Extracellular electron transfer systems fuel cellulose oxidative degradation

See allHide authors and affiliations

Science  27 May 2016:
Vol. 352, Issue 6289, pp. 1098-1101
DOI: 10.1126/science.aaf3165

The fuel for fungal enzymes

Many microorganisms have specialized enzymes to target and break down plant biomass. In fungi, these enzymes, called lytic polysaccharide monooxygenases (LPMOs), partner with electron transfer partners to oxidatively cleave the polysaccharide backbone of lignocellulosic polymers. Kracher et al. examined several potential extracellular electron transfer partners for LPMO, including other enzymes and small redoxactive metabolites (see the Perspective by Martínez). All three were able to donate electrons to the single-copper active site. Such versatility helps these fungi adapt to a range of redox conditions and potentially use other extracellular electron donors to fuel biomass degradation.

Science, this issue p. 1098; see also p. 1050


Ninety percent of lignocellulose-degrading fungi contain genes encoding lytic polysaccharide monooxygenases (LPMOs). These enzymes catalyze the initial oxidative cleavage of recalcitrant polysaccharides after activation by an electron donor. Understanding the source of electrons is fundamental to fungal physiology and will also help with the exploitation of LPMOs for biomass processing. Using genome data and biochemical methods, we characterized and compared different extracellular electron sources for LPMOs: cellobiose dehydrogenase, phenols procured from plant biomass or produced by fungi, and glucose-methanol-choline oxidoreductases that regenerate LPMO-reducing diphenols. Our data demonstrate that all three of these electron transfer systems are functional and that their relative importance during cellulose degradation depends on fungal lifestyle. The availability of extracellular electron donors is required to activate fungal oxidative attack on polysaccharides.

Fungi play a central role in the terrestrial carbon cycle; their elaborate enzymatic machineries deconstruct plant biomass into its molecular building blocks (1). Cellulose, the most abundant and economically valuable polymer in plant cell walls, is broken down to glucose by endo- and exo-acting hydrolases through acid-base catalysis (2). Lytic polysaccharide monooxygenase [LPMO; family AA9 in the Carbohydrate-Active Enzymes (CAZy) database] has been shown to enhance such degradation by attacking crystalline cellulose and providing new chain ends as starting points for the hydrolases (3). Originally discovered in chitin-degrading bacteria (4), LPMO activity has since been found to degrade all major polysaccharides, including cellulose (5, 6), hemicellulose (7), and starch (8). High LPMO expression by fungi growing on lignocellulose substrates (9, 10) underscores the importance of these auxiliary redox activities. LPMOs use a single active-site copper ion to activate molecular oxygen and hydroxylate the polysaccharide backbone, eventually leading to chain breaks (11). Oxygen activation in the active site depends on exogenous electron donors (4, 12), several of which have been reported, including cellobiose dehydrogenase [CDH; CAZy subfamily AA3.1 (11, 13)], small-molecule reductants (4, 6), and photosynthetic pigments (14). Existing data are inconclusive, however, regarding the nature, availability, efficiency, and physiological relevance of extracellular electron donors.

To understand the nature and relevance of extracellular LPMO-reducing systems, we studied the LPMO-activating potential of phenols (mostly benzenediols) derived from plants and fungi together with redox enzymes, such as CDH, or other members of the glucose-methanol-choline (GMC) family of oxidoreductases (CAZy subfamily AA3.2). We compared these redox enzyme­–based electron systems with small-molecule reductants such as ascorbate (4), sulfur-containing species (5), gallic acid (6), and pyrogallol (15) by biochemical and electrochemical methods (supplementary materials). We also analyzed 97 fungal genomes to elucidate the relative importance of the various LPMO-activating mechanisms and their correlation with fungal lifestyle.

At least three electron transfer systems reduce LPMOs during fungal lignocellulose attack (Fig. 1). The extracellular flavocytochrome CDH (system 1) can reduce LPMO directly, whereas members of the GMC oxidoreductases (system 3) use plant-derived or fungal diphenols as redox mediators. The diphenols (system 2) efficiently reduce LPMO in the absence of regenerating enzymes but are irreversibly depleted by LPMO activity. The co-occurrence of lpmo genes with the genes of these LPMO-regenerating enzymes indicates the incidence of the three electron transfer systems in fungi.

