Identifying new lignin bioengineering targets: 1. Monolignol-substitute impacts on lignin formation and cell wall fermentability

Background Recent discoveries highlighting the metabolic malleability of plant lignification indicate that lignin can be engineered to dramatically alter its composition and properties. Current plant biotechnology efforts are primarily aimed at manipulating the biosynthesis of normal monolignols, but in the future apoplastic targeting of phenolics from other metabolic pathways may provide new approaches for designing lignins that are less inhibitory toward the enzymatic hydrolysis of structural polysaccharides, both with and without biomass pretreatment. To identify promising new avenues for lignin bioengineering, we artificially lignified cell walls from maize cell suspensions with various combinations of normal monolignols (coniferyl and sinapyl alcohols) plus a variety of phenolic monolignol substitutes. Cell walls were then incubated in vitro with anaerobic rumen microflora to assess the potential impact of lignin modifications on the enzymatic degradability of fibrous crops used for ruminant livestock or biofuel production. Results In the absence of anatomical constraints to digestion, lignification with normal monolignols hindered both the rate and extent of cell wall hydrolysis by rumen microflora. Inclusion of methyl caffeate, caffeoylquinic acid, or feruloylquinic acid with monolignols considerably depressed lignin formation and strikingly improved the degradability of cell walls. In contrast, dihydroconiferyl alcohol, guaiacyl glycerol, epicatechin, epigallocatechin, and epigallocatechin gallate readily formed copolymer-lignins with normal monolignols; cell wall degradability was moderately enhanced by greater hydroxylation or 1,2,3-triol functionality. Mono- or diferuloyl esters with various aliphatic or polyol groups readily copolymerized with monolignols, but in some cases they accelerated inactivation of wall-bound peroxidase and reduced lignification; cell wall degradability was influenced by lignin content and the degree of ester group hydroxylation. Conclusion Overall, monolignol substitutes improved the inherent degradability of non-pretreated cell walls by restricting lignification or possibly by reducing lignin hydrophobicity or cross-linking to structural polysaccharides. Furthermore some monolignol substitutes, chiefly readily cleaved bi-phenolic conjugates like epigallocatechin gallate or diferuloyl polyol esters, are expected to greatly boost the enzymatic degradability of cell walls following chemical pretreatment. In ongoing work, we are characterizing the enzymatic saccharification of intact and chemically pretreated cell walls lignified by these and other monolignol substitutes to identify promising genetic engineering targets for improving plant fiber utilization.


