Characterization of Miscanthus cell wall polymers

Abstract Efficient utilization of lignocellulosic Miscanthus biomass for the production of biochemicals, such as ethanol, is challenging due to its recalcitrance, which is influenced by the individual plant cell wall polymers and their interactions. Lignocellulosic biomass composition differs depending on several factors, such as plant age, harvest date, organ type, and genotype. Here, four selected Miscanthus genotypes (Miscanthus sinensis, Miscanthus sacchariflorus, Miscanthus × giganteus, Miscanthus sinensis × Miscanthus sacchariflorus hybrid) were grown and harvested, separated into stems and leaves, and characterized for their non‐starch polysaccharide composition and structures, lignin contents and structures, and hydroxycinnamate profiles (monomers and ferulic acid dehydrodimers). Polysaccharides of all genotypes are mainly composed of cellulose and low‐substituted arabinoxylans. Ratios of hemicelluloses to cellulose were comparable, with the exception of Miscanthus sinensis that showed a higher hemicellulose/cellulose ratio. Lignin contents of Miscanthus stems were higher than those of Miscanthus leaves. Considering the same organs, the four genotypes did not differ in their Klason lignin contents, but Miscanthus × giganteus showed the highest acetylbromide soluble lignin content. Lignin polymers isolated from stems varied in their S/G ratios and linkage type distributions across genotypes. p‐Coumaric acid was the most abundant ester‐bound hydroxycinnamte monomer in all samples. Ferulic acid dehydrodimers were analyzed as cell wall cross‐links, with 8‐5‐coupled diferulic acid being the main dimer, followed by 8‐O‐4‐, and 5‐5‐diferulic acid. Contents of p‐coumaric acid, ferulic acid, and ferulic acid dimers varied depending on genotype and organ type. The largest amount of cell wall cross‐links was analyzed for Miscanthus sinensis.

Lignocellulosic biomass is made up of plant cell wall polymers. Their composition is complex and often determines biomass recalcitrance. Cell walls of grasses are mainly composed of cellulose, hemicelluloses (arabinoxylans and mixed-linked β-glucans in particular grasses), and lignin (Vogel, 2008). In addition, hydroxycinnamates, such as ferulic and p-coumaric acids are bound to polysaccharides and lignin. Ferulic acid is usually attached to arabinoxylans and is responsible for the formation of cross-links between polysaccharides, and polysaccharides and lignin, whereas p-coumaric acid is usually bound to lignin (Bunzel, 2010;Ralph, 2010) with only minor amounts of p-coumaric acid being attached to arabinoxylans (Allerdings, Ralph, Steinhart, & Bunzel, 2006). In order to obtain biochemicals and biofuels from lignocellulosic biomass, efficient pretreatment procedures, followed by enzymatic hydrolysis, and microbial fermentation processes are required (Rastogi & Shrivastava, 2017). Biomass pretreatment removes or alters lignin, a key factor of biomass recalcitrance, and loosens the cell wall polymer network. Enzymatic hydrolysis converts cellulose and hemicelluloses into monomers, which are fermented using various microorganisms. The efficiency of these processes varies considerably depending on biomass composition. Thus, cell wall composition data of various Miscanthus genotypes is crucial to select suitable genotypes for potential biochemical and biofuel production.
The gross composition of Miscanthus biomass, i.e., cellulose, hemicellulose, and lignin contents, has been described earlier (Arnoult & Brancourt-Hulmel, 2015;da Costa et al., 2017;Hodgson et al., 2011;Lygin et al., 2011;Zhao et al., 2014). A few authors also report some structural details such as lignin monomer composition and/ or lignin linkage type profiles (Cheng, Sorek, Zimmermann, Wemmer, & Pauly, 2013;Le Ngoc Huyen, Rémond, Dheilly, & Chabbert, 2010;Villaverde, Li, Ek, Ligero, & Vega, 2009), and acetylation degree of polysaccharides (Van der Weijde et al., 2017). However, these data are mostly limited to Miscanthus × giganteus. Studies on structural details of cell wall polymers such as lignin linkage type profiles and ferulic acid dehydrodimer contents specific to other Miscanthus genotypes are rarely available. In addition, published studies often describe the relationship between individual constituents and fermentation yields without analyzing the whole cell wall composition in detail. In general, polysaccharide contents mainly determine the fermentation yield. However, efficient biomass utilization is determined by the whole cell wall composition because structural features of cell wall polymers and polymer interactions also affect biomass recalcitrance. Lignin is a key factor in determining biomass recalcitrance, and the biomass needs to be pretreated to remove lignin. Besides lignin contents, the structural composition of lignin polymers may influence pretreatment processes, e.g. high portions of β-O-4-ethers can be easily cleaved during pretreatment processes. Therefore, it is important to characterize all cell wall polymers of different Miscanthus genotypes in detail instead of limiting the analytical work to specific polymers to better evaluate the genotypes for biomass utilization. For this reason, we analyzed four Miscanthus genotypes for their cell wall composition, and all relevant polymers were characterized in detail.

