Abstract
The discovery of Lytic Polysaccharide Monooxygenases (LPMOs) has been instrumental for the development of economically sustainable lignocellulose biorefineries. Despite the obvious importance of these exceptionally powerful redox enzymes, their mode of action remains enigmatic and their activity and stability under process conditions are hard to control. By using enzyme assays, mass spectrometry and experiments with labeled oxygen atoms, we show that H2O2, and not O2 as previously thought, is the co-substrate of LPMOs. By controlling H2O2 supply, stable reaction kinetics and high enzymatic rates are achieved, the LPMOs work under anaerobic conditions, and the need for adding stoichiometric amounts of reductants is alleviated. These results offer completely new perspectives regarding the mode of action of these unique mono-copper enzymes, the enzymatic conversion of biomass in Nature, and industrial biorefining.
- Abbreviations
- (AscA)
- ascorbic acid
- (CDH)
- cellobiose dehydrogenase
- (Chl)
- chlorophyllin
- (GMC)
- glucose-methanol-choline oxidoreductase
- (GH)
- glycoside hydrolase
- (HAA)
- hydrogen atom abstraction
- (LPMO)
- lytic polysaccharide monooxygenases
- (pMMO)
- particulate methane monooxygenases
- (O2•−)
- superoxide
- (SOD)
- superoxide dismutase
- (XTH)
- xanthine
- (XOD)
- xanthine oxidase
The depolymerization of complex plant biomass, primarily composed of cellulose, various hemicelluloses and lignin, relies on a network of enzymatic and chemical reactions that is still full of mysteries. Until recently, the degradation of the recalcitrant polysaccharides in plant biomass was thought to be achieved by an arsenal of hydrolytic enzymes called glycoside hydrolases (GHs) (1). In some ecosystems, the enzymatic deconstruction process is supported by Fenton chemistry, i.e. transition metal-driven in situ generation of H2O2-derived hydroxyl radicals, one of the most powerful oxidizing species found on Earth (2), which can oxidize both polysaccharides and lignin in plant biomass (3). In 2010, a new class of enzymes was discovered, which carry out oxidative cleavage of polysaccharides (4). These enzymes, today known as lytic polysaccharide monooxygenases (LPMOs) (5), are single-copper redox enzymes (6, 7), that can hydroxylate the C1 or C4 positions of scissile glycosidic bonds (4, 8–10).
Despite their abundance in Nature (5, 11) and their obvious industrial importance, for example in the production of cellulosic ethanol (12), the mode of action of LPMOs remains enigmatic, although some catalytic mechanisms have been proposed (7, 8, 10, 13, 14). It is well-established that one LPMO reaction cycle requires the recruitment of two electrons (4, 14–16). The first electron is often thought to be acquired via reduction of the LPMO’s Cu(II) center to Cu(I) (11). When and how oxygen and the second electron are recruited remains an enigma. It appears impossible that an electron provider such as cellobiose dehydrogenase (CDH) (11) carries out direct reduction of the active site copper while the LPMO is bound to the substrate, whereas it is unlikely that the protein unbinds during catalysis to allow such a direct second reduction step. The existence of an internal electron channel that would allow electron delivery to a substrate-bound enzyme has therefore been postulated (10, 17,18).
Interestingly, a recent study has shown that unprecedented high levels of LPMO activity may be obtained when the enzyme is exposed to visible light in the presence of chlorophyllin (Chl) and ascorbic acid (AscA) (19). Although this study fell short of mechanistic explanations, the effect was attributed to the generation of high-energy electrons provided by photoexcited Chl, with AscA regenerating the Chl. From the increasing amount of publicly available data it appears clear that LPMO catalytic rates are indeed dependent on the nature of the redox partner (11, 20), which is intriguing, since it has been shown that the rate of LPMO reduction in solution is much higher (11, 16) than reported overall rates for LPMO action (4, 14, 21). These observations made us postulate that a chemical species, common to all known reaction systems but accumulating at different rates, plays an unsuspected key role in the LPMO mechanism. Looking for a potential culprit for LPMO activity, we studied the Chl/light, Chl/light-AscA and AscA systems for LPMO activation. A bacterial C1-specific cellulose-active LPMO10 from Streptomyces coelicolor (ScLPMO10C) was used as primary model enzyme.
