Discovery of LPMO activity on hemicelluloses shows the importance of oxidative processes in plant cell wall degradation

Significance

Plant cell walls are recalcitrant copolymeric structures mainly comprising polysaccharides and lignin. Enzymatic degradation of the polysaccharides is a crucial step in biorefining of biomass. Recently, it was discovered that nature employs copper-dependent redox enzymes called lytic polysaccharide monoxoygenases (LPMOs) to promote degradation of the most recalcitrant and crystalline of these polysaccharides, cellulose. By carrying out oxidative cleavage of otherwise inaccessible cellulose chains, LPMOs create access points for classical hydrolytic enzymes such as cellulases. Intriguingly, the genomes of biomass degrading microorganisms encode a plethora of LPMOs (up to over 40). To our knowledge, we demonstrate for the first time that LPMOs act on hemicelluloses. This finding dramatically widens the scope of LPMOs and oxidative processes in plant cell wall degradation and biorefining.

Abstract

The recently discovered lytic polysaccharide monooxygenases (LPMOs) are known to carry out oxidative cleavage of glycoside bonds in chitin and cellulose, thus boosting the activity of well-known hydrolytic depolymerizing enzymes. Because biomass-degrading microorganisms tend to produce a plethora of LPMOs, and considering the complexity and copolymeric nature of the plant cell wall, it has been speculated that some LPMOs may act on other substrates, in particular the hemicelluloses that tether to cellulose microfibrils. We demonstrate that an LPMO from Neurospora crassa, NcLPMO9C, indeed degrades various hemicelluloses, in particular xyloglucan. This activity was discovered using a glycan microarray-based screening method for detection of substrate specificities of carbohydrate-active enzymes, and further explored using defined oligomeric hemicelluloses, isolated polymeric hemicelluloses and cell walls. Products generated by NcLPMO9C were analyzed using high performance anion exchange chromatography and multidimensional mass spectrometry. We show that NcLPMO9C generates oxidized products from a variety of substrates and that its product profile differs from those of hydrolytic enzymes acting on the same substrates. The enzyme particularly acts on the glucose backbone of xyloglucan, accepting various substitutions (xylose, galactose) in almost all positions. Because the attachment of xyloglucan to cellulose hampers depolymerization of the latter, it is possible that the beneficial effect of the LPMOs that are present in current commercial cellulase mixtures in part is due to hitherto undetected LPMO activities on recalcitrant hemicellulose structures.

Plant cell walls constitute the largest source of biomass on earth and have evolved over some 450 million years to provide support and protection to the plant body. The effective deconstruction of these highly complex and recalcitrant structures is dependent on efforts to discover and understand the batteries of enzymes evolved by microorganisms for cell wall degradation. One important class of enzymes is the copper-dependent lytic polysaccharide monooxygenases (LPMOs) (1–6). Several LPMOs have been characterized, and for all of these activity on either cellulose or chitin has been demonstrated, indicating that LPMOs act on highly crystalline substrates (1, 4, 5, 7, 8). LPMOs cleave the β-(1→4) glycosidic bonds in these substrates, leaving the C1 or the C4 carbon oxidized (1, 9–12). The only exception so far emerged in a recent study showing that the strictly C4-oxidizing NcLPMO9C from Neurospora crassa is active on water-soluble, cellulose-derived oligosaccharides (12). This finding suggests that NcLPMO9C, and possibly other LPMOs, might be active on other noncrystalline polysaccharides, for example hemicelluloses, which are important structural components of plant cell walls (13).

Hemicelluloses are among the most abundant and diverse class of plant cell wall polysaccharides and are typically based on a backbone of β-(1→4) glycosidic linkages. In contrast to cellulose, hemicelluloses are usually substituted with a variety of sugar residues and the backbone may be interspersed with other linkage types instead of being strictly β-(1→4) as in cellulose (13, 14). Both these structural features prevent chain-association and crystallization, and enable hemicelluloses to form a highly resistant coextensive load-bearing network with cellulose microfibrils (10, 15). Several studies have shown that xylan and xyloglucan can adsorb to cellulose, forming a partial outer coating to microfibrils, which likely provides protection against cell wall deconstruction by glycoside hydrolases (16, 17).

