Importance of Lignin Coniferaldehyde Residues for Plant Properties and Sustainable Uses

Abstract Increases in coniferaldehyde content, a minor lignin residue, significantly improves the sustainable use of plant biomass for feed, pulping, and biorefinery without affecting plant growth and yields. Herein, different analytical methods are compared and validated to distinguish coniferaldehyde from other lignin residues. It is shown that specific genetic pathways regulate amount, linkage, and position of coniferaldehyde within the lignin polymer for each cell type. This specific cellular regulation offers new possibilities for designing plant lignin for novel and targeted industrial uses.

Lignins constitute a class of water-insoluble phenolic polymers of variable size accumulating in plant cell walls.These polymers have different compositions and concentrations, depending on cell type, their developmental state, and environmental conditions, in order to ensure the chemical and mechanical properties required for the function of each cell type.[3] These large quantities make lignin an ideal resource for future sustainable bio-economy, but only if we can fully understand, predict and exploit its formation and structure.Lignins are synthesized by the oxidative polymerization of secreted C 6 C 3 phenylpropenoid compounds differing in their C 6 aromatic substitution (hydroxyl and methoxyl groups) as well as their C 3 sidechain terminal function (mostly alcohol) (Scheme 1).[3] The main lignin C 6 C 3 aldehyde monomers derive from coniferaldehyde (C 6 with 1 hydroxyl and 1 methoxyl group) and sinapaldehyde (C 6 with 1 hydroxyl and 2 methoxyl groups).It however remains unknown whether pcoumaraldehyde (C 6 with 1 hydroxyl and C 3 with aldehyde) also forms residues in developmental lignin.[4] The industrial valorization of lignin aromatic structure, through its depolymerization by methods such as catalytic fractionation, or biogas production, [5,6] offers promising opportunities to sustainably exploit the lignin in plant biomass by biorefineries.However, the efficiency of these future uses depends on a clear understanding of the different lignin residues incorporated, their distribution and homogeneity between cell types, their position within the polymers as well as their genetic regulation to allow for the optimal utilization of available biomass, and the design of improved plants for biorefineries.
Mutagenesis of CINNAMYL ALCOHOL DEHYDROGENASE (CAD) genes has allowed modulating C 6 C 3 aldehyde residue levels in lignin (Scheme 1).[8][9] Saccharification and catalytic fractionation yields of cad4xcad5 stem biomass were increased approximately two-and threefold respectively, compared to wild-type (WT) plants. [6,8]][12][13][14][15][16][17][18] In fact, natural mutants in CADs thrive in the wild and have been readily identified, such as the CAD-null mutant of pine. [19,20]Natural mutants in CAD have also been selected and preferentially used in agriculture more than 100-years ago, like the Sekizaisou variety of mulberry trees, which improved both silkworm growth and silk quality when used for feed. [21]The far reaching effects of aldehyde concentration on biomass properties suggest that these residues, despite being considered minor lignin constituents, have a determining role in diversifying the biological functions and industrial uses of lignin in plants.However, although different methods have been previously used to quantify coniferaldehyde residues in lignin, their position, amount and linkage have never been compared to obtain a full picture of how these less abundant residues are accumulated.
Indeed, synthetic lignins, or dehydrogenation polymers (DHPs), synthesized by directly incubating coniferaldehyde (G CHO ) monomers with peroxidases (Figure S1 in the Supporting Information), were more hydrophobic and less soluble in a range of solvents than those made from coniferyl alcohol (G CHOH ). [4]The artificial lignification of isolated primary cell walls only with G CHO moreover decreased the cross-bridging between lignin and other cell wall polysaccharides. [22]][25] The impact on whole plant physical properties suggests that lignin composition, such as in G CHO residues, alters the overall cell wall organization and its interconnections.Incorporated G CHO , but not sinapaldehyde, within the lignin polymer can moreover specifically cross-react in acid conditions and covalently bind other free phenolic compounds, such as phloroglucinol. [7]These G CHO residues can also react with NaHSO 3 /Na 2 SO 3 to form sulfonic acid derivatives. [26]This suggests that the lateral functionalization of  lignin polymers, using internal G CHO residues as anchors, could be used similarly to the lateral functionalization of cellulose using 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) oxidative treatment. [27]To this end, the reliable quantification of G CHO residue amounts in lignin as well as the quantification of their proportion at the end and/or within the lignin polymer are necessary, but has not yet been demonstrated.