When using the Chl/light-AscA system, with relatively high light intensities, a strong increase in LPMO activity was indeed observed, notably accompanied by an almost immediate inactivation of the enzyme (Fig. 1A). Since light-exposed chlorophyll may produce superoxide (O2•−) (22), we investigated whether addition of superoxide dismutase (SOD) or superoxide-consuming chemicals to the Chl/light-AscA system would allow better control of the reaction, which turned out not to be the case (Fig. S1). On the other hand, we found that both the catalytic rate and apparent inactivation of the enzyme could be modulated by varying the amount of AscA (Fig. 1A; Fig. S2) or the light intensity (Fig. S3). Interestingly, in the absence of AscA, the Chl/light system yielded good LPMO activity and apparent inactivation of the enzyme was much reduced, as illustrated by a more linear progress curve for LPMO activity (Fig. 1A). Under these latter conditions, low concentrations of SOD were beneficial for LPMO activity, whereas high concentrations of SOD were detrimental due to rapid inactivation of the enzyme (Fig. 1A; Fig. S4). These results show that the levels of superoxide and/or the products of SOD, O2 and H2O2, affect LPMO activity.
Figure 1C shows that, in the absence of an LPMO, the Chl/light system produces H2O2 and that production is strongly increased by adding SOD, which enzymatically converts superoxide to H2O2, or AscA, which chemically reduces superoxide to H2O2 (Fig. S5). These H2O2 production levels in the absence of the LPMO (Fig. 1C) correlate well with the initial rates observed in the LPMO reactions (Fig. 1A). Moreover, rapid enzyme inactivation in the LPMO reactions (Fig. 1A) correlates with the H2O2 production potential (Fig. 1C) of the system used and is associated with accumulation of H2O2 in the reaction mixture (Fig. 1B). Notably, in the control reaction with only Chl/light, yielding relatively stable reaction kinetics (Fig. 1A), accumulation of H2O2 was not observed (Fig. 1B), whereas the Chl/light system does produce H2O2 (Fig. 1C). Addition of catalase reduced the detrimental effect of adding high amounts of SOD, reflected in slower inactivation of the LPMO (Fig. 1A) and reduced accumulation of H2O2 (Fig. 1B). All together, these results suggest that H2O2 is an unsuspected co-substrate for LPMOs and that too high levels of H2O2 are detrimental. The high initial LPMO rate observed when using the Chl/light+AscA system (Fig. 1A) is likely related to fast H2O2 production (up to 200 μM within the 12 first minutes of the reaction; Fig. 1C), which leads to rapid inactivation of the enzyme.
Control reactions with only AscA, well known for its ability to drive LPMO activity, yielded more modest H2O2 levels (< 40 μM within the first 60 min, Fig. S6C), which is likely related to AscA being less capable of engaging in the thermodynamically challenging reduction of O2, compared to Chl/light. Reactions similar to those in Fig. 1 but only using AscA generally yielded less clear results (Fig. S6A&B), which is likely due to the many possible redox reactions involving AscA, superoxide and H2O2 (Fig S5). However, the same overall trend stood out: both higher LPMO activity and faster apparent enzyme inactivation were correlated with higher H2O2 levels.
Reactions with the Chl/light system (i.e. no AscA) seemingly lack a reductant needed to reduce the LPMO copper, which led us to speculate that O2•− could be involved in LPMO reduction (pathway (iv) Fig. S5). Indeed, chemical (KO2) or enzymatic (xanthine/xanthine oxidase) O2•− generating systems could drive LPMO activity, albeit at low levels (Fig. S7). Control experiments without any reductant but with exogenous H2O2 did not lead to cellulose oxidation (Fig. S8). This latter observation (Fig. S8) indicates that only the reduced LPMO can react with H2O2 and is crucial for the discussions below.
To determine the role of H2O2, we then analyzed initial LPMO rates in the presence of a reductant and varying concentrations of exogenous H2O2 (Fig. S9). A spectacular increase in initial LPMO rates was observed at the lower H2O2 concentrations, with up to 26-fold more soluble oxidized products being released from Avicel by ScLPMOl0C after 2 minutes when incubated in the presence of 200 μM H2O2 (Fig. S9C). This increase in activity is in the same order of magnitude as the increases reported for the Chl/light+AscA system (Fig. S3E and (19)). At higher H2O2 concentrations, the LPMO reactions stopped very rapidly. Exogenous H2O2 affected the activity of a fungal LPMO9 from Phanerochaete chrysosporium K-3 (PcLPMO9D) (Fig. S9D-F), another type of cellulose-active bacterial LPMO10, ScLPMO10B (Fig. S9G-I), and a chitin-active LPMO10, CBP21 (Fig. S9K-L), in a similar manner, but significant differences were observed in terms of the degree of activity enhancement and the sensitivity to H2O2 (note that the rate enhancement for CBP21 is >100-fold; Fig. S9L) Control reactions in which the enzyme was replaced by Cu(II)SO4 did not show any oxidized products (Fig. S10).