Fig. 1 Three electron transfer systems reduce the LPMO (blue) active-site copper (cyan) to initiate cellulose attack.

System 1: CDH transfers electrons from its catalytic dehydrogenase domain (DH, yellow; CAZy subfamily AA3.1) via its mobile cytochrome domain (CYT, red) to the LPMO (32). System 2: Various plant-derived and fungal phenols can act as direct sources of electrons. The figure shows a LPMO electron donor identified in this study, 2,6-dimethoxy-1,4-benzenediol, and the oxidized quinoid reaction product. System 3: GMC oxidoreductases, such as GDH, GOx, or PDH (CAZy subfamily AA3.2), and the dehydrogenase domain of CDH regenerate the quinoid form of the redox couple (redox mediator) to reduce the LPMO active-site copper.

Genes encoding LPMO were found in 92% of the fungal genomes, whereas 58% contained cdh genes. Although white rots, soft rots, and plant pathogens had an overlap between these two gene families of 63 to 85%, brown rots, plant symbionts, and animal pathogens had a much lower co-occurrence of both genes (22 to 30%; Fig. 2A, figs. S1 to S4, and table S1). Ascomycota and Basidiomycota had a similar co-occurrence of cdh and lpmo genes, suggesting that the observed difference is independent of the species’ phylogenetic origin. On average, the number of lpmo genes was slightly higher when gmc genes (cdh excluded) were present in the genome (fig. S5) but much higher in genomes that also contained cdh. The latter genomes contained between 9 and 23 lpmo genes on average, which indicates the presence of a diversified set of LPMOs that are likely to carry out a range of catalytic tasks. The factor map generated by principal component analysis shows a high correlation between the LPMO and CDH variables (Fig. 2B and fig. S6). The clustering of individual data sets corresponds to the reported lifestyles for most of the investigated fungal species. The clusters of brown rots and plant symbionts correlate negatively with all extracellular oxidoreductase variables. The loss of oxidoreductases in brown rots has been explained by an evolutionary contraction (16).

Fig. 2 Occurrence and analysis of oxidoreductases linked to lignocellulose degradation.

(A) Relative occurrence of cdh, lpmo, and gmc (cdh excluded) genes in 97 genomes grouped by fungal lifestyles. The relative occurrence is indicated by color-coded bars, normalized to the number (n) of analyzed genomes given above [cdh and lpmo genes, red; gmc (cdh excluded) and lpmo genes, yellow; only lpmo genes, blue; none of these, gray]. None of the genomes contained cdh without lpmo. Numbers in the bars indicate the average number of lpmo genes for fungi in each category. The lifestyle of a fungus is often uncertain, and published data allowed the assignment of only 81 genomes (table S1). (B) Principal component analysis. The numbers of genes encoding CDH, GMC (CDH excluded), LPMO (CAZy family AA9), cellulase (Cel), laccase (Lac), and ligninolytic activities (POD) per genome were used as variables (small symbols; large symbols indicate group-specific centers of gravity). A high correlation among the analyzed genes is indicated by a similar orientation of the dark blue vectors. The contribution of the variables to the overall data variance is indicated by the line style of the vectors. The individual contribution (percent) to each dimension is further specified in the axis legend. Outliers in the lifestyle-dependent clustering of individual data sets could originate from an incorrect assignment.

CDH is a reported activator of LPMO (11, 13). To investigate the catalytic efficiency of CDH and thus the biological relevance of LPMO-CDH co-occurrence, the interactions of both enzymes were tested by rapid spectroscopy. The Neurospora crassa genome has two cdh and 14 lpmo genes, and both CDHs [CDH IIA, carrying a family 1 carbohydrate binding module (CBM1), and CDH IIB, lacking such a module] and four LPMOs from this soft rot were recombinantly produced and characterized (figs. S7 and S8 and tables S2 to S4). Fast electron transfer from the CDH was observed for all investigated enzyme combinations, albeit with different rates (kobs at pH 6.0 was between 0.9 s−1 and 20.6 s−1) and pH dependencies (Fig. 3, A and B, and tables S5 and S6). The observed interprotein electron transfer rates between CDH and LPMO at pH 6.0 exceeded the side activities of CDH with oxygen or ferric iron by two to three orders of magnitude. This shows that a previously suggested function of CDH, namely, the generation of hydroxyl radicals in a Fenton reaction through concomitant production of ferrous iron and H2O2 (17), is kinetically unfavorable. However, at pH 4.0, the production of H2O2 and ferrous iron became more competitive with LPMO reduction (table S7). Preferences of LPMOs for particular CDHs were demonstrated by higher electron transfer rates; such preferences could be due to variation in surface complementarity at the protein/protein interface. An inspection of isoelectric points showed no correlation to interprotein electron transfer rates (tables S3 and S6).