0]28] Pyrolysis coupled to gas chromatography and mass spectroscopy (PyÀ GC/MS), which enables the quantitative measurement of G CHO , G CHOH and sinapyl alcohol (S CHOH ) residues, [29][30][31][32] showed that total G CHO content in lignin represented ~9 % in spruce, ~5 % in Arabidopsis, ~2 % in eucalyptus and ~0.2 % in poplar. [7,33,34]The content variability in lignin of G CHO , G CHOH and S CHOH residues was further examined using pyrolysis/GCÀ MS on a set of Arabidopsis thaliana mutants affected in one or several genes encoding for enzymes responsible of changing the C 6 and/or C 3 parts of lignin monomers (Figure S1).8][9][10][11][12][13][14][15][16][17][18][19][20][21] In contrast to total G CHOH and S CHOH residue levels which could only be reduced or annulled compared to WT plants, total G CHO residue content varied by roughly threefold changes in either direction in Arabidopsis with specific genetic changes, namely increasing in the cad4xcad5 mutant and decreasing in the 4cl1x4cl2 mutant (Figure 2A).Arabidopsis natural ecotype variant Wassilewskija (WS) presented ~60 % more G CHO residues than the Columbia-0 (Col-0) ecotype, although their G CHOH and S CHOH residue amounts did not differ (Figure S2).The G CHO over-accumulation due to the cad4xcad5 mutations was even more accentuated in the WS than in the Col-0 ecotype, also without affecting G CHOH and S CHOH residue amounts (Figure S2).Overall, our results highlight that the accumulation of G CHO residues in lignin follows a specific regulation differing from the one controlling the abundant G CHOH and S CHOH residue amounts.
We then evaluated the relative positions of G CHO residues in the lignin polymer linked by β-O-4 (Scheme 1) using thioacidolysis coupled to gas chromatography and detection with mass spectroscopy and flame ionization (thioacidoylsis/GC-MS-FID).][37] However, the formation of this link in DHPs also depends on the relative proportion of Terminal and internal residues correspond to the sum of the different relative peak contributions as shown in Figure 3.The respective position of each mutation in the metabolic pathway is indicated in Figure S1.Different letters for each residue category indicate significant differences according to a one-way analysis of variance (ANOVA) with Tukey test (with a 95% confidence level α = 0.05).Linear correlations between the methods for each residue are presented in (C).Note that R 2 value of the linear regression between thioacidolysis and pyrolysis for G CHO residues is reduced to 0.5685 when removing the cad4xcad5 mutant (extreme value).enzyme to substrate. [38]We simplified lignin as a linear polymer [39] with terminal residues at one end keeping their C 3 sidechain unaltered (Scheme 1).Thioacidolyzed products of the internal and terminal residues were identified using DHPs made of only G CHOH , S CHOH or G CHO residues (Figure 3).[42][43][44][45] We then measured the positional proportion of β-O-4 linked G CHOH , S CHOH and G CHO residues in stems of our Arabidopsis mutant series with modified lignins.Each type of β-O-4 linked residues showed specific genetic control: (i) G CHOH content decreased in all mutants except for fah1 and omt1; (ii) S CHOH amounts decreased in ccr1-3 and cad4xcad5, and were absent in fah1 and omt1; and (iii) G CHO levels decreased in 4cl1, 4cl1x4cl2, ccoaomt1, and ccr1-3 but increased in cad4xcad5 (Figure 2B).Comparing the levels of β-O-4 linked G CHOH , S CHOH and G CHO determined by thioacidolysis with the total amounts of these residues measured by PyÀ GC/ MS showed different correlation strengths for each residue (Figure 2C).G CHO and G CHOH content correlated strongly between the two methods, suggesting that β-O-4 represented the main linkage for G CHO and G CHOH in lignin (Figure 2C).In contrast, S CHOH residue content correlated to a lesser extent between the methods, suggesting fewer β-O-4 linkages exist for S CHOH (Figure 2C).This lower proportion of β-O-4 for S CHOH residues