Background
Recent discoveries highlighting the metabolic pliability of plant lignification indicate that lignin can be engineered to dramatically alter its composition. Perturbing single or multiple genes in the monolignol pathway of angio-sperms can lead to dramatic shifts in the proportions of normal monolignols (e.g. coniferyl 1 and sinapyl alcohol 2, Figure 1) and pathway intermediates polymerized into lignin [1,2]. The malleability of lignification is further illustrated in some angiosperms by the pre-acylation of monolignols with acetate, p-hydroxybenzoate, or p-coumarate [2,3] and the oxidative coupling of ferulate and diferulate xylan esters into lignin [4][5][6].
Recent efforts in lignin bioengineering are primarily aimed at manipulating the normal monolignol biosynthetic pathway [7], but in the future apoplastic targeting of phenolics from other metabolic pathways may provide exciting opportunities for designing lignin that is less inhibitory toward polysaccharide hydrolysis and fermentation or easier to remove by biological or chemical pretreatments. Recent model studies with maize cell walls demonstrated that partial substitution of coniferyl alcohol with coniferyl ferulate (a monolignol conjugate) dramatically enhanced the alkaline extractability of lignin and the enzymatic hydrolysis of fiber [8]. Based on these results, bioengineering of plants to copolymerize coniferyl or sinapyl ferulate with monolignols is being pursued as a means for enhancing biomass saccharification or pulping for paper production.
To identify other promising avenues for lignin bioengineering, we are conducting a series of experiments to assess how the inclusion of phenolics derived from various metabolic pathways may alter lignin formation and the utilization of plant cell walls. One path to explore is reducing the hydrophobicity of lignin to permit greater penetration and hydrolysis of fiber by polysaccharidases. Lignin hydrophobicity could be modulated by the incor-    13 14 12 15 poration of phenolics with extensive sidechain or aromatic ring hydroxylation (e.g., guaiacyl glycerol 4 or epigallocatechin gallate 11) or substitution with hydrophilic groups (e.g., feruloylquinic acid 8 or 1-O-feruloyl glycerol 14). Another approach would be to incorporate phenolics with o-diol functionality (e.g., methyl caffeate 5, caffeoylquinic acid 6, epicatechin 9, epigallocatechin 10, and epigallocatechin gallate 11). The presence of odiols provides an intramolecular pathway to trap lignin quinone methide intermediates which form cross-links between lignin and structural polysaccharides [9]; such cross-links appear to limit the enzymatic hydrolysis of cell walls [10,11]. Another route, first illustrated by our work with coniferyl ferulate [8], would be to incorporate readily cleaved bi-phenolic conjugates (e.g., epigallocatechin gallate 11,1,3-di-O-feruloyl glycerol 15, or 1,4-di-Oferuloyl threitol 16) to facilitate lignin depolymerization during pretreatment of biomass for subsequent saccharification.
In this study, we used a well-characterized biomimetic cell wall model [12] to explore how various monolignol substitutes influence the formation of lignin and the enzymatic degradability of cell walls. Degradability was assessed by monitoring gas production during the in vitro hydrolysis and fermentation of cell walls by rumen microflora and by the analysis of residual nonfermentable polysaccharides (NP). Because gas production is directly linked to structural carbohydrate hydrolysis and fermentation by rumen microflora [13,14] and highly correlated with biomass fermentation to ethanol [15], our findings should have broad application in plant selection and engineering programs aimed at improving the utilization of fibrous feeds by livestock and cellulosic biomass for biofuel production

General
Quinic acid, ferulic acid, and other reagents were obtained from Aldrich. NMR spectra of synthesized compounds were run on a Bruker (Billerica, MA) DRX-360 fitted with a 5 mm 1H/broadband gradient probe with inverse ( 1 H-detected) geometry. The solvent used for synthetic compounds was acetone-d 6 unless otherwise specified; referencing was to the central solvent peak (δ C 29.80, δ H 2.04 ppm for acetone).

5-Feruloylquinic acid 8
The synthesis follows that of chlorogenic acid (caffeoylquinic acid 6) and other analogs described by Hemmerle et al. [20]. Briefly, methyl ferulate 7 was protected via reaction with 2-[(trimethylsilyl)ethoxy]methyl chloride to give compound 17 as an orange oil, which was saponified directly to give 18 as a solid (58% of theoretical yield based on methyl ferulate); recrystallization from hexane afforded colorless crystals, mp 80-83°C. The amide 19 was prepared and used directly in coupling with 23. Protection of (1R, 3R, 4R, 5R)-(-)-quinic acid 20 with cyclohexanone gave solid 21 in 92% yield. Recrystallization from ether/hexane gave a solid with mp 142. 5 Compounds 13-16 all derived from coupling reactions between an alcohol and the phenolic acetate of feruloyl chloride 26, were made from ferulic acid 24, via the 4-Oacetylated ferulic acid 25, as described previously [22]. Phenolic deacetylation of final products was via pyrrolidine as described previously [23]; the deacetylation to compound 13 is fully described as an example. The remaining steps were as follows.