Fig. 3 CDH-catalyzed LPMO reduction.

(A) Electron transfer rates from N. crassa CDH IIA to LPMOs were measured by stopped-flow spectroscopy. LPMOs with different oxidative regioselectivity and cellulose-binding properties were tested (LPMO-02916, which has a CBM1, oxidizes cellulose at position C4; LPMO-01867 and LPMO-08760 oxidize at C1 and have a CBM1; LPMO-03328 oxidizes at C1 and lacks a CBM1). The observed heme b oxidation rates (0.9 to 20.6 s−1) at pH 6.0 were much higher than the reoxidation rate with oxygen (100 to 2100 times as high; table S6) or ferric iron (30 to 660 times as high; table S7). (B) The pH dependency of the observed electron transfer rates from CDH IIA or CDH IIB to each of the four LPMOs [same color code as in (A)] indicates a slightly acidic optimum pH. LPMO-08760 shows a preference for CDH IIA as an electron donor, whereas LPMO-02916 and LPMO-01867 were reduced faster by CDH IIB. Circles represent mean values of at least three replicates, and error bars represent standard deviations. (C and D) Electron transfer from a glassy carbon electrode to LPMO, mediated by (C) CDH IIA or (D) its isolated cytochrome domain (I, current; E, redox potential). Cyclic voltammetry showed a charging current for LPMO in the absence of oxygen; the onset potential is limited by the reduction potential of the heme b cofactor. In the presence of oxygen, a catalytic current was observed. The higher current observed for the cytochrome domain–mediated process may be due to its higher mobility compared with that of CDH. Data for LPMO-02916 are presented in fig. S9.

We also investigated electron transfer to LPMOs by cyclic voltammetry, which allows the direct monitoring of the flow of electrons between an electrode and a biomolecule (Fig. 3, C and D, and fig. S9). The LPMOs did not directly interact with a glassy carbon cathode but rather required either CDH or its isolated cytochrome domain as mediators to obtain electrons. LPMO catalytic currents were observed only in the presence of oxygen, thus demonstrating catalytic reduction of oxygen at the LPMO copper site. These results prove that the cytochrome domain alone is the structural element responsible for electron transfer. Several fungal genomes encode extracellular cytochrome domains carrying a cellulose-binding CBM1 (18), which could play a role in electron transfer to cellulose-bound LPMOs. Our data show that CDH is part of a specific extracellular redox chain, which is scarcely affected by oxygen and which comprises the FAD (flavin adenine dinucleotide) cofactor in CDH’s catalytic dehydrogenase domain [~0 mV versus the standard hydrogen electrode (SHE)], heme b in the cytochrome domain [~93 to 163 mV (19)], and the active-site copper in the LPMO [~224 mV in (20); ~250 mV in this work].

A variety of small-molecule reductants, including plant-derived compounds such as gallic acid and low-molecular-weight lignin products, can initiate LPMO activity under laboratory conditions (2123). The action of such reductants is essential in cases in which CDH is not expressed or is absent from the genome. Although several reductants have been found, it is not clear whether these compounds are biologically relevant, in terms of either their occurrence in biological systems or their reducing efficiency on LPMOs compared with that of CDH. Another uncertainty concerns the limited stability of some of these reductants (fig. S10) and their availability in decaying plant material. Screening of a large number of phenolic compounds revealed substantial differences in their LPMO-reducing efficiency (fig. S11 and tables S8 and S9). The highest product release occurred in experiments that used ascorbic acid and the diphenols 2,5-dimethoxy-1,4-benzenediol and 2,6-dimethoxy-1,4-benzenediol. Cellulose conversion by LPMO-02916 and various phenols resulted in typical degradation products (fig. S12).

