al., 2008). Besides its reliability, NIRS is also a fast clean and non-destructive method for cell wall determination. Compared to tr aditional chemical methods, su ch as wet chemistry and mass spectrometry, which usually require hours of ope ration, it only takes approximately one minute for near infrared spectrum acquisition. Other main advantages of NIRS are easy sample preparation, small sample amount required. Anothe r main advantage is that materials remain non-destructive during the NIR opera tion. Therefore, same material could be used for subsequent wet chemistry determination, such as pyrolysis-mass spectrometry. Utilization of a Pseudo-backcross Family fo r Quantita tive Trait Loci Determination It is important to understand the mechanisms that regulate the car bon flux into the plant cell wall in order to enhance the sink strength of roots. However, the ge netic control of cell wall biosynthesis still remains one of the biggest re search questions, especially for woody plants (Geisler-Lee et al., 2006; Li et al., 2006; Tusk an et al., 2006; Jansson and Douglas, 2007; Jamet et al., 2008; Taylor, 2008). The use of quantitative trait locus (QTL) analys is using pedigrees enables the identification of potential regulatory genes, but this requires measurement of la rge numbers of samples within the pedigree. QTL mapping of the genetic basi s of carbon allocation and partitioning in leaves,
35 stems and roots is now being carri ed out at the University of Fl orida, Gainesville (Drs. Kirst, Peter and Davis) as part of the DOE-funded proj ect, Genomic mechanisms of carbon allocation and partitioning in poplar. This project relied on the study of the UMD-UF-1 pedigree generated by backcrossing of Populus deltoides and Populus trichocarpa the two parental species that vary extensively for genetic traits controlling both source and sink organs. This high variation among the parental genotypes and the subsequent segregation in the seedlings allow for extensive quantitative trait analysis. QTL for the diameter and height for P. deltoides and P. trichocarpa hybrids have been identified (Wu et al 1994, 2003; Wu, 1998), as well as QTL for the chemical composition (cellulose, hemicellulo se and lignin). Total lignin content was shown to vary between 21 and 28% in the hybrid. This ge netic diversity also facilitates the identification of QTL for root mass and root architect ure-related traits (Wu, 1998; Friend et al 2000). Therefore, a wide segregation for root carbon allocation and partitioning traits is expected. The objective of this research is to test the feasibility of NIRS to characterize cell wall in poplar adventitious root tissues, and further to derive proper NI R models to quantify these cell wall traits, such as lignin and carbohydrate related tra its. Combining with the available poplar genome marker data, this research also intends to identify QTLs that are responsible for cell wall variation. The feasibility of NIRS was tested and proved to be an accurate and reliable method to quantify six root cell wall trai ts, syringyl lignin monomer content, guaiacyl lignin monomer content, total lignin content, syringyl-/guaiacyl lignin monome r content ratio, total carbohydrate content and lignin/carbohydrate cont ent ratio. Besides showing the NIRS as a fast screening tool in this research, chromosome regions are identified for variations in these traits under two nitrogen availabilities.
36 CHAPTER 2 MATERIALS AND METHODS Plant Materials Two Populus genotypes, Clone 52-225 (a fem ale Populus trichocarpa x Populus deltoides hybrid) and Clone D109 (a pure Populus deltoides male) were crossed to generate a pseudobackcross pedigree, Family UMD/UF-1 Clone 52-225 was generated from a cross between clones Populus trichocarpa 93-968 and Populus deltoides ILL-101. Rooted cuttings of the pseudo-backcross pedigree UMD/UF-1 were produced and grown in the greenhouse kept between 18 and 26 Tree samples were distributed in the greenhouse (University of Florida, Gainesville, FL) according to the experimental design. Two different nitrogen levels were supplied by watering the plants with Hockings complete nutrient solution with ammonium nitrate at 0 mM and 25 mM as final concen trations (Hocking, 1971). These two different nitrogen treatments were applied to the four ramets of same genotype from the pedigree. Adventitious root samples were separated from soil, freeze dried, and stored at 4 for further processing. Root samples were ground into a fi ne powder with particle s passing through a 1-mm filter. Ground root tissues we re kept in an oven at 55 A set of 286 SSR and 63 AFLP markers were used to map the Family 331 and 462 SSR and 434 AFLP markers were used to map Family 13 (Kirst et al, 2006). Family 13 shares the same female parent with Family UMD/UF-1 ( 52-225). Parents of Family UMD/UF-1 ( 52-225 and D125) have been initially genotyped with over 500 microsatellite markers at UF and DOE/Oak Ridge National Lab (Kirst proposal 2006). A framework map with 230-250 SSR markers spaced 10 cM apart was clustered based on parental genotypes for the analysis proposed.
37 Chemical Analysis of Root Cell Wall Near Infrared Spectroscopy A FieldSpec Pro spectrometer (Model No. FSP350-2500P, Analytical Spectral Devices, Inc., Boulder, CO) was connected to sampli ng accessory (Model A122100, Analytical Spectral Devices, Inc., Boulder, CO). Reflectance spectra in the range between 350 and 2500 nm were collected at a spectral resolution of 1 nm. Each spectrum was averaged from 30 measurements. The Spectra were saved and processed using FieldSpec Pro RS3, Version 2.0 (Analytical Spectral Devices, Inc ., Boulder, CO). Dried root samples were taken from the oven and loaded onto the sample holder. Root particles were firmly pressed in the holder and their reflectance spectra (350-2500nm) were acquired. The raw spectrum was truncated to 1003 to 2403 nm, and normalized using WINDAS software (Kemsley, 1998) through baseli ne correction and area normalization. NIR spectra were corrected with a full multiplicative scatter correction (MSC) model (X=(X-a)/b) using the Unscrambler software (Version 7.8 and 9.7 trial version, CAMO Software Inc., Woodbridge NJ). The MSC model wa s saved to allow future large scale screening of the entire pedigree Family UMD/UF-1 Pyrolysis-Mass Spectrometry The Pyrolysis-m ass spectrometer was a combin ation of a direct e xposure probe controller (Model PC-3, Scientific Instrument Services Inc., Ringoes, NJ) a nd Varian 1200 quadrupole mass spectrometer (Walnut Creek, CA). A small sample of ground root was placed onto the pyrolysis probe blade, and inserted into the pyrolysis chamber of a Varian 1200 quadrupole mass spectrometer. The root sample was pyrolyzed at the rate of 5.0 A/min to 1.5 A for 1.5 min. A cleaning procedure was used betw een every sample measurement ( 2.0 A at the heating rate of 5.0A/min for 1 min).
38 The mass spectrum of each root sample was recorded during the pyrolysis. The analysis was performed on an average mass spectrum, represented by the spectra acquired between the time points associated with 50% of the peak tota l ion current (TIC). The data for m/z 84 were excluded from the analysis because they app eared to represent background noise. Each mass spectrum was normalized based on its intensity divi ded by the total ion counts of the spectrum. A matrix was constructed that contained root sa mple names and normalized mass spectra. The mass spectra were processed with MS WorkStat ion Version 6.8 (Varian, Walnut Creek, CA). Mass spectra obtained after pyrolysis represen t the chemical composition profile of cell wall tissues. Six cell wall traits were estimated based on certain mass-to -charge values of the mass spectra obtained after pyrolysis. The m/z values used to quantity these six traits were listed in Table 4.1. Mass spectra of each genotype grown in either of the two nitrogen availabilities were acquired in triplicate. For each genotype, the va lue of six cell wall traits was calculated from each of the triplicate mass spectra, and the average was recorded as the trait value. Lignin Measurement Lignin content in root tissues was m easur ed with the Klason method (Schwanninger and Hinterstoisser, 2002). One hundred mg of root sample was weighed, and hydrolyzed in 1.5 mL of 12 M H2SO4 at 4 for 30 min, and then at 30 stirring at 100 rpm. The solution was diluted with 9.75 mL water to 1.6 M H2SO4 concentration and autoclaved at 121 for 1 hour. The reaction was then filtered through 4.25 cm Fisherbrand glass fiber filter circles (G6, Fisher Scientific Co., Pittsburgh, PA), and washed with 5 mL hot water three times. The filtrate with the glass fiber filter was then dried at 55 for 48 hours. The filtrate with the filter was weighed ( Weight1 ) and transferred into Isotemp Muffle Furnace (Fisher Scientific, Ltd, Pittsburgh, PA)
39 for 5 hours at 500 The residual and fiber filter were weighed again ( Weight2 ). The Klason lignin content for each sample was calculated by Weight1 Weight2. Statistical Data Analysis Near Infrared Spectra Processing and Multivariate Data Analysis Multiv ariate linear regression were performed in the Unscrambler software (Version 9.7 trial version, CAMO Software Inc., Woodbridge NJ). The whole data were centered with full model size. Partial least squares-1 (PLS1) a nd principal component regression (PCR) were performed by cross validation with uncertainty test using optimal numbers of principal components. All independent variates were a ssigned with equal wei ght of 1.0 during the regression processes. Quantitative Trait loci Analysis Quantita tive trait locus mapping was perf ormed using Windows QTL Cartographer Version 2.5 (Statistical Genetics, North Carolina State Universi ty, USA). Composite interval mapping method was employed with a LOD score threshold of statistical significance at =0.05, set by a permutation test using 1000 iterations. A precision selection of 2.0 cM was used as walk speed for all 19 chromosomes and all traits evaluated.
40 CHAPTER 3 CONSISTENCY TEST OF NEAR INFRARED SPECTRAL AND PYROLYSIS MASS SPECT ROMETRY ACQUISITION Near infrared spectroscopy is sensitive to environmental effects (Burns and Ciurczak, 2008). Variation could be observed from the NIR spect ra of different samples, and even the same sample measured at different time points. These variances in analysis were likely due to the fluctuation of moisture level in the lab and possi ble variation in temperature. NIR spectra also vary between different samples due to distinct part icle size. Therefore, in order to utilize the NIR spectra of root samples, the consistency of th e NIR spatial acquisition ne eded to be tested. The consistency of NIR spectral acquisition in the lab was confirmed by PCA analysis on NIR spectra obtained at f our different time points from 20 genotypes randomly selected from the population. From the PCA score plot (PC1XPC2), NIR spectra representing the same genotype were tightly clustered together. The PCA score plot showed that genotypes explained the most variation in NIR spectra within the 20-sample population. This confirmed the consistency of NIRS acquisition in the lab. The consistency wa s shown by PCA analysis of the NIR spectra obtained from three control samples UF493, UF 804 and UF840, as examples, grown under lownitrogen condition (Figure 3.1). The effect of nitrogen availability was visible based on the score plot obtained with PCA, showing how the NIR spectra representing these two growth conditions could be distinguished. The main difference between the two populations appeared to be due to the nitrogen treatment and explained 90% of variance (Figure 3.2). The process of mass spectral ac quisition also needed to be optimized. The quality of the mass spectra after pyrolysis was mainly affected by the electron energy during ionization and the detector voltage of mass spectrometer. The op timal setting of these two parameters were obtained by comparing mass spectral acquisition us ing combinations of two electron levels of
41 energy and two detector voltages. Combinations of two electron beam energylevels, 20eV and 70eV, and two detector voltages, 1200V and 1700V were utilized to obtain mass spectra from the 20 genotypes used to evaluate for NIRS cons istency. The higher the electron energy is, the more likely the ions will be fragmented, which reduce the m/z range. The detector voltage affects the signal-to-noise ratio. Mass spectra from these 20 samples showed that 20eV and 70eV yielded a similar degree of frag mentation. Therefore, the more typical 70eV was used as the electron energy in the subsequent experiments. Signal-to-noise ratio was compared between mass spectra obtained with detector voltage s of 1200V and 1700V. Utilization of 1700V as detector voltage easily led to detector saturati on. Therefore, 1200V was us ed in the subsequent experiments as the detector voltage setting. The consistency of mass spectrum acquisiti on after pyrolysis was confirmed by PCA analysis with mass spectra obtained from adven titious root tissues of two control genotypes, UF493 and UF804, grown under low nitrogen conditions. The mass spectrum of ground root sample of each genotype was acquired in triplicate. It took three days to obtain one replicate of root samples. Within each repli cate, these two control genotypes were incorporated to test the consistency of pyrolysis-mass spectrometer setti ngs. Principal component analysis with full cross-validation method plus uncertainty test of eight principal components (PCs) showed the consistency of mass spectral acquisition. A three-dimensional score plot (PC1(X) by PC2 (Y) by PC3 (Z)) showed the mass spectra clustered into two groups by genotype (Figure 3.3 A, B and C). A two-dimensional score plot by the firs t two principal components showed the first principal component explained al most 70% of total variance (F igure 3.3D), which is by the genotype.