confirmed previous studies showing higher capacity of S CHOH residues to form other bonds, such as β-β, in DHPs as well as  [40][41][42][43][44][45] retention time (RT in min) and FID/TIC fold-ratio (see also Figure S3).42][43][44][45] the low correlation between β-O-4 proportion and the relative S residue content in poplar natural variants. [46,47]Altogether, our results showed that different residues are subjected to a specific proportion of β-O-4 linkages within the lignin polymer.Positional analyses of G CHOH , S CHOH and G CHO residues moreover revealed a clear decoupling between the proportion of terminal and internal residues depending on the mutation.WT plants had G CHO terminal residues representing ~83 % of the normalized chromatogram compared to ~17 % for internal G CHO residues (Figure 2B).Specific mutants exhibited distinct changes differently affecting the positional proportion: (i) terminal G CHO were specifically decreased by the 4cl1, 4cl2, 4cl1x4cl2, ccoaomt1, and ccr1-3 mutations, but increased in the fah1 and cad4xcad5 mutants; whereas (ii) internal G CHO were only increased by the cad4xcad5 mutation (Figure 2B).Analyses of the positional proportion of β-O-4 linked G CHOH and S CHOH residues revealed a different genetic control: (i) terminal G CHOH residues were increased by the fah1 and omt1 mutations and decreased in the other mutants, in contrast to internal G CHOH residues which were unaltered in the fah1 and omt1 mutations but decreased in the other mutants; and (ii) S CHOH residues were completely absent from the fah1 and omt1 mutants, but terminal S CHOH residues increased in the ccoaomt1 mutant, although internal S CHOH residues only decreased in the ccr1-3 and cad4xcad5 mutants (Figure 2B).Overall, our results show that the different lignin monomers are subjected to specific incorporation genetically controlling their amount, their position as well as their linkagetypes within the lignin polymer.
One essential characteristic of lignin generally overlooked in the context of biomass optimization is its heterogeneity at the cellular and sub-cellular levels. [1,7,48]By neglecting this crucial aspect, opportunities are missed to design plant biomass with homogeneous lignin composition to improve its valorization potential.To characterize this cellular heterogeneity of lignin composition, the cell type specific accumulation of G CHO , G CHOH and S residues were measured using two in situ quantitative methods.These included the histochemical Wiesner test [7] as well as Raman confocal microspectroscopy, [49][50][51] both recently reported to enable quantitative measurement of G CHO , G CHOH and S residues in the cell walls of the different cell types (Figure 4A-C).A recent study however showed that the 1625 and 1141 cm À 1 Raman bands, previously suggested to reflect lignin G CHO residues, [49][50][51][52][53] did not correlate strongly with either the Wiesner test [7] or pyrolysis/GC-MS [51] quantification of total G CHO residues.Raman microspectra of G CHO residues as monomers and DHPs confirmed the presence of these two characteristic 1625 and 1141 cm À 1 Raman bands (Figure 4C).Comparison of Raman microspectra obtained from cross-sections also showed an increased 1625 cm À 1 and to a lesser extent 1141 cm À 1 Raman bands in cad4xcad5 mutant compared to WT plants (Figure 4C).We therefore hypothesized that the differences between the Wiesner test and 1625/1141 cm À 1 Raman bands depended on the position of G CHO residues within the lignin polymer, as they exhibited distinct cell type values (Figure 4D,E).The Wiesner test intensity showed strong correlation with the total β-O-4 linked G CHO residues measured by thioacidolysis/GC-MS-FID, but weaker correlations with terminal or internal G CHO residues (Figure 5).These results confirmed that the Wiesner test detects all G CHO residues in the lignin polymer, thus providing the most precise in situ quantitative method currently available.In contrast, correlation