Our current understanding of microbial degradation of the cellulose/hemicellulose network envisages that a wide variety of glycoside hydrolase enzymes (GHs) degrade hemicelluloses, whereas LPMOs target crystalline cellulose which is then susceptible to further degradation by cellulases. However, the vast sequence variation among LPMOs (10, 11) and the fact that multiple LPMOs are often cosecreted by biomass degrading microorganisms (18–20) are inconsistent with cellulose and chitin being the only LPMO substrates. It seems likely that additional LPMO activities exist, and if that is the case, this would be a significant new insight into the biological roles of LPMOs and also extend the enzyme toolbox available for studying plant cell wall structures and valorization.

We report here important findings about the activity of NcLPMO9C on a number of α- and β-linked polysaccharides, obtained using glycan microarray substrate screening combined with chromatography and mass spectrometry. We show unequivocal activity against the biologically important noncellulosic soluble polysaccharides tamarind xyloglucan, mixed linkage (1→3),(1→4)-β-d-glucan and, more weakly, against glucomannan. Furthermore, we demonstrate, to our knowledge for the first time, that NcLPMO9C is active on two naturally occurring and structurally distinct xyloglucan substrates from Arabidopsis thaliana and Solanum lycopersicum (tomato) and also on (1→3),(1→4)-β-d-glucan from the lichen Cetraria islandica. These findings have significant implications both for our understanding of NcLPMO9C in nature and for our utilization of LPMOs for plant biomass degradation.

Results

Activity Screening of NcLPMO9C Using Glycan Microarrays.

We used glycan microarray technology (21) to rapidly screen NcLPMO9C toward an extensive collection of putative substrates including the most abundant polysaccharides present in plant and fungal cell walls (Fig. 1 and Fig. S1). This analysis clearly showed that NcLPMO9C was active on xyloglucan, (1→3),(1→4)-β-d-glucan, and, to a lesser extent, also on glucomannan, and that activity was increased in the presence of the reductant. These data indicate that NcLPMO9C requires short stretches of contiguous (1→4)-β–linked glucose units for activity. Indeed, there was no indication of activity on polysaccharides not containing some degree of (1→4)-β-d-glucan, exemplified by (1→4)-β-d-xylan, (1→4)-α-d-polygalacturonan, (1→4)-β-d-mannan (without glucose in the backbone), (1→4)-β-d-galactan, yeast β-d-glucan [(1→3),(1→6)-β-d-glucan], callose [(1→3)-β-d-glucan], and arabinoxylan (Fig. 1 and Fig. S1).

Fig. 1.

Glycan microarray screening of NcLPMO9C activity on various polysaccharide substrates. Activities were detected by the loss or reduction of polysaccharide-borne epitopes recognized by substrate-specific monoclonal antibodies (mAb). (A) Heatmap showing degradation of the substrates listed above the panel following treatment with NcLPMO9C at different concentrations (listed left), in the presence (+) or absence (−) of ascorbic acid. Control means no enzyme treatment. The mAbs used to recognize the substrates are in parentheses, top row. The heatmap shows relative mean spot signal intensities; the lower the signal when comparing controls and NcLPMO9C treated samples, the greater the degree of degradation. (B) Fold-change heatmap of the data shown in A. This heatmap shows the degree of change in mAb binding resulting from NcLPMO9C treatment of the substrates. A cutoff of 5 was imposed before the fold changes were calculated. (C) Images of the arrays used to produce the heatmaps, showing consistent spot morphologies and depleted signals caused by NcLPMO9C. * denotes that final concentration of xyloglucan was 1 mg/mL, whereas the other substrates were 0.1 mg/mL.

Activity on Xyloglucan.

The data obtained from the glycan microarray screening indicate that NcLPMO9C acts on (1→4)-β-d-glucans, independent of substitution of the polysaccharide backbone (as in, for example, xyloglucan). We selected a set of biologically relevant polysaccharides to confirm the findings from the microarrays and to analyze the products. We compared the product profiles generated by NcLPMO9C to those generated by two endo-glucanases, TaCel5A (22) and AfCel12A (23).