42 Figure 3-1. Consistency of NIRS settings with analysis of NIR spectra of control root genotypes. (A) Plotting NIR spectra obtaine d from three genotypes at four time points. (B) Score plot of principal component analysis of th e NIR spectra. Scores were plotted with the first and second principal components. A B
43 Figure 3-2. Principal component an alysis of NIR spectra obtained from adventitious root tissues grown in lowand highnitrogen conditions. (A) Score pl ot of NIR spectra obtained from samples under two nitrogen regimes. (B) Cumulative explained variance of principal component numbers. (C) Load ing of principal component one. nm A B C Wavelength 1003 nm 2403 nm
44 Figure 3-3. Principal component analysis of mass spectra obtained from two control genotypes for consistency test. (UF493-3: Data 1-9; UF804-3, Data 10-18) (A) three dimensional plot by the first three principa l components. (B) two dimensional plot by the first two principal compone nts. (C) influential plot of mass spectra with residual x-variance and leverage. (D) cumulative explained variance by component numbers (calibration and va lidation curves). A D C B
45 CHAPTER 4 CHARACTERIZATION OF CELL W ALL TRAITS USING NIRS AND PYROLYSIS-MASS SPECTROMETRY COMBINED WITH MULTIVARIATE STATISTICAL ANALYSIS It is essential to examine the NIR spectra and mass spectra obtained from the entire sample population before generating NIR models to predic t the cell wall traits. Th is step identifies regions of NIR spectra with the most variation and the general fragmentation pattern in the mass spectra obtained after pyrolysis. A general view of the NIR spectra can also identify potential outliers. Near infrared spectra were obtained from ground adventitious root tissues as described. A general view of the NIR spectra was shown by ma trix plot in four formats. The bar format (Figure 4.1A) and landscape format (Figure 4.1B) de monstrated the general stacked view of raw spectra. Generally, all the NIR spectra obtained fo llowed the same pattern. However, in certain wavelength ranges of several samples, such as in samples UF251-3, UF493-3, UF587-3 and UF840-3, the intensities of NIR spectra had extrem e values compared to other samples. These could be better observed from the contour form at (Figure 4.1C) and map format (Figure 4.1D). While these variations may represent extreme values of certain cell wall traits, it is more likely that they may result from variations during NIR acquisition. The variation in acquisition was likely due to scattering effects due to different pa rticle size of root samples. The large variation would abolish the recognition sensitivity of subseque nt statistics analysis to the subtle changes of the rest of NIR spectra. The variations of reflectance recorded by the detector include two types of variations among the sample population, chemical differenc e and physical difference. A multiplicative scatter correction (MSC) process was used to minimize the effects of physical differences between the spectra data, such as the multiplicative and additive scattering effects which resulted from the particle size variation. Each NIR sp ectrum was plotted to the average spectrum
46 calculated from the entire spectra (Figure 4.2A). Correction coefficients for multiplicative ( a) and additive scattering ( b) effects were calculated individually using Unscrambler software and used in computation. In this plot, each NIR spect rum was adjusted in the plot based on these two parameters (using a to adjust intercept and b for slope) to generate its MSC-transformed spectrum (Figure 4.2B). Therefore, comparison of MSC-transformed NIR spectra could be used to reveal the differences of chemical component s more accurately and reliably. Matrix plots in bar (Figure 4.3A) and landscape (F igure 4.3B) formats generally demonstrated the overview of MSC-transformed NIR spectra. MSC transformation revealed subtle variations along the NIR range that were previously invisible. This wa s better observed from the contour (Figure 4.3C) and map formats (Figure 4.3D). MSC-transformed NIR spectra were used for the subsequent analysis. Mass spectra of the entire population were obtai ned after pyrolysis with the combination of electron energy at 70eV and detector voltage at 1200V. Intensities of mass-to-charge values of peak spectra were normalized. An overview of normalized mass spectra is shown in bar (Figure 4.4A) and landscape (Figure 4.4B) formats. Vari ation among the mass spectra was basically within the mass-to-charg e range of 55 to 210 as observed in the contour (Figure 4.4C) and map (Figure 4.4D) formats. The entire population was sorted ascending ba sed on the trait value for each of the six traits, and further split into th ree portions so that each portion would cover one third of specific trait value range. Two thirds of samples within each portion were randomly selected to construct the calibration set, while the rest samples were us ed as the validation set. The calibration set was used to generate the NIR multivariate linear regression model, and the validation set was used to test the derived model. NIR models were built based on MSC-transformed NIR spectra.
47 Partial least squares-1 (PLS1) and principa l component regression (PCR) methods were used and compared with each other to derive the multivariate NIR model to predict the trait value with full cross validation method and uncertainty test at the optimal PC number. The ideal NIR model would be the model that could be used reli ably, accurately and repr oducibly to predict the trait value. The ideal model shoul d be insensitive to any other ch anges, but very sensitive to the variation of this trait only. Thr ee criteria were used to dia gnose the derive d NIR model: (1) regression coefficient. In the predicted-measured value 2D scatter plot, the most accurate model would represent a straight regression line with predicted and measured trait values tightly clustered around the regression line. The higher the value of corre lation coefficient, the more sample variance is explained by the regr ession model. Correlation coefficient (R2 %) above 0.80 is desirable; (2) Root mean s quare error of calibrati on (RMSEC) and (3) root mean square error or prediction (RMSEP). These two parameters represent the NIR model performance. RMSEC and RMSEP are the average residuals between pred icted and measured values at the calibration and prediction stages, repectively. Therefore, th ey represent the calibra tion and prediction error individually. The lower of the values of RM SEC and RMSEP are, the more accurate the NIR model is. During the derivation of the NIR model from the modeling group, sample outliers were excluded in order to derive the pr oper NIR model. Currently there is no software to automatically identify outliers perfectly; therefore, the id entification was subjective and varied between different NIR procedures and esti mates for different traits. Four criteria were used to identify these outliers. (1) Influence plot with the optimal principal com ponent number. NIR spectra were plotted based on sample residual y-variances agai nst their leverage values, which characterize their influences on the model. Samples with high er residual variance were those that failed to
48 accurately predict the tra it value, and, were therefore potentia l outliers. The higher the leverage of a sample, the farther the projected sample woul d be from the mean value, and the harder it is for the model to describe this sample. Therefore, samples with higher leve rage tend to distort the model, decrease the prediction ability for normal sa mples, and thus may be potential outliers. (2) Score plot of principal component analysis (PCA). The samples th at do not cluster with most of the other samples in the PCA score plot of the mo st two PCs would likely to be outliers. (3) X-Y relation 2D scatter plot (t-u plot). Within a partial least squares1 (PLS1) model, the t-score was the score in x-matrix, while u-score was the one in y-matrix. The sa mples that reside farther from the regression line in scatter plots of most PCs tend to be potential outlie rs since they couldnt predict the X-Y relation relatively. (4) Predicted-Measured 2D scatter plot. The farther a sample lies from the regression line, the more likely it is an outlier. However, these potential outliers cannot be simply deleted from the modeling sin ce they may represent the samples with extreme value of that trait. In the validation step, once one sample has poor prediction ability, then it would be added to the previous calibration set, and new NIR models would be developed. Only if the potential outlier was not with extrem e value, it was delete d from model building. NIRS models were developed to accurately de termine these six cell wall traits in poplar adventitious roots using PLS1 and PCR (Figure). These two regr ession methods showed similar abilities to determine these traits, however, PLS1 method generally us ed fewer independent variates (wavelengths) to derive the regression models. This is consisted with the mathematics processes of PCR procedure which uses pr incipal components which are consisted of wavelengths. For traits syringyl lignin monomer content, guai acyl lignin monomer content and total carbohydrate content determ ination, PLS1 showed slightly advantages to PCR method.
49 While, PCR method gives better prediction to determine total carbohydr ate content (Figure 4.14). Total lignin content of a subpopulation of ca libration set was obtai ned through the Klason method(Theander and Westerlund, 1986). Plot between the values of Klason lignin and Predicted lignin calculated from mass spectra was shown in Figure 4.19. For the lignin/carbohydrate conten t ratio trait, neither of th e PLS1 nor PCR method yields acceptable NIR models using MSC-transformed NIR spectra and the trait value (Figure 4.15, 4.16). Even though both of the correlation coefficient (R2) exceeded 0.92, the RMSEC and RMSEP values were higher than 0.05 (Figur e 4.15A, B and 4.16A, B). This suggested the prediction models tend to be over-f itted. It is likely that unifor m noise, which was not readily in the validation set, was generally taken into the modeling building. This also reflects the possibility that there is not a straight linear relationship existed between the MSC-transformed NIR spectra with the trait value. Therefore, a logarithm transformation of the trait value successfully yielded an acceptable NIR model to characterize the li gnin/carbohydrate content ratio with PLS1 method (Figure 4.17). In conclusion, NIRS could be used to accura tely determine the six cell wall traits in adventitious root cell wall materials from poplar UMD/UF-1 family. It is feasible to use the NIR models derived in this research to determine th e cell wall trait values in the entire pedigree in order to determine the genetic mechanisms of carbon sequestration and partitioning in poplar roots.