of the 1625 cm À 1 Raman band did not show any strong association with the total β-O-4 linked G CHO measured by thioacidolysis/GC-MS-FID (Figure 5).Instead, the 1625 cm À 1 Raman band reflected more the concentration of terminal β-O-4 linked G CHO residues (Figure 5).This result confirmed previous hypotheses which suggested that the 1625 cm À 1 band originated predominantly from G CHO units with an unsaturated and unlinked C 3 , such as lignin terminal residues. [53]The influence of residue position in the lignin polymer on Raman scattering was however specific to G CHO and was not observed for G CHOH or S residues (Figure 5).The 1141 cm À 1 Raman band had also been used to quantify G CHO residues in milled wood lignin, [52] but showed weaker correlations than the 1625 cm À 1 band, probably due to the presence in cross-sections of other cell wall polymers removed by the milling process (Figure S4).Altogether, our results show that the different in situ imaging methods allowed distinguishing and quantifying G CHO residues in different positions within the lignin polymer at the cellular level.
A recent study has shown that specific genetic regulation controls G CHO residue amounts in the different cell types of Arabidopsis and poplar stems. [7]The accumulation of terminal and total G CHO residues in specific cell types was thus monitored in Arabidopsis stems for protoxylem vessels (PX), metaxylem vessels (MX) and interfascicular fibers (IFs).The relative positional proportion of G CHO residues was estimated using the ratio of the 1625 cm À 1 Raman band, reflecting β-O-4 linked terminal residues, to the Wiesner test intensity, to account for all G CHO residues.Arabidopsis WT plants showed that the three cell types presented different positional proportions in their lignin: PX presented low amount of terminal residue whereas MXs and IFs had similar higher amounts (Figure 6).Analyses of crosssections from the Arabidopsis mutant series revealed that the genetic regulation controlling the position of G CHO residues differed between the three cell types.PXs and MXs, which have respectively the lowest and highest concentrations of total and terminal G CHO residues in their cell wall (Figure 4D,E), [7] were not affected by the different genetic regulation altering G CHO biosynthesis (Figure 6).In contrast, IFs were the most susceptible to large changes in the positional proportion of G CHO residues, with large increases in the 4cl1x4cl2 and ccr1-3 mutants, compared to slight to no decreases in the cad4xcad5, fah1 and omt1 mutants (Figure 6).These results represent an unsuspected discovery on the genetic regulation of the distribution of G CHO residues within the lignin polymer in specific cell types.This specific regulation of G CHO residues in lignin was anticipated from previous analyses using NMR spectroscopy, which only detected β-O-4 linked G CHO with syringyl (S) residue, but not other guaiacyl (G) residues in CAD down-regulated angiosperm tobacco, poplar and mulberry -all species with wood composed of more IFs than PXs/MXs. [27,31,54,55]n contrast, CAD down-regulated plants from gymnosperms, such as the CAD-null pine, having wood composed of mostly PXs/MXs, or angiosperms devoid of S residues, such as the fah1 mutant, are nevertheless capable of linking G CHO residues by β-O-4 links to other G residues. [8,37]The proportions of the different lignified cell types vary between plant organs and their developmental state, thus allowing one to harvest biomass with distinct coniferaldehyde profiles.This aspect highlights the advantages of plant biomass as a multipurpose renewable resource for sustainable uses.Although the exact molecular mechanisms enabling the positional control of G CHO residues yet remain unclear, such specific genetic control suggests that the molecular nature by which G CHO monomers are secreted and/or oxidatively polymerized, depending on their positions in the polymer, are differently regulated in each cell type.
