Cellobiose dehydrogenase: An essential enzyme for lignocellulose degradation in nature – A review / Cellobiosedehydrogenase: Ein essentielles Enzym für den Lignozelluloseabbau in der Natur – Eine Übersicht

To date, the literature reports the characterization of about 30 CDHs from their native fungal producers. The formation of CDH in vitro requires the presence of insoluble cellulose or soluble cello-oligosaccharides, which act as an inducer (Hori et al., 2012). CDH is secreted into the extracellular space and typically represents about 0.5-2.5% of the total secreted protein on a mass basis (Zámocký et al., 2006). Most of the characterized CDHs were isolated from basidiomycetous white-rot fungi, which are ubiquitous wood degraders in boreal forests. They secrete a battery of oxidative and hydrolytic enzymes allowing them to completely mineralize complex lignocellulosic substrates. Notable examples of CDH-producing strains are the extensively studied model organisms P. chrysosporium and Trametes versicolor, which are also the first reported CDH-producers (Westermark and Eriksson, 1974). A recent, unexpected finding was the formation of CDH by the selective delignifier Ceriporiopsis subvermispora (Harreither et al., 2009) when grown on cellulose as the sole carbon source. A secretome analysis of the fungus grown on more complex lignocellulosic carbon sources showed that CDH is secreted in the late stages of growth, and most likely supports the weak cellulose-degrading machinery of the fungus after the depletion of lignin (Hori et al., 2014). In contrast to white-rot fungi, the cellulose-specific brown-rots contain a drastically reduced set of lignocellulose degrading enzymes. To date, CDH secretion by these fungi has only been shown for Coniophora puteana (Schmidhalter and Canevascini, 1993). Most brown-rot fungi lack cdh-encoding genes, which supports the view that they rely on other, non-enzymatic mechanisms for wood decay (Floudas et al., 2012). CDH has also been isolated from the secretomes of some plant-pathogenic species, suggesting a role in the host-pathogen interaction. The phytopathogenic fungus Sclerotium rolfsii, which has a host-spectrum of approx. 600 plants, is recognized as a potent producer of CDH, whose secretion has been studied and optimized (Sachslehner et al., 1997; Ludwig and Haltrich, 2003). Moreover, a proteome analysis showed the presence of CDH in the “early” secretome of the grape-infecting fungus Botrytis cinerea, which was collected 16 hrs post-infection. The enzymes isolated at this stage of growth presumably play an important role in establishing a successful infection (Espino et al., 2010). Another recent example is the ascomycete Mycosphaerella fijiensis, which secreted CDH during in vitro cultivation on banana leaves (Escobar-Tovaret et al., 2015). In contrast, a mass spectrometric analysis did not detect CDH in the secretomes of the ascomycetes Aspergillus nidulans and Aspergillus niger during plant infection (Couturier et al., 2012), although these organisms contain putative cdh sequences in their genomes. The precise role of CDH in phytopathology therefore remains unclear, but may be connected to the initial penetration of plant cell walls. In recent years a growing number of CDHs from ascomycete fungi has been reported. In contrast to mesophilic basidiomycetes, ascomycetes commonly degrade lignocellulose in more extreme habitats at elevated temperatures, higher moisture contents or at alkaline pH (Worrall et al., 1997). Ascomycetes preferentially metabolize cellulose and hemicellulose, while lignin remains generally unaffected, or is only slightly modified (Worrall et al., 1997). A comprehensive screening of 40 ascomycete fungi showed significant CDH activity in 13 strains using cellulose-based media and shaking flask cultures (Harreither et al., 2011). The cellulose concentration in the cultivation medium had the highest impact on CDH activity, while the availability of the nitrogen source (peptone) seemed less important in most cases. The genome of the ascomycete model organism Neurospora crassa contains two active CDHs (IIA and IIB), both of which were previously expressed and characterized (Zhang et al., 2011; Sygmund et al., 2012). A proteome analysis showed a high prevalence of CDH IIA in the secretome (2.4% mass abundance) when cellulose was the sole carbon source (Phillips et al., 2011a). The corresponding cdh gene was significantly (at least three-fold) upregulated when the fungus was grown on complex substrates such as barley, corn, rice, soybean and wheat (Wang et al., 2015). Thus, cdh belongs to a core-set of genes in N. crassa which are induced under lignocellulolytic conditions. In contrast, growth on xylan and pectin did not enhance transcription of cdh (Wang et al., 2015). Among the known ascomycetous CDHs, several show elevated (neutral or alkaline) pH optima and a high thermostability. Examples are CDHs isolated from the thermophilic fungi Corynascus thermophilus (Harreither et al., 2011; 2012b) and Chaetomium sp. INBI 2-26(–) (Karapetyan et al., 2006), which have maximal catalytic activities at a neutral pH around 7. This is considerably higher than the pH optima of “classical” basidiomycete CDHs, which usually exhibit highest catalytic activities in an acidic milieu (Table 1).