50 Table 4-1. Cell wall traits characterized with mass spectrum obtained after pyrolysis. The m/z values used for quantificati on were listed respectively. Cell Wall Traits m/z Values Used For Characterization Syringyl Lignin Monomer Content 154, 167, 168, 182, 194, 210 Guaiacyl Lignin Monomer Content 124, 137, 138, 150, 164, 178 Syringyl-/Guaiacyl Lignin Monomer Content Ratio (Syringyl Lignin Monomer Content)/(Guaiacyl Lignin Monomer Content) Total Lignin Content (Syringyl Lignin Monomer Content) + (Guaiacyl Lignin Monomer Content) Total Carbohydrate Content 85, 96, 98, 114, 126, 144 Lignin/Carbohydrate Content Ratio (Total Lignin Content)/(Total Carbohydrate Content)
51 Figure 4-1. Diagrams of NIR sp ectra acquired in (A) bar, (B) landscape, (C) contour and (D) map formats. Statistics of the spectra were shown in (E) mean and standard derivation and (F) percentiles. A B C D E F
52 Figure 4-2. Plots showing scatte ring effects. Each spectrum was plotted with the average spectrum. (A) before MSC-transformation. (B) after MSC-transformation. A B
53 Figure 4-3. Diagrams of NIR spectra acquired after MSC tran sformation in (A) bar, (B) landscape, (C) contour and (D) map formats. Statistics of the sp ectra were shown in (E) mean and standard deri vation and (F) percentiles. A B C D E F
54 Figure 4-4. Diagrams of mass sp ectra obtained after pyrolysis in (A) bar, (B) landscape, (C) contour and (D) map formats. A B C D
55 Figure 4-5. Prediction of syringyl lignin monomer content with PLS1 method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
56 Figure 4-5. (Continued) C D
57 Figure 4-6. Prediction of syri ngyl lignin monomer content with PCR method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A
58 Figure 4-6. (Continued). C D
59 Figure 4-7. Prediction of guaiacy l lignin monomer content with PLS1 method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
60 Figure 4-7. (Continued). C D
61 Figure 4-8. Prediction of guai acyl lignin monomer content with PCR method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
62 Figure 4-8. (Continued). C D
63 Figure 4-9. Prediction of syri ngyl-/guaiacyl lignin ratio with PLS1 method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D) B A
64 Figure 4-9. (Continued). C D
65 Figure 4-10. Prediction of syringyl-/guaiacyl ligni n ratio with PCR method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
66 Figure 4-10. (Continued) C D
67 Figure 4-11. Prediction of tota l lignin content with PLS1 met hod with statistical parameters shown in the inner window. Model building at calibration stage wa s shown in (A) and (B), validation stage was shown in (C). Devi ation of samples in the prediction set was shown in (D). A B
68 Figure 4-11. (Continued). C D
69 Figure 4-12. Prediction of tota l lignin content with PCR method with statistical parameters shown in the inner window. Model building at calibration stage wa s shown in (A) and (B), validation stage was shown in (C). Devi ation of samples in the prediction set was shown in (D). A B
70 Figure 4-12. (Continued). C D
71 Figure 4-13. Prediction of to tal carbohydrate conten t with PLS1 method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
72 Figure 4-13. (Continued). C D
73 Figure 4-14. Prediction of to tal carbohydrate conten t with PCR method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
74 Figure 4-14. (Continued). C D
75 Figure 4-15. Prediction of li gnin/carbohydrate ratio with PL S1 method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
76 Figure 4-15. (Continued). C D
77 Figure 4-16. Prediction of li gnin/carbohydrate content with P CR method with statistical parameters shown in the inner window. Model building at ca libration stage was shown in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
78 Figure 4-16. (Continued). C D
79 Figure 4-17. Prediction of to tal lignin content with PLS1 method using the logarithmic transformed NIR spectra with statistical parameters shown in the inner window. Model building at calibration stage was show n in (A) and (B), validation stage was shown in (C). Deviation of samples in the prediction set was shown in (D). A B
80 Figure 4-17. (Continued). C D
81 Figure 4-18. Plot of klason lignin and predi cted lignin obtained from mass spectra. Inner window shows the regression equation betw een these two lignin contents with regression coefficient shown (x, predicted lignin content. y, klason lignin content).
82 CHAPTER 5 QUANTATITIVE TRAIT LOCI DETERMINA TION FOR POPLAR ROOT CELL WALL TRAIT VARIATION Tree deve lopment is regulated by the interac tion of genetic networks and environmental effects. Numerous studies have indicated the physiological traits are controlled by quantitative trait locus. These QTL regions were identified to explain variations in many phenotypic traits, such as biomass (Semel et al., 2006; Brewer et al., 2007; Collins et al., 2 008; Maccaferri et al., 2008), plant architecture (Wu et al., 2003; Loudet et al., 2005; Luquez et al., 2006), and flowering time (Botto and Co luccio, 2007; Lou et al., 2007). A high-density genetic linkage map with 458 markers and 167 microsatellites was constructed from the pseudobackcross pedigree, Family UMD/UF-1 in the Forest Genomics Lab at the University of Florida, Gainesvill e, Florida (Figure 5.1, by courtesy of Dr. Matias Kirst). Quantitative trait loci were predicted for the cell wall traits estimated from the mass spectra obtained after pyrolysis for two nitrogen availabilities. The QTL determined are listed with the information respectiv ely (shown in Table 5.1). For the cell wall trait variati on under low nitrogen condition, six chromosome regions were identified to regulate the cell wall traits (F igure 5.2). Only one chromosome region was identified to regulate syringyl lig nin monomer content in adventi tious root tissues along the 19 poplar chromosomes (Figure 5.3). This region resides in Linkage Group (LG)-14 between markers rs1554 and G1866 (Figure 5.4). Two chromosome regions were identified for total carbohydrate content trait (Figure 5.5), locating LG-1 (Figur e 5.6) and -18 (Figure 5.7) under low nitrogen condition. The chromosome regi on between markers P2221 and rG2281 in LG-6 was revealed to regulate lignin/carbohydrate content ratio (Fi gure 5.8, 5.9). The guaiacyl lignin monomer content was regulated by one genetic re gion in LG-1 (Figure 5.10), and is different from the chromosome region which was found to re gulate total lignin conten t trait in the same
83 chromosome (Figure 5.11, 5.12 and 5.13). Overlappi ng QTL identified in this study for QTL determined for traits in the stem tissue from samples of the same pedigree. The chromosome region on LG-14 affecting the variation in syring yl lignin monomer content overlapped with the QTL previously identified for variation in leaf biomass. Another chromosome region in Linkage Group 10 which was identified for variation of li gnin/carbohydrate content ratio overlapped with QTL previously identified fo r internode num ber trait. For high nitrogen conditions, genome regions were also identified that regulate the six cell wall traits in adventitious root tissues (Figure 5.14). One region on LG-11 was found to regulate syringyl lignin monomer content (Figure 5.15, 5.16). Another region in LG-15 was shown to be the only region which is responsible for varia tion in guaiacyl lignin monomer content (Figure 5.17, 5.18). For the syringyl-/guaiacyl ratio, almost 15% variation could be explained by genes residing on LG-11 (Figure 5.19, 5.20). Only one genome region on LG-15 was identified to regulate total lignin content (Figure 5.21). This re gion is different from the region identified for the same trait under low nitrogen condition (F igure 5.22). Another genome region in LG-11 was also the only region for total ca rbohydrate content tra it (Figure 5.23). The gene(s) regulating this trait resides between markers G 3981 and S598, and explain almost 8% of the variation (Figure 5.24). For the cell wall tr ait lignin/carbohydrate content rati o, another region in LG-17 was identified. It explain almost 18% of the varia tion (Figure 5.25), and this region resides between markers E408 and rG880 (Figure 5.26).
84 Figure 5-1. Genetic map of 19 linkage groups. Po sitions of genetic markers are shown on the right of each chromosome, with distances in centimorgans (cM) on the left. (by courtesy of Dr. Matias Kirst)
85 Table 5-1. Quantitative trait loci determined fo r cell wall traits under low nitrogen condition in stem and root tissues. C6, six carbon cellulo se sugar. (Data of stem tissues were by courtesy of Forest Genomics Laboratory, University of Florida, Gainesville, FL) Quantitative Trait Loci Determined under Low Nitrogen Condition Trait Tissue Linkage Group Genetic Markers Maximum R2 Value Syringyl Lignin Monomer Content Root LG 14 rS554, G1866 0.22 Stem N/A N/A N/A Guaiacyl Lignin Monomer Content Root LG 1 G1782, rP2385 0.26 Stem LG 1 G3205, G834 0.05 LG 6 rE918, rG139 0.07 LG 13 G2577, G2218 0.075 LG 17 G641, G3580 0.062 Total Lignin Content Root LG 1 G3784, G947 0.25 Stem LG 6 rS613, G2126 0.04 LG 17 G641, G3580 0.07 Total Carbohydrate Content Root LG 1 G1782, rP2385 0.29 LG 18 O534, rS89 0.23 Stem (C6 ) LG 1 G3205, G834 0.08 LG 6 rS613, G2126 0.04 LG 6 G2126, E538 0.074 LG 18 G1089, rG1244 0.048 Lignin/Carbohydrate Content Ratio Root LG 6 P2221, rG2281 0.23 Stem (C6/Lignin) LG 1 G3205, G834 0.053 LG 6 rS613, G2126 0.044
86 Table 5-2. Quantitative trait loci determined for cell wall traits under high nitrogen condition in stem and root tissues. C6, six carbon cellulo se sugar. (Data of stem tissues were by courtesy of Forest Genomics Laboratory, University of Florida, Gainesville, FL) Quantitative Trait Loci Determined under High Nitrogen Condition Trait Tissue Linkage Group Genetic Markers Maximum R2 Value Syringyl Lignin Monomer Content Root LG 11 G3981, G3037 0.26 Stem LG 1 P2731, G1782 0.035 LG 10 G1946, rG938 0.034 Guaiacyl Lignin Monomer Content Root LG 15 rG1454, G4047 0.15 Stem LG 13 G2577, G2218 0.068 Total Lignin Content Root LG 15 rG1454, G4.47 0.15 Stem LG 17 G641, P648 0.05 LG13 G2577, G2218 0.096 Syringyl /Guaiacyl Lignin Ratio Root LG 11 G3981, S598 0.17 Stem LG 1 P2731, G1782 0.039 LG 10 G1946, rG93 8 0.043 LG 14 P2515, G674 0.028 LG 15 G4047, rP2585 0.05 Total Carbohydrate Content Root LG 11 G3981, S598 0.06 Stem (C6) LG 1 P575, P2786b 0.075 LG 6 E538, G3600 0.041 LG 13 G2577, G2218 0.112 Lignin/Carbohydrate Content Rati o Root LG 17 E408, rG880 0.2 Stem (C6/Lignin) LG 1 P575, P2786b 0.082 LG 13 G2577, G2218 0.102 LG 17 G641, P648 0.048
87 Figure 5-2. LOD score plot of overview of quantitative trait loci analysis fo r 6 cell wall traits under low nitrogen condition along the 19 poplar linkage groups with R2 value shown at the bottom.
Figure 5-3. LOD score plot of quantitative trait loci analysis for syringyl lignin monomer content under low nitrogen condition tra it along the 19 poplar linkage groups with R2 value shown at the bottom.
89 Figure 5-4. LOD score plot of quantitative trait loci analysis for syringyl lignin monomer content trait, under low nitrogen cond ition, on linkage group 14 (LG-14) with R2 value shown at the bottom.
90 Figure 5-5. LOD score plot of qua ntitative trait loci analysis fo r total carbohydrate content trait, under low nitrogen condition, with R2 value shown at the bottom.
91 Figure 5-6. LOD score plot of qua ntitative trait loci analysis fo r total carbohydrate content trait, under low nitrogen condition, on li nkage group 1 (LG-1) with R2 value shown at the bottom.
92 Figure 5-7. LOD score plot of qua ntitative trait loci analysis fo r total carbohydrate content trait, under low nitrogen condition, on linkage group 18 (LG-18) with R2 value shown at the bottom.
93 Figure 5-8. LOD score plot of quantitative trait loci analysis for lignin/carbohydrate content ratio trait along the 19 poplar linkage gr oups under low nitrogen condition with R2 value shown at the bottom.
94 Figure 5-9. LOD score plot of quantitative trait loci analysis for lignin/carbohydrate content ratio trait, under low nitrogen cond ition, on linkage group 6 (LG-6) with R2 value shown at the bottom.
95 Figure 5-10. LOD score plot of quantitative tr ait loci analys is for guaiacyl lignin monomer content trait, under low nitrogen condition, with R2 value shown at the bottom.