Fig. 2A shows that NcLPMO9C and AfCel12A generated different products from tamarind xyloglucan. Products generated by NcLPMO9C include two major oxidized products corresponding to the keto-form of single oxidized species, the most probable product configurations being G^oxXXL (m/z 1,245.3) and G^oxXLL (m/z 1,407.3). Note that the relative position of the X and L units can vary [ref. 24; we use the generally accepted nomenclature of xyloglucan according to ref. 24, where G = β-d-Glc; X = α-d-Xyl-(1→6)-β-d-Glc; L = β-d-Gal-(1→2)-α-d-Xyl-(1→6)-β-d-Glc; F = α-l-Fuc-(1→2)-β-d-Gal-(1→2)-α-d-Xyl-(1→6)-β-d-Glc); S = α-l-Araf-(1→5)-α-d-Xyl-(1→6)-β-d-Glc]. These assignments were made based on the known primary structure of tamarind xyloglucan with a nonsubstituted G appearing every fourth residue in the glucan backbone (25). The appearance of oxidized products indicates that a single glucose unit in the (1→4)-β-d-glucan backbone is sufficient for the LPMO to access and cleave xyloglucan. In comparison, the reaction with AfCel12A generated native products only appearing as m/z 1,085.1, 1,247.2, and 1,409.2, most probably corresponding to XXXG, XLXG/XXLG and XLLG, respectively (25). These assignments were made based on the notion that glycoside hydrolases degrading xyloglucan tend to release oligosaccharides with an unsubstituted glucose unit at the reducing end (24).

Fig. 2.

MALDI-ToF MS analysis of product profiles. The spectra show products generated from tamarind xyloglucan (A; see Inset regarding nomenclature; blue, glucose; orange, xylose; yellow, galactose), konjac glucomannan (B), lichenan and mixed linked glucan (MLG) from barley (C) treated with either endo-glucanase (magenta) or LPMO (black, turquoise for MLG). Product profiles upon endo-glucanase treatment of lichenan and MLG were essentially the same and therefore only one of the spectra is shown. Brackets in B indicate product clusters of same DP, indicated by the number. In the main spectra, only sodium adducts are labeled, whereas the inserts also show potassium adducts (marked *) and various forms of oxidized species where both the keto-group formed upon C4 oxidation (−2 Da) and its gemdiol form (marked #, i.e., addition of H₂O, +18 Da) appear. Abbreviations: G, X. and L, see A Inset; Glc, glucose (+ 162 Da); Hex, hexose (+ 162 Da); Ac, acetyl group (+ 42 Da); ox, oxidized (−2 Da if keto form). Analysis of control reactions, without enzyme addition, showed no signals related to carbohydrates.

Although Fig. 2A shows the generation of oxidized products from xyloglucan by NcLPMO9C, the assignment of oligosaccharide sequences to the observed products is somewhat arbitrary, due to the heterogeneity of the substrate, and leaves questions as to where exactly the LPMO and the endo-glucanase cleave. Therefore, we studied the degradation of XXXGXXXG^OH (XG14^OH), a reasonably pure 14-mer generated by endo-glucanase treatment of tamarind xyloglucan, with known sequence and with a reduced glucose (glucitol) in the reducing end (+2 Da). Reaction products were analyzed using both high-performance anion exchange chromatography (HPAEC) (Fig. S2 and ref. 26) and multidimensional mass spectrometry (MS) (Fig. 3 A and B), again revealing clear differences between NcLPMO9C and AfCel12A product profiles. The product profile generated by NcLPMO9C shows that the lytic reaction primarily happened at the nonsubstituted internal glucosyl unit and caused an oxidation at the nonreducing end, since oxidations were only observed on fragments carrying the reduced terminal glucose (Fig. 3 A and B). Accordingly, the only native fragment detected in the reaction with NcLPMO9C corresponds to XXX. The endo-glucanase only hydrolyzed XG14^OH at the reducing end of the internal nonsubstituted glucosyl (G) unit, generating two products, XXXG (m/z 1,085) and XXXG^OH (m/z 1,087) (Fig. 3 A and B and Fig. S2). The experiments with XG14^OH confirmed that xyloglucan cleavage by NcLPMO9C occurs at a single unsubstituted glucosyl unit.

Fig. 3.