The results described above suggest a catalytic mechanism in which an H2O2-derived oxygen atom, rather than an O2-derived oxygen atom, would be introduced into the polysaccharide chain. In the proposed mechanism (Fig. 2), a priming reduction of the LPMO-Cu(II) to LPMO-Cu(I) occurs first. H2O2 would then bind to the Cu(I) center and homolytic bond cleavage, similar to what happens during Fenton chemistry, would produce a hydroxyl radical. This likely leads to formation of a Cu(II)-hydroxide intermediate and a substrate radical by one of several possible pathways (Fig. S11). In each of these mechanisms, the reaction between a copper-hydroxyl intermediate and the substrate radical leads to hydroxylation of the substrate and to regeneration of the Cu(I) center, which can enter a new catalytic cycle.
To test this pathway and obtain final proof of H2O2 being the catalytically relevant co-substrate of LPMOs, additional experiments were carried out (Fig. 3). Figures 3A&B show that LPMO-dependent consumption of H2O2 (Fig. 3A) correlates with the release of oxidized products (Fig. 3B). Importantly, these experiments were done using catalytic (rather than putatively stoichiometric) amounts of reductant (10 μM; i.e. 100 times lower than commonly used concentrations; Fig. S12) to assess the concept of a “priming reduction” and to reduce the effect of AscA on H2O2 stability (Fig. S13). Fig. 3B shows that product levels are much higher than the total amount of AscA added. This is in agreement with the proposed mechanism in which a reduced LPMO can catalyze several reactions provided that the co-substrate, H2O2, is supplied. Analogous results were obtained when using a glucose oxidase from Aspergillus niger (AnGOX), for controlled in situ generation of H2O2. Fig. S14 shows that the glucose/AnGOX system boosts ScLPMO10C activity in a dose-dependent manner, but only if the LPMO is reduced by a reductant added in small amounts (Fig. S14).
As a consequence of the above findings, LPMOs should be able to work under anaerobic conditions, which indeed was observed (Fig. 3C; Fig. S15). Fig. 3C shows that stable kinetics are obtained by adding H2O2 and reducing equivalents gradually to the reaction mixture and that the reaction rate is independent of the presence of O2. Finally, experiments with a labeled co-substrate, H218O2, showed that indeed, the oxygen introduced into the polysaccharide chain comes from H2O2 and not from O2 (Figs. 3D, S16-S18). For example, Fig. 3D shows that when using H218O2, the characteristic peaks for sodium adducts of the aldonic acid form of an oxidized cellohexaose (m/z 1029.7 &; 1051.7) shifted by +2 Da. Similar observations were made for the chitin-active LPMO10 CBP21 (Fig. S16) and a fungal cellulose-active LPMO9 (Fig. S17).Reactions with lower concentrations of H218O2 showed that even in the presence of a 10-fold surplus of 16O2, the oxidized products carry 18O (Fig. S18). Finally, a competition experiment with a peroxidase and an LPMO showed that the peroxidase completely inhibited LPMO activity, despite the presence of O2 and reducing power (AscA or lactose/CDH; Fig. S19). Altogether, the experiments depicted in Fig. 3 and Figs S15-S19 unequivocally show that H2O2 is the catalytically relevant co-substrate for LPMO-catalyzed oxidation of a polysaccharide.