| Field trial description
Miscanthus material was harvested from a field trial established as a part of the European project OPTIMISC to compare 15 Miscanthus genotypes at six sites across Europe. A detailed description of the field trials and genetic material is provided in Kalinina et al. (2017). For this study four Miscanthus genotypes were selected from the location Stuttgart (lat. 48°74′N; long. 8°93′E). The field trial was planted from in-vitro propagated plants on 22nd May 2012 on former arable land. Before planting out the plugs, the ground was prepared as follows: Weeds were removed with glyphosate, inversion ploughed and harrowed to produce a fine tilth. After planting, the plugs were watered to provide a good hydraulic contact between the soil and plug. The genotypes were established in randomized complete block design with three replications and each plot comprised of 5 × 5 m with 49 plants (plugs) per plot (resulting in a density close to 2 plants per m 2 ).
Post-planting herbicide was not applied in the first year, and weeds were controlled mechanically. In the first year (2012), fertilizer was applied at all the sites at rates of 44 and 110 kg/ha phosphorus (P) and potassium (K), respectively. No nitrogen (N) fertilizer was applied that year to avoid stimulating weed growth. From the second year onwards fertilizer was applied at a rate of 100 kg/ha P, 140 kg/ha K and 60 kg/ha N to ensure non-limiting crop nutrition.

| Genotype description
The selected genotypes belong to the species of Miscanthus sinensis (OPM-11), Miscanthus sacchariflorus (OPM-3), or their hybrids Miscanthus × giganteus (OPM-9) and Miscanthus sinensis × Miscanthus sacchariflorus (OPM-6; Table 1). The sterile clone Miscanthus × giganteus is a naturally occurring interspecific hybrid (Miscanthus sinensis × Miscanthus sacchariflorus) that has been used commercially in Europe for biomass production for over a decade and is therefore selected as standard reference clone. In general, Miscanthus sinensis interspecific hybrids have thinner and shorter stems than Miscanthus sacchariflorus and their hybrids, which combined lead to lower yields in trials with the scientific standard planting density of 20,000 plants/ha (Iqbal & Lewandowski, 2014). Additionally, lower yields of Miscanthus sinensis hybrids are caused by earlier senescence than Miscanthus sacchariflorus, which also impacts biomass composition . Genotypes OPM-3, -6 and -11 stem from the Miscanthus breeding program led by the University of Aberystwyth. Over the past decade, it focused on producing interspecific Miscanthus sinensis × M. sacchariflorus hybrids with high yield, cold or other stress tolerance, and seed production (Clifton-Brown et al., 2017). The standard genotype Miscanthus × giganteus was also provided by University of Aberystwyth.

| Sample collection
For this study, biomass grown in the fifth year of the plantation was harvested on 20th March 2017, which is the preferred harvest time for Miscanthus biomass. Samples were harvested at the same time to minimize environmental effects on the cell wall composition. For biomass yield assessment samples were collected using manual cutters through harvesting the middle 2 m 2 of a plot at a cutting height of~5 cm. Sub-samples were taken randomly from the 2 m 2 cut and separated into leaves and stems. The flowers were considered as part of leaves. The collected samples were chopped and dried at 60°C for 48 hr in a cabinet dryer. The dried leave and stem fractions were milled through a cutting mill (SM 200-Retsch, Haan, Germany) equipped with 1 mm sieve size.