Feruloyl ethylene glycol 13
2''-Hydroxyethyl-3-(4'-acetoxy-3'-methoxyphenyl)-2-propenoate 28. 3-(4'-Acetoxy-3'-methoxyphenyl)-2-propenoyl chloride 26 (1.17 g, 4.6 mmol) and 5 mL of dichloromethane were placed in a 25 mL round-bottom flask equipped with a magnetic stir bar. The mixture was cooled in an ice/water bath. A mixture of 0.02 g of 4-dimethyl-aminopyridine, 0.5 mL of pyridine, and 5 mL of 1,2dihydroxyethane 27 were added dropwise to the cooled, stirred acid chloride solution. After the addition, the ice/ water bath was removed and the stirred mixture was allowed to warm to ambient temperature. After standing overnight, the mixture was poured into 50 mL of dilute hydrochloric acid (1 M). The mixture was extracted with dichloromethane (3 × 25 mL). The combined organic extracts were washed with saturated sodium bicarbonate solution (1 × 50 mL) and saturated sodium chloride solution (1 × 50 mL) and then dried over anhydrous magnesium sulfate. The solvent was removed using a rotary evaporator to afford 1.12 g of viscous oil (87%). The crude material was deacetylated in the next step without further purification.
Feruloyl ethylene glycol 13 (2''-Hydroxyethyl-3-(4'hydroxy-3'-methoxyphenyl)-2-propenoate). Compound 28 and 10 mL of pyrrolidine were placed in a 50 mL round-bottom flask and stirred for 10 minutes. The mixture was poured into a separatory funnel containing 75 mL of 1 M hydrochloric acid and 50 mL of ethyl acetate. The pH of the aqueous phase of the mixture was adjusted to below 3 by addition of 1 M hydrochloric acid. The phases were separated and the aqueous phase was extracted with ethyl acetate (3 × 25 mL). The combined organic phases were washed with saturated sodium chloride until the pH was close to neutral (3 × 50 mL). The extracts were dried over anhydrous magnesium sulfate and the solvents removed by rotary evaporator. Removal of residual volatile material on a high-vacuum line afforded 0.838 g of a colorless foam (76% based on the acid chloride 26).

1-O-feruloyl glycerol 14
2,2-Dimethyl-4-(hydroxymethyl)-1,3-dioxolane 29 from Aldrich (1.36 g, 10.3 mmol), 0.8 g of pyridine, 0.02 g of 4dimethylaminopyridine, and 20 mL of dichloromethane were placed in a 100 mL round-bottom flask with a magnetic stir bar. The stirred mixture was cooled in an ice/ water bath. A solution of 2.55 g (10.0 mmol) of chloride 26 in 20 mL of dichloromethane was added dropwise to the stirred mixture. After addition, the cooling bath was removed and the mixture was allowed to warm to ambient temperature. The stirring was continued overnight (14 h). The mixture was poured into 20 mL of water contained in a separatory funnel. The phases were separated and the aqueous phase was extracted with dichloromethane (2 × 20 mL). The combined organic phases were washed with saturated ammonium chloride solution (1 × 50 mL), saturated sodium bicarbonate solution (1 × 50 mL), and saturated sodium chloride solution (1 × 50 mL). After drying the organic phase over anhydrous magnesium sulfate, the solvent was removed to afford 3.49 g (99.5%) of 30 as a white solid. To remove the dioxolane protecting group, compound 30 was combined with 50 mL of 80% acetic acid in 100 mL round-bottom flask. The mixture was heated on a hot water bath (~90°C) for one hour and then allowed to stand overnight. The mixture was extracted with ethyl acetate (4 × 25 mL). The com-bined extracts were washed with saturated sodium bicarbonate solution until most of acetic acid was removed (4 × 50 mL) and then washed with saturated sodium chloride solution (1 × 50 mL). The mixture was dried over anhydrous magnesium sulfate. Evaporation of the solvent with a rotary evaporator yielded compound 31 as a white foam. Phenolic deacetylation of compound 31, using 20 mL pyrrolidine as described above, yielded 2.61 g of compound 14 as a white foam (97%).  13

