As a final investigation of xylo-oligomer formation and destruction, samples of powdered corn stover were hydrolyzed in the microwave reactor vessels. These samples were Pioneer 34M95 maize harvested in 2002, which was ground to about 1 mm particle size. Earlier studies have shown that these corn stover samples contain approximately 20% by weight of xylan (10). Approximately, 85 mg of corn stover was used in 20 mL of 1.2% sulfuric acid so that the amount of xylan matched that in the experiments discussed above. Figure 9.11 shows the amount of xylose, xylobiose, and furfural measured as a function of reaction temperature for residence times of 10 and 30 minutes. These experiments were conducted to compare the results using biomass with those using pure xylobiose and xylan. Note that these residence

Temperature/°C Temperature/°C

Figure 9.11 Formation of xylose, xylobiose, and furfural as a function of temperature from the acid treatment of corn stover in a microwave reactor. Residence time was (a) 10 minutes and (b) 30 minutes.

times are longer than what is realistic for a commercial scale reactor (2 min). We observed that xylose formation was again maximized at low temperature (135°C), whereas higher temperatures result in the dehydration of xylose and the formation of furfural. Xylobiose formation was minor. These results are completely consistent with the experimental results observed for xylobiose and xylan, and with the theoretical results for xylobiose.

However, it should be noted that the results observed here are different from what has been measured in larger scale pretreatment studies. In those examples, higher biomass loadings have often been used and in some cases no stirring was possible. In these larger scale studies, optimal yields of xylose were found at higher temperatures (>150°C). It seems likely that higher temperatures were required because the mass transport was more restricted in those studies. In this study, a low loading, small particle size, and high agitation of the solution helped reduce mass transport limitations. In our experiments, there was more effective delivery of the catalyst (protons) to the substrate.

9.6 Conclusions

The results of this study suggest that the rates of direct dehydration of xylo-oligomers to form unwanted products during pretreatment is likely to be insignificant compared to hydrolysis to form smaller oligomers and xylose. This was shown both by experiment and theory. Quantum mechanical calculations showed that the barrier for hydrolysis is much smaller than the barriers for possible dehydration reactions. Experimental measurements showed that at low temperatures, the glycosidic linkages in xylo-oligomers can be completely hydrolyzed with little formation of dehydration products.

The results of our experiments with the microwave reactor also showed that the temper­ature required for complete conversion of xylan and xylo-oligomers to xylose is much lower than required for optimal xylose formation in larger scale studies with corn stover, although the residence times used in our study were larger (5 min and longer compared to 2 min). In this study, optimized yields of xylose were obtained at about 135°C, whereas temperatures greater than 180°C are typically used in pilot scale studies. For our experiments at 135°C,
little conversion to furfural was observed, whereas in pilot scale studies, high conversion at 180°C is accompanied by significant loss of xylose. The most likely explanation for these different observations is the better mass transport available in our microwave reactor-based study. For example, the particle sizes were smaller, the slurries had more water, and the samples were vigorously stirred. Thus, it is likely that the measurements in this study are more likely to result from the intrinsic kinetics of xylan hydrolysis.

An important consequence of these observations is that higher yields of xylose and lower losses to dehydration can be achieved if mass transport can be improved. It appears that further investigations into this area are warranted.

9.7 Future studies

Since mass transport appears to be an important limitation in the hydrolysis of xylan and xylo-oligomers, studies should be conducted on different approaches to increase the access of the acid to the hemicellulose in biomass. It is unlikely that the mass transport available in the experiments in this study could be obtained in a practical pretreatment reactor; however, there might be ways to add mechanical energy during pretreatment that could offset the reaction severity and lead to higher xylose yields and less xylose loss. Experiments and cal­culations investigating the effects due to increased particle size, concentration and agitation should be conducted. These studies should include molecular modeling to understand pro­ton transfer and sugar recombination reactions and computational fluid dynamics to study fluid flow and mass transfer in relevant particle sizes. The goal of future work should be to develop quantitative kinetic parameters that can be used to design improved pretreatment reactors and pretreatment economics.

Fiber degradation is coordinated by a multitude of bacterial and fungal enzymes. In contrast to the free cellulase systems of the aerobic microorganisms, the particularly efficient cellulose degradation by anaerobic bacteria is achieved by the relatively small quantities of enzymes that they produce. This enigma was clarified in part by the identification of the cellulosome, produced by certain anaerobes: a multienzyme complex specialized in cellulose degradation, first studied in Clostridium thermocellum (13,16,18-26). The C. thermocellum cellulosomes reside primarily within cell surface protuberances and the culture medium (27-34).