In recent years large-scale genome sequencing projects, such as the 1000 Fungal Genomes Project (Hibbett and Taylor, 2013; Hori et al., 2013), uncovered a plethora of putative cdh sequences which show a high overlap with lpmo and cellulase-encoding genes (Beeson et al., 2015). CDH sequences are abundant in both Ascomycota and Basidiomycota, with sequence identities as low as 35%. A generally high sequence homology of the CYT domains in all cdh sequences, including the rare His/Met iron coordination (see Chapter 5.1 for details), suggests a common ancestor that probably existed before the separation of Dikarya into the phyla Ascomycota and Basidiomycota (Zámocký et al., 2004) about 600 mio. years ago (Floudas et al., 2012). While basidiomycete genomes contain single copies of cdh genes, up to three copies are found in ascomycetes. A general classification of these sequences based on maximum likelihood phylogenetic reconstruction has been proposed (Zámocký et al., 2006; 2008). As shown in Figure 1, CDHs form three well-supported phylogenetic branches: Class-I CDHs from basidiomycetes and Class-II or Class-III CDHs from ascomycetes. Class-II CDHs can be further divided into Class-IIA harboring a CBM1, with the remainder of sequences forming Class-IIB. The existence of Class-III CDHs was suggested previously (Zámocký et al., 2008), but to the best of our knowledge, none of these sequences has been functionally expressed or isolated so far.

CDHs show a high preference for soluble, β-(1,4)-interlinked saccharides, but scarcely oxidize monosaccharides. Highest catalytic efficiencies are observed for β-D-cellobiose, the dimeric subunit of cellulose, and higher soluble cello-oligosaccharides. Lactose, an epimer of cellobiose, displays comparable catalytic constants. The catalytic capabilities of many CDHs partially correlate to their phylogenetic classification, as is exemplified in Table 1. Generally, Class-I CDHs exhibit higher turnover numbers (k_cat, s-1) for the native substrate cellobiose, while higher affinities (lower K_M values) are reported for Class-II CDHs. The substrate spectrum of Class-II CDHs appears to be more relaxed, and also extends to the (starch-derived) β-(1,4)-linked maltose which is oxidized at a reasonable rate. In addition, the discrimination against monosaccharides in this class seems to be less pronounced compared to basidiomycetous CDHs. Notable examples are Thielavia terrestris and C. thermophilus CDHs, both of which have 4 – 70-fold lower K_M values for glucose relative to other characterized CDHs. Members from both classes (I and II) are reported to oxidize hemicellulose-derived sugars, such as xylobiose and higher xylooligosaccharides, and also conversion of galactosyl-mannose was reported (Henriksson et al., 1998). P. chrysosporium CDH also oxidized the β-(1,4)-linked disaccharide 2α-mannobiose (Bao et al., 1993), which was not observed for N. crassa CDH IIA and IIB. However, the latter enzymes readily oxidized mannopentaose (Sygmund et al., 2012). Several studies also showed the potential of CDH to oxidize the reducing end groups of insoluble cellulose fibers. Some authors reported catalytic activity of P. chrysosporium CDH in the presence of cellulose (Morpeth, 1985; Kremer and Wood, 1992b). Pricelius et al. (2010) observed that incubation of pure M. thermophilum CDH with cotton or microcrystalline cellulose led to the formation of minor amounts of hydrogen peroxide in the absence of soluble CDH substrates.