| Preparation of cell wall material
Preparative amounts of cell wall material were obtained using the principles of AOAC method 985.29 (Prosky et al., 1985) with minor modifications. Dried material was suspended with 1.5 ml of α-amylase in 200 ml of sodium phosphate buffer (0.08 M, pH 6.2) and stirred for 20 min at 92°C. Samples were cooled to room temperature, the pH was adjusted to 7.5 with 0.275 M NaOH, protease (600 µl) was added, and the samples were incubated for 30 min at 60°C. After cooling the samples to room temperature, the pH was adjusted to 4.5 using 0.325 M HCl. Amyloglucosidase (700 µl) was added, and the samples were incubated for 30 min at 60°C. A fourfold amount (v/v) of ethanol (99%) was added, and precipitation was completed overnight. Following centrifugation, residues were washed three times with 78% ethanol and acetone, and dried at 60°C in a vacuum oven. The isolated cell wall material was further Soxhlet extracted for 6 hr each with ethanol, ethyl acetate, and n-hexane.

| Polysaccharide analysis
Monomer composition of cell wall polysaccharides was analyzed after H 2 SO 4 hydrolysis (Saeman, Bubl, & Harris, 1945); Klason lignin filtrates (see Lignin analysis) were used for this purpose. The filtrates were diluted and analyzed by high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD; Wefers, Gmeiner, Tyl, & Bunzel, 2015). Methylation analysis (Pettolino, Walsh, Fincher, & Bacic, 2012) was used to determine interunit linkages of the polysaccharide constituents as detailed in Wefers et al. (2015). In brief, methylation was performed in DMSO/NaOH using 1 ml of methyl iodide followed by dichloromethane extraction. The methylation step was performed twice. Hydrolysis of the methylated polysaccharides was performed using 2 ml of 2 M trifluoroacetic acid for 90 min at 120°C, followed by reduction with 20 mg of NaBD 4 , acetylation with acetic anhydride (3 ml) and 1-methylimidazole (450 µl), and extraction of the partially methylated alditol acetates (PMAAs) into dichloromethane. PMAAs were identified by GC-MS and semiquantitatively determined by GC-FID using molar response factors (Sweet, Shapiro, & Albersheim, 1975

| Determination of lignin contents
Lignin contents were determined as Klason lignin (Adams, 1965;Theander & Westerlund, 1986) and as acetyl bromide soluble lignin (ABSL; Hatfield, Grabber, Ralph, & Brei, 1999;Iiyama & Wallis, 1990;Johnson, Moore, & Zank, 1961) as detailed in Bunzel, Schüßler, and Tchetseubu Saha (2011) and Bunzel, Seiler, and Steinhart (2005). To determine Klason lignin contents, cell wall material was suspended in 5 ml of 12 M H 2 SO 4 , and the suspensions were stirred on ice for 30 min and additionally for 2 hr at room temperature. The acid was diluted with 32.5 ml of H 2 O, and hydrolysis was performed at 100°C for 3 hr. After filtration through glass microfiber filters (1.6 µm; Whatman, Little Chalfont, England), the residues were washed acid free and dried at 60°C. The filtrates were used for monosaccharide determination (see Polysaccharide analysis). Klason lignin contents were corrected for residual ash and protein contents. Ash contents were analyzed gravimetrically after incineration of the samples at 550°C for 5 hr. Nitrogen contents of the Klason lignin samples were determined after Kjeldahl digestion. Liberated ammonium-N was analyzed using an ammonium selective electrode (Thermo Scientific, Waltham, MA). Protein contents were calculated using the general conversion factor 6.25. In preparation of the ABSL assay, cell wall material was ball-milled using a PM100 planetary ball mill with zirconium dioxide balls (Retsch, Haan, Germany; 1 g, 600 rpm, 3.25 hr, 5 min interval with 5 min break between millings), following enzyme digestion using the carbohydrate mixture Driselase (60 mg of Driselase and 100 ml of H 2 O per gram of milled sample material). Enzyme-digested material (20-25 mg) was suspended in 4 ml of 25% acetyl bromide in glacial acetic acid and stirred for 2 hr at 50°C. The solution was diluted with 12 ml of glacial acetic acid. Sample solution (1 ml or 0.5 ml), 2.5 ml of glacial acetic acid, 1.5 ml of 0.3 M NaOH, and 0.5 ml of a 0.5 M hydroxylamine hydrochloride solution were mixed, the volume was adjusted to 10 ml with glacial acetic acid, and the absorbance of the mixture was read at 280 nm on a Jasco V-550 spectrometer (Jasco, Groß-Umstadt, Germany). Lignin contents were calculated using an absorption coefficient of 20 ml cm −1 mg −1 (Iiyama & Wallis, 1990).