96 Figure 5-11. LOD score plot of quantitative tr ait loci analys is for guaiacyl lignin monomer content trait, under low nitrogen cond ition, on linkage group 1 (LG-1) with R2 value shown at the bottom.
97 Figure 5-12. LOD score plot of quantitative trait loci analysis for total lignin monomer content trait, under low nitrogen condition with R2 value shown at the bottom.
98 Figure 5-13. LOD score plot of quantitative trait loci analysis for total lignin monomer content trait, under low nitrogen condition, on linkage group 1 (LG-1) with R2 value shown at the bottom.
99 Figure 5-14. LOD score plot of quantitative trait loci analysis for 6 cell wa ll traits, under high n itrogen condition, on 19 l inkage groups with R2 value shown at the bottom.
100 Figure 5-15. LOD score plot of quantitative tr ait loci analys is for syringyl lignin monomer content trait, under high nitrogen condition, on 19 linkage groups with R2 value shown at the bottom.
101 Figure 5-16. LOD score plot of quantitative tr ait loci analys is for syringyl lignin monomer content trait, under high nitrogen cond ition, on linkage group-11 (LG-11) with R2 value shown at the bottom.
102 Figure 5-17. LOD score plot of quantitative tr ait loci analys is for guaiacyl lignin monomer content trait, under high nitrogen condition, on 19 linkage groups with R2 value shown at the bottom.
103 Figure 5-18. LOD score plot of quantitative tr ait loci analys is for guaiacyl lignin monomer content trait, under high nitrogen cond ition, on linkage group-15 (LG-15) with R2 value shown at the bottom.
104 Figure 5-19. LOD score plot of quantitative tr ait loci analys is for syringyl/guaiacyl lignin monomer content ratio trait, under high nitrogen condition, on 19 linkage groups with R2 value shown at the bottom.
105 Figure 5-20. LOD score plot of quantitative tr ait loci analys is for syringyl/guaiacyl lignin monomer content ratio trait, under high n itrogen condition, on linkage group-11 (LG11) with R2 value shown at the bottom.
106 Figure 5-21. LOD score plot of quantitative trai t loci analysis for total lignin c ontent trait, under high nitrogen condition, on 19 linkage group with R2 value shown at the bottom.
107 Figure 5-22. LOD score plot of quantitative trai t loci analysis for total lignin c ontent trait, under high nitrogen condition, on linkage group-15 (LG-15) with R2 value shown at the bottom.
108 Figure 5-23. LOD score plot of quantitative tr ait loci analys is for total carbohydrate content trait, under high nitrogen condition, on 19 linkage groups with R2 value shown at the bottom.
109 Figure 5-24. LOD score plot of quantitative tr ait loci analys is for total carbohydrate content trait, under high nitrogen condition, on linkage group-11 (LG-11) with R2 value shown at the bottom.
110 Figure 5-25. LOD score plot of quantitative tr ait loci analys is for lignin/carbohydrate content ratio trait, under high nitrogen condition, on 19 linkage groups with R2 value shown at the bottom.
111 Figure 5-26. LOD score plot of quantitative tr ait loci analys is for lignin/carbohydrate content ratio trait, under high nitrogen cond ition, on linkage group-17 (LG-17) with R2 value shown at the bottom.
112 CHAPTER 6 DISCUSSION AND FUTURE WORK Since the 19th century, economic growth has resulted in large amounts of CO2 deposited into the atmosphere. This has induced climate cha nge also described as th e greenhouse effect. This resulted in an urgent call for a closer examina tion of the impact of CO2 emissions on global ecosystems, especially in terms of carbon and nitrogen cycles. In order to reduce carbon emissions, it is essential to examine the role of forests in terrestrial ecosystems and to enhance their carbon sink strength. Besides the effects of climate change on ecosystems, the current energy crisis resulting from the use of fossil oils also led to calls for alternative energy. Among all of the strategies, bioenergy appears as a pr omising strategy because it is renewable and is considered to have a small carbon footprint. To enhance the carbon storage and conversion efficiency of bioenergy feedstocks, it is also essential to examine the genetic mechanisms of carbon sequestration and partitioning in trees. Extensive efforts have been spent in order to understand the mechanisms of carbon allocation and sequestration in plan ts, especially cell wall biogenesi s, in Arabidopsis as well as other species. Looking back into th e strategies that have been utilized, they could be generalized into three types: (1) transcripti onal profiling (Pear et al., 1996; Dhugga et al., 2004; Samuga and Joshi, 2004; Aspeborg et al., 2005; Geisler-Lee et al., 2006; Cocuron et al., 2007; Burton et al., 2008), (2) heterologous expression (Liepman et al., 2005; Burton et al., 2006) and (3) genetic studies, including forward and reverse genetic sc reening (Favery et al., 2001; Wang et al., 2001; McCarty et al., 2005; Sindhu et al ., 2007) and population genetics (T sarouhas et al., 2002; Bao et al., 2007; Keurentjes et al., 2008; Lisec et al., 2008). The release of complete genome of Populus trichocarpa made poplar as an important model plant and the first tree model ready for investigation of molecular and genetics
113 mechanisms. Studies on the genetic mechanisms of carbon sequestration and allocation in poplar can help improve the role of trees as carbon sinks and as bioenergy feedstocks. The poplar pseudo-backcross pedigree, UMD/UF-1 maintained at the University of Florida (Gainesville, FL) has already been utilized to reveal genetic mechanisms of various traits, such as leaf area and cell wall characteristics in stem tissues. However, in order to fully elucidate the mechanisms of car bon sequestration and allocation besides the knowledge in stem tissues, the genetic networks regulating carbon flux into root tissues are yet to be revealed. Therefore, this calls for a method to characterize root cell wall traits reliably, accurately, rapidly and reproducibly. The first part of this study investigated the use of NIRS as a method to quantify six cell wall traits in adventitious root tissues that were based on a combination of pyrolysis mass spectrometry and wet chemical analysis. A s ubpopulation was selected from the entire UMD/UF-1 pedigree based on its genetic diversity. It is assumed that the selected population with the most diverse genetic variation would cover most of most phe notypic variation in the pedigree. In the first part of th is study, NIRS was proven to be ab le to predict the trait values both qualitively and quantitatively. Combing the trait values determined from mass spectra acquired after pyrolysis from plants grown under two nitrogen regimens, the correlation coefficients of prediction models for all the tr aits exceeded 0.85, with th e error of prediction all within 0.4%. With the regression methods of multivariate analysis, NIRS has a better prediction ability towards the traits, such as total carbohydrate content, tota l lignin content and lignin/carbohydrate ratio. For th e prediction of traits syringyland guaiacyl content and their ratio, the correlation coefficients are smaller th an other traits. All the models derived were validated by an independent sample set. The quality of validation and the prediction ability were
114 examined by the RMSEP value, which were all below 0.4%. In conclusion, the NIRS models derived could be utilized to quantify and predict these six root cell wall traits accurately and reliably. The first part of this study also compared two regression methods to derive the NIRS models, PLS1 and PCR methods. Generally, both methods gave si milar quantitative prediction ability for all six traits. The PCR method showed relative advantages in th e prediction of syringyl content, whereas PLS1 method exceeded in pr edicting guaiacyl content and syringyl/guaiacyl ratio. This was evaluated by the RMSEP of the validation set. During validation of the NIR model, the models with simila r correlation coefficients but hi gher RMSEP suggested over-fitting in this model. When predicting the same trait, the PCR method always utilizes more independent variates to derive the model with similar prediction abil ity as the PLS1 method. This is consistent with the mathematical processing of PCR me thod, since it takes the principal components consisting of combinations of wavelengths to build the model. In summary, both of the multivariate methods, PLS1 and PCR, showed accura te prediction ability to derive NIR model to quantify these six root cell wall traits. When predicting the trait of lignin/carbohydrat e ratio, the models derived with PLS1 and PCR methods using native MSC-transformed NIR sp ectra showed correlation coefficients both above 0.92. However, the errors of prediction (RMSEPs) in the va lidation step are both too big to consider the corresponding NIR model useful. The logarithmic transf ormation of the data resulted in a decrease in the RMSEPs from 0.068 to 0.045, which is within the acceptable range, and with a similar correlation co efficient (0.94) as before. This suggested either there is no straight linear relationship between the native MSC-transf ormed NIR spectra and the
115 lignin/carbohydrate ratio, or the logarithm tran sformation procedure mitigated the uniform noise which is in the calibration set but absent from validation set. The NIR models derived in this study could be feasibly utilized to predict trait values of the remaining samples from the UMD/UF-1 pedigree. The NIR models have been saved, and protocols are available for the mathematical proc essing of raw NIR spectra, such as NIR spectra truncation, MSC-transformation and PLS1/PCR m odel equations. Therefore, when using NIR spectra collected from new samples from the fam ily, all of the six root trait values can be automatically calculated using the NIR models derived from this study. This will greatly enhance large-scale phenotype screening. Since NIRS is a non-destructive method, the samples with predicted trait values could be selected for further validation. Several precautions should be taken into c onsideration when utilizing the NIR models derived in this study. First, the predicted value is not the absolute chemi cal value of each root cell wall trait. The models were developed based on values calculated from mass spectra acquired after sample pyrolysis. Once the relatio nship between these calculated values and the absolute chemical values is determined, either in linear or logarithmic relationship, the NIR models would be able to genera te absolute chemical trait va lues. Second, the models derived were based on strategies by addi ng non-fitting samples (outliers) of validation set into the previous calibration set until the most possible number of samples in the validation set fit the newly derived model. Currently, there is no universa l standard or software that can automatically identify these non-fitting samples. Therefore, the process used in th is study is somewhat subjective. There are different NI R models with similar prediction abilities. Third, when utilizing the NIR models derived in this study for larg e-scale screening, the pr ediction abilities are sensitive to experiment environment. This is due to the sensitivity of NIRS. Therefore, before
116 large-scale screening, NIR spectra from several samples used in this study and a number of the rest samples should be acquired in advance, in order to check the consistency for model usage. At the same time, chemical values of the samples should be determined from wet chemistry to examine the accuracy of NIR models. Extra efforts need to be spent to further determine the relationship between the Klason lignin content and the lignin content estimated from pyrolysis-mass spectra. The lack-of-fit relationship between these two values may be due to several reasons (Figure 4.18), such as smaller number of samples measured, or narro wer range of trait va riation. Therefore, incorporating more samples would determine the relationship between these two contents. It may be also resulted from the co-variation relationshi p between S/G ratio trait and total lignin trait. Since the higher G content in the sample, the more linkages would be in the samples, therefore, the higher G content results in less efficient pyrolysis, and leads to low lignin content estimation from the mass spectra. The second part of this study identified the chromosome regions al ong the poplar genome that could explain the variations in these six root cell wall traits under both nitrogen conditions. The QTLs identified were listed in Table 5.1. Co mpared to the values determined from mass spectra representing two nitrogen regimens, as th e nitrogen availability increases, the calculated values for syringyl lignin monomer content, guaiacyl lignin monomer content, total lignin content and syringyl/guaiacyl rati o all decrease. However, the calculated values of carbohydrate content as well as the lignin/car bohydrate ratio increase. The range in these trait values showed only limited overlap between both nitrogen cond itions. This indicated that the nitrogen availability affects the carbon allocation and se questration greatly, refl ecting the contrasting properties of root cell wall under these two nitrogen conditions. Also, the QTL identified for the
117 same trait were different under these two conditions. This also indicated that the nutrient availability in the environment could greatly affect regulatory gene expression, and there might be different regulatory genes th at are responsible for cell wa ll biogenesis under these two contrasting nitrogen conditions. Th is is consistent with other st udies on nitrogen availabilities (Cooke et al., 2005; Pitre et al ., 2007; Pitre et al., 2007). The QTL identified in this st udy for each cell wall tr ait are also different from the QTL determined for the same trait in stem tissues (by Forest Genomics Laboratory, University of Florida). This indicates distinct expression profiling in different tissues at same or different developmental stages, and tissue-specific re gulatory networks for carbon allocation and sequestration. This is consistent with the studies sugges ting tissue-specific expression levels of carbohydrate-active enzymes in poplar (Aspeborg et al., 2005; Geisler-Lee et al., 2006; Rolland et al., 2006) and tissue-specific biomass partitioning (Hancock et al., 2007). Chromosome regions have been identified wh ich are responsible for the variations of different traits of poplar adventi tious root cell wall. Therefore, ge ne(s) regulating these traits are likely located in these regions. However, ther e are hundreds of protein-encoding genes in each QTL. Knowledge of the expression profiling of adventitious root at the developmental stage at which the samples were collected for this st udy would help to narrow the list of potential candidate genes. In future study, the roles of candidate gene(s) regulating each trait could be identified either by reverse genetics, or the functions could be dete rmined by heterologous expression where there is no native cell wall biogenesis.