The screening showed that plant-derived phenols are capable of efficiently reducing LPMOs. Spectroscopy and cyclic voltammetry were used to further investigate LPMO-catalyzed oxidation of monophenols to their corresponding phenoxy radicals and of diphenols to their corresponding quinones (figs. S13 and S14 and tables S10 to S13). We found no evidence for the oxidation of monophenols by LPMO, probably because of their high oxidation potential (>400 mV versus SHE; fig. S15), which is higher than the LPMO midpoint potential. In contrast, most diphenols have an oxidation potential close to or below that of the LPMO copper site. Among the investigated naturally occurring diphenols (Fig. 4A), methoxylated and methylated diphenols have oxidation potentials below 250 mV versus SHE and result in the fastest LPMO reduction rates (Fig. 4B). In 24-hour experiments, cellulose degradation by LPMO in combination with the reductants ascorbic acid, pyrogallol, and gallic acid, as well as diphenols or CDH, showed that all of these substances can reduce LPMO efficiently. The product release was highest for the CDH-catalyzed LPMO reduction in this experimental setup (Fig. 4C).

Fig. 4 Activation of LPMO by different reducing systems.

(A) LPMO-reducing substances from plants or fungi identified in this study are either di- or triphenols. Often, only the quinoid form of a diphenol was available. Ascorbic acid (17) is a commonly used LPMO reductant. (B) Plot of diphenol and quinone midpoint redox potentials (Ered) at pH 6.0 versus their reactivity with LPMO, as assessed by the turnover number for hydroquinone oxidation, which was obtained by means of spectrophotometric assays (table S13). The dashed line indicates the limiting oxidation potential, suggesting an LPMO midpoint potential close to 250 mV versus SHE (circles, mean values of at least three replicates; error bars, standard deviations; circle colors are a guide to the eye). (C) Oligosaccharides with a degree of polymerization from 2 to 5 that were released from cellulose during 24-hour conversion experiments in the presence of 1 mM reductant [ascorbic acid (17), pyrogallol (13), gallic acid (14), 3-methoxycatechol (5), catechin (16), caffeic acid (15), and catechol (2)]. In the presence of CDH (0.6 or 0.3 μM; orange bars), the quantification of oxidized cellodextrin reaction products is difficult. Substrate turnover was therefore estimated by the turnover of the cosubstrate lactose (right axis). Bars indicate means (error bars, standard deviations of at least three replicates). (D) LPMO reduction by diphenol and/or quinone redox mediators at final concentrations of 100 μM. The tested substances include 2,5-dimethoxy-1,4-benzoquinone (9), 2-hydroxymethyl-6-methoxy-1,4-benzoquinone (10), methoxy-1,4-benzoquinone (4), 1,2-naphthquinone-4-sulfonic acid (12), methyl-1,4-benzoquinone (3), and wood (hot water) extract in a concentration of 0.42 (high) and 0.084 (low) mg ml−1. LPMO activity was analyzed as in (C). GDH, its substrate glucose, and the redox mediators establish an electron transfer system that performs as well as the LPMO reducing substances in (C).

We then hypothesized that plant-derived diphenols and quinones (24, 25) and diphenols secreted by fungi (26, 27) can act as redox mediators by accepting electrons from GMC oxidoreductases of CAZy family AA3, which are abundantly present in the genomes and secretomes of biomass-degrading fungi. In the investigated fungal genomes, genes encoding non-CDH GMC oxidoreductases (CAZy subfamily AA3.2) showed a strong positive correlation with the presence of lpmo genes (fig. S5). Extending this observation, we demonstrated the ability of glucose oxidase (GOx), glucose dehydrogenase (GDH), pyranose dehydrogenase (PDH), and the isolated flavodehydrogenase domain of CDH to reduce these quinones (fig. S16 and tables S10 and S11). GDH was used together with LPMO to investigate the electron mediating process in more detail (Fig. 4D). The results demonstrate that diphenolic-quinoid redox couples perform as redox mediators, provided that there is efficient regeneration by a GMC oxidoreductase (fig. S17). In the presence of the redox mediators 2,5-dimethoxy-1,4-benzenediol, methoxy-1,4-benzenediol, or 2-hydroxymethyl-6-methoxy-1,4-benzenediol, 6 nM GDH regenerated a LPMO concentration that was 1670 times as high, and it did so as efficiently as 1-mM concentrations of the typically used reductants ascorbic acid and gallic acid. This finding suggests a role for secreted fungal oxidoreductases and could explain why “-omics” studies tend to lead to the conclusion that these enzymes are associated with biomass conversion. To compare the efficiency of wood-derived phenols with that of the isolated redox mediators described above, we investigated a beech wood (hot water) extract. The LPMO activity on microcrystalline cellulose (Fig. 4D and table S14) showed that beech wood extract reduced LPMO directly and in combination with GDH with the same efficiency as the isolated diphenols did. The addition of small amounts of GDH resulted in about a twofold increase in product formation, nearly reaching the product concentration obtained with 1 mM gallic acid.