| Determination of phenolic monomers and ferulic acid dehydrodimers and -trimers
Cell wall material (15-75 mg) was saponified (5 ml of 2 M NaOH) for 18 hr under nitrogen and protected from light (Dobberstein & Bunzel, 2010). Following acidification, the internal standard compounds trans-o-coumaric acid (20-150 µg, depending on the sample) and 5-5 (methylated)-dehydrodiferulic acid (0.5-3 µg, depending on the sample) were added. Extraction was performed three times using 2 ml of diethylether. Extracts were dried under nitrogen, redissolved in 0.5 ml of tetrahydrofuran/H 2 O (50:50, v/v), and analyzed by HPLC-DAD on a 250 mm × 4.6 mm, 5 µm Luna phenylhexyl column (Phenomenex, Torrance, CA; Dobberstein & Bunzel, 2010). Quantitation of phenolic monomers was performed at 308 nm for transp-coumaric-acid, at 294 nm for cis-p-coumaric acid, at 321 nm for trans-ferulic acid, at 310 nm for cis-ferulic acid, and at 323 nm for trans-o-coumaric acid. Quantitation of ferulic acid dehydrodimers and -trimers was conducted at 280 nm using the correction factors described in literature (Dobberstein & Bunzel, 2010). Because the concentration range of some diferulic acids (DFA) exceeded the concentration range tested by Dobberstein and Bunzel (2010), the linear response was tested and confirmed for the concentrations measured in the samples.

| RESULTS AND DISCUSSION
To study cell wall structural details, four Miscanthus genotypes were selected. The main criterion for the selection of the four genotypes tested in this study was a high dry matter yield potential and contrasting biomass composition .