118 LIST OF REFERENCES Adler PR, Del Grosso SJ, Parton WJ (2007) Life-cycle assessm en t of net greenhouse-gas flux for bioenergy cropping systems. Eco Appl 17: 675-691 Alves A, Schwanninger M, Pereira H, Rodrigues J (2006) Calibration of NIR to assess lignin composition (H/G ratio) in maritime pine wood us ing analytical pyrolysis as the reference method. Holzforschung 60: 29-31 Andersen JR, Zein I, Wenzel G, Darnhofer B, Eder J, Ouzunova M, Lubberstedt T (2008) Characterization of phenylp ropanoid pathway genes within European maize ( Zea mays L.) inbreds. BMC Plant Biol 8: 2 Appenzeller L, Doblin M, Barreiro R, Wang HY Niu XM, Kollipara K, Carrigan L, Tomes D, Chapman M, Dhugga KS (2004) Cellulose synthesis in maize: isolation and expression analysis of the cellulose synthase ( CesA ) gene family. Cellulose 11: 287-299 Arabidopsis-Genome-Initiative (2000) Analysis of the genome sequence of the flowering plant Arabidopsis thaliana Nature 408: 796-815 Aspeborg H, Schrader J, Coutinho PM, Stam M, Kallas A, Djerbi S, Nilsson P, Denman S, Amini B, Sterky F, Master E, Sandberg G, Mellerowicz E, Sundberg B, Henrissat B, Teeri TT (2005) Carbohydrate-active enzymes involved in the secondary cell wall biogenesis in hybrid aspen. Plant Physiol 137: 983-997 Atalla RH, Vanderhart DL (1984) Native cellulose: a composite of two distin ct crystalline forms. Science 223: 283-285 Bai QL, Chen SJ, Dong XL, Meng QX, Yan YL, Dai JR (2004) Prediction of NDF and ADF concentrations with near in frared reflectance spectrosc opy (NIRS). Guang Pu Xue Yu Guang Pu Fen Xi 24: 1345-1347 Balatinecz JJ, Kretschmann DE, Leclercq A (2001) Achievements in the utilization of poplar wood guideposts for the future. Forestry Chron 77: 265-269 Bao JS, Jin L, Shen Y, Xie JK (2007) Genetic mapping of quantitat ive trait loci associated with fiber and lignin content in rice. Cereal Res Commu 35: 23-30 Baucher M, Monties B, Van Montagu M, Boerjan W (1998) Biosynthesis and genetic engineering of lignin. Crit Rev in Plant Sci 17: 125-197 Bell-Lelong DA, Cusumano JC, Meyer K, Chapple C (1997) Cinnamate-4-hydroxylase expression in Arabidopsis. Regulation in response to development and the environment. Plant Physiol 113: 729-738 Betz C, McCollum TG, Mayer RT (2001) Differential expression of two cinnamate 4hydroxylase genes in 'Valencia' orange ( Citrus sinensis Osbeck). Plant Mol Biol 46: 741748
119 Boerjan W, Ralph J, Baucher M (2003) Lignin biosynthesis. Annu Rev Plant Biol 54: 519-546 Bonan GB (2008) Forests and climate change: forcings feedbacks, and the climate benefits of forests. Science 320: 1444-1449 Botto JF, Coluccio MP (2007) Seasonal and plant-density de pendency for quantita tive trait loci affecting flowering time in multiple populations of Arabidopsis thaliana Plant Cell Environ 30: 1465-1479 Brett CT (2000) Cellulose microfibrils in plants: bi osynthesis, deposition, and integration into the cell wall. Int Rev Cytol 199: 161-199 Brewer MT, Moyseenko JB, Monforte AJ, van der Knaap E (2007) Morphological variation in tomato: a comprehensive study of quantitat ive trait loci controlling fruit shape and development. J Exp Bot 58: 1339-1349 Brown RM (1996) The biosynthesis of cellulose. Jour of Macro Sci-Pure and Appl Chem A33: 1345-1373 Brown RM (2004) Cellulose structure and biosynthesis : What is in store for the 21st century? Jourl of Polymer Sci Part A-Polymer Chem 42: 487-495 Brown RM, Jr., Montezinos D (1976) Cellulose microfibrils: visualization of biosynthetic and orienting complexes in association with the plasma membrane. Proc Natl Acad Sci U S A 73: 143-147 Burns DA, Ciurczak EW (2008) Handbook of Near-infrared analysis, Ed Third. CRC Press, Taylor & Francis Group, LLC, Boca Raton, FL Burton RA, Jobling SA, Harvey AJ, Shirley NJ, Mather DE, Bacic A, Fincher GB (2008) The genetics and transcri ptional profiles of the cellulose synthase-like HvCslF gene family in barley. Plant Physiol 146: 1821-1833 Burton RA, Shirley NJ, King BJ, Harvey AJ, Fincher GB (2004) The CesA gene family of barley. Quantitative analysis of transcripts reveals two groups of co-expressed genes. Plant Physiol 134: 224-236 Burton RA, Wilson SM, Hrmova M, Harvey AJ, Shirley NJ, Medhurst A, Stone BA, Newbigin EJ, Bacic A, Fincher GB (2006) Cellulose synthase-like CslF genes mediate the synthesis of cell wall (1,3;1,4)-beta-D-glucans. Science 311: 1940-1942 Canadell JG, Le Quere C, Raupach MR, Fiel d CB, Buitenhuis ET, Ciais P, Conway TJ, Gillett NP, Houghton RA, Marland G (2007) Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intens ity, and efficiency of natural sinks. Proc Natl Acad Sci U S A 104: 18866-18870 Carpita N, Vergara C (1998) A recipe for cellulose. Science 279: 672-673
120 Celen I, Harper D, Labbe N (2008) A multivariate approach to the acetylated poplar wood samples by near infrared spectroscopy. Holzforschung 62: 189-196 Cocuron JC, Lerouxel O, Drakakaki G, Alonso AP, Liepman AH, Keegstra K, Raikhel N, Wilkerson CG (2007) A gene from the cellulose synthase-like C family encodes a beta1,4 glucan synthase. Proc Natl Acad Sci U S A 104: 8550-8555 Coleman HD, Park JY, Nair R, Chapple C, Mansfield SD (2008) RNAi-mediated suppression of p-coumaroyl-CoA 3'-hydroxylase in hybrid poplar impacts lignin deposition and soluble secondary metabolism. Proc Natl Acad Sci U S A Collins NC, Tardieu F, Tuberosa R (2008) Quantitative trait loci and crop performance under abiotic stress: Where do we stand? Plant Physiol 147: 469-486 Cooke JE, Martin TA, Davis JM (2005) Short-term physiological and developmental responses to nitrogen availabil ity in hybrid poplar. New Phytol 167: 41-52 Cosgrove DJ (2005) Growth of the plant ce ll wall. Nat Rev Mol Cell Biol 6: 850-861 Costa MA, Collins RE, Anterola AM, Cochrane FC, Davin LB, Lewis NG (2003) An in silico assessment of gene function and organization of the phenylpropanoid pathway metabolic networks in Arabidopsis thaliana and limitations thereof. Phytochem 64: 1097-1112 Cutler S, Somerville C (1997) Cellulose synthesis: cloning in silico. Curr Biol 7: R108-R111 de Vernal A, Hillaire-Marcel C (2008) Natural variability of Greenland climate, vegetation, and ice volume during the past million years. Science 320: 1622-1625 Desprez T, Juraniec M, Crowell EF, Jouy H, Pochylova Z, Parcy F, Hofte H, Gonneau M, Vernhettes S (2007) Organization of cellulose synt hase complexes involved in primary cell wall synthesis in Arabidopsis thaliana Proc Natl Acad Sci U S A 104: 15572-15577 Dhugga KS, Barreiro R, Whitten B, Stecca K, Hazebroek J, Randhawa GS, Dolan M, Kinney AJ, Tomes D, Nichols S, Anderson P (2004) Guar seed beta-mannan synthase is a member of the cellulose synthase super gene family. Science 303: 363-366 Djerbi S, Lindskog M, Arvest ad L, Sterky F, Teeri TT (2005) The genome sequence of black cottonwood ( Populus trichocarpa) reveals 18 conserved cellulose synthase ( CesA ) genes. Planta 221: 739-746 Dunkley TP, Hester S, Shadforth IP, Runions J, Weimar T, Hanton SL Griffin JL, Bessant C, Brandizzi F, Hawes C, Watson RB, Dupree P, Lilley KS (2006) Mapping the Arabidopsis organelle proteome Proc Natl Acad Sci U S A 103: 6518-6523 Ebringerova A, Heinz e T (2000) Xylan and xylan derivatives biopolymers with valuable properties, 1 Naturally occurring xylans structures, procedures and properties. Macromol Rapid Commu 21: 542-556
121 Ehlting J, Mattheus N, Aeschliman DS, Li E, Hamberger B, Cullis IF, Zhuang J, Kaneda M, Mansfield SD, Samuels L, Ritland K, Ellis BE, Bohlmann J, Douglas CJ (2005) Global transcript profiling of primary stems from Arabidopsis thaliana identifies candidate genes for missing links in lignin bios ynthesis and transcriptional regulators of fiber differentiation. Plant J 42: 618-640 Ehlting J, Provart NJ, Werck-Reichhart D (2006) Functional annotation of the Arabidopsis P450 superfamily based on large-scale co -expression analysis. Biochem Soc Trans 34: 1192-1198 Emons AM (1985) Plasma-membrane rose ttes in root hairs of Equisetum Hyemale Planta 163: 350-359 Faix O (1986) Investigation of lignin polymer models (DHPS) by FTIR Spectroscopy. Holzforschung 40: 273-280 Fargione J, Hill J, Tilman D, Polasky S, Hawthorne P (2008) Land clearing and the biofuel carbon debt. Science 319: 1235-1238 Farrell AE, Plevin RJ, Turner BT, Jones AD, O'Hare M, Kammen DM (2006) Ethanol can contribute to energy and e nvironmental goals. Science 311: 506-508 Farrokhi N, Burton RA, Brownfield L, Hrmova M, Wilson SM, Bacic A, Fincher GB (2006) Plant cell wall biosynthesis: geneti c, biochemical and functional genomics approaches to the identification of key genes. Plant Biotechnol J 4: 145-167 Favery B, Ryan E, Foreman J, Linstead P, Boudonck K, Steer M, Shaw P, Dolan L (2001) KOJAK encodes a cellulose synthase-like pr otein required for root hair cell morphogenesis in Arabidopsis. Genes Dev 15: 79-89 Field CB, Campbell JE, Lobell DB (2008) Biomass energy: the scale of the potential resource. Trends in Ecol & Evo 23: 65-72 Franke R, Hemm MR, Denault JW, Ruegger MO, Humphreys JM, Chapple C (2002) Changes in secondary metabolism and deposition of an unusual lignin in the ref8 mutant of Arabidopsis. Plant J 30: 47-59 Franke WW, Ermen B (1969) Negative staining of plant slim e cellulose: an examination of the elementary fibril concept. Z Naturforsch B 24: 918-922 Frey-Wyssling A (1954) The fine structure of cellulose microfibrils. Science 119: 80-82 Fry SC, York WS, Albersheim P, Darvill A, H ayashi T, Joseleau JP, Kato Y, Lorences EP, Maclachlan GA, Mcneil M, Mort AJ, Reid JSG, Seitz HU, Selvendran RR, Voragen AGJ, White AR (1993) An unambiguous nomenclat ure for xyloglucan-derived oligosaccharides. Physiol Plant 89: 1-3