Later, crystal structures of separated CYT and DH domains were reported. In 2000, the X-ray crystal structure of the isolated CYT domain of P. chrysosporium CDH was resolved at 1.9 Å (pdb: 1D7C) and revealed a scaffold similar to the antibody fold of the heavy-chain Fab fragment (Hallberg et al., 2000). The 190 amino-acid long CYT has a β-sandwich topology forming a compact globular structure with a diameter of approx. 35 Å. The hydrophobic heme b-binding pocket is located at the surface of CYT. Notably, the propionate-A arm of the heme cofactor is partly exposed to the surrounding solvent. The heme iron is hexa-coordinated by Met and His as axial ligands (Cox et al., 1992). The replacement of either of these ligands in P. chrysosporium CDH by alanine resulted in a weaker binding of the heme and destabilized the CYT fragment, which was rapidly degraded in the cultivation medium due to proteolytic cleavage (Rotsaert et al., 2001). Besides CDH, only few examples of heme proteins with Met/His ligation exist in literature. Among these are the ubiquitous electron transfer protein cytochrome c (Klein-gardner and Bren, 2015), which is part of the mitochondrial electron transport chain, and cytochrome b₅₆₂ found in the periplasm of Escherichia coli (Mathews et al., 1979). In addition, a secreted cytochrome b₅₆₂ domain featuring a cellulose-binding CBM1 has also been identified in the supernatant of P. chrysosporium (Yoshida et al., 2005). While its function remains to be clarified, a high homology to CDH`s CYT fragment, including a conserved pair of His/ Met heme ligands, suggests a role in extracellular electron transfer. The cytochrome moiety of the pyrroloquinoline quinone (PQQ)-dependent pyranose dehydrogenase isolated from the fungus Coprinopsis cinerea (Matsumura et al., 2014) also shows a well-conserved His/Met pair analogous to the CYT domain in CDH (Takeda et al., 2015).

The crystal structure of the P. chrysosporium DH domain (1.6 Å resolution, pdb: 1KDG) was reported in 2002. DH is a member of the large, structurally related family of glucose-methanol-choline (GMC) sugar oxidoreductases. The structure shows a global resemblance with the p-hydroxybenzoate hydroxylase (PHBH)-fold, which is common for flavin-containing enzymes (Hallberg et al., 2002). Two structurally distinct subdomains are observed: a flavin-binding domain (F-domain) and a substrate-binding domain (S-domain). The F-domain features a typical, flavin-binding βαβ-motif, the Rossman-fold, which is a common structural motif in FAD or NAD-binding proteins. The FAD moiety is non-covalently bound to the enzyme and shows a butterfly-like bending with a torsion angle of 22° in P. chrysosporium DH. It is well buried in the interior of the enzyme, approximately 12 Å below the protein surface. However, the active site can be accommodated via a substrate channel located at the subdomain interface.

The recently reported full-length crystal structures of two CDHs from the ascomycetes M. thermophilum and N. crassa allowed for a more detailed insight into the dynamic interactions between CYT and DH (Tan et al., 2015). Two structures of N. crassa CDH IIA showed an “open” conformation in which DH and CYT were spatially separated, whereas a structure of M. thermophilum CDH showed a “closed” conformation (Figure 2A). Analysis by SAXS also suggested a number of possible intermediate conformers that exist in solution. The crystal structureof M. thermophilum CDH was determined at a resolution of 3.2 Å and revealed a tight association of DH and CYT. CYT was oriented with the heme b facing the active site of DH, and its propionate-A chain protruding into the substrate binding pocket. The observed close distance of 9 Å between the heme b and FAD cofactors (edge-to-edge) in this conformation would theoretically allow direct electron transfer between the redox centers. Interestingly, the substrate channel leading to the active site was not blocked by CYT, and substrates as well as products could spatially accommodate the active site in closed conformation. The site-directed mutagenesis of amino acids along the interface of DH and CYT provided further evidence for the relevance of the “closed” state of CDH. Mutation of Ser298, Met309 and Arg698 on the surface of DH barely affected the substrate oxidation, but selectively weakened the interaction between the FAD and heme b cofactors as was shown by a kinetic analysis. These amino acids also formed the docking platform for CYT in the “closed” structure. Furthermore, two “open” conformations of N. crassa CDH IIA were resolved at a resolution of 2.9 Å. In both the structures, the domains showed different relative orientations, as well as different conformations of the linker peptides. The maximal edge-to-edge distance between FAD and heme b in these conformers was approx. 60 Å, which indicates a high mobility of the two subdomains. In both structures the heme b propionate-A was fully solvent-exposed. These observations were corroborated by the SAXS analysis of full-length CDH using the ensemble-optimization method (EOM). In solution, intermediate conformers were detected which included the closed and open state, and the addition of the inhibitor cellobionolactam (CBLM) restricted the number of observed conformers. While the “closed” conformation of CDH favors interaction between the redox centers, the open conformation may allow for interaction of CYT with external electron acceptors, such as LPMO.