1,3-di-O-feruloyl glycerol 15
Chloride 26 (2.55 g, 10.0 mmol) and 5 mL of dichloromethane were placed in a 25 mL round-bottom flask equipped with a magnetic stir bar. The mixture was cooled in an ice/water bath. A mixture of 0.02 g of 4-dimethyl-aminopyridine, 0.9 mL of pyridine, and 0.496 g of glycerol 32 was added dropwise to the cooled, stirred acid chloride solution. After the addition, the ice/water bath was removed and the stirred mixture was allowed to warm to ambient temperature. After standing overnight, the mixture was poured into 50 mL of dilute hydrochloric acid (1M). The mixture was extracted with dichloromethane (3 × 25 mL). The combined organic extracts were washed with saturated sodium bicarbonate solution (1 × 50 mL) and saturated sodium chloride solution (1 × 50 mL) and then dried over anhydrous magnesium sulfate. The solvent was removed using a rotary evaporator to afford 2.16 g of compound 33 as a viscous oil (82%). The crude material and 20 mL of pyrrolidine were placed in a 50 mL round-bottom flask and stirred for 10 minutes to effect deacetylation. Workup as described above yielded 1.65 g of product 15 as a viscous oil. The material was purified by flash chromatography on silica gel (30% ethyl acetate in hexanes) to afford 0.724 g of amorphous solid (32%).

1,4-di-O-feruloyl threitol 16
2,3-Di-O-isopropylidene-L-threitol 34 was prepared by the method described in Organic Syntheses, Coll. Vol. VIII, p. 155, 1993. Compound 34 (1.01 g, 6.2 mmol), 0.01 g of 4-dimethylpyridine, 1.2 mL of pyridine, and 10 mL of 1,2-dichloroethane were placed in a 100 mL round-bottom flask with a magnetic stir bar. The mixture was cooled in an ice/water bath. A solution of 3.46 g (13.6 mmol) of chloride 26 in 25 mL of dichloromethane was added dropwise to the stirred mixture. After addition, the cooling bath was removed and the mixture was allowed to warm to ambient temperature and the mixture was allowed to stand 14 h. The mixture was poured into 20 mL of water contained in a separatory funnel. The phases were separated and the aqueous phase was extracted with dichloromethane (2 × 25 mL). The combined organic phases were washed with saturated ammonium chloride solution (2 × 50 mL), saturated sodium bicarbonate solution (2 × 50 mL), and saturated sodium chloride solution (1 × 50 mL). After drying the organic phase over anhydrous magnesium sulfate, the solvent was removed to afford 3.48 g (93%) of compound 35 as a white foam. Trituration of the foam with methanol yielded 2.71 g of white crystals. Deacetylation of compound 35 (2.01 g, 3.4 mmol) with 20 mL of pyrrolidine, as above, yielded 1.77 g of the deacetylated 35, compound 36, as a white foam (100%). The material from above was dissolved in 25 mL of 1,4-dioxane in 125 mL Erlenmeyer flask. Fifty mL of 2N hydrochloric acid was added to remove the dioxolane protecting group, and the mixture was stirred using a magnetic stirrer. Cloudiness in the mixture was removed by dropwise addition of 1,4-dioxane. After stirring overnight at ambient temperature, the mixture was extracted with ethyl acetate (4 × 25 mL). The combined extracts were washed with saturated sodium bicarbonate solution until most of acetic acid was removed (4 × 50 mL) and then washed with saturated sodium chloride solution (1 × 50 mL). The mixture was dried over anhydrous magnesium sulfate. Evaporation of the solvent with a rotary evaporator yielded 1.37 g of compound 16 as a white foam (85%).