| Polysaccharide composition
Being a monocotyledonous plant, Miscanthus cell wall polysaccharide composition was described to show typical traits of grasses in the past (De Souza et al., 2015;Domon et al., 2013;Le Ngoc Huyen et al., 2010;Lygin et al., 2011). Generally, type II primary cell walls of grasses are mainly composed of cellulose and arabinoxylans as predominant hemicellulosic polymers. Other hemicelluloses besides arabinoxylans as well as pectins are present in small amounts only (Carpita & Gibeaut, 1993;Vogel, 2008). However, detailed structural analyses of Miscanthus cell wall polysaccharides are rarely reported in literature (da Costa et al., 2017).
Here, cell wall polysaccharides of leaves (including flowers) and stems from different Miscanthus genotypes were characterized by monosaccharide analysis after H 2 SO 4 hydrolysis and by methylation analysis (Tables 2 and 3). The dominant neutral sugars were glucose (52.6 mol%-63.0 mol%), xylose (34.1 mol%-40.4 mol%), and arabinose (2.6 mol%-6.6 mol%), suggesting cellulose and arabinoxylans as main cell wall polysaccharides. Methylation analysis data showed that most of the glucopyranose units were 1,4-linked, confirming cellulose as dominant polysaccharide. H 2 SO 4 hydrolysis of stem polysaccharides resulted in slightly higher portions of glucose compared to glucose portions of leaf polysaccharides, suggesting higher cellulose portions in Miscanthus stems, which was already shown for different Miscanthus sinensis genotypes (Van der Weijde et al., 2017). In addition, low portions of PMAAs representing 1,3-linked glucopyranose units were identified in both stems and leaves, suggesting either mixed-linked β-glucans and/or callose in Miscanthus cell walls. Also, the identification of low portions of PMAAs representing 1,4,6-linked glucopyranose units demonstrates low amounts of xyloglucans in all samples analyzed.
Polysaccharide characterization analyses reveal differences among the Miscanthus genotypes studied. Both leaf and stem polysaccharides of genotype OPM-11 had higher portions of xylose (39.7 mol% for leaves and 40.4 mol% for stems) and less portions of glucose (52.6 mol% for leaves and 55.1 mol% for stems) than the other genotypes, suggesting higher portions of arabinoxylans but lower portions of cellulose for OPM-11 cell wall polysaccharides compared to those of genotypes OPM-3, OPM-6, and OPM-9. Hodgson et al. (2011) showed that Miscanthus × giganteus and Miscanthus sacchariflorus genotypes contain higher contents of cellulose and lower contents of hemicelluloses than the Miscanthus sinensis genotypes, too. The xylan substitution degree was comparably low among the Miscanthus genotypes analyzed here as demonstrated by similar, low arabinose/xylose ratios.
It is important to note, that polysaccharide characterization data do not represent absolute amounts of polysaccharides in the lignocellulosic biomass because the principles of the methods used in this study do not allow for a quantitative polysaccharide analysis. This is mostly due to

Notes.
Galactose was detected in leaves of OPM-3 and OPM-6, and in the stems of all genotypes but concentrations were below the calibration range. Monosaccharide yields (g/100 g dry weight) of Miscanthus varying susceptibility of glycosidic linkages to acid hydrolysis, and due to varying stability of the liberated monosaccharides in hot acidic solutions. However, calculated monosaccharide amounts liberated on a dry weight basis after H 2 SO 4 hydrolysis can be found in Table 2. Polysaccharide contents determine theoretical yields of fermentable monosaccharides. Efficient biomass utilization requires extensive hydrolysis of both cellulose and hemicelluloses. However, the efficiency of polysaccharide hydrolysis is affected, among others, by the overall biomass composition and interactions of cell wall polymers. The degradability of cellulose to glucose units is influenced by different factors, including its crystallinity and its interactions within the cell wall network (Chang & Holtzapple, 2000;Puri, 1984). Hemicelluloses are often described to act as physical barriers that limit enzymatic hydrolysis of cellulose . In general, different pretreatments are used to solubilize hemicelluloses (and to reduce lignin contents), which enhance enzymatic degradability of cellulose (Yang, Chen, Chen, Wang, & Sun, 2015). However, hemicelluloses were also found to be positively associated to cell wall degradability and saccharification efficiency, depending on their structural features (Li et al., 2013;Van der Weijde et al., 2017;Xu et al., 2012). According to those studies, the higher hemicellulose/cellulose ratio of OPM-11 may positively affect the polysaccharide degradability of this genotype. Hemicelluloses such as arabinoxylans are efficiently hydrolyzed using acidic pretreatments. However, liberated sugars are not completely stable and enzyme inhibitors such as furfural derivatives may be formed (Palmqvist & Hahn-Hägerdal, 2000). In addition, both interactions of cell wall polymers and lignin contents may have greater impact on biomass utilization than simple hemicellulose to cellulose ratios.