The cellulosome comprises a set of multi-modular components — some structural and some enzymatic (see Figure 13.1). A pivotal noncatalytic subunit called scaffoldin secures the various enzymatic subunits into the complex, via the cohesin-dockerin interaction (19, 35-39). For this purpose, the primary scaffoldin possesses a series of functional mod­ules, called cohesins, involved in enzyme attachment. In addition, the scaffoldin contains a cellulose-specific carbohydrate-binding module (CBM) for substrate targeting. The vari­ous enzyme subunits (notably, the cellulases and hemicellulases) each contain a specialized dockerin module, which is complementary to the scaffoldin-based cohesins. The specificity characteristics and tenacious binding between the scaffoldin-based cohesin modules and the enzyme-borne dockerin domains dictate the supramolecular architecture of the cellulosome.

Figure 13.1 Current view of the C. thermocellum cellulosome and its attachment to the bacterial cell surface. The CBM borne by the CipA scaffoldin is believed to mediate the binding of the cell to the cellulose substrate and anchoring proteins SdbA, Orf2p, and OlpB mediate attachment of increasing numbers of CipA scaffold і ns and their enzymes to the cell surface.

The organization of cellulases into a cellulosome is considered to be the most efficient of all microbial cellulose-degrading systems thus far studied (22, 23, 26). The cellulosomal scaffoldin of C. thermocellum also contains a dockerin of its own — a type-II dockerin that binds to type-II cohesins on one of three known anchoring scaffoldins (40-43). The latter contain one, two, or four type-II cohesins which incorporate the complementary number of scaffoldins onto the cell surface. The type-II dockerin is exquisitely specific for the type-II cohesins, and no interaction with its type-I counterpart takes place, thus ensuring correct cellulosome assembly and architecture (Figure 13.1).

The discovery and description of cellulosomes in different bacteria revealed marked di­versity in the structure and architecture of their component parts. Although the existence of cellulosomes in a wide variety of different anaerobic bacteria has long been established (44), molecular details of various scaffoldin proteins, and, in particular, the interactions between the cohesin and dockerin, are only now beginning to emerge (35, 37, 45-52). There is also a limited amount of information that suggests that the composition and disposition of the cellulosome can be affected by carbon source (27, 53-62). Much remains to be learned about the extent to which cellulosome composition and location are affected by factors such as polysaccharide complexity and composition.

Recent work on Ruminococcus flavefaciens strain 17 has revealed an especially intricate cellulosome complex comprising numerous cohesin-containing scaffoldins, together with interacting enzymes and other unidentified dockerin-bearing proteins (Figure 13.2) (63­67). The assembly of these components differs markedly from the proposed molecular architecture in the Clostridial cellulosomes. In R. flavefaciens, ScaA incorporates a group of dockerin-containing enzymes into its three resident cohesin repeats (66, 67). In addi­tion, a small ScaC scaffoldin serves as an adaptor protein that enhances the repertoire of cellulosomal subunits by binding both ScaA via its dockerin and to a range of as yet uniden­tified polypeptides via its single divergent cohesin (65). In turn, ScaA binds to any of seven cohesin repeats of ScaB via a specific cohesin-dockerin interaction, and ScaB is attached to the cell surface via a specialized cohesin-dockerin interaction with ScaE (68). ScaE in­cludes an N-terminal cohesin and a C-terminal LPXTG-like motif, which suggests that it is positioned covalently on the cell surface via proteolytic cleavage and sortase-mediated attachment mechanism (69-72). This mechanism differs from the previously defined mode of cellulosome attachment in C. thermocellum via S-layer homology modules (41). Intrigu­ingly, neither ScaA nor ScaB contains a CBM; it was thus unclear how the cellulosome or bacterial cell binds to cellulosic substrates. More recently, another important component of the R. flavefaciens sca gene cluster, cttA, has been shown to encode for a protein that also bears a C-terminal XDoc, which interacts with the ScaE cohesin (Rincon etal., unpublished results). The gene product, CttA, contains two putative CBMs that bind to crystalline cel­lulose, thus providing, at least in part, insight into the molecular mechanism that accounts for the observed binding of the R. flavefaciens cell to cellulose. The system in R. flavefaciens therefore appears more intricate than those in Clostridium species (20, 22).