Cell Wall Lignification
Nonlignified primary cell walls (~1.8 g dry weight) isolated from maize cell suspensions [24] were stirred in 250 mL of Homopipes buffer (75 mM, pH 5. In a separate study, nonlignified primary cell walls (~2 g dry weight) isolated from maize cell suspensions [24] were stirred in 300 mL of Homopipes buffer (25 mM, pH 5.5) and H 2 O 2 (0.35 mmol in 3.5 mL of water, ~3 eq/mol of cell wall ferulate) was added dropwise over 30 min to dimerize ferulates by wall-bound peroxidases. After stirring for an additional 30 min, cell walls were artificially lignified with coniferyl alcohol 1 (1. In both studies, treatments were replicated twice in independent experiments and nonlignified controls were stirred in a solvent mixture similar to the final makeup of the lignification reaction media. Cell wall peroxidase activity at the end of monolignol addition was visually assessed with guaiacol-H 2 O 2 staining [25]. Following additions, cell walls were stirred for an additional 24 h, pelleted (5,000 × g, 15 min) and twice resuspended in 600 mL of water for 60 min, and pelleted (5,000 × g, 15 min) to remove low molecular weight dehydrogenation products. The cell wall pellets were then resuspended in 650 mL of 9:1 (v/v) acetone:water for 30 min, collected on glass-fiber filters (1.2 μm retention) and washed with 300 mL of acetone:water to remove non-bound lignins. After repeated washing with acetone, cell walls were set overnight in a hood to evaporate off acetone and then dried at 55°C and weighed.

Cell wall analyses
Acid-insoluble Klason lignin in cell wall samples (75 mg) was determined in duplicate by a two-stage hydrolysis in 12 M H 2 SO 4 at 25°C for 2 h followed by 1.6 M H 2 SO 4 at 100°C for 3 h [26]. Whole cell walls (~50 mg) from selected lignification treatments were sonicated in DMSO-d 6 with pyridine-d 5 and subjected to gel-state 2-D NMR using a Bruker-Biospin (Billerica, MA) 500 MHz Avance spectrometer equipped with an inverse gradient 5-mm TCI cryoprobe, as previously described [27].
Gas production during microbial hydrolysis and fermentation of cell walls (100 mg) at 39°C in 60 mL sealed bottles was monitored with pressure transducers for 45 h following addition of 5.7 mL of phosphate-bicarbonate buffer, 0.3 mL of reducing agent, and 4 mL of diluted rumen inoculum [28]. Filtered inoculum was prepared with a 1:2 ratio (v/v) of rumen fluid and blended bufferextracted rumen solids collected from two Holstein cows fed a total mixed ration of corn silage, corn grain, alfalfa hay, soybean meal, and supplemental vitamins and minerals [28]. Blank-corrected gas production data from two to four independent fermentation runs were fitted with a dual-pool logistic model to estimate the kinetics of microbial hydrolysis and fermentation of cell walls [28]. Freeze-dried residues remaining after cell-wall degrada-tion by rumen microflora were dissolved in 12 M H 2 SO 4 at 25°C for 2 h and analyzed for NP by the phenol-sulfuric acid assay [29] with corrections for inoculum contamination and sugar recovery. The recovery of sugars from NP was estimated by running unfermented nonlignified cell walls through the 12 M H 2 SO 4 dissolution/phenol-sulfuric acid assay procedure.

Statistical methods
Data for monolignol treatments were subjected to an analysis of variance by PROC GLM according to a randomized complete block design with two replications [30,31]. If F-tests were significant (P ≤ 0.05), then differences among monolignol treatment means were tested by the LSD procedure (P = 0.05). Unless otherwise noted, all reported treatment differences were significant at P = 0.05.