122 Fukushima RS, Hatfield RD (2004) Comparison of the acetyl bromide spectrophotometric method with other analytical lignin methods for determining lignin concentration in forage samples. J Agric Food Chem 52: 3713-3720 Gardiner JC, Taylor NG, Turner SR (2003) Control of cellulose synthase complex localization in developing xylem. Plant Cell 15: 1740-1748 Geisler-Lee J, Geisler M, Coutinho PM, Segerm an B, Nishikubo N, Takahashi J, Aspeborg H, Djerbi S, Master E, Andersson-Gunnera s S, Sundberg B, Karpinski S, Teeri TT, Kleczkowski LA, Henrissat B, Mellerowicz EJ (2006) Poplar carbohydrate-active enzymes. Gene identification and expression analyses. Plant Physiol 140: 946-962 Goff SA, Ricke D, Lan TH, Presting G, Wang R, Dunn M, Glazebrook J, Sessions A, Oeller P, Varma H, Hadley D, Hutchison D, Martin C, Katagiri F, Lange BM, Moughamer T, Xia Y, Budworth P, Zhong J, Miguel T, Paszkowski U, Zhang S, Colbert M, Sun WL, Chen L, Cooper B, Park S, Wood TC, Mao L, Quail P, Wing R, Dean R, Yu Y, Zharkikh A, Shen R, Sahasrabudhe S, Thom as A, Cannings R, Gutin A, Pruss D, Reid J, Tavtigian S, Mitchell J, Eldredge G, Scholl T, Miller RM, Bhatnagar S, Adey N, Rubano T, Tusneem N, Robinson R, Feldhaus J, Macalma T, Oliphant A, Briggs S (2002) A draft sequence of the rice genome (Oryza sativa L. ssp. japonica ). Science 296: 92-100 Ha MA, MacKinnon IM, Sturcova A, App erley DC, McCann MC, Turner SR, Jarvis MC (2002) Structure of cellulose-defic ient secondary cell walls from the irx3 mutant of Arabidopsis thaliana Phytochem 61: 7-14 Haigler CH, Brown RM (1986) Transport of rosettes from the Golgi-apparatus to the plasmamembrane in Isolated mesophyll cells of Zinnia elegans during differentiation to tracheary elements in suspension culture. Protoplasma 134: 111-120 Hall DA, Saxl H (1960) Human and other an imal cellulose. Nature 187: 547-550 Hamberger B, Ellis M, Friedmann M, Souza CDA, Barbazuk B, Douglas CJ (2007) Genome-wide analyses of phe nylpropanoid-related genes in Populus trichocarpa, Arabidopsis thaliana and Oryza sativa : the Populus lignin tool box and conservation and diversification of angiosperm gene families. Can J Bot-Revue Canadienne de Botanique 85: 1182-1201 Hancock JE, Loya WM, Giardina CP, Li L, Chiang VL, Pregitzer KS (2007) Plant growth, biomass partitioning and soil carbon formation in response to altered lignin biosynthesis in Populus tremuloides New Phytol 173: 732-742 Harding SA, Leshkevich J, Chiang VL, Tsai CJ (2002) Differential substrate inhibition couples kinetically distinct 4-coumarate:co enzyme a ligases with spatially distinct metabolic roles in quaking aspen. Plant Physiol 128: 428-438 Hatfield R, Fukushima RS (2005) Can lignin be accurately measured? Crop Sci 45: 832-839
123 Herth W (1983) Arrays of plasma-membrane rose ttes involved in cellulose microfibril formation of Spirogyra. Planta 159: 347-356 Herth W (1985) Plasma-membrane rosettes involved in localized wall thickening during xylem vessel formation of Lepidium Sativum L. Planta 164: 12-21 Hill J, Nelson E, Tilman D, Polasky S, Tiffany D (2006) Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofue ls. Proc Natl Acad Sci U S A 103: 11206-11210 Holland N, Holland D, Helentjaris T, Dhu gga KS, Xoconostle-Cazares B, Delmer DP (2000) A comparative analysis of the plant cellulose synthase ( CesA ) gene family. Plant Physiol 123: 1313-1324 Holtman KM, Chang HM, Jameel H, Kadla JF (2003) Elucidation of lignin structure through degradative methods: Comparison of modifi ed DFRC and thioacidolysis. J Agric Food Chem 51: 3535-3540 Hu WJ, Harding SA, Lung J, Popko JL, Ra lph J, Stokke DD, Tsai CJ, Chiang VL (1999) Repression of lignin biosynthesis promotes cellulose accumulation and growth in transgenic trees. Nat Biotechnol 17: 808-812 Imai T, Sugiyama J (1998) Nanodomains of I-alpha and I-beta cellulose in algal microfibrils. Macromol 31: 6275-6279 Jacob-Wilk D, Kurek I, Hogan P, Delmer DP (2006) The cotton fiber zi nc-binding domain of cellulose synthase A1 from Gossypium hirsutum displays rapid turnover in vitro and in vivo Proc Natl Acad Sci U S A 103: 12191-12196 Jamet E, Albenne C, Boudart G, Irshad M, Canut H, Pont-Lezica R (2008) Recent advances in plant cell wall proteomics. Proteomics 8: 893-908 Jansson S, Douglas CJ (2007) Populus: A model system fo r plant biology. Annu Rev Plant Biol 58: 435-458 Jarvis MC (2000) Interconversion of the I alpha and I beta crystalline forms of cellulose by bending. Carbohydr Res 325: 150-154 Jones JK, Purves CB, Timell TE (1961) Constitution of a 4-0-methylglucuronoxylan from wood of trembling aspen ( Populus tremuloides Michx). Can J Chem -Revue Canadienne de Chimie 39: 1059-& Jones L, Ennos AR, Turner SR (2001) Cloning and characterization of irregular xylem4 ( irx4 ): a severely lignin-deficient mutant of Arabidopsis. Plant J 26: 205-216 Joshi CP, Bhandari S, Ranjan P, Kalluri UC, Liang X, Fujino T, Samuga A (2004) Genomics of cellulose biosynthesis in poplars. New Phytol 164: 53-61
124 Jumanova Z, Sdykov T, Seitmuratov E, Dalimova GN (2006) Lignins from Oryza sativa. Chem Nat Comp 42: 724-726 Kao YY, Harding SA, Tsai CJ (2002) Differential expression of two distinct phenylalanine ammonia-lyase genes in condensed tannin-accumulating and lignifying cells of quaking aspen. Plant Physiol 130: 796-807 Keegstra K, Walton J (2006) Plant science. Beta-glucans-brewer's bane, dietician's delight. Science 311: 1872-1873 Kemsley EK (1998) Discrimintant analysis and class modelling of spectroscopic data. In John Wiley and Sons Ltd., Chichester, UK Keurentjes JJ, Koornneef M, Vreugdenhil D (2008) Quantitative genetics in the age of omics. Curr Opin Plant Biol 11: 123-128 Kimura S, Itoh T (1995) Evidence for the role of the glom erulocyte in cellulose synthesis in the tunicate, Metandrocarpa uedai Protoplasma 186: 24-33 Kimura S, Laosinchai W, Itoh T, Cui X, Linder CR, Brown RM, Jr. (1999) Immunogold labeling of rosette terminal cellulose-synthesizing complexes in the vascular plant Vigna angularis Plant Cell 11: 2075-2086 Kirst M, Johnson AF, Baucom C, Ulrich E, Hubbard K, Staggs R, Paule C, Retzel E, Whetten R, Sederoff R (2003) Apparent homology of expressed genes from woodforming tissues of loblolly pine ( Pinus taeda L.) with Arabidopsis thaliana Proc Natl Acad Sci U S A 100: 7383-7388 Koutaniemi S, Warinowski T, Karkonen A, Alatalo E, Fossdal CG, Saranpaa P, Laakso T, Fagerstedt KV, Simola LK, Paulin L, Rudd S, Teeri TH (2007) Expres sion profiling of the lignin biosynthetic pathway in Norway spruce using EST sequencing and real-time RT-PCR. Plant Mol Biol 65: 311-328 Koyama M, Helbert W, Imai T, Sugiyama J, Henrissat B (1997) Parallel-up structure evidences the molecular direc tionality during biosynthesis of bacterial cellulose. Proc Natl Acad Sci U S A 94: 9091-9095 Kumar A, Ellis BE (2001) The phenylalanine ammonia-lyase gene family in raspberry. Structure, expression, and evolution. Plant Physiol 127: 230-239 Kurek I, Kawagoe Y, Jacob-W ilk D, Doblin M, Delmer D (2002) Dimerization of cotton fiber cellulose synthase catalytic subunits occurs via oxidation of the zinc-binding domains. Proc Natl Acad Sci U S A 99: 11109-11114 Lange BM, Lapierre C, Sandermann H, Jr. (1995) Elicitor-induced spruce stress lignin (stru ctural similarity to early developmental lignins). Plant Physiol 108: 1277-1287
125 Lenoir J, Gegout JC, Marquet PA, de Ruffray P, Brisse H (2008) A significant upward shift in plant species optimum elevati on during the 20th century. Science 320: 1768-1771 Lepl JC, Dauwe R, Morreel K, Storme V, Lapierre C, Pollet B, Naumann A, Kang KY, Kim H, Ruel K, Lefebvre A, Joseleau JP, Grima-Pettenati J, De Rycke R, Andersson-Gunneras S, Erban A, Fehrle I, Petit-Conil M, Kopka J, Polle A, Messens E, Sundberg B, Mansfield SD Ralph J, Pilate G, Boerjan W (2007) Downregulation of cinnamoyl-coenzyme a reductase in poplar: Multiple-level phenotyping reveals effects on cell wall poly mer metabolism and structure. Plant Cell 19: 3669-3691 Li LG, Lu SF, Chiang V (2006) A genomic and molecular view of wood formation. Crit Rev in Plant Sci 25: 215-233 Liepman AH, Wilkerson CG, Keegstra K (2005) Expression of cellulose synthase-like ( Csl ) genes in insect cells reveals that CslA family members encode mannan synthases. Proc Natl Acad Sci U S A 102: 2221-2226 Lisec J, Meyer RC, Steinfath M, Redestig H, Becher M, Witucka-Wall H, Fiehn O, Torjek O, Selbig J, Altmann T, Willmitzer L (2008) Identification of metabolic and biomass QTL in Arabidopsis thaliana in a parallel analysis of RIL and IL populations. Plant J 53: 960-972 Lou P, Zhao J, Kim JS, Shen S, Del Carpio DP, Song X, Jin M, Vreugdenhil D, Wang X, Koornneef M, Bonnema G (2007) Quantitative trait loci for flowering time and morphological traits in multiple populations of Brassica rapa. J Exp Bot 58: 4005-4016 Loudet O, Gaudon V, Trubuil A, Daniel-Vedele F (2005) Quantitative tra it loci controlling root growth and architecture in Arabidopsis thaliana confirmed by heterogeneous inbred family. Theor Appl Genet 110: 742-753 Lu F, Ralph J (1999) Detection and determination of pcoumaroylated units in lignins. J Agric Food Chem 47: 1988-1992 Lu F, Ralph J (2002) Preliminary evidence for sinapyl acetate as a lignin monomer in kenaf. Chem Commun (Camb) : 90-91 Lu FC, Ralph J (1997) Derivatization followed by reductive cleavage (DFRC method), a new method for lignin analysis: protocol for anal ysis of DFRC monomers J Agric Food Chem 45: 2590-2592 Lundquist K, Kirk TK (1971) Acid degradation of lignin. IV. Analysis of lignin acidolysis products by gas chromatography, using trimet hylsilyl derivatives. Acta Chem Scand 25: 889-894 Luquez VM, Sasal Y, Medrano M, Martin MI, Mujica M, Guiamet JJ (2006) Quantitative trait loci analysis of leaf and plant longevity in Arabidopsis thaliana J E xp Bot 57: 13631372