| Lignin contents and structures
Lignin contents of Miscanthus leaves and stems were determined as both Klason lignin and ABSL. Both methods have limitations in analyzing absolute lignin contents, which also becomes obvious from the different lignin contents analyzed here using these methods (Figure 1). Klason lignin is measured gravimetrically as an acid insoluble residue and can be overestimated due to non-hydrolyzed nonlignin compounds such as proteins, waxes, and cutin (Bunzel et al., 2011). To minimize these effects, dried plant samples were Soxhlet extracted using different solvents, and the Klason lignin preparations were corrected for residual ash and protein contents as suggested by most protocols. ABSL is measured spectrophotometrically at 280 nm after solubilization, and may be overestimated due to polysaccharide-bound phenolic compounds that absorb at 280 nm, too. For this reason, cell wall material was enzyme digested using a cell wall polysaccharide degrading enzyme mixture. However, ABSL contents may also be underestimated due to incomplete solubility of the lignin polymer in acetyl bromide and glacial acetic acid. Thus, comparison of absolute lignin contents with literature data T A B L E 3 Ratios (mol%) of partially methylated alditol acetates (PMAAs) resulting from methylation analysis of Miscanthus leave and stem cell wall material (n = 1) | 197 is often difficult due to different methods and/or protocols used for lignin determination. As expected, Klason lignin and ABSL contents of Miscanthus stems were higher than lignin contents of Miscanthus leaves. Klason lignin contents ranged from 20.14 ± 0.62 g/100 g dry weight to 22.14 ± 1.10 g/100 g dry weight in cell wall material from Miscanthus stems and from 17.31 ± 0.96 g/100 g dry weight to 18.85 ± 0.67 g/ 100 g dry weight in cell wall material from Miscanthus leaves. ABSL contents ranged from 13.32 ± 0.47 g/100 g dry weight to 17.49 ± 0.41 g/100 g dry weight in cell wall material from Miscanthus stems and from 7.78 ± 0.03 g/ 100 g dry weight to 9.16 ± 0.07 g/100 g dry weight in cell wall material from Miscanthus leaves. Klason lignin contents of the different genotypes were not statistically different within the two groups (leaves and stems; ANOVA, Tukey test, α = 0.05). In contrast, differences in the ABSL contents of the analyzed genotypes are statistically significant within the stem and leave fractions (ANOVA, Tukey test, α = 0.05). In the leave fraction, genotypes OPM-6 and OPM-11 had slightly higher ABSL contents than OPM-3 and OPM-9. In the stem fraction, which is the quantitatively dominant fraction, the highest ABSL content was analyzed for OPM-9. Considering leave/stem ratios of about 0.20 (OPM-3, OPM-9), 0.31 (OPM-11), and 0.48 (OPM-6) it can be suggested that Miscanthus plants of OPM-9 contain the highest ABSL contents among the genotypes studied here.
Besides lignin contents, structural composition of lignin polymers is suggested to play an important role for efficient biomass saccharification (Li, Pu, & Ragauskas, 2016;Simmons, Loqué, & Ralph, 2010). Lignin is a complex polymer, which is mainly composed of three monomers: pcoumaryl, coniferyl, and sinapyl alcohol. These monomers form p-hydroxyphenyl (H), G-, and S-units through oxidative radical coupling . However, lignin structures vary among plant species, tissues, cell types, and maturation stage of the plants, and other phenolic compounds may be incorporated into the polymer, too.
Using whole cell wall 2D-NMR analysis, lignins of Miscanthus leaves were characterized as being rich in Gunits (81.4%-92.1%) with lower portions of S-units (7.9%-18.6%), resulting in S/G ratios of 0.09-0.23. In contrast, the portions of S-units (25.0%-41.1%) were higher in Miscanthus stems (S/G ratios 0.33-0.70; Table 4). In literature, a few authors describe the monolignol composition using F I G U R E 1 Klason lignin and acetylbromide soluble lignin (ABSL) contents of different Miscanthus genotypes (Table 1) separated into stems and leaves. Means labeled with different letters are statistically different (ANOVA, Tukey test, α = 0.05, leaves and stems are tested separately). n = 3 NMR spectroscopy to analyze whole cell wall material and milled wood lignins from Miscanthus × giganteus plants (Cheng et al., 2013;El Hage et al., 2009;Villaverde et al., 2009). Analyzed S/G ratios of about 0.7-0.8 are in agreement with our data. El Hage et al. (2009) and Cheng et al. (2013) also determined small portions of H-units (about 5%). Here, small signals were detected in the particular region of the spectra where the carbon-2/carbon-6 correlations of H-units usually appear. However, due to the difficult, not fully unambiguous assignment of the characteristic signal (Kim et al., 2017), H-units were not determined.
Within the leave and stem fractions, different monolignol compositions were observed for the different genotypes. The S/G ratio of lignins from OPM-3 stems (0.70) was significantly higher than the S/G ratios of lignins from the other genotypes (0.33-0.54; ANOVA, Tukey test, α = 0.05). The lowest S/G ratio was analyzed for lignins from OPM-11 stems (0.33). The S/G ratios of lignins from OPM-6 and OPM-9 stems were similar (0.51 and 0.54). The monolignol composition of lignins from leaves of genotypes OPM-6 and OPM-11 (S/G ratios: 0.23 and 0.20) differs significantly from the monolignol composition of lignins from leaves of OPM-3 and OPM-9 (S/G ratio: 0.09). Considering the low leave/stem ratios, it can be suggested that lignins from Miscanthus plants of genotype OPM-3 contain the highest portions of sinapyl alcohol.
Only slight differences in linkage type distributions were observed for lignins isolated from Miscanthus leaves across genotypes. In contrast, lignins isolated from