Cell wall lignification
Previous work demonstrated that artificial lignins formed in primary cell walls isolated from maize cell suspensions are structurally similar to those naturally formed in grasses [24]. In the current study, isolated maize cell walls containing bound peroxidases were artificially lignified with normal monolignols (coniferyl and sinapyl alcohols 1, 2) added alone or in combination with various phenolic monolignol substitutes 3-16. Each monolignol substitute comprised about one-third by weight of the precursor mixture, a shift in lignin composition comparable to that often observed in mutant or transgenic plants with altered lignin biosynthesis. After dissolution in a minimal volume of dioxane, normal monolignols and most monolignol substitutes readily formed aqueous solutions, but ethyl ferulate 12 and polyol esters, especially feruloyl ethylene glycol 13 and 1,3-di-O-feruloyl glycerol 15, precipitated unless a high proportion of dioxane (~25%) was used.
At the conclusion of lignification, guaiacol staining indicated weak residual peroxidase activity following addition of normal monolignols with monoferuloyl esters, very weak to no activity with diferuloyl esters, but good activity for all other treatments (data not shown). While peroxidase activity is gradually lost during lignification [32], partial substitution of monolignols with mono-or especially diferuloyl esters greatly accelerated this process. Among possible inactivation pathways [33][34][35], active site attack by ferulate phenoxy radicals or enzyme precipitation due to sorption or cross-linking to polymeric products are most plausible given the extremely high substrate preference of maize peroxidase for ferulate esters [36], the bi-phenolic cross-linking capability of diferuloyl polyol esters, and their aforementioned poor aqueous solubility.
Based on Klason lignin analysis, inclusion of feruloylquinic acid 8, methyl caffeate 5, or especially caffeoylquinic acid 6 considerably depressed lignin formation while methyl ferulate 7 readily copolymerized with monolignols to form wall bound lignin (Table 1). Methyl and quinic acid esters of ferulic and caffeic acids are plant extractives, of which caffeoylquinic acid (also known as chlorogenic acid) and feruloylquinic acid are especially abundant in coffee and some other beverages and foods [37]. To tentatively investigate the cause of depressed lignification, cell walls from selected treatments were subjected to gel-state 2D NMR. Spectra of cell walls lignified with normal monolignols (Figure 3a) or monolignols plus feruloylquinic acid (Figure 3b) both contained only correlations of guaiacyl units (derived from coniferyl alcohol) and syringyl units (derived from sinapyl alcohol); the absence of ferulate correlations in the latter spectrum suggests that feruloylquinic acid was not incorporated into wall bound lignin. It must, however, be pointed out that correlations of native cell wall ferulates comprising ~10% of the lignin polymer are also missing from spectra of control cell walls (Figure 3a). Thus extensive polymerization of ferulates into a wide array of coupling products [4][5][6] may sufficiently disperse their correlations or coalesce them with adjacent guaiacyl correlations to render them undetectable in our 2D NMR experiments. In any case, it seems that depressed lignin formation with these monolignol substitutes was associated with the presence of quinic or caffeic acid moieties. Highly hydrophilic quinic acid moieties might hinder the association and copolymerization of hydroxycinnamate radicals with hydrophobic lignin polymers, but this or alternative mechanisms require further study. Peroxidase-generated radicals of caffeate can undergo homodimerization or cross-coupling reactions with ferulate and sinapate [38], but in the current study any coupling products involving caffeate apparently did not become incorporated into polymeric wall-bound lignin. Alternatively, poor incorporation of caffeate into lignin may be due to its tendency to form quinones, which avoid radical coupling reactions characteristic of monolignols [39]; to our knowledge oxidative coupling reactions between caffeate and monolignols have not been reported and call for additional study.