Co-crystallization with the competitive inhibitor CBLM, which resembles a possible transition state, revealed the active site architecture and identified key amino acid residues for substrate binding and catalysis (Hallberg et al., 2003; Tan et al., 2015). A series of ligand interactions with conserved amino acid residues was observed, which formed a dense hydrogen bonding network with CBLM (Figure 2B). Generally, the active site is composed of two sub-sites, a catalytic subsite C and a substrate-binding subsite B. CBLM was positioned in the active site with the C1 carbon (corresponding to the reducing end in cellobiose) facing the N5 atom of the isoalloxazine ring at a distance of approximately 2.9 Å. This orientation would allow oxidative attack of cellobiose by a catalytic histidine (His701 in M. thermophilum CDH), which is the only basic residue in the vicinity of C1. The binding pose of CBLM strongly supports a general hydride transfer mechanism, in which the catalytic His as the general base initially abstracts a proton from the C1 hydroxyl group. The transfer of the anomeric (C1) hydrogen to the N5 atom of the FAD cofactor then results in a 2-electron reduced FAD (hydroquinone form), while cellobiose is oxidized to cellobionolactone. The role of His as a catalytic base was confirmed by mutation of the structurally conserved His689 in P. chrysosporium CDH, which resulted in kinetically impaired enzymes (Rotsaert et al., 2003). Later, it was also shown that mutations of His701 in M. thermophilum had the same detrimental effect (Sygmund et al., 2013). The rational engineering of amino acid residues in the substrate binding subsite B also allowed changing the substrate specificity of the enzyme. In P. chrysosporium CDH, Glu279 specifically interacts with the 2^nd hydroxyl group of the non-reducing end sugar moiety (Hallberg et al., 2003). Replacement of this residue by Gln had little effect on cellobiose turnover, but completely abolished the activity towards lactose (Desriani et al., 2010). Using the CDH of M. thermophilum, Sygmund et al. (2013) demonstrated that replacement of Asn700 (Asn688 in P. chrysosporium) by Ser enhanced the re-oxidation of FADH₂ by artificial electron acceptors and oxygen. This amino acid is located in-between subsite B and subsite C, and steric changes at this position may influence the accessibility of the FAD cofactor for diffusible substrate or acceptor molecules (Tan et al., 2015).

Oxidative enzymes greatly contribute to the overall degradation capacity of lignocellulolytic fungi (Horn et al., 2012). A distinct reaction catalyzed by the so-called lytic polysaccharide monooxygenase (LPMO, CAZy: AA9) introduces chain breaks into otherwise recalcitrant crystalline cellulose surfaces, which creates isolated glucan chains susceptible for attack by cellulases (Eibinger et al., 2014). Consequently, LPMOs came to be known as “cellulase booster”, and have since been employed in commercial enzyme cocktails to increase the degradation of lignocellulosic biomass.

