Polysaccharide oxidation by lytic polysaccharide monooxygenase is enhanced by engineered cellobiose dehydrogenase

The catalytic function of lytic polysaccharide monooxygenases (LPMOs) to cleave and decrystallize recalcitrant polysaccharides put these enzymes in the spotlight of fundamental and applied research. Here we demonstrate that the demand of LPMO for an electron donor and an oxygen species as cosubstrate can be fulfilled by a single auxiliary enzyme: an engineered fungal cellobiose dehydrogenase (CDH) with increased oxidase activity. The engineered CDH was about 30 times more efficient in driving the LPMO reaction due to its 27 time increased production of H2O2 acting as a cosubstrate for LPMO. Transient kinetic measurements confirmed that intra‐ and intermolecular electron transfer rates of the engineered CDH were similar to the wild‐type CDH, meaning that the mutations had not compromised CDH’s role as an electron donor. These results support the notion of H2O2‐driven LPMO activity and shed new light on the role of CDH in activating LPMOs. Importantly, the results also demonstrate that the use of the engineered CDH results in fast and steady LPMO reactions with CDH‐generated H2O2 as a cosubstrate, which may provide new opportunities to employ LPMOs in biomass hydrolysis to generate fuels and chemicals.


Introduction
Lytic polysaccharide monooxygenases (LPMOs, EC: 1.14.99.53-56, CAZy IDs: AA9-11, AA13-16) are copper-containing enzymes found in chitin-, starch-, and plant cell wall-degrading organisms [1][2][3][4][5]. LPMOs use an unprecedented oxidative reaction mechanism to cleave and locally decrystallize these biopolymers, which enhances the activity of associated hydrolases [1,2,[6][7][8]. The reaction cycle starts with the reduction of LPMO's mononuclear type-II copper by external electron donors. After binding of a cosubstrate (O 2 or H 2 O 2 ), an oxygen atom is inserted into the C-H bond at the C1 or C4 carbon of a glycosidic bond in the substrate [9][10][11], which is broken by a spontaneous elimination reaction. The detailed catalytic mechanism of LPMOs and the nature of their cosubstrate are currently debated. Both O 2 [12][13][14] and H 2 O 2 [15,16] are reported to interact with reduced LPMO, but with different implications on the reaction pathway. The monooxygenase reaction using O 2 requires two sequential electron transfer steps and is likely to proceed via an intermittent dioxo [9,12] or oxyl [17] species. In contrast, a single reduction step to yield Cu(I) could be followed by binding of H 2 O 2 to directly form an oxyl-or a hydroxyl species [15]. The reported turnover numbers of LPMO with H 2 O 2 as cosubstrate are one to two orders of magnitude higher than those reported with O 2 depending on the substrate and reaction conditions [15,16,18]. However, exceeding H 2 O 2 concentrations also lead to enzyme inactivation via autooxidation of, primarily, the Cu-coordinating histidine residues [15]. This makes a controlled supply of H 2 O 2 essential to maintain LPMO stability during catalysis [16,19].
In lignocellulose-degrading fungi, one known route for LPMO reduction involves the electron transfer protein cellobiose dehydrogenase (CDH; EC: 1.1.99. 18; CAZy ID: AA3_1) [6,11,20]. CDHs are widely distributed extracellular flavocytochromes comprising a flavin adenine dinucleotide (FAD)-containing dehydrogenase domain (DH) for substrate oxidation that transfers electrons to a mobile cytochrome b domain that can reduce the LPMO active site [9,11,20,21]. CDHs have a reported low oxidase activity [22][23][24][25], for example, the Crassicarpon hotsonii (syn. Myriococcum thermophilum) CDH (ChCDH) used in this study has a ca. 200 times lower turnover number for O 2 than for flavin-dependent 1,4-benzoquinone reduction [26]. Still, under physiological conditions, CDH might provide catalytically relevant amounts of H 2 O 2 for LPMOs, which have a µM affinity for this cosubstrate [16,27]. To scrutinize the significance of H 2 O 2 in CDH-driven LPMO activity, we enhanced the H 2 O 2 production rate of ChCDH by means of genetic engineering. Our results demonstrate that a single synthetic enzyme, hereafter termed CDH oxy+ , is able to fulfill a dual function as LPMO reductase and as a supplier of H 2 O 2 for the LPMO reaction. This allowed reducing the effective concentration of CDH during oxidative polysaccharide degradation by a factor of~30 relative to the wild-type ChCDH, while achieving similar LPMO activity. The results shed light on the role of H 2 O 2 in LPMO catalysis and the CDH-LPMO interplay.