Cell wall hydrolysis and fermentation by rumen microflora
Because maize cell walls used in our lignification experiments were fully disaggregated and extremely thin (~1 μm), they exposed a large surface area that could be quickly colonized by rumen microflora [24]. Furthermore, rumen microflora produce a diverse array of hydrolases capable of rapidly degrading essentially all organic components except lignin [46]. Together, these factors allowed us assess how lignin alterations affect cell wall degradation at the molecular level, free from other confounding constraints that also hinder cell wall degradation such as the anatomical structure of plant tissues or the use of enzyme preparations lacking key hydrolase activities [12]. In addition to measuring residual NP at the endpoint of digestion, we continuously monitored gas production by rumen bacteria to estimate lag time before fermentation commenced as well as the rate and extent of cell wall fermentation. Although the production of fermentation gasses by rumen bacteria is substrate dependent [14], our use of a common source of cell walls for all lignification treatments insured that shifts in gas production mirrored the enzymatic degradation of cell walls.
As in a previous study [47], cell walls exhibited a biphasic production of fermentation gasses that was best described by a two-pool logistic model with two discrete lag times. For nonlignified cell walls, gas production for both pools commenced after a lag of <2.5 h, but the primary pool had roughly a 5-fold greater rate and extent of gas production than the secondary pool (Table 1 and 2). Nonlignified cell walls were extensively degraded, leaving only 21 mg/g of NP. Based on the polysaccharide composition and fermentation characteristics of primary maize walls, we previously speculated the larger pool represented rapidly degraded noncellulosic polysaccharides, while the smaller pool represented slowly degraded cellulose [47]. The relative rate and extent of gas production Table 2: Klason lignin, in-vitro ruminal fermentation kinetics a , nonfermentable polysaccharides (NP), gas reduction per unit lignin (GRL) b , and nonfermentable polysaccharide accumulation per unit lignin (NPAL) c for nonlignified and artificially lignified cell walls of maize.

Monolignols
Lignin mg/g L 1 (h) k 1 (h -1 ) A (mL/g) L 2 (h) k 2 (h -1 ) B (mL/g) AB (mL/g) NP (mg/g) GRL NPAL  for these pools were, however, strikingly similar to two fiber digestion pools observed in graminaceous crops [48], which undoubtedly possess a markedly different structural polysaccharide makeup than our primary maize cell walls. Clearly, further work is warranted to reveal the exact structural nature of the rapidly and slowly degraded pools in grass cell walls.
In both experiments, total gas AB declined 24% and NP increased by ~100 mg/g when cell walls were artificially lignified with normal monolignols (coniferyl and sinapyl alcohols 1, 2) to a Klason lignin content of 150 mg/g ( Table 1 and 2). Lignification with normal monolignols had little effect on lag L 1 , but rate k 1 and gas production A for the large rapidly degraded pool decreased by 57 and 24%, respectively. Conversely for the smaller pool, lignification with normal monolignols dramatically increased lag L 2 by 10.7 to 17.5 h and decreased gas production B by 30 to 70%, with comparatively minor effects on rate k 2 . Although effects on the rapidly and slowly degraded pools differed, the overall impact of lignification was to obstruct cell wall hydrolysis and fermentation by rumen bacteria. Similar findings were previously reported in studies utilizing coarsely ground grass stems and isolated tissues from grasses [49,50]. Thus, lignification is an impediment to the enzymatic hydrolysis and fermentation of cell walls at both the molecular and tissue level in plants.
In the first experiment, substituting one-third of normal monolignols with methyl caffeate 5, caffeoylquinic acid 6, or feruloylquinic acid 8 strikingly improved total gas AB and reduced NP for lignified cell walls, primarily through increasing rate k 1 and gas production A from the large rapidly digested pool (Table 1). These monolignol substitutes also reduced lag times for both pools, particularly lag L 2 of the small slowly digested pool. The impact of these shifts on gas production from cell walls lignified with feruloylquinic acid is shown in Figure 4. These monolignol substitutes also lessened gas reduction per unit lignin (GRL) and nonfermentable polysaccharide accumulation per unit of lignin (NPAL), indicating that factors in addition to reduced lignin content contributed to enhanced cell wall hydrolysis and fermentation (Table  1). Caffeate, like other benzene-1,2-diols, might internally trap lignin quinone-methide intermediates [51,52] to limit the cross-linking of lignin to polysaccharides and enhance the enzymatic hydrolysis of cell walls [10,11]. Incorporation of hydrophilic caffeoyl or quinic acid moieties into lignin might also enhance penetration or limit irreversible binding of hydrolytic enzymes to the lignocellulosic matrix, but identifying actual causative factors requires further study.
While not conclusively shown by the kinetics of gas production, reductions in NP and NPAL indicated a modest improvement in cell wall hydrolysis occurred when one-third of normal monolignols were substituted with guaiacylglycerol 4, epigallocatechin 10, or epigallocatechin gallate 11 (Table 1). In contrast, dihydroconiferyl alcohol 3 had no effect, while incorporation of epicatechin 9 and especially methyl ferulate 7 adversely affected cell wall hydrolysis and fermentation (Figure 4) by decreasing rate k 1 and gas production A and AB, and by increasing NP, GRL, and NPAL. Among the catechins, those with 1,2,3-triol functionality enhanced cell wall hydrolysis and fermentation, primarily by increasing rate k 1 and gas production AB and by decreasing lag L 2 , NP, GRL, and NPAL--again implicating reduced hydrophobicity or cross-linking as playing a role--but underlying mechanisms await revelation. In contrast, increased sidechain hydroxylation of incorporated phenylpropanoids (i.e., dihydroconiferyl alcohol vs. guaiacylglycerol) slightly increased cell wall hydrolysis as indicated by modest reductions (P <0.1) in NP or NPAL.
In the second experiment, substituting one-third of a normal monolignol (coniferyl alcohol) with various monoferuloyl esters did not alter cell wall hydrolysis and   (Table 2). But within the monoferuloyl ester treatments, increases in ester group hydroxylation slightly increased rate k 1 and a reduced lag L 2 , NP and NPAL. Incorporating 1,4-di-O-diferuloyl threitol 16 and particularly1,3-di-O-feruloyl glycerol 15 into lignin, however, modestly increased rate k 1 and reduced both NP and NPAL compared to cell walls lignified solely with coniferyl alcohol. While having relatively small effects on the inherent degradability of non-pretreated cell walls, incorporation of diferuloyl polyol esters (and other readily cleaved bi-phenolic conjugates like epigallocatechin gallate) could substantially boost pretreatment efficacy for removing lignin prior to biomass saccharification. Other biomimetic model studies by our group recently demonstrated the potential for bi-phenolic conjugates to improve cell wall delignification and saccharification [8].