126 Maccaferri M, Sanguineti MC, Corneti S, Ortega JL, Salem MB, Bort J, DeAmbrogio E, del Moral LF, Demontis A, El-Ahmed A, Maalouf F, Machlab H, Martos V, Moragues M, Motawaj J, Nachit M, Nsera llah N, Ouabbou H, Royo C, Slama A, Tuberosa R (2008) Quantitative trait loci for grai n yield and adaptation of durum wheat ( Triticum durum Desf.) across a wide range of water availability. Genetics 178: 489-511 Matthews JF, Skopec CE, Mason PE, Zuccato P, Torget RW, Sugiyama J, Himmel ME, Brady JW (2006) Computer simulation studies of microcrystalline cellulose I beta. Carbohydr Res 341: 138-152 McCann MC, Carpita NC (2005) Looking for invisible phe notypes in cell wall mutants of Arabidopsis thaliana Plant Biosys 139: 80-83 McCann MC, Defernez M, Urbanowicz BR, Tewa ri JC, Langewisch T, Olek A, Wells B, Wilson RH, Carpita NC (2007) Neural network analys es of infrared spectra for classifying cell wall architectures. Plant Physiol 143: 1314-1326 McCann MC, Wells B, Roberts K (1990) Direct visualization of cross-links in the primary plant cell wall. J Cell Sci 96: 323-334 McCartney L, Blake AW, Flint J, Bola m DN, Boraston AB, Gilbert HJ, Knox JP (2006) Differential recognition of plant cell wa lls by microbial xylan-specific carbohydratebinding modules. Proc Natl Acad Sci U S A 103: 4765-4770 McCartney L, Gilbert HJ, Bola m DN, Boraston AB, Knox JP (2004) Glycoside hydrolase carbohydrate-binding modules as molecular probes for the analysis of plant cell wall polymers. Anal Biochem 326: 49-54 McCarty DR, Settles AM, Suzuki M, Tan BC Latshaw S, Porch T, Robin K, Baier J, Avigne W, Lai J, Messing J, Koch KE, Hannah LC (2005) Steady-state transposon mutagenesis in inbred maize. Plant J 44: 52-61 Mcdougall GJ, Morrison IM, Stewar t D, Weyers JDB, Hillman JR (1993) Plant fibers botany, chemistry and processing for industrial use. J Sci Food Agri 62: 1-20 Mellerowicz EJ, Baucher M, Sundberg B, Boerjan W (2001) Unravelling cell wall formation in the woody dicot st em. Plant Mol Biol 47: 239-274 Moller I, Sorensen I, Bernal AJ, Blaukopf C, Lee K, Obro J, Pettolino F, Roberts A, Mikkelsen JD, Knox JP, Bacic A, Willats WG (2007) High-throughput mapping of cell-wall polymers within and between pl ants using novel microarrays. Plant J 50: 11181128 Morreel K, Ralph J, Kim H, Lu FC, Goeminne G, Ralph S, Messens E, Boerjan W (2004) Profiling of oligolignols reveals monolignol coupling conditions in lignifying poplar xylem. Plant Physiol 136: 3537-3549
127 Nair RB, Xia Q, Kartha CJ, Kurylo E, Hirji RN, Datla R, Selvaraj G (2002) Arabidopsis CYP98A3 mediating aromatic 3-hydroxylation. Developmenta l regulation of the gene, and expression in yeast. Plant Physiol 130: 210-220 Nobles DR, Romanovicz DK, Brown RM, Jr. (2001) Cellulose in cyanobacteria. Origin of vascular plant cellulose synthase? Plant Physiol 127: 529-542 Northcot D (1972) Chemistry of plant-cell wall. Ann Rev Plant Physiol 23: 113-132 Osakabe Y, Ohtsubo Y, Kawai S, Katayama Y, Morohoshi N (1995) Structure and tissuespecific expression of genes for phenylalanine ammonia-lyase from a hybrid aspen, Populus kitakamiensis Plant Sci 105: 217-226 Paredez AR, Somerville CR, Ehrhardt DW (2006) Visualization of cellulose synthase demonstrates functional associa tion with microtubules. Science 312: 1491-1495 Pear JR, Kawagoe Y, Schrecken gost WE, Delmer DP, Stalker DM (1996) Higher plants contain homologs of the bacterial celA genes encoding the catalytic subunit of cellulose synthase. Proc Natl Acad Sci U S A 93: 12637-12642 Pearson PN, Palmer MR (2000) Atmospheric carbon dioxide concentrations over the past 60 million years. Nature 406: 695-699 Peng L, Kawagoe Y, Hogan P, Delmer D (2002) Sitosterol-beta-glucoside as primer for cellulose synthesis in plants. Science 295: 147-150 Petisco C, Garcia-Criado B, Mediavilla S, Vazquez de Al dana BR, Zabalgogeazcoa I, Garcia-Ciudad A (2006) Near-infrared reflectance spectroscopy as a fast and nondestructive tool to predict foliar organic constituents of several woody species. Anal Bioanal Chem 386: 1823-1833 Petit JR, Jouzel J, Raynaud D, Barkov NI, Ba rnola JM, Basile I, Bender M, Chappellaz J, Davis M, Delaygue G, Delmotte M, Kotlyakov VM, Legrand M, Lipenkov VY, Lorius C, Pepin L, Ritz C, Saltzman E, Stievenard M (1999) Climate and atmospheric history of the past 420,000 years from th e Vostok ice core, Antarctica. Nature 399: 429436 Pettolino FA, Hoogenraad NJ, Ferguson C, Bacic A, Johnson E, Stone BA (2001) A (1 4)beta-mannan-specific monoclonal antibody a nd its use in the immunocytochemical location of galactomannans. Planta 214: 235-242 Philippe S, Robert P, Barron C, Saulnier L, Guillon F (2006) Deposition of cell wall polysaccharides in wheat endosperm duri ng grain development: Fourier transforminfrared microspectroscopy study. J Agric Food Chem 54: 2303-2308 Philippe S, Tranquet O, Utille JP, Saulnier L, Guillon F (2007) Investigation of ferulate deposition in endosperm cell walls of mature and developing wheat grains by using a polyclonal antibody. Planta 225: 1287-1299
128 Piao SL, Ciais P, Friedlingstein P, Peylin P, Reichstein M, Luyssaert S, Margolis H, Fang JY, Barr A, Chen AP, Grelle A, Hollinger DY, Laurila T, Lindroth A, Richardson AD, Vesala T (2008) Net carbon dioxide losses of northern ecosystems in response to autumn warming. Nature 451: 49-U43 Pitre FE, Cooke JEK, Mackay JJ (2007) Short-term effects of nitrogen availability on wood formation and fibre properties in hybrid poplar. Trees-Struc Func 21: 249-259 Pitre FE, Pollet B, Lafarguette F, Cooke JE, MacKay JJ, Lapierre C (2007) Effects of increased nitrogen supply on the lignificat ion of poplar wood. J Agric Food Chem 55: 10306-10314 Plomion C, Leprovost G, Stokes A (2001) Wood formation in trees. Plant Physiol 127: 15131523 Preston RD, Wardrop AB, Nicolai E (1948) Fine structure of cell walls in fresh plant tissues. Nature 162: 957-959 Raes J, Rohde A, Christensen JH, Van de Peer Y, Boerjan W (2003) Genome-wide characterization of the lignification toolbox in Arabidopsis. Plant Physiol 133: 1051-1071 Ralph J, Lapierre C, Marita JM, Kim H, Lu F, Hatfield RD, Ralph S, Chapple C, Franke R, Hemm MR, Van Doorsselaere J, Sedero ff RR, O'Malley DM, Scott JT, MacKay JJ, Yahiaoui N, Boudet A, Pean M, Pilate G, Jouanin L, Boerjan W (2001) Elucidation of new structures in lignins of CADand CO MT-deficient plants by NMR. Phytochem 57: 993-1003 Read SM, Bacic T (2002) Plant biology. Prime time for cellulose. Science 295: 59-60 Richmond TA, Somerville CR (2000) The cellulose synthase superfamily. Plant Physiol 124: 495-498 Richmond TA, Somerville CR (2001) Integrative approaches to determining Csl function. Plant Mol Biol 47: 131-143 Ro DK, Mah N, Ellis BE, Douglas CJ (2001) Functional characte rization and subcellular localization of poplar ( Populus trichocarpa x Populus deltoides ) cinnamate 4hydroxylase. Plant Physiol 126: 317-329 Rolland F, Baena-Gonzalez E, Sheen J (2006) Sugar sensing a nd signaling in plants: conserved and novel mechanisms. Annu Rev Plant Biol 57: 675-709 Roumet C, Picon-Cochard C, Dawson LA, Joffre R, Mayes R, Blanchard A, Brewer MJ (2006) Quantifying species composition in r oot mixtures using two methods: nearinfrared reflectance spectroscopy an d plant wax markers. New Phytol 170: 631-638 Samuga A, Joshi CP (2004) Cloning and characterization of cellulose synt hase-like gene, PtrCslD2 from developing xylem of aspen trees. Physiol Plant 120: 631-641