Our data show that LPMO action—and thus cellulose degradation—can be fueled by different electron transfer systems. When LPMO and CDH are both present, interprotein interactions ensure specificity and efficiency of the extracellular electron transfer chain. Alternatively, plant-derived phenols and lignin fractions can serve as electron donors in fresh plant material through an unspecific reduction mechanism. Lastly, plant-derived or fungal diphenols and/or quinones serve as redox mediators between LPMO and GMC oxidoreductases, thus establishing an efficient electron transfer system. The lattermost observation may in part explain the abundance of GMC oxidoreductases, which supply H2O2 to peroxidases and hydroxyl radical–producing systems but whose roles remain enigmatic. The redox-mediating systems described in detail here bear resemblance to similar mediator-based systems that were hypothesized as early as 1977 to play a role in lignin conversion (28). Although we cannot yet pinpoint the biological relevance, our data reveal connections between enzymes involved in lignin degradation, Fenton reactions, and LPMO-catalyzed cellulose degradation. The quinone reductases of CAZy family AA6, which have been implicated in the recycling of quinones during the production of extracellular oxidizers by wood-rotting fungi (29), were present in 93 out of 96 screened genomes and may play a role in fueling LPMO action. Also, the recently discovered PQQ (pyrroloquinoline quinone)–containing dehydrogenase belonging to CAZy family AA12 (30) contains a cytochrome domain that could transfer electrons generated in its AA12 dehydrogenase domain to a LPMO.

The ability of LPMOs to versatilely accept electrons from different donors may allow a fungus to better adapt to different substrates, thus supporting LPMO activity during different stages of biomass degradation. This capacity to interact with different electron donors could include the ability to interact with electron transfer systems from different fungi, as a possible adaptation to co-occurrences of fungal species (31). The interaction of CDHs and LPMOs from different fungi was shown in previous work (8, 11, 13). Arguably, fungi secreting CDH or other GMC oxidoreductases as alternative electron transfer systems to plant-derived phenolic reductants can better regulate LPMO activity independently of changes in biomass composition, which may be caused by variation in available plant material or by depletion of reductants during the degradation process. We propose plant-derived phenols as the first electron donors that were available during the evolution of LPMOs. The advance of extracellular GMC oxidoreductases connected to enzymes involved in lignin degradation then introduced an additional electron transfer system based on plant-derived diphenols and/or quinones or diphenols secreted by fungi as redox mediators. Last, the fusion of one of these GMC oxidoreductases with a cytochrome domain evolved CDH as a specific and efficient electron transfer system.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S17

Tables S1 to S14

References (33110)

References and Notes

  1. Acknowledgments: This work was supported by the European Commission (project INDOX FP7-KBBE-2013-7-613549). D.K., M.P., and E.B. acknowledge support from the Austrian Science Fund (project BioToP; grant FWF W1224); S.S. acknowledges support from the BMWFW (Austrian Federal Ministry of Science, Research and Economy) IGS BioNanoTech; and A.K.G.F. acknowledges support from the Austrian Academy of Sciences (doctoral grant recipient). We thank C. Lorenz for technical assistance, M. Ravber for his generous gift of beech wood (hot water) extracts, and C. Keuschnig for discussing statistical topics. Data are available in the Knowledge Network for Biocomplexity data repository at
View Abstract

Navigate This Article