Several of the reaction progress curves discussed above show that LPMOs are readily inactivated and under some conditions, such as when using the Chl/light-AscA system (Fig. 1A, Fig. S3D), inactivation seems to occur within a few minutes. Enzyme inactivation was confirmed by a series of experiments where the LPMO was pre-incubated and then tested for remaining activity (Fig. S20). Enzyme inactivation was similar in the presence of EDTA, showing that inactivation is not due to free metal-catalyzed generation of hydroxyl radicals. Importantly, inactivation was partly avoided by the presence of substrate (Fig. S20). Using proteomics technologies, we found that the inactivated LPMO had undergone several oxidative modifications that were confined to the catalytic histidines and, to a lesser extent, neighboring residues (Figs. 4, S21, S22). Other residues prone to oxidative damage, such as surface exposed residues in the LPMO domain, the linker or the CBM were not modified (Fig. S23). This leads to the important conclusion that oxidative damage is not caused by ROS in solution, as has been suggested (23), but by ROS generated in the catalytic center, i.e. in situ, by enzyme-generated hydroxyl radicals with diffusion-limited timescale reactivity. The protective effect of the substrate (Fig. S20, S24), was reflected in reduced oxidative damage of the N-terminal catalytic histidine (Fig. 4B). The higher sensitivity of the N-terminal histidine may be related to the orientation of the reactive oxygen species during catalysis.
The present findings unequivocally show that H2O2, and not O2, is the catalytically relevant co-substrate of LPMOs, implying that they should perhaps be called peroxygenases (or LPPO). Basically, LPMOs, after a priming reduction, carry out Fenton-type chemistry in a controlled and substrate-associated manner. From a biological point of view, such a scenario makes more sense than the seemingly somewhat random generation of dangerous hydroxyl radicals in the classical Fenton concept. Although the use of H2O2 by redox enzymes (e.g. peroxidases) is well-known, to the best of our knowledge, the biochemistry of the LPMO reaction as unraveled in this study, is unprecedented in Nature (25, 26).
LPMOs were initially classified as monooxygenases because experiments with 18O2 and using AscA as reductant, showed that one labeled oxygen atom was incorporated in the oxidized product (4). The proposed monooxygenase mechanism is well known (26, 27), seems “logical”, and has not been challenged until now. Strictly spoken, however, the seminal 2010 experiment with labeled oxygen did not show that O2 is the catalytically relevant co-substrate. In fact, other reactive oxygen species generated from O2 may have been the co-substrate, including H2O2, which indeed is produced under the conditions used (as shown in Fig. S6C). Importantly, the H2O2-pathway described here should not be confused with the “peroxide shunt” pathway that has been described for several monooxygenase reactions carried out by binuclear iron/copper enzymes, non-coupled binuclear copper enzymes and mononuclear iron enzymes containing additional co-factors such as porphyrin or biopterin. Such a “peroxide shunt” normally refers to a slow, rather artificial reaction that requires high concentrations of H2O2 (10-100 mM) and that is sometimes harnessed to avoid the use of reductants and O2 (28, 29). These shunt pathways involve the oxidized resting state of the enzyme and tend to lead to unstable reactions with a limited number of turnovers. The situation for LPMOs is very different and truly unique. LPMOs are mono-copper enzymes with no other co-factors, that, after an essential priming reduction, display stable reaction kinetics with multiple turnovers at low (sub mM) H2O2 concentrations.
Our findings explain several hitherto unexplained phenomena in LPMO biochemistry: (i) The consecutive delivery of two external electrons to the catalytic center is difficult to envisage, but with H2O2 being the co-substrate, recruitment of two electrons is not needed. (ii) The widely observed non-linearity of process kinetics is partly due to the self-inactivation of the LPMOs. (iii) The fact that most published catalytic rates for LPMOs are low and similar, and, most remarkably, independent of the LPMO or the substrate used (4, 14, 21), is likely due to the fact that the rate-limiting factor in most experiments was H2O2 formation. This point is well illustrated by a recent study demonstrating that the rate of a chitin-active LPMO fueled by the lactose/CDH system and the rate of H2O2 production by the latter system (in the absence of an LPMO) are similar (16). (iv) The increase in LPMO rate observed by Cannella et al. in their study on light-activation of LPMOs is due to production of hydrogen peroxide, not to the generation of some sort of “high energy electron” (19). (v) The observation that dehydrogenases can drive LPMO activity while strict oxidases cannot ((30) & Fig. S14) is due to the fact that, while both these enzyme types can produce H2O2, only the former can reduce the LPMO (11).