Analysis of products generated from XG14^OH. (A) ESI-MS spectra highlighting Na-adducts of products generated by NcLPMO9C (black) or AfCel12A (magenta). ESI-MS spectra show primarily the gemdiol form (marked #) of oxidized products. (B) MALDI-ToF-MS spectra highlighting Na-adducts of XG14^OH (blue) and of products generated from XG14^OH by NcLPMO9C (black) or AfCel12A (magenta). MALDI-ToF-MS spectra show primarily the keto-form of oxidized products, which in the case of a reduced substrate has the same m/z as native products (keto group gives m/z -2, and reduction gives m/z +2). Note that panel B shows that XG14^OH (XXXGXXXG^OH) is contaminated with other species containing one or more additional hexoses, probably galactoses coupled to one or more of the X units as this is a very common moiety in xyloglucan from tamarind (hence annotation as L in the figure). Some products derived from these contaminations are annotated in the mass spectra. (C) MS2 fragmentation of an m/z 1,249 species generated upon lithium doping of the product mixture shown in panel A (m/z 1,249 corresponds to the Li-adduct of the m/z 1,265 species in A). Gemdiol-fragments readily lose a water molecule thus occurring in the spectrum as B or C −18 Da (fragmentation nomenclature as in ref. 30); this has been observed previously for gemdiol products (12). The substrate, XG14^OH, and possible products of m/z 1,249 are shown as cartoons according to the nomenclature of (31): blue circle, glucose; orange star, xylose; yellow circle, galactose. Parenthesis surrounding galactosyl-units denote that the position of these units may vary. “Ox” denotes the position of the oxidation. “Red” denotes the position of reduction. Note that dominating fragmentation reactions lead to removal of substitutions from the glucan backbone, explaining why several oligo-G products are detected.

Activity on Glucomannan and β-Glucan.

The activity of NcLPMO9C on konjac glucomannan, a linear (1→4)-β–linked mannan comprising randomly distributed glucose (∼35%; ref. 27), generated clusters of products up to at least DP15, each cluster containing the keto-, gemdiol and native form of hexose oligosaccharides and acetylated species (Fig. 2B). The fact that relatively high DP species appeared indicates that larger parts of the substrate could not be cleaved, which is likely due to the occurrence of mannoses, because galactomannan is not cleaved by the enzyme (Fig. 1). The endo-glucanase TaCel5A seemed rather efficient in degrading konjac glucomannan, generating primarily DP3 and DP4 and the corresponding mono- and diacetylated species, i.e., a totally different product profile compared with NcLPMO9C. The differences may be due to TaCel5A being less affected by the high mannose content of the backbone or by the enzyme being able to hydrolyze (1→4)-β-d-mannan linkages (28).

MALDI-ToF analysis of products generated from barley β-glucan and lichenan (Fig. 2C) confirmed NcLPMO9C activity on these substrates. The MALDI-ToF spectrum of barley β-glucan showed dominance of products increasing by three sugars in size, i.e., DP4, DP7, DP10, DP13, and so forth (Fig. 2C), which correlates with the known repetitive structure of this polysaccharide, from barley with every third linkage being a (1→3)-β-D linkage (29). Apparently, (1→3)-β-D linkages affect the substrate’s suitability for cleavage by NcLPMO9C. Products generated from lichenan did not show this repetitive pattern, indicating that the organization of its backbone is less repetitive. Activity of NcLPMO9C seemed lower on lichenan compared with barley β-glucan, which is also indicated by the heatmap in Fig. 1B.

In-Depth Analysis of Substrate Specificity of NcLPMO9C on Xyloglucan.

The assignment of MS signals to certain products with certain sugar compositions is somewhat arbitrary, due to the many overlapping masses and contaminations in XG14^OH (Fig. 3B). The data discussed so far demonstrate occurrence of oxidation and show that NcLPMO9C can accommodate a considerable degree of substitution of the glycan backbone close to the cleavage point. In fact, the data discussed suggest that every glucose near the cleavage point may be substituted, except, perhaps, the one that becomes oxidized. To gain further insight into these matters, MS2 experiments were conducted. Fig. 3C shows MS2 fragmentation of an ion with m/z 1,249 generated by NcLPMO9C from XG14^OH and assigned in Fig. 3A as G^oxXXXG^OH (m/z 1,249 represents the Li-adduct, whereas m/z 1,265, in Fig. 3A, is a Na-adduct). Many signals in the spectrum in Fig. 3C provide little structural information because they result from (dominating) fragmentation reactions leading to removal of substitutions (primarily xyloses, −132/150 Da) from the glucan backbone. However, the presence or absence of several key ions does provide useful information. First, the m/z 853/835 species represents a pentahexose carrying both an oxidized and a reduced end; this can only be a backbone glucose pentamer, confirming that NcLPMO9C cuts at the nonreducing end of the intermediate G-unit of XG14^OH and produces G^oxXXXG^OH, as indicated in Fig. 3 A and B. Second, the occurrence of an m/z 939 fragment implies the loss of an oxidized X-unit, whereas m/z 1,069 implies the loss of a terminal native hexose. The latter ion can only arise if an l-unit, which has a terminal galactose, is present within the m/z 1,249 product pool from XG14^OH, which again would mean that the oxidation is carried on an X-unit, as indeed indicated by the presence of the m/z 939 species (so the m/z 1,249 ion would then be X^oxXLG^OH, where the position of the L may vary). Fragmentation of the corresponding ion (m/z 1,247) generated from a xyloglucan polymer also supports that oxidation of substituted glucose may occur (Fig. S3). All in all, the data show that NcLPMO9C is promiscuous when it comes to accepting substitutions in or close to the catalytic site. The only absolute requirement, indicated by the substrate screening experiments described above, is that the backbone contains (1→4)-β-d-glucan.