Serial analysis of ribosomal sequence tags (SARST) is a novel, powerful approach for high — throughput profiling of microbial diversity in medical, industrial, or environmental samples (1). With SARST, ribosomal sequence tags (RSTs) from hyper-variable V1-region of the bacterial small subunit 16S rRNA gene are amplified using the biotinylated primers Bac64f — BpmI and Bac104r-BsgI. After amplification, the PCR products are digested with BpmI and BsgI, both types of IIS restriction enzymes that will selectively remove the conserved priming regions flanking the V1 region. Subsequently, DNA linkers with compatible ends are ligated to each site of the SARST tags, this to introduce SpeI and NheI restriction sites to flank the SARST tags. Enzymatic digestion with SpeI and NheI is used to create compatible overhangs for SARST tag ligation into concatemers. Subsequently, concatemers are cloned and sequenced, resulting in up to 20 SARST tags per sequencing reaction, hence offering a significant increase in throughput over traditional rDNA clone libraries. The concatemers contain predicted border sequences that demarcate the position and polarity of the SARST tags.

In addition to the protocol by Neufeld and coworkers (109), a simplified method based on the same principles was developed by (110) that uses the V6 variable region to generate SARST tags. SARST libraries were successfully generated from a defined mixture of pure cultures and from duplicate arctic soil DNA samples. The results showed that SARST is functional, reproducible, and that the distribution of RSTs in the defined community li­brary reflects the composition of the original sample. Specific primers could be designed based on sequence data from abundant soil SARST tags for further phylogenetic analy­sis. However, the distribution of sequences in SARST libraries is biased by multiple copies and similarities between 16S rRNA gene sequences of different species, thereby frustrating accurate quantitative inferences. Nevertheless, the SARST approach is accurate enough to reflect population trends and make comparisons between samples or treatments based on increase or decrease in specific sequence representation.

An alternative for SARST is Single Point Genome Signature Tags (SP-GST), a generally applicable, high-throughput sequencing-based method that targets specific genes to gen­erate identifier tags from well defined points in a genome, was developed (111). The tech­nique yields identifier tags that can distinguish between closely related bacterial strains and allow for the identification of microbial community members. SP-GSTs are determined by three parameters: (i) the primer designed to recognize a conserved gene sequence; (ii) the anchoring enzyme recognition sequence; (iii) the type IIS restriction enzyme which defines the tag length. The SP-GST method was evaluated in silico for bacterial identification using rpoC, uvrB, recA and the 16S rRNA gene. The best distinguishing tags were obtained with the restriction enzyme Csp6I upstream of the 16S rRNA gene, which discriminated all or­ganisms in the data set to at least the genus level, and most organisms to the species level. The method was successfully used to generate Csp6I-based tags upstream of 16S rRNA gene and allowed to discriminate between closely related strains of Bacillus cereus and Bacillus anthracis.

In bacteria, cellulose synthase appears to be constitutively produced and is activated by the regulatory molecule Bis-(3/-5/)-cyclic dimeric guanosine monophosphate (9, 57). C — di-GMP has not been found in plants but cotton fibers were reported to have a binding protein (58). However, comparison of the sequence of the apparent binding protein with the Arabidopsis proteome indicates that the putative binding protein is a-tubulin or something that copurified with it.

As noted above, recent results suggest that plant cellulose synthase is activated by a process associated with secretion. In principle, a plasma membrane-localized kinase or phosphatase could alter the activation state of cellulose synthase following transfer from the Golgi, pro­viding a mechanism for keeping it inactive in the Golgi but rapidly activating it upon arrival in the plasma membrane. A proteomics survey of plasma membrane phosphoproteins re­vealed that CESA1, CESA3, and CESA5 proteins were phosphorylated at a number of sites, and several of the peptides had more than one residue phosphorylated (59). The sites were clustered in the N-terminal domain and in the hypervariable region of the central domain (3). Analysis of the CESA7 protein by mass spectrometry showed that two serine residues in the hypervariable region are phosphorylated (60). Cell extracts catalyzed phosphorylation of these residues and the phosphorylated polypeptide region was rapidly degraded by a proteosome-dependent pathway, leading to the suggestion that phosphorylation may regu­late protein turnover. The KOR protein, which is required for cellulose synthase activity in vivo, also had at least two phosphorylated peptides.