Engineering of ChCDH for increased oxygen reactivity
We used a previously published variant of C. hotsonii CDH with a moderately (three times) increased H 2 O 2 production rate (N700S) [28] as a template for semirational engineering. Further site-saturation mutagenesis was performed on residue N748 since studies on other FAD-dependent oxidases and monooxygenases had shown that this position on top of the isoalloxazine C4a position, which is adjacent to the reactive N5, influences the oxygen reactivity of flavoenzymes [29]. The proposed reaction mechanism for the reduction of O 2 proceeds via an initial electron transfer from the flavin hydroquinone to O 2 forming a superoxide anion (O ÀÁ 2 ) followed by a spin inversion of the resulting caged radical pair. After spin inversion H 2 O 2 is formed after a second electron transfer from the flavin semiquinone to the oxygen intermediate, which can proceed via the formation of either a transient C4ahydroperoxy-flavin and subsequent decay to H 2 O 2 or an outer sphere electron transfer [30,31]. The closest amino acid in CDH to the isoalloxazine C4a position is N748, which is also a neighbor of the catalytic H701 and therefore a hotspot for mutagenesis by influencing either the access of O 2 to the FAD or the electron transfer steps (Fig. 1). A phylogenetic analysis of 56 cdh sequences showed that N748 is highly conserved in CDHs and only one sequence from a hitherto uncharacterized CDH contained a serine at the same position [32].
A site-saturation library of N748 was transformed in Pichia pastoris X-33 cells and was subjected to a high-throughput screening using the peroxidase-coupled ABTS (2,2'-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid)) assay for detection of H 2 O 2 production [28]. A second screening was based on the activity of CDH toward the artificial electron acceptor 2,6dichloroindophenol to normalize the expression level of the enzyme variants [28].
In active ChCDH variants, the substitution N748G showed the highest activity with O 2 . The resulting double variant N748G/N700S was produced in both its full-length form (CDH oxy+ ) and in a truncated form (DH oxy+ ), lacking the electron transferring cytochrome b domain [21] (Table 1). Steady-state kinetic experiments showed a 27-and 36-time higher specific activity of CDH oxy+ and DH oxy+ , respectively, over the wildtype enzymes (ChCDH and ChDH) when using the natural substrate cellobiose and O 2 as electron acceptor ( Table 1). The turnover of the non-natural substrate lactose, which was used in later parts of this study for analytical reasons, was similarly enhanced. Interestingly, DH oxy+ showed a higher turnover than CDH oxy+ , which indicates that the cytochrome domain can kinetically impair the two-electron reduction of O 2 by removing one electron from FADH 2 .