129 Saxena IM, Brown RM, Jr., Dandekar T (2001) Structure--functi on characterization of cellulose synthase: relationship to other glycosyltransferases. Phytochem 57: 1135-1148 Saxena IM, Brown RM, Jr., Fevre M, Geremia RA, Henrissat B (1995) Multidomain architecture of beta-glycosy l transferases: implications for mechanism of action. J Bacteriol 177: 1419-1424 Schultz TP, Templeton MC, Mcginnis GD (1985) Rapid-determination of lignocellulose by diffuse reflectance Fourier-transform infrared spectrometry. Anal Chem 57: 2867-2869 Schwanninger M, Hinterstoisser B (2002) Klason lignin: Modi fications to improve the precision of the standardized determination. Holzforschung 56: 161-166 Searchinger T, Heimlich R, Houghton RA, Dong F, Elobeid A, Fabiosa J, Tokgoz S, Hayes D, Yu TH (2008) Use of U.S. croplands for biof uels increases gree nhouse gases through emissions from land-use change. Science 319: 1238-1240 Semel Y, Nissenbaum J, Menda N, Zinder M, Krieger U, Issman N, Pleban T, Lippman Z, Gur A, Zamir D (2006) Overdominant quantitative tra it loci for yield and fitness in tomato. Proc Natl Acad Sci U S A 103: 12981-12986 Siegenthaler U, Stocker TF, Monnin E, Luthi D, Schwander J, Stauffer B, Raynaud D, Barnola JM, Fischer H, Masson-Delmotte V, Jouzel J (2005) Stable carbon cycleclimate relationship during the late Pleistocene. Science 310: 1313-1317 Sindhu A, Langewisch T, Olek A, Multani DS, McCann MC, Vermerris W, Carpita NC, Johal G (2007) Maize Brittle stalk2 encodes a COBRA-like protein expressed in early organ development but required for tissue flexibility at maturity. Plant Physiol 145: 1444-1459 Somerville C (2006) Cellulose synthesis in hi gher plants. Annu Rev Cell Dev Biol 22: 53-78 Somerville C, Bauer S, Brininstool G, Facette M, Hamann T, Milne J, Osborne E, Paredez A, Persson S, Raab T, Vorwerk S, Youngs H (2004) Toward a systems approach to understanding plant cell walls. Science 306: 2206-2211 Sturm M, Racine C, Tape K (2001) Climate change. Increasi ng shrub abundance in the Arctic. Nature 411: 546-547 Sugiyama J, Persson J, Chanzy H (1991) Combined infrared a nd electron diffraction study of the polymorphism of native celluloses. Macromolecules 24: 2461-2466 Sugiyama J, Vuong R, Chanzy H (1991) Electron diffraction study on the 2 crystalline phases occurring in native cellulose from an algal cell wall. Macromolecules 24: 4168-4175 Suzuki S, Li L, Sun YH, Chiang VL (2006) The cellulose synthase gene superfamily and biochemical functions of xylem-specific cellulose synthase-like genes in Populus tricho carpa. Plant Physiol 142: 1233-1245
130 Taiz L and Zeiger E. (2006) Plant Physiology. Ed. Fourth Sinauer Associates. Inc. Tanaka K, Murata K, Yamazaki M, Onosato K, Miyao A, Hirochika H (2003) Three distinct rice cellulose synthase catalytic subunit gene s required for cellulose synthesis in the secondary wall. Plant Physiol 133: 73-83 Taylor NG (2008) Cellulose biosynthesis and de position in higher plants. New Phytol 178: 239252 Taylor NG, Howells RM, Huttly AK, Vickers K, Turner SR (2003) Interactions among three distinct CesA proteins essential for cellulose sy nthesis. Proc Natl Acad Sci U S A 100: 1450-1455 Taylor NG, Laurie S, Turner SR (2000) Multiple cellulose s ynthase catalytic subunits are required for cellulose synthesi s in Arabidopsis. Plant Cell 12: 2529-2539 Taylor NG, Scheible WR, Cutler S, Somerville CR, Turner SR (1999) The irregular xylem3 locus of Arabidopsis encodes a cellulose synthase required for secondary cell wall synthesis. Plant Cell 11: 769-780 Theander O, Westerlund EA (1986) Studies on dietary fiber .3. Improved procedures for analysis of dietary fiber. J Agric Food Chem 34: 330-336 Tsai CJ, Harding SA, Tschaplin ski TJ, Lindroth RL, Yuan YN (2006) Genome-wide analysis of the structural genes regula ting defense phenylpropanoid metabolism in Populus. New Phytol 172: 47-62 Tsarouhas V, Gullberg U, Lagercrantz U (2002) An AFLP and RFLP linkage map and quantitative trait locus (QTL) analysis of growth traits in Salix. Theor Appl Genet 105: 277-288 Turner SR, Somerville CR (1997) Collapsed xylem phenotype of Arabidopsis identifies mutants deficient in cellul ose deposition in the seconda ry cell wall. Plant Cell 9: 689-701
131 Tuskan GA, Difazio S, Jansson S, Bohlmann J, Gr igoriev I, Hellsten U, Putnam N, Ralph S, Rombauts S, Salamov A, Schein J, Sterck L, Aerts A, Bhalerao RR, Bhalerao RP, Blaudez D, Boerjan W, Brun A, Brunner A, Busov V, Campbell M, Carlson J, Chalot M, Chapman J, Chen GL, Cooper D, Coutinho PM, Couturier J, Covert S, Cronk Q, Cunningham R, Davis J, Degroeve S, Dejardin A, Depamphilis C, Detter J, Dirks B, Dubchak I, Duplessis S, Ehlt ing J, Ellis B, Gendler K, Goodstein D, Gribskov M, Grimwood J, Groover A, Gunter L, Hamberger B, Heinze B, Helariutta Y, Henrissat B, Holligan D, Holt R, Huang W, Islam-Faridi N, Jones S, Jones-Rhoades M, Jorgensen R, Joshi C, Kangasjarvi J, Karlsson J, Kelleher C, Kirkpatrick R, Kirst M, Kohler A, Kalluri U, Larimer F, Leebens-Mack J, Lepl JC, Locascio P, Lou Y, Lucas S, Martin F, Montanini B, Napoli C, Nelson DR, Nelson C, Nieminen K, Nilsson O, Pereda V, Peter G, Philippe R, Pilate G, Poliakov A, Razumovskaya J, Richardson P, Rinald i C, Ritland K, Rouze P, Ryaboy D, Schmutz J, Schrader J, Segerman B, Shin H, Siddiqui A, Sterky F, Terry A, Tsai CJ, Uberbacher E, Unneberg P, Vahala J, Wall K, Wessler S, Yang G, Yin T, Douglas C, Marra M, Sandberg G, Van de Peer Y, Rokhsar D (2006) The genome of black cottonwood, Populus trichocarpa (Torr. & Gray). Science 313: 1596-1604 Vergara CE, Carpita NC (2001) Beta-D-glycan synthases and the CesA gene family: lessons to be learned from the mixed-lin kage (1-->3),(1-->4)beta-D-glucan synthase. Plant Mol Biol 47: 145-160 Vermerris W, Saballos A, Ejeta G, Mosier NS, Ladisch MR, Carpita NC (2007) Molecular breeding to enhance ethanol production from corn and sorghum stover. Crop Sci 47: S142-S153 Veytsman BA, Cosgrove DJ (1998) A model of cell wall e xpansion based on thermodynamics of polymer networks. Biophys J 75: 2240-2250 Vian B, Reis D, Darzens D, Roland JC (1994) Cholesteric-like crystal analogs in glucuronoxylan-rich cell-wall composites e xperimental approach of Acellular reassembly from native cellulo sic suspension. Protoplasma 180: 70-81 Vincken JP, York WS, Beldman G, Voragen AG (1997) Two general branching patterns of xyloglucan, XXXG and XXGG. Plant Physiol 114: 9-13 Wang X, Cnops G, Vanderhaeghen R, De Block S, Van Montagu M, Van Lijsebettens M (2001) AtCslD3 a cellulose synthase-like gene im portant for root hair growth in Arabidopsis. Plant Physiol 126: 575-586 Whetten RW, Sederoff RR (1992) Phenylalanine ammonialyase from loblolly pine : purification of the enzyme and isolation of complementary DNA clones. Plant Physiol 98: 380-386 Willats WG, Gilmartin PM Mikkelsen JD, Knox JP (1999) Cell wall antibodies without immunization: generation and use of de-e sterified homogalacturonan block-specific antibodies from a naive phage display library. Plant J 18: 57-65
132 Willats WG, Knox JP (1999) Immunoprofiling of pectic polysaccharides. Anal Biochem 268: 143-146 Willats WG, Marcus SE, Knox JP (1998) Generation of monocl onal antibody specific to (1->5)-alpha-L-arabinan. Carbohydr Res 308: 149-152 Willats WG, Rasmussen SE, Kristensen T, Mikkelsen JD, Knox JP (2002) Sugar-coated microarrays: a novel slide surface for the high-t hroughput analysis of glycans. Proteomics 2: 1666-1671 Wu R, Ma CX, Yang MC, Chang M, Littell RC, Santra U, Wu SS, Yin T, Huang M, Wang M, Casella G (2003) Quantitative trait loci for grow th trajectories in Populus. Genet Res 81: 51-64 Yamada T, Yeh TF, Chang HM, Li LG, Kadla JF, Chiang VL (2006) Rapid analysis of transgenic trees using transmittance near-i nfrared spectroscopy (NIR). Holzforschung 60: 24-28 Yamamoto H, Horii F, Odani H (1989) Structural changes of native cellulose cr ystals induced by annealing in aqueous alkaline and acidic solutions at high-temperatures. Macromole 22: 4130-4132 Yeh TF, Yamada T, Capanema E, Chang HM, Chiang V, Kadla JF (2005) Rapid screening of wood chemical component variations using transmittance near-infrared spectroscopy. J Agric Food Chem 53: 3328-3332 Yu J, Hu S, Wang J, Wong GK, Li S, Liu B, Deng Y, Dai L, Zhou Y, Zhang X, Cao M, Liu J, Sun J, Tang J, Chen Y, Huang X, Lin W, Ye C, Tong W, Cong L, Geng J, Han Y, Li L, Li W, Hu G, Li J, Liu Z, Qi Q, Li T, Wang X, Lu H, Wu T, Zhu M, Ni P, Han H, Dong W, Ren X, Feng X, Cui P, Li X, Wang H, Xu X, Zhai W, Xu Z, Zhang J, He S, Xu J, Zhang K, Zheng X, Dong J, Zeng W, Tao L, Ye J, Tan J, Chen X, He J, Liu D, Tian W, Tian C, Xia H, Bao Q, Li G, Gao H, Cao T, Zhao W, Li P, Chen W, Zhang Y, Hu J, Liu S, Yang J, Zhang G, Xi ong Y, Li Z, Mao L, Zhou C, Zhu Z, Chen R, Hao B, Zheng W, Chen S, Gu o W, Tao M, Zhu L, Yuan L, Yang H (2002) A draft sequence of the rice genome ( Oryza sativa L. ssp. indica ). Science 296: 79-92 Zenoni S, Reale L, Tornielli GB, Lanfalon i L, Porceddu A, Ferrarini A, Moretti C, Zamboni A, Speghini S, Ferranti F, Pezzotti M (2004) Downregulation of the Petunia hybrida alpha-expansin gene PhExp1 reduces the amount of crys talline cellulose in cell walls and leads to phenotypic changes in petal limbs. Plant Cell 16: 295-308
133 BIOGRAPHICAL SKETCH Jianfei Zhao was born in 1980, in Jilin Provin ce, China. He entered the U niversity of Science and Technology of China, Hefei, in 1999 and earned his Bachelor of Science degree in the Department of Biochemistry, Molecular and Cell Biology in July 2004. During his last year of undergraduate study, he did his bachelors thesis at the Institute of Biochemistry and Cell Biology of Shanghai, Chinese Academy of Sciences Shanghai, with Dr. Lin Li. Then he worked as the research assistant in the National La boratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beiji ng, from 2004, with Drs. Xujia Zhang and Fuyu Yang. In 2005, Jianfei Zhao began his graduate trai ning in the Graduate Program in Plant Molecular and Cell Biology at the Un iversity of Florida, Gainesvill e, FL. The work detailed in this masters thesis was conducted in th e laboratory of Dr. Wilfred Vermerris.