During cell expansion, cellulose synthesis is a major consumer of fixed carbon. Thus, it seems likely that whatever regulates cellulose synthesis is coordinated with other aspects of primary carbon metabolism. In plants, UDP-glucose is thought to be largely synthesized by sucrose synthase (SuSy) (61). Amor and coworkers (62) observed a form of SuSy that was associated with the plasma membranes. They also observed that sucrose supported much higher rates of cellulose synthesis by extracts from developing cotton fibers than UDP — glucose and that sucrose synthase is very strongly upregulated in cotton fibers at the onset of fiber elongation. Haigler and coworkers (61) have presented an extensive review of the hypothesis that SuSy might channel UDP-glucose to cellulose synthesis. This is an attractive idea but direct evidence is lacking. Arabidopsis has six SuSy genes but no two isoforms have the same pattern of expression (63). Mutant plants lacking individual isoforms, or double mutants of closely related isoforms, had no alteration in cellulose content. Thus, these studies did not provide support for the idea that SuSy is an important factor in controlling cellulose synthesis. By contrast, transgenic suppression of several SuSy genes in developing cotton fibers prevented formation of fiber cells (64). The effect was more profound than could be attributed solely to an inhibition of cellulose synthesis, obscuring a mechanistic interpretation of the effects. Increased expression of various forms of SuSy in transgenic tobacco plants did not result in increased cellulose per cell, suggesting that UDP-glucose is not the limiting factor in cellulose accumulation in that system (65).

Analysis of the steady-state level of mRNA in major tissues of Arabidopsis with gene chips showed that the CESA1, 2, 3, 5, and 6 genes are expressed in all tissues at moderately high levels that differ by about fourfold at most (66). Similar results can be compiled from the large number of public microarray datasets that are now available for Arabidopsis from sites such as Genevestigator (67). As noted below, CESA1,2,3, and 6 have been implicated in pri­mary wall synthesis by mutant analysis. Analyses of expression of CESA genes in Arabidopsis embryos revealed that CESA1, 2, 3, and 9 are the only CESAs expressed there (68). Thus, following the nomenclature of Burton and coworkers (69) CESA1, 2, 3, 5, 6, 9 are probably involved in primary wall synthesis and are referred to at Group-I CESAs. By contrast CESA4, 7, 8 are mostly or only expressed in tissues such as stems where secondary cell walls are found and are designated Group-II (21, 66). CESA4 promoter:GUS expression studies confirmed that the CESA4 gene was mostly or only expressed in the vascular tissues (13). Similarly, im­munological staining of tissue prints with antibodies against CESA7 and CESA8 showed that the corresponding genes were only expressed in the xylem and interfascicular region (70).

Maize has at least 12 CESA genes (71). PCR analysis of transcript levels of six of the genes in various tissues indicated that all of the genes were expressed in all of the tis­sues examined (13). Analysis of eight of the maize genes by massively parallel signature sequencing indicated that the levels of several of the CESA genes varied from one tissue type to another, but no conclusions were reached concerning functional specialization (72). A subsequent analysis that included three additional genes resulted in the identification of three genes that were specifically associated with secondary cell wall formation (71). Thus, maize also shows evidence for specialization of primary and secondary cell wall synthases.

Quantitative information about the relative levels of expression of the Arabidopsis CESA genes is lacking because the gene chips used for most studies have not been calibrated for the various CESA genes. By contrast, Burton and coworkers (69) used quantitative PCR to measure the expression of the eight known barley CESA genes. They observed that the CESA genes could be grouped into two expression patterns (i. e., Group I and II) that were generally consistent with roles in primary and secondary wall synthesis. Additionally, they observed that there were large differences in the relative abundance of transcripts for the various members of a CESA group. If the CESA genes are translated with similar efficiency, this observation would suggest that the various CESA proteins are not present in identical amounts in the CESA complexes.

Consistent with genetic evidence that at least three CESA proteins are required to produce a functional cellulose synthase complex, correlation analysis of public and private DNA chip datasets revealed that expression of the Arabidopsis CESA4, 7, 8 gene were indeed very highly correlated (73, 74). The expression of a number of other genes was also very highly correlated with these genes and insertion mutations in several of these genes resulted in cellulose deficient phenotypes. Mutations in some highly correlated genes did not result in obvious effects on cellulose synthases but resulted in other defects in secondary wall synthesis. Thus, the evidence is compatible with the idea that the CESA genes that participate in secondary wall synthesis are under developmental control along with other genes required for secondary wall synthesis. The CESA genes implicated in primary wall synthesis were less highly correlated. This is consistent with the observation that there are more than three CESA genes associated with primary wall synthesis. This presumably indicates that some of the Group-I CESAs are functionally redundant and, therefore, their expression may vary from one tissue to another for unknown reasons. For instance, as noted above, CESA9 appears to be specifically expressed in embryos.