| 199
Miscanthus stems of OPM-9 and OPM-11 showed different linkage type profiles than those from stems of genotypes OPM-3 and OPM-6, which have similar linkage type distributions. Lignins isolated from stems of OPM-11 had the lowest portion of β-O-4 linkages (59.7%). In contrast, the highest portion of β-O-4 linkages (76.7%) was observed for lignins isolated from OPM-9 stems. In addition, phenylcoumaran units (3.4%) were less abundant in this genotype compared to phenylcoumaran portions of the other genotypes (8.1%-10.7%). β-O-4-Ethers are more labile than carbon-carbon bonds (e.g., β-5, β-β, β-1) and can be cleaved during pretreatment of the biomass, with the extent being dependent on the pretreatment type (Pu, Hu, Huang, & Ragauskas, 2015). For this reason, genotypes with high portions of β-O-4-ethers may be favorable to achieve efficient pretreatment of biomass resulting in high fermentation yields. Often, high syringyl lignins contain higher portions of β-ether linkages because during cross-coupling monolignols couple to a syringyl unit (from oligolignols) only through β-O-4 linkages . However, in our study lignin isolated from stems of genotype OPM-3, which had the highest S/ G ratio, did not contain the highest β-O-4 portion.

| Phenolic monomers and ferulic acid dehydrodimers and -trimers
Grasses usually incorporate ferulic and p-coumaric acid into their cell walls. Ferulic acid is mainly attached to the arabinofuranose side-chains of arabinoxylans. Formation of ferulate dimers, trimers, or tetramers results in cross-linked cell wall polysaccharides (Bunzel, 2010). In addition, ferulates play an important role in the formation of cross-links between polysaccharides and lignin (Jacquet, Pollet, & Lapierre, 1995;Ralph, Grabber, & Hatfield, 1995). Although p-coumaric acid is found to be linked to polysaccharides to a lesser extent, too (Allerdings et al., 2006), it is predominantly attached to lignins (Ralph, 2010). p-Coumaric acid is described to be ester-linked to the γ-position of lignins, which indicates the formation of p-coumaric acid-monolignol conjugates prior to the lignification process (Crestini & Argyropoulos, 1997;. Here, p-coumaric and ferulic acid were identified as the most abundant hydroxycinnamate monomers (Table 6), which is in accordance with literature data (Belmokhtar, Habrant, Lopes Ferreira, & Chabbert, 2013;da Costa et al., 2017;Le Ngoc Huyen et al., 2010;Lygin et al., 2011). The total hydroxycinnamate monomer contents of Miscanthus stems of OPM-3, -6, and -9 (about 2,102-2,450 mg/ 100 g dry weight) were about 1.5-2.0-fold higher than in the leaves of the corresponding genotypes (about 1,196-1,486 mg/100 g dry weight), which results mainly from higher p-coumaric acid contents in the stems of these genotypes. In contrast, contents of total hydroxycinnamate monomers were comparable (about 1720 and 1814 mg/ 100 g dry weight) in stems and leaves of OPM-11. Transp-coumaric acid was the most abundant phenolic acid in all samples (about 721-1,096 mg/100 g dry weight in the leave fraction; about 1,174-1,914 mg/100 g dry weight in the stem fraction), whereas trans-ferulic acid contents were about 2-3-fold lower (about 278-530 mg/100 g dry weight). p-Coumaric and ferulic acid were found in their cis-configuration, too. As expected, the portion of cis-configured hydroxycinnamates was higher in the leave fraction, most likely due to light exposure that supports trans/cis-isomerization. However, formation of small amounts of cis-isomers during sample preparation cannot be excluded.
Significant differences of p-coumaric and ferulic acid contents among the genotypes were observed (ANOVA, Tukey test, α = 0.05). Genotype OPM-11 had the lowest content of p-coumaric acid but the highest ferulic acid content in the quantitatively significant stem fraction. Highest amounts of p-coumaric acid were found in stems of genotypes OPM-3 and OPM-9. The opposite trend was observed for the leave fraction. Comparison of the hydroxycinnamate levels of leaves from the different genotypes shows the highest p-coumaric and ferulic acid contents for OPM-11, but the lowest p-coumaric acid contents for OPM-3 and OPM-9.