The extent of the possibilities for the sustainable uses of plant cell wall biomass depends on the compositional homogeneity and predictability of the lignin polymer structure in the feedstock used.Our study details how the amount of G CHO residues in lignin differs between the cell types making up the plant biomass.Such cellular specificity, with large differences in G and S residue levels, had already been reported between MXs and IFs. [1]These specificities appear to depend on the cell type itself as genetic engineering or monomer feeding to force both angiosperm and gymnosperm MXs to incorporate S residues only slightly changed their lignin composition. [48]We also showed that the positional distribution of the G CHO residues within the lignin polymer varied between the cell types.It yet remains unknown whether similar cell-specific regulation mechanisms also exist for the more abundant C 6 C 3 alcohol monomers.The apparent complexity and evolutionary conservation [48] of these regulatory systems is understandable from a biological perspective as both the proportions and positions of specific residues will diversify the lignin polymer's chemical and mechanical properties to vary its physiological functions.Future studies to decipher the underlying genetic and molecular mechanisms will thus allow defining to which extent plants can be selected or genetically designed to control the G CHO residue distribution within lignin and/or between cell types without hindering agronomical yields for future sustainable uses in biorefineries.

Dehydrogenation polymers (DHPs):
DHPs were synthesized according to the Zutropf method as previously described. [51]10 mL of a solution with 1 mg of horseradish peroxidase (Sigma, P8375-10KU) in 0.1 M NaHPO 4 buffer at pH 6 was mixed under magnetic stirring with 10 mL solutions of 14 mM H 2 O 2 (Sigma-Aldrich, 95299) and 12 mM of monomer in 3 : 7 methanol/0.1 M NaHPO 4 buffer at pH 6 at a rate of 0.5 mL h À 1 using a Legato 200 syringe pump (KdScientific, USA).After 24 h, DHPs in the mixture were isolated by centrifugation at 10000 g for 10 min, the supernatant was removed, and the pellet was washed three times in ultrapure water and freeze-dried.
Cell wall isolation: Extract-free cell wall material were isolated from stems ground in liquid nitrogen using ceramic mortar and pestle.Proteins and membranes were removed by three washes using vortex mixer agitation with a solution containing 140 mM Tris-base (Sigma-Aldrich, T1503), 105 mM tris acetate (Sigma-Aldrich, T1258), 0.5 mM ethylenediamine tetraacetic acid (EDTA, Scharlau Chemie, AC0965), and 8 % w/v lithium dodecyl sulfate (LDS, Sigma-Aldrich, L4632) combined with centrifugation (10000 g, 10 min) and the removal of supernatant.Pellets were then successively washed/ centrifuged with water, 100 % methanol and finally chloroform/ methanol (1 : 1).Pellets were then washed in acetone and air dried overnight.
In situ quantitative lignin analysis: Quantitative Wiesner data was taken from Ref. [7], and is available in the Supporting Information of that publication.Briefly, 50 μm stem cross-sections were imaged before and after staining with 0.5 % phloroglucinol (Sigma, P3502) in 1 : 1 ethanol/HCl (37 %).The acquired images were transformed into absorbance using ImageJ, aligned, and measured in 50 circular points per plant and cell type.Finally, the unstained background absorbance of each point was subtracted from the stained absorbance.Quantitative Raman microspectroscopy data was partly taken from Ref. [51] and extended using the same experimental setup.Briefly, spectra from stem cross-sections were acquired using a Raman Touch-VIS-NIR (Nanophoton, Japan) equipped with a 532 nm laser.Spectra (1.6 cm À 1 resolution) were baseline corrected using an asymmetric least-squares algorithm and normalized to the total Raman signal (area under the curve) between 300 and 1700 cm À 1 .

Scheme 1 .