There is sparse evidence suggesting that cellulose synthesis may be regulated in response to stimuli other than developmental programs. Transgenic trees in which 4-coumarate:coenzyme A ligase expression was reduced by expression of an antisense gene exhibited up to a 45% reduction of lignin and a 15% increase in cellulose (75). However, the apparent increase in cellulose my have been due to a decrease in total mass caused by the reduced lignin content. Conversely, antisense-mediated reduction in expression of an a-expansin in petunia caused a significant reduction in cellulose accumulation in petals (76). According to current theories of expansin action (77) this presumably reflects an indirect effect from a defect in cell expansion. The properties of this mutant raise the pos­sibility that many or all mutants with defects in cell expansion may have reduced cellulose content due to some form of feedback regulation of cellulose synthesis.

Sucrose, a major carbon source for growing cells, is delivered between cells via plasmod — esmata (symplastic route) and by long-range transport from source to sink cells via the phloem. SuSy, sucrose synthase (EC 2.4.1.13), catalyzes the reversible conversion of su­crose and UDP into UDP-Glc and fructose. But in vitro SuSy also converts sucrose to form TDP-Glc and ADP-Glc from TDP and ADP, respectively (403, 418), as well as GDP-Glc and CDP-Glc (419). SuSy isoforms have been identified in many plant species. In pea, three SuSy isoforms (Sus) were functionally isolated and found to have different kinetic properties. For example, Sus1 was strongly inhibited by Frc (420). In Arabidopsis, six distinct gene members of SuSy are known, and the tissue expression pattern for each SuSy transcript isoform is complex (421) and does not provide clue to their distinct biological functions. Part of the complexity in assigning a biological role for each isoform is the fact that SuSy isoforms differ in protein length and in their amino acid sequence. In addition, cell fractionation studies

and immunogold labeling demonstrate that SuSy isoforms are associated with different sub­compartments. For example, distinct SuSy isoforms fractionate with the Golgi apparatus, the tonoplast, and the plasma membrane (422). A recent report also identifies SuSy associ­ated with the actin cytoskeleton (423). Subbaiah and coworkers (424) reported recently that of the three maize SuSy isoforms, SH1 contains a mitochondrial targeting peptide that is required for its intramitochondrial localization. This isoform was proposed to be involved in the regulation of solute fluxes into and out of mitochondria. The association of SuSy with membranes is often observed in growing cells. A specific phosphorylation site on the amino terminal region of SuSy regulates movement of the enzyme between a cytosolic form and a plasma membrane-associated form (425). Biochemical characterization and substrate specificities of each SuSy isoform will be required to elucidate their role in either supporting the flux of carbon to cell wall (i. e., A/T/UDP-Glc) or in carbohydrate storage (ADP-Glc). Lastly, an understanding of the relationship between UDP-Glc PPase isoforms and SuSy isoforms in photosynthetic and non-photosynthetic cells will be needed to understand how the flux of sugars is directed into growth or storage.

With the patterns of aggregation indicated for native celluloses in living plants, two stages in the isolation processes will influence the final pattern of aggregation. The first is elevation of temperature during most such processes. The second is the effect of drying the sample as the last stage in the processes of isolation of native celluloses.

In the methods most often used for purification of celluloses for investigation, two methodologies stand out. For tissues that are relatively pure celluloses, boiling in dilute caustic under nitrogen is a very common practice. And the mildest procedure used for isolating cellulose from highly lignified woody tissue begins with the acid chlorite method, usually carried out at 70°C. Both of these methods involve temperature elevation to levels that can alter the patterns of aggregation in ways that make the cellulose more recalcitrant during hydrolysis.

Though of course we know that cellulose oligomers are essentially insoluble in water at the octamer and beyond, we expect it to be hydrated in its native state at the level of the ele­mentary nanofibrils or at least at the first level of aggregation. We also know that all cellulose derivatives that do not have any ionic substitution have negative correlations of solubility with temperature (45), with most of them precipitating between 40 and 60°C. This is likely also true of cellulose at the nanoscale level. So, temperature elevation is likely to change the state of aggregation of native celluloses. The only path to avoiding this effect is isolation at ambient temperatures. Such procedures are possible, though they require patience.

Another important factor is that celluloses are deformed during drying procedures. The more common methods are drying samples in an oven at 105°C or freeze-drying them. The first hint of distortion during the drying process came during the early solid state 13C NMR studies, where researchers observed that moistening samples to a level of 20% moisture

300 400 500 600

Figure 6.13 Raman spectra of bleached spruce kraft pulp, before and after freeze drying.

resulted in enhancement of the resolution of spectra. It was thought that this effect could be due to enhancement of cross-polarization efficiency or distortions during the drying process that are relaxed by the added moisture. Definitive answers to this question on the basis of SS 13C NMR were not pursued (5).