Note.
Means labeled with different letters are statistically different (ANOVA, Tukey test, α = 0.05, leaves and stems are tested separately). p, para.
2 Semiquantitative determination of 8-5-coupled DFA contents in Miscanthus stems, and of 8-5-non-cyclic-DFA contents in Miscanthus leaves: values were below the tested concentration range, but >limit of quantitation.
Significant differences of the total DFA content and individual DFAs were observed in both leaves and stems (ANOVA, Tukey test, α = 0.05, Table 5). OPM-11 had the highest total DFA contents in both stems (35.1 ± 3.4 mg/ 100 g dry weight) and leaves (115.4 ± 2.6 mg/100 g dry weight). In contrast, the lowest total DFA contents in both stems (21.7 ± 0.5 mg/100 g dry weight) and leaves (55.6 ± 3.3 mg/100 g dry weight) were analyzed for OPM-9.
Biomass recalcitrance may also be influenced by the presence of hydroxycinnamates, especially by the formation of cell wall cross-links. Considering leave/stem ratios, genotype OPM-11 contained the highest content of total DFAs, whereas the lowest content was found for OPM-9 (approximately half of the DFA content of OPM-11).

| CONCLUSION
The recalcitrance of lignocellulosic biomass against hydrolysis and fermentation is influenced by its gross composition, structural features of its cell wall polymers, and their interactions. Here, different parts of the Miscanthus plant (leaves vs. stems) and different genotypes (Miscanthus sacchariflorus, Miscanthus sinensis × Miscanthus sacchariflorus hybrid, Miscanthus × giganteus, Miscanthus sinensis "Goliath") were shown to partially differ in the overall cell wall composition and in structural details of the cell wall polymers demonstrating some genetic variability of Miscanthus. Although specific parameters such as reduced lignin contents, specific types of linkages within the lignin polymers, etc., are generally associated with biomass saccharification efficiency, an unambiguous prediction of the fermentation yields of biomass from the different Miscanthus genotypes appears difficult as none of the studied genotypes combines all favorable traits. For example, the high lignin content (determined as ABSL) of Miscanthus × giganteus is supposed to reduce the saccharification efficiency of this genotype. However, lignins isolated from this genotype contained the highest portions of β-O-4-linkages, which may be more easily cleaved during pretreatment processes to efficiently reduce the lignin content. Because Klason lignin and ABSL contents of Miscanthus stems were higher than lignin contents of Miscanthus leaves, the recalcitrance of stems is potentially higher than that of leaves. However, Miscanthus leaves contained higher DFA contents which is suggested to increase biomass recalcitrance.
Although the usefulness of cell wall composition data for the evaluation of biomass utilization is undisputed, this data need to considered in conjunction with potential biomass pretreatments to combine plant material and processes for biochemical use in an optimal way.