Previous studies have shown that NcLPMO9C has low activity on cellotetraose, yielding dimeric products (12), and this may be considered a minimal substrate. We addressed both the effect of substitution and the type of glycosidic bond by testing the enzyme on substrates with four backbone glucosyl units. It was not capable of cleaving between two X-units in XG7 (XXXG), but was able to very slowly convert this substrate to XXX and G^ox (Fig. S4). Similarly, the activity on mixed linked cellotetraosyls was also very low and the data clearly showed that 1→3 linkages close to the catalytic site limited enzyme activity (Fig. S4). Of the tetraoses G4G3G4G and G4G4G3G, only the latter was cleaved slowly and only between the first two glucoses (producing G + G^ox4G3G).

To evaluate the relative importance of xyloglucan and cellulose as substrates, we compared degradation rates for cellopentaose versus XG14^OH and polymeric xyloglucan versus phosphoric acid swollen cellulose (Fig. S5). The oligomeric substrates may be considered as “minimal” substrates as they have only one major cleavage point (Fig. 3A and ref. 12). These experiments showed that the enzyme degraded XG14^OH approximately twice as fast as cellopentaose (0.06 s⁻¹ and 0.03 s⁻¹, respectively, at 40 °C). A similar difference was not observed when using the polymeric substrates; in this case, the rates were very similar (0.11 s⁻¹ at 50 °C).

NcLPMO9C Activity on Plant Cell Walls.

To explore the activity of NcLPMO9C on hemicellulose polymers in biologically relevant contexts, we then studied product release from various plant materials. Fig. 4 shows that NcLPMO9C released a variety of oxidized xyloglucan fragments from Arabidopsis thaliana and tomato (Solanum lycopersicum), including fragments with more complex substitution patterns than those obtained with the model substrates described above, i.e., fragments containing fucosylations, acetylations, and arabinosyl substitutions. Acetylations in Arabidopsis xyloglucan are positioned on the galactose units that in addition can be further substituted by a fucosyl unit (32). Acetylations in the Solanaceae may be positioned either directly on the glucan backbone or on arabinosyl units (33). The product mixtures thus further illustrate the ability of NcLPMO9C to accept a variety of substitutions. Although the mass spectra clearly show the occurrence of oxidation in the case of the LPMO only, they do not allow firm conclusions about differences in substrate specificity between the LPMO and the endo-glucanase, due to the many overlapping masses.

Fig. 4.

MALDI-ToF spectra of products generated from natural substrates. Arabidopsis (A) or tomato stem (B) cell wall material was incubated with either NcLPMO9C (black line) or AfCel12A (magenta line) and product mixtures were analyzed. Peaks were assigned by combining the mass information with knowledge on the substrate and the preferences of the enzymes in question; note that these assignments to some extent are arbitrary, because the exact positions of modifications/substitutions cannot always be inferred. Also note that additional sugars and modifications occur in these natural substrates: F, fucosylation on an l-unit (+146 Da); S, arabinosyl substitution on an X-unit (+132 Da). Acetylations (+42 Da) are also observed and marked by underlined G, F, or S units. All assigned masses are sodium adducts, except when indicated otherwise. Potassium adducts are marked by an asterisk. §, background signal. Control reactions showed that ascorbic acid alone did not release any products and did not change the product profile generated by the endo-glucanase.

The product profiles obtained with Icelandic moss (Fig. S6) showed clear differences between NcLPMO9C and the endo-glucanase (TaCel5A), similar to what was seen for purified lichenan (Fig. 2).