However, in a retrospective review of Raman spectra recorded over many years in Atalla’s laboratory, one set of spectra stood out as clear evidence of the occurrence of such distortions during dehydration. The spectra are shown in Figure 6.13. They are spectra of a sample of never-dried bleached spruce kraft pulp recorded before and after freeze drying, thus avoiding further effects of temperature. In the course of the kraft-pulping process, the fibers had been subjected to temperatures in the range of 170-180°C, which resulted in a high degree of aggregation of cellulose well beyond the level in its native state. In spite of that, an effect of freeze drying is observed.

The upper spectrum of the never-dried sample bands are broadened dramatically as a result of the drying process; the spectrum has many weak bands that appear to be broadened and drowned into the background after freeze drying. These effects suggest that the freeze­drying process, by removing too much of the water needed to lubricate the motion of the nanofibrils relative to each other, results in distortions of the nanofibrils as they hydrogen — bond with each other rather than remaining isolated in a vacuum.

From these observations, we conclude that almost all studies of structures of cellulose undertaken so far have been of celluloses that have been modified to varying degrees in the course of isolation. In addition to the evidence presented here, X-ray diffraction studies and solid state 13C NMR measurements on cotton fibers have shown that cotton fibers in the unopened boll tend to decline in degree of order upon opening of the boll and subsequent dehydration of the cotton fibers; the X-ray diffraction patterns are broadened and the spectra loose resolution. This can only be interpreted as resulting from some distortion of the native order as the native moisture is removed.

More recently, we have observed in our laboratory that bacterial celluloses manifest changes in their spectra after dehydration. The changes are not dramatic, but they reveal a broadening of the bands that are more sharply resolved in the never-dried samples. Samples for the spectra in Figure 6.9 were treated with acid chlorite at ambient temperature. The samples used in all of the prior studies of structure (10,11,46) were boiled in NaOH or KOH solutions for extended periods, followed by extended periods of exposure to very concen­trated sulfuric acid, presumably to remove the “amorphous fractions" It is useful to consider the effect of boiling on an aggregate of cellulose nanofibrils, given the effects of temperature on cellulose hydration noted above. Figure 6.14 shows stages of further aggregation likely to occur at elevated temperatures in native higher-plant celluloses. The secondary aggregate shown as A in panel B of Figure 6.11 has been subjected to tighter compaction to varying degrees.

Thus, a previously uniform aggregate with low curvature is forced into a form where it has extended parallel primary structures, interrupted by a highly twisted connection; this is not unlike the nanofibril of Micrasterius denticulata shown in the left panel of Figure 6.1. The highly twisted region is then regarded as amorphous. A previously uniform helical natural structure that has a distinctive long period and limited curvature has been transformed by thermal dehydration into an approximation of the fringed micelle model introduced in the first half of the last century. It is our view that the effect illustrated in Figure 6.14 is an inevitable consequence of boiling in caustic solutions for extended periods.

Since our primary concern in this volume is cellulose in higher plants, we have developed another model structure intended to approximate fibrils in higher plants. It consists of four elementary fibrils ina2by2 arrangement. The elementary fibrils are approximately 4 nm in diameter. These are shown in Figure 6.15 where now the long period is represented as 300 nm, and the segment shown is 75 nm long and undergoes a twist of 90°. This approximation of the native form is represented in pattern A as it is expected to occur in the native state in the living plant. The effect of dehydration, taken to be tight aggregation resulting in

approximately 85% conversion to the more tightly aggregated form, is shown in pattern A’ of the figure.

Thus, a consequence of the dehydration, whether by drying or by temperature elevation, is the creation of more tightly aggregated domains that are likely to be more recalcitrant to hydrolytic action, whether by enzymes or acids. Furthermore, we suggest that the linear

parallel segments, which are artifacts of isolation processes are easily mistaken for naturally occurring crystalline domains.