LPMOs activate molecular oxygen, which is inserted at the C1 or C4 atom adjacent to a glycosidic bond to create an instable intermediate that eventually leads to strand breaks (Phillips et al., 2011b). LPMO employs a mononuclear type-II copper atom in the active site, which is located on a surface-exposed flat patch oriented towards the cellulose surface during catalysis (Figure 4A).The precise reaction mechanism by which LPMOs cleave cellulose and the type of the intermediate oxygen species are presently not known. Importantly, the activity of LPMO depends on the reductive activation by an external electron donor, which could be an artificial small-molecule reductant such as ascorbic acid (Quinlan et al., 2011) or potentially an electron transferring partner enzyme. Since its discovery, the importance of CDH for the degradation of cellulose was noted, and an increasing number of transcriptome and proteome studies report the co-expression of CDH and LPMO during ligno-cellulose degradation (see, e.g., Phillips et al., 2011a; Hori et al., 2014; Couturier et al., 2015). Furthermore, putative cdh and lpmo genes show a high overlap in fungal genomes, although some fungi contain a smaller number of solitary lpmo genes (Rytioja et al., 2014). A notable example is the industrially exploited fungus Trichoderma reesei, which does not contain putative CDH-encoding genes but actively secretes LPMOs under cellulolytic conditions (Adav et al., 2013). Lignin-derived compounds have recently been identified as additional electron sources for LPMO (Dimarogona et al., 2012; Westereng et al., 2015). Such soluble lignin breakdown products are released during fungal lignocellulose degradation, and could potentially augment or substitute CDH activity. The exact physiological role of these alternative reducing agents still remains elusive, but is currently of high scientific interest.

Before the discovery of LPMO activity, Bao and Rengana-than realized in 1992 that the addition of purified CDH to crude extracts of T. reesei enhanced the depolymerization of crystalline cellulose by up to 49% (Bao and Renganathan, 1992). The same behavior was observed for T. viride (Bao and Renganathan, 1992), which contains putative LPMOs in its genome, but lacks CDH-encoding genes. Likewise, a CDH-deficient strain of T. versicolor showed a severely impaired cellulose catabolism (Dumonceaux et al., 2001; Canam et al., 2011). In 2011, Langston et al. (2011) were the first to conclusively link the electron transfer properties of CDH to the enigmatic activity of LPMO. In a landmark paper, they reported that a combination of LPMO from T. terrestris (TtGH61E) and CDH from the same organism increased the performance of various cellulases approximately two-fold; an effect that was seen for crystalline and amorphous (phosphoric acid swollen) cellulose substrates (Langston et al., 2011). However, the specific mechanism underlying the synergy of the CDH/LPMO system remained unclear. These results were soon after corroborated by a study which showed that deletion of a single cdh gene in the wood-degrading fungus N. crassa reduced its cellulose-degrading abilities by approximately two-fold (Phillips et al., 2011a). The cellulolytic capabilities of the deletion strain could be fully recovered by supplementation with exogenous Myceliophthora thermophila CDH. The same study also showed that purified LPMOs from N. crassa in combination with full-length CDH directly oxidized cellulose (Phillips et al., 2011b). Thereafter, a number of reports showed that CDH instigates LPMO, leading to the formation of soluble oxidized cello-oligosaccharides which are not observed when adding only one of the enzymes (Table 2).

Interestingly, many studies have shown that single CDHs can activate multiple LPMOs. This observation is in line with transcriptome or secretome studies; for example, the secretome of the ascomycete N. crassa contains four LPMOs along with one CDH when cultivated in the presence of cellulose (Phillips et al., 2011a). The interaction across species borders was also observed in many cases. The multitude of lpmo sequences found in many fungal genomes (e.g., 30 in C. cinerea; Harris et al., 2011) might be explained by a functional and catalytic diversity of LPMOs. For example, LPMO9C from N. crassa was shown to cleave hemicellulosic polymers as well, such as xyloglucan (Agger et al., 2014), and LPMO9A from M. thermophila catalyzed the oxidative cleavage of xylan (Frommhagen et al., 2015). A new family of starch-degrading LPMOs (CAZy: AA11) found in N. crassa was recently introduced (Vu et al., 2014). Along these lines, it has to be noted that the substrate spectrum of CDHs largely coincides with the polysaccharide-degrading capabilities of these fungal LPMOs. This provides the possibility for a functional regulation mechanism, in which the differential expression of LPMOs with diverse catalytic capabilities may allow a fungus to specifically adapt to different growth conditions or habitats (Figure 5). The catalytic promiscuity of some CDHs, for example, towards starch-derived malto-oligosaccharides or hemicellulosic sugars, may allow activation of LPMOs during the degradation of these polysaccharides.