Discussion

The discovery of LPMO activity was a significant breakthrough in our understanding of how nature degrades recalcitrant biomass and has already had significant implications for industrial enzymatic conversion of lignocellulosic biomass. Until recently, LPMO activity had been demonstrated for chitin and cellulose only. The “flat” substrate-binding surfaces of these enzymes (3, 34, 35) suggest that they are indeed optimized for binding the surfaces of crystalline materials, rather than single polysaccharide chains. However, several observations indicate that LPMOs may have a wider role in biomass degradation. First, LPMOs also act on rather amorphous insoluble substrates such as phosphoric acid swollen cellulose. Second, the genomes of biomass-degrading microbes often contain many LPMO-encoding genes (over 40 in some cases; ref. 10), and expression studies show that the upregulation of these genes varies with the substrates, including hemicelluloses (20, 36, 37). Third, apart from a conserved catalytic center, LPMOs show large sequence variation (10, 11). All these observations indicate that LPMO activity on plant cell walls may not be restricted to cellulose but could include a variety of other polysaccharides, including xylan, xyloglucan, mannan, pectin and even starch. In this study, we provide, to our knowledge, the first evidence for LPMO activity on xyloglucans, β-glucans, and glucomannan.

By using high-throughput glycan microarray technology, we were able to rapidly screen putative hemicellulosic substrates, leading to the discovery of hitherto unidentified activities. This initial screening effort was then followed by detailed product analysis using model substrates and, importantly, also natural plant materials. Although quantification of substrate degradation is difficult, our data indicate that the primary hemicellulose targeted by NcLPMO9C is xyloglucan. The data also indicate that the enzyme is able to accommodate a wide variety of modifications on the glucan backbone. The structural basis of the substrate-specificity of NcLPMO9C needs further research. Interestingly, a structural model of NcLPMO9C suggests that the enzyme has a more extended and more polar substrate-binding surface relative to PcLPMO9D, which is not active on xyloglucan. The additional protruding polar side chains in NcLPMO9C may engage in hydrogen bonding with the carbohydrate moieties of branched substrates such as xyloglucan (Fig. S7).

A main biological function of hemicellulose is to tether cellulose microfibrils, thereby strengthening the cell wall (13). Xyloglucan in particular plays a major role in building the load-bearing network in type I primary plant cell walls and interacts very closely with cellulose (14, 16, 38). The biological implications of NcLPMO9C having broad substrate specificity on xyloglucan could be to enable N. crassa to “shave off” hemicelluloses and thereby increase accessibility to otherwise recalcitrant regions of cellulose. Perhaps these enzymes attack junction zones between cellulose and hemicellulose, which may not be good substrates for glycoside hydrolases (39), because the activity of these enzymes depends on interactions with accessible single polysaccharide chains. LPMOs may be better suited to attack these zones because they are adapted to act on surfaces.

To our knowledge, the present study provides the first example of LPMO activity on hemicellulose and it is likely that more such activities will be discovered in the future. Considering the present observations and the fact that LPMO-encoding genes are abundant in biomass-degrading microorganisms, it seems likely that LPMO-catalyzed oxidative processes play a major role in natural biomass conversion. Considering the broad specificity on xyloglucan, it is intriguing that NcLPMO9C is not active on xylan because the (1→4)-β-d-xylan structure in many aspects resembles the (1→4)-β-d-glucan structure. Xylans are another prominent and structurally important class of hemicelluloses and it seems plausible that other LPMOs are targeted toward this polysaccharide.

Another important implication of the present findings concerns current worldwide efforts in establishing efficient methods for enzymatic saccharification of lignocellulosic biomass. Recent improvements in the efficiency of commercial cellulase mixtures are in part due to the addition of LPMOs (40). The present findings raise questions as to why these LPMOs actually work in practice. The LPMO effect may of course be due to higher efficiency of the degradation of cellulose itself (2, 41), but might also be due to hitherto nondetected LPMO activities on recalcitrant hemicellulose structures that block cellulase access (38). Notably, whereas cellulose is homogeneous and similar in all types of plant biomass, hemicelluloses vary a lot within the plant kingdom. So, if hemicellulose degradation by LPMOs already is or eventually will be a factor determining the efficiency of commercial enzyme mixtures for biomass saccharification, it is likely that the LPMO content of such mixtures needs to be adapted to the substrate.

Materials and Methods

Enzymes.