As to the level of H2O2 under reaction conditions, it is important to note that, notwithstanding the current findings, LPMOs are capable of activating molecular oxygen, albeit at low apparent rate (8, 31). It is well known that LPMOs generate H2O2 in the absence of substrate, which leads to the remarkable conclusion that LPMOs can generate their own co-substrate from O2. This property may have biological implications since H2O2 generated by unbound enzymes (several LPMOs display low substrate-binding) may be used by the substrate-bound population to degrade the substrate, explaining why H2O2 production by LPMOs is not observed in the presence of substrate (16, 32). It is conceivable that H2O2 interacts more strongly with substrate-bound, reduced LPMOs, compared to LPMOs in solution, which would explain why low concentrations of exogenous H2O2 are beneficial for activity, whereas higher concentrations lead to self-destructive reactions on unbound enzymes. Notably, the assumption that substrate-affinity has an impact on H2O2 management and self-destruction by the LPMOs sheds new light on the role of the CBMs that are appended to some LPMOs, including ScLPMO10C.
The link between H2O2, Fenton-type systems and enzymatic biomass depolymerization has been a matter of debate, controversy and investigations for several decades. The present findings reveal a novel role for H2O2. Glucose-methanol-choline (GMC) oxidoreductases are known H2O2 producers and, like LPMOs, abundant in fungal secretomes (11). Some GMC oxidoreductases can reduce LPMOs (11, 30), but their ability to produce H2O2, perhaps in a controlled manner, could be another important biological function, as suggested by our experiment showing that a reduced LPMO can be fueled by the H2O2 generating glucose/glucose oxidase system. Along the same line, a recent study of the secretome of Aspergillus nidulans grown on starches revealed co-secretion of LPMOs, catalase and H2O2-producing oxidoreductases (AA3, AA7) in the absence of known H2O2-consuming partners such as peroxidases (33). It is noteworthy that the present findings may also be relevant for understanding host-pathogen interactions since for instance LPMO-producing necrotrophic bacteria are known to benefit from H2O2 generated by the plant defense system (34).
The present findings will have far-reaching implications for the design of biorefining processes, including the production of cellulosic ethanol. LPMOs are important components of current commercial cellulase cocktails (12) but proper aeration and delivery of electrons at industrial scale are considered major challenges as is the instability of LPMOs. We show here that LPMO performance and stability can be controlled by regulating the supply of H2O2, a liquid, cheap and easy-to-handle industrial bulk chemical. We further show that LPMOs can act in the presence of only catalytic amounts of reductant, which abolishes reductant-induced undesirable redox side reactions, and in the absence of oxygen, which eliminates the need for aeration. So far, the application of LPMOs has likely been hampered by suboptimal process conditions and it seems evident that further process improvements may be achieved now that the role of H2O2 has been uncovered. Notably, overdosing LPMOs can be a problem, since lack of LPMO binding sites on the substrate may lead to LPMO inactivation. It is conceivable that careful balancing of LPMOs and hydrolytic enzymes (e.g. cellulases) is needed, with the cellulases “peeling off” LPMO-disrupted polymer chains from the substrate surface, thus exposing new LPMO binding sites. As to LPMO stability, it is interesting to note that one of the residues most vulnerable to oxidation, the N-terminal catalytic histidine, is methylated in fungal LPMOs. Perhaps this methylation helps protecting the fungal LPMOs from oxidative self-destruction.
In the six years after their discovery (4), the role of H2O2 in LPMO catalysis has been overlooked, despite intense worldwide research on these enzymes. It is tempting to speculate that a similar situation may exist for other enzymes, in particular for copper monooxygenases that are thought to require two electrons and molecular oxygen. It has not escaped our notice that the still enigmatic particulate methane monooxygenase (pMMO), whose active site bears some resemblance to LPMO active sites, displays LPMO-like H2O2-related features: it has been reported that production of H2O2 by pMMO is lower in presence of substrate (35) and that H2O2 binds to and can oxidize the pMMO active site (36). It is conceivable that the present findings have implications beyond understanding and optimizing the enzymatic conversion of recalcitrant polysaccharides by LPMOs.
Acknowledgments
We thank Bjørge Westereng at NMBU, Ås and Mats Sandgren at SLU, Uppsala, Sweden for providing a sample of a purified recombinant fungal AA9 (PcLPMO9D). We thank Jennifer Loose at NMBU, Ås for providing the catalase katE. B.B. has received the support of the EU in the framework of the Marie-Curie FP7 COFUND People Programme, through the award of an AgreenSkills fellowship (under grant agreement n° 267196). The postdoctoral fellowship of B.B. was also supported by the French Institut National de la Recherche Agronomique (INRA) [CJS]. This work was also supported by the Research Council of Norway through grants 214613, 240967, 243950 and 249865, and by the Vista programme of The Norwegian Academy of Science and Letters through grant 6510.