Scheme 1. Schematic representation of lignin C 6 C 3 monomers, coniferaldehyde, and coniferyl alcohol, interconverted by the activity of the NADP + /NADPH + H + -dependent CADs as well as β-O-4-linked oxidative polymerization lignin products.Enzyme-catalyzed steps are shown by large grey arrows.Note that terminal residues are indicated in red, internal residues colored in blue, and β-O-4 linkages indicated by dotted circles.Trimethylsilylated (TMS)-derivatized thioacidolyzed products corresponding to the different lignin residues are indicated by black dotted lines in green.Black plain arrows indicate pyrolytic products obtained from lignin polymer irrespective of residue position.

Figure 1 .
Figure 1.Impact of aldehyde residue over-accumulation in lignin on Arabidopsis plant productivity.Phenotypic differences between wild-type (WT) and cad4xcad5 double mutant Columbia-0 plants on stem height (A, n = 10 plants), stem weight (B, n = 5 plants), number of fruit per plant (C, n = 5 plants) as well as fruit size (D, n = 5 stems and 5 fruits each).Different letters indicate significant differences according to a student t-test with Tukey test (α = 0.05).

Figure 2 .
Figure 2. Residue proportions in a set of Arabidopsis mutants differently altered in lignin monomer biosynthesis.Analysis of the relative proportion of G CHO , G CHOH , and S CHOH in lignins of stems using pyrolysis-GC/MS (A) and thioacidolysis-GC/MS-FID (B), with n = 2-6 independent biological replicates per genotype.Terminal and internal residues correspond to the sum of the different relative peak contributions as shown in Figure3.The respective position of each mutation in the metabolic pathway is indicated in FigureS1.Different letters for each residue category indicate significant differences according to a one-way analysis of variance (ANOVA) with Tukey test (with a 95% confidence level α = 0.05).Linear correlations between the methods for each residue are presented in (C).Note that R 2 value of the linear regression between thioacidolysis and pyrolysis for G CHO residues is reduced to 0.5685 when removing the cad4xcad5 mutant (extreme value).

Figure 4 .
Figure 4.In situ quantitative detection of G CHO content in cell walls of specific cell types in stem cross-sections of Arabidopsis.Sample response before (A) and after (B) staining with to the Wiesner test (phloroglucinol/HCl).Bars = 30 μm.Protoxylem vessels (PXs), metaxylem vessels (MXs) and interfascicular fibers (IFs) are indicated by arrows.Standard average Raman spectra of G CHO monomers, DHPs and MXs in WT and cad4xcad5 Arabidopsis stem cross-sections (C).The 1141 and 1625 cm À 1 bands, previously suggested to reflect G CHO residues, are indicated by grey line.Cell type-specific responses in WT plant cross-sections for the Wiesner test (D) and Raman (E), n = average of each cell type in 3-5 independent biological replicates.Different letters for each category indicate significant differences according to a one-way ANOVA with Tukey test (α = 0.05).

Figure 5 .
Figure 5. Linear regression analyses between specific Raman band heights and thioacidolysis-GC/MS-FID for G CHO , G CHOH , and S CHOH residues connected by β-O-4 linkages at different positions with the lignin polymers of stem tissues in a set of Arabidopsis with differently modified lignins.Note that instead of Raman, regression analyses between thioacidolysis with Wiesner intensity for G CHO residues are indicated in the right y axis of the upper row in grey.Note that the R 2 value of linear regressions between internal G CHO and 1625 cm À 1 Raman band is reduced to 0.0001 when removing the cad4xcad5 mutant (extreme value), and between terminal S CHOH and 1334 cm À 1 Raman band is reduced to 0.1339 when removing the ccoaomt1 mutant (extreme value).

Figure 6 .
Figure 6.Genetic regulation of the G CHO positional proportion in lignin of different cell types in stem cross-sections of a set of Arabidopsis mutants differently altered in lignin.Cell types include protoxylem vessels (PX), metaxylem vessels (MX), and interfascicular fibers (IFs).Different letters for each residue category indicate significant differences according to a oneway ANOVA with Tukey test (α = 0.05), n = 2-6 cells from 2-3 individual plants per genotype for Raman divided by n = cellular average of 5 individual plants per genotype for the Wiesner test.