Endo-glucanases TaCel5A from Thermoascus aurantiacus (UniProt:Q8TG26) and AfCel12A from Aspergillus fumigatus (UniProt:Q8TG26) were expressed in Pichia pastoris (ade2, prb1, pep4) under the AOX1 promoter using the PichiaPink expression system (Invitrogen, Life Technologies), and purified according to procedures described in SI Materials and Methods. Lytic polysaccharide monooxygenase 9C from Neurospora crassa, NcLPMO9C, was expressed in Pichia pastoris X33 and subsequently purified according to Kittl et al. (42).

Microarray-Based Detection of LPMO Activity.

Microarray construction.

NcLPMO9C was screened against nine polysaccharide substrates: xyloglucan (tamarind), β-glucan (oat), β-glucan (barley), lichenan (icelandic moss), glucomannan (konjac), galactomannan (carob), polygalacturonic acid, and callose (1,3-β-d-glucan) (Megazyme International Ireland) and birch wood xylan (X0502, Sigma-Aldrich). Additional substrates tested are described in SI Materials and Methods. Solutions were prepared in 100 mM sodium phosphate buffer, pH 8.0, and added into a microtiter plate in the following order (final concentrations in parenthesis): polysaccharide (0.1 mg/mL, except xyloglucan which was 1 mg/mL), ascorbic acid (2 mM) and NcLPMO9C (5 µg/mL, 2.5 µg/mL or 1.25 µg/mL) solutions. Each reaction was performed in triplicate. The plate containing the reaction mixtures was incubated for 2 h at 50 °C following 10 min at 80 °C. Afterward, printing buffer (55.2% glycerol, 44% water, 0.8% Triton X-100) was added in each well as 1/4 of the final volume. Reaction mixtures were then printed using a piezoelectric noncontact microarray robot (Sprint, Arrayjet) onto a nitrocellulose membrane with a pore size of 0.45 μm (Whatman).

Probing and analysis of microarrays.

Nitrocellulose microarrays were probed as described (21). Briefly, the printed arrays were probed with a panel of anti-rat and anti-mouse monoclonal antibodies (PlantProbes and Biosupplies). Antibodies were diluted in PBS containing 5% wt/vol low fat milk powder (MPBS) to 1/10 and 1/1,000, respectively. For secondary antibodies, anti-rat and anti-mouse secondary antibodies conjugated to alkaline phosphatase (Sigma) were diluted in MPBS to 1/5,000. A complete list of antibodies may be found in the supplementary information. Developed microarrays were scanned at 2,400 dpi (CanoScan 8800F), converted to TIFFs and signals were measured using appropriate software (Array-Pro Analyzer 6.3, Media Cybernetics). Data are shown in a heatmap where color intensity is correlated to mean spot signal value. A cut off of 5 arbitrary units was applied. A fold-change heatmap was produced by calculating the following ratio: control signal:enzyme-treated signal.

Substrates for Enzyme Reactions.

The following substrates were used for exploring enzymatic activities of the LPMO and endo-glucanases: xyloglucan (XG) from tamarind seed (Tamarindus indica); glucomannan from konjak (Amorphophallus konjac); mixed linked β-glucan from barley (Hordeum vulgare); lichenan from Icelandic moss (Cetraria islandica); cellopentaose (Glc₅); xyloglucan-heptamer (XG7) with known composition XXXG; a reduced 14-mer (XG14^OH) derived from tamarind xyloglucan mainly containing XXXGXXXG^OH (the reducing end glucose is reduced to a glucitol); two types of mixed linked cellotetraosyl (G4G4G3G and G4G3G4G). All of these substrates were from Megazyme International Ireland. Finally, enzymes were tested against three natural sources of mixed linked β-glucan and xyloglucan: a suspension of ball-milled Icelandic moss and alcohol insoluble residues from Arabidopsis thaliana and from stems of tomato (Solanum lycopersicum).

Enzyme Reactions.

Reaction mixtures of 200-µL total volume contained 0.4 mg/mL substrate, 0.05 mg/mL enzyme and 1 mM ascorbic acid in 10 mM Na-acetate, pH 6.0. Samples were incubated at 40 °C with shaking at 600 rpm for 18 h. Products were analyzed using MALDI-ToF-MS, direct infusion ESI-MS or HPAEC (SI Materials and Methods). Determination of enzyme rates is described in SI Materials and Methods.