The detailed mechanism of electron transfer between CDH and LPMO is currently a subject of great scientific interest. The work of Phillips et al. (2011b) indicated for the first time that the cellulolytic machinery of N. crassa requires full-length CDH and oxygen for full activity, but does not respond to the isolated DH fragment. This observation represents the first, albeit indirect, evidence for a CYT-mediated activation of LPMO. Sygmund et al. (2012) later provided an assay-based indication for interaction, which showed that the reduction of the one-electron acceptor cytochrome c by CDH was slowed down considerably in the presence of N. crassa LPMO9C. The suggested reason was a competition for electrons between LPMO and cytochrome c at CDH’s CYT domain. The strong inhibition of activity towards cytochrome c, which is an electron acceptor with a high bimolecular reaction rate, suggested a similarly fast interaction for CDH with LPMO. A direct spectroscopic approach published by Vu et al. (2014) indeed showed that fully reduced CDH re-oxidized considerably faster in the presence of the starch-active LPMO NCU08746 than by oxygen alone. The wavelength used for monitoring the redox state of CDH (430 nm) was specific for the heme b, which indicated that LPMO was a better electron acceptor for CDH than atmospheric oxygen. These important findings laid the foundation for further research. Tan et al. (2015) used the separated domains of N. crassa CDH IIA to show that direct electron transfer between CDH and LPMO occurs via CYT, but not via DH. Rapid-kinetics experiments revealed that the CYT-to-LPMO electron transfer proceeded with an apparent rate of 67 s⁻¹, which was about 100,000 fold faster than the rate measured for oxidation of CYT by atmospheric oxygen alone. No apparent interaction between DH and LPMO could be detected using the same experimental setup. These studies, however, were performed in homogenous solution in the absence of a suitable substrate for LPMO. The precise mechanistic and kinetic details of this interaction in a heterogeneous environment in the presence of cellulose await experimental clarification. A question of particular interest is how LPMO could receive electrons during catalysis. The ambiguity-driven docking experiments with the CYT domain of N. crassa CDH IIA and LPMO9C favored direct heme-to-copper interaction in solution (Tan et al., 2015). To date, the precise catalytic mechanism of LPMO is not known with certainty, but the suggested reaction pathways involve the formation of an intermediate oxyl radical (Kim et al., 2013) or a superoxo species (Phillips et al., 2011b; Beeson et al., 2012; 2015), both of which have a high oxidation potential. In both catalytic mechanisms, two separate single-electron transfers are required to facilitate hydroxylation and subsequent glycosidic bond cleavage in cellulose. While direct interaction between the CYT and LPMO cofactors in solution is plausible, the transfer of the second electron during catalysis would require the detachment of LPMO from the cellulose surface and subsequent re-attachment to cellulose to complete the reaction (Beeson et al., 2015). A single docking study predicted a CDH-binding site on the LPMO surface (Li et al., 2012) (Figure 4A, B) which is well-conserved in fungal cellulose-active LPMOs. Such interaction requires long-range electron transfer from the protein surface to the active site via conserved amino acid residues. In this scenario, CDH could donate electrons to LPMO during catalysis, while the active site of LPMO is oriented towards cellulose (Beeson et al., 2015). A high overall conservation of putative electron transferring amino acids, spanning the distance from the CDH binding site to the active-site copper (~20 Å) was noted (Beeson et al., 2015).

Several questions concerning the mechanistic context of this interaction remain to be addressed. For example, no structurally conserved surfaces on CYT domains are found; rather, the heme b propionate appears to be the only absolutely conserved surface feature among different CDHs. Finally, it remains to be elucidated how the charged propionate group of CYT could interact with the proposed (hydrophobic) “PGP” binding patch on the LPMO surface (Figure 4B). Also, two electron-donors such as ascorbic- or gallic acid were repeatedly shown to efficiently reduce the LPMO copper center and efficiently drive the oxidative cellulose degradation (Quinlan et al., 2011), which may indicate a 2-electron transfer to the copper before substrate binding. Mutational studies and further structural investigation of the CDH/LPMO couple will be needed to fully understand their interaction during oxidative cellulose degradation.