Downregulation of the currently annotated “CAD1” homologues, NtCAD1-1 and NtCADl — 7, in tobacco using an RNAi technology (141) recently led to selection of two lines, L11 and L14: expression of the “NtCADl” genes was barely detectable in line L11, whereas it was reduced in line L14. Lignin contents, as estimated by the Klason method, were though basically unchanged in both lines as compared to wild type. While these researchers proposed that the G contents in the lignin were reduced in the transformants, the body of evidence was scant for any effect on lignification proper. Indeed, the observations made may yet again be another consequence of pleiotropic effects. These data are in marked contrast to the striking effects of “knocking-out” both AtCAD4/5, the corresponding bona fide CADs present in Arabidopsis, the result of which was that neither the corresponding monolignols nor lignin were formed to any extent (71). Thus, the involvement of the putative tobacco “CAD1s” in both coniferyl alcohol (3) and lignification has not been established; other physiological roles need to be considered and examined as indicated earlier.

In this respect, a recent study of two genes (LePAR1 and LePAR2) from tomato (Solanum lycopersicum), with high degree of similarity/identity to both the Eucalyptus and tobacco CAD1s resulted in a different metabolic role being identified: Tieman etal. (245) concluded that these genes encoded a 2-phenylacetaldehyde reductase required for formation of 2- phenylethanol (64, Figure 7.11A). Interestingly, both Eucalyptus and tobacco accumulate 2-phenylacetaldehyde (65)/2-phenylethanol (64) as volatile constituents; these data thus suggest that their CAD1s may instead be 2-phenylacetaldehyde reductases.

Finite difference methods work by splitting the integration into many small steps, each separated by a fixed time bt. At a time t, the force on each atom is calculated from the vector sum of the forces arising from its interactions with the other atoms. From this the acceleration (d2xi/dt2) can be calculated. The accelerations are then combined with the position and velocity data for time t in order to calculate the positions and velocities at a time t + bt. The iteration is repeated until sufficient time steps have been sampled.

Over the period of the time step the force is assumed to be constant. This places severe constraints on the length of a time step. Ideally the larger time steps allow more phase space to be explored for a given computational effort. However, if the time step is too long the assumption of constant force will break down and lead to instabilities. This causes unrealistic oscillations in the system that can rapidly multiply resulting in an unstable molecular dynamics trajectory. In practice, the time step is limited to an order of magnitude lower than the highest frequency motions. In flexible molecules, these are typically bond stretches involving hydrogen (e. g., C-H ca. 10 fs period). Several methods such as SHAKE (48) and RATTLE (49) exist that, via the use of constraints on bonds with high-frequency oscillations, allow longer time steps to be used. For accurate dynamics, it is generally accepted that only motions involving hydrogen can be constrained and thus the maximum time step is effectively limited to a maximum of 2 fs. This means that to simulate a biological system for 100 ns the complete energy and forces of the entire system have to be evaluated a total of 50 x 109 times in a linear fashion. Since step n + 1 cannot be evaluated before step n, it should be immediately apparent that such MD simulations are not applicable to distributed or grid-based computation and instead require very tightly coupled supercomputers with fast processors and extremely low latency and high bandwidth interconnects.

The dominant hemicelluloses in softwoods are glucomannan and galactoglucomannan (GGM) (65). The GGM structure varies by species and cell wall location, but is gener­ally built upon a p-(1—4)-D-mannopyranose and p-(1—4)-D-glucopyranose backbone. The ratio and sequence of the two backbone sugars vary, but typically several mannose units are linked together and periodically interrupted by a glucose unit. The usual man:glc ratio is 3:1. Side chains are dominated by a-D-galactopyranose units, typically linked to the O-6 position of either backbone sugar, though O-2 and O-3 linkages on the mannose units have been reported for some species. The galactose side units are typically single, but may be double, with the second terminal galactose linked p-(1—>2) to the first unit. The other common substituent is an ester-linked acetyl group. Typically, these are linked at the O-2 or O-3 positions of the backbone sugars and reported content and substitution patterns vary widely. As with acetylated xylans, one of the primary effects of acetyl content is the effect on the solubility of these polymers in water. High acetyl content GMMs tend to be water soluble, as do GMMs with high galactose content. The alkali-soluble GMMs appear to have fewer side chains, though the extraction conditions remove any ester-linked acetyl groups by saponification. The ratio of galactose to glucose is often much lower in these as well and they are frequently referred to as glucomannans (65).

Despite a lack of systematic studies on enzymatic degradation of softwood, much can be deduced from the structural information regarding the enzyme activities necessary for hydrolysis. Acetyl esterase and a-galactosidase are the dominant debranching enzymes, with the requirement for either being dictated by the specific type of softwood glucomannan or galactoglucomannans being hydrolyzed (12, 76). Endomannanases and p-glucanases are the dominant main-chain depolymerases and p — mannosidases and p — glucosidases are likely needed for reduction of the oligomers to monomer sugars.