Conclusion
To identify new targets for lignin bioengineering, we artificially lignified maize cell walls with normal monolignols and a variety of phenolic monolignol substitutes normally associated with other plant metabolic pathways. Lignification with normal monolignols severely impeded structural polysaccharide hydrolysis and fermentation by rumen microflora at the molecular level in cell walls. Inclusion of feruloylquinic acid, methyl caffeate, or caffeoylquinic acid with normal monolignols considerably depressed lignin formation and substantially improved cell wall hydrolysis and fermentation. In contrast, epicatechin, epigallocatechin, and epigallocatechin gallate readily formed copolymer lignins with monolignols as did the phenylpropanoids dihydroconiferyl alcohol and guaiacylglycerol. Cell wall fermentability was moderately enhanced by catechins with 1,2,3-triol functionality, while inclusion of phenylpropanoids with extensive sidechain hydroxylation had little impact. Mono-or diferuloyl esters with various aliphatic or polyol groups readily formed copolymers with monolignols, but their tendency to accelerate peroxidase inactivation slightly diminished lignin formation in some cases. Relative to cell walls lignified with normal monolignols, copolymerization of mono-or diferuloyl esters into lignin had negative to weakly positive effects on cell wall hydrolysis and fermentability, which were in part dependent on the quantity of lignin formed and the degree of ester group hydroxylation. Overall, monolignol substitutes improved the inherent degradability of cell walls by depressing lignin formation or possibly by reducing lignin hydrophobicity or cross-linking to structural polysaccharides. Some monolignol substitutes, chiefly readily cleaved biphenolic conjugates like epigallocatechin gallate or diferuloyl polyol esters are expected to greatly boost enzymatic saccharification of cell walls following chemical pretreatment. In a forthcoming paper, we will examine the impact of the aforementioned monolignol substitutes on the enzymatic saccharification of intact and chemically pretreated cell walls. In subsequent model studies, we will examine these and other monolignol substitutes in greater detail to identify promising bioengineering targets for improving plant fiber utilization.