10.5.1 Arabinogalactan, xyloglucan, and p — glucan

Other minor hemicelluloses found in biomass feedstocks include arabinogalactan, xy­loglucan, and p-glucans. Arabinogalactan, found mainly in softwoods, is comprised of a linear p-(1—3)-D-galactopyranose backbone highly substituted at the C-6 position, though p-(1—>4) galactans have also been identified (5,96). These side chains include p-D-galactose, a-L-arabinose, and p-D-glucuronic acid in either monomer or dimer configurations. From the configuration, debranching with p-galactosidases, a-L-arabinofuranosidase, and p-glucuronidases is likely to be supplemental to depolymerization by p-galactanases.

Xyloglucan is a linear p-(1—4)-D-glucopyranose polymer with a(1 — 6)-d — xylopyranose side chains, typically present as a monomer, but also extended by (1—2) linkages to p-galactose which maybe further extended by (1—2) a-fucose. Specific xyloglu — canases which depolymerize the p-(1—4)-D-glucopyranosyl backbone have been identified and most fall into the GH74 family (62,63, 97). The distinction between these endoxyloglu — canases and endocellulases may lie in the requirement for specific side groups in the case of the xyloglucanases and synergy has been demonstrated between these enzymes (61, 62, 98, 99). Little work has been done on the debranching enzyme requirements in xyloglucan, though a-xylosidases are obviously important candidates.

The p-glucans of softwoods are typically p-(1—3)-D-glucopyranose chains with a few p-(1——4) linkages and the rare glucuronic or galacturonic side chain. Other potential, though minor, p-glucans from biomass feedstocks include polymers such as endosperm p-glucan, a mixed (1—3,4)-p-glucan, where the (1—4) linkages are periodically replaced every three or four residues by a (1—3) linkage. As with cellulose, endo — and exoglucanases are the major contributors, however, the specificity of the enzymes may be to a specific linkage or series of linkages. The so-called lichenanases (3.2.1.73) hydrolyze the p-(1——4) bond but require an adjacent 3-O-substituted glucose and will not hydrolyze homogeneous p-(1——4) glucans (100).Laminarinases (3.2.1.39) hydrolyze p-(1——3) bonds. Homogeneous stretches of p -(1—4) links are hydrolyzed by cellulases. As these polymers are nearly exclusively linear, debranching enzymes are not likely required.

14.5.2.1 Steam explosion

Steam explosion processes with no added chemical catalysts have been practiced for nearly a century, dating back to the development of the Masonite process on wood chips in the 1920s (29). In steam explosion, chipped or coarsely shredded biomass is contacted with high-pressure saturated steam at high solids loadings (generally >20% solids) in a pressure vessel for a residence time that is generally 20 minutes or less (12-14, 30). Depending on the feedstock used and the objective of the pretreatment, steam explosion pretreatment temperatures are generally in the range of 140-260°C. At the end of the pretreatment time, the pressure vessel contents are rapidly decompressed into an atmospheric pressure flash tank, which causes significant disruption and defibration of the biomass. The use of specially designed orifices or other shear-enhancing devices to increase the mechanical disruption of the pretreated biomass is often practiced.

Even without the addition of any chemical catalysts, hydrolysis reactions in steam ex­plosion are catalyzed by the release of organic acids that are liberated from acetyl func­tional groups associated with hemicellulose. This results in some lignin solubilization and hemicellulose hydrolysis, although yields of xylose from the hemicellulose fraction of most biomass types are typically no higher than 65% of theoretical, primarily due to extensive sugar degradation reactions that occur under typical uncatalyzed steam explosion reaction conditions (12, 31, 32). Compared to uncatalyzed steam explosion pretreatment, the ad­dition of acid catalysts, such as sulfur dioxide or sulfuric acid, to biomass feedstocks in steam explosion, has been shown to improve the yield of released carbohydrates by re­ducing sugar degradation (13, 14, 33) and to improve the enzymatic digestibility of the resulting pretreated biomass (11, 13, 34). In general, steam explosion pretreatments have been shown to be effective in increasing the enzymatic digestibility of a wide range of feed­stocks, although it is less effective on softwoods (35). This technique has also been tested in continuous pilot-scale reactor systems (36, 37), but it does result in some sugar degradation losses, incomplete disruption of the lignin-carbohydrate matrix, and the requirement to wash the pretreated biomass or otherwise condition the liquid portion of the pretreated slurry to remove inhibitory products prior to fermentation to ethanol or other products (11, 14,33).