Activation of bacterial and fungal LPMOs by ChCDH
We initially tested the ability of CDH oxy+ to activate the bacterial C1-oxidizing chitin-active LPMO10A from Serratia marcescens (Fig. 2), which releases C1-oxidized chito-oligosaccharides as a product of the chitin oxidation reaction. Using lactose as a CDH substrate allowed a clear distinction between oxidation products generated by the LPMO (chito-oligosaccharides) and those generated by CDH (lactobionic acid) since the latter does not oxidize chito-oligosaccharides [33]. CDH oxy+ was more effective in driving the LPMO reaction when compared to the wild-type ChCDH. For example, after 1 h, we observed a 2.8-time higher LPMO product formation at a 12-time lower CDH oxy+ concentration ( Fig. 2A; compare bars for 3000 nM ChCDH vs. 250 nM CDH oxy+ ), which translates into CDH oxy+ being~35 times more efficient in driving the LPMO reaction. Comparison with product concentrations obtained after 24 h showed a~30 times higher efficiency ( Fig. 2B; 1500 nM ChCDH vs. 31.25 nM CDH oxy+ ) but suggests considerable enzyme inactivation in reactions with higher CDH oxy+ concentrations. The nonlinear increase in the amount of reaction products from the 1-to 24-h experiments and the lower optimal CDH : LPMO ratios found in the 24-h experiment point toward a fast inactivation of one or both biocatalysts. We, therefore, analyzed the time-dependent formation of oxidation products from both CDH (lactobionic acid) and LPMO (oxidized chito-oligosaccharides). Data presented in Fig. 2C show that a high concentration of CDH oxy+ caused fast inactivation of the LPMO, while CDH activity was less affected (Fig. 2D). At low CDH oxy+ : LPMO ratios the LPMO activity was maintained throughout the reaction. This suggests LPMO inactivation when the rate of H 2 O 2 production exceeded LPMO's capacity to convert its reactive cosubstrate in a productive manner, leading to H 2 O 2 -induced auto-oxidation reactions [15].
In all conversion experiments, the initial LPMO activity was linearly dependent on the H 2 O 2 formation rate by CDH (Fig. 3). Quantitative analyses of both soluble and insoluble oxidized sites indicated that LPMO Fig. 3. Oxidized sites generated during chitin degradation by SmLPMO10A. Panel (A) shows a time-independent plot of the degree of solubilization of oxidized products (i.e., the percentage of products obtained in the soluble fraction compared to the total number of oxidized sites) as a function of the quantity of detected solubilized oxidized products. The total number of oxidized sites was determined in reactions containing 10 gÁL À1 b-chitin and 1 µM SmLPMO10A with varying concentrations of CDH oxy+ (31.25 or 125 nM) and 15 mM lactose. Samples were taken at different stages of the reaction (15-180 min). The samples were diluted twice and heat-inactivated (100°C, 15 min). Half of the reaction was filtrated and 1 µM SmGH20 chitobiase was added for quantitation of solubilized products (i.e., the standard procedure for quantitation of solubilized oxidized sites). To the other half of the reaction, a cocktail of chitinases (containing 2 µM SmChi18A, 2 µM SmChi18C and 1 µM SmGH20) was added followed by incubation for 20 h at 37°C until the chitin was fully degraded. The resulting chitobionic acid concentration represents the total amount of oxidized sites. Rates derived from the initial phases of the progress curves in Fig. 2C,D only included time points where the solubilization was < 200 µM which means that the total number of oxidized sites is underestimated by~50%. Panel (B) shows the LPMO rate without and with correction for this underestimation. The corrected curve shows that 1 molecule of lactose converted by CDH oxy+ leads to approximately 0.8 molecules of chitobionic acid being produced by SmLPMO10A. catalyzed approximately 0.8 reactions per CDH-oxidized lactose molecule. The substoichiometric number in this experiment reflects CDH's dual role as a source of the cosubstrate and as an electron donor for LPMO. When using the fungal NcLPMO9C in combination with ChCDH or CDH oxy+ , product chromatograms were dominated by double (C1/C4) oxidized products ( Fig. 4) since CDH oxidizes the reducing end of the C4oxidized products formed by the LPMO. The complex product spectrum of this reaction and the lack of suitable standards did not allow for a thorough analysis of the reaction kinetics. However, when comparing overall product peaks it becomes obvious that compared to reactions with ChCDH an~24 time lower concentration of CDH oxy+ generated a similar product concentration (compare Fig. 4B vs. C). Note that in the reaction with high concentrations of CDH oxy+ (1.5 µM; Fig. 4B) NcLPMO9C became rapidly inactivated, which corroborates the findings with SmLPMO10A.

Electron transfer rates of ChCDH and CDH oxy+
We conducted rapid mixing experiments to probe CDH's internal electron transfer from the reduced FADH 2 to heme b. To this end, we mixed ChCDH and CDH oxy+ with an excess of cellobiose and monitored both the reduction of the FAD (449 nm) and the heme b (563 nm) cofactors. For both CDH oxy+ and ChCDH the intramolecular electron transfer rates were similar at cellobiose concentrations higher than 1 mM (0.35-0.48 s À1 ) (Fig. 5A). The lower electron transfer rates for CDH oxy+ at low cellobiose concentrations indicate competition for electrons between O 2 and the cytochrome domain at the DH domain (Fig. 5B).
The second, intermolecular electron transfer step from the reduced heme b to LPMO was probed by reacting prereduced CDH oxy+ with oxidized LPMO (Fig. 5C-E). Prereduction of CDH with cellobiose was performed under atmospheric conditions with or without catalase to investigate the effect of the by-product H 2 O 2 on the observed rate. The presence of H 2 O 2 had no effect on the LPMO reduction rate and in both cases, the apparent bimolecular rate constant for the CDH oxy+ /LPMO interaction was 1.27 9 10 6 M À1 Ás À1 , which is similar to the reported value for ChCDH/ LPMO (0.63 9 10 6 M À1 Ás À1 ; [33]).
These experiments show that both ChCDH and CDH oxy+ have similar electron transferring capabilities under the tested conditions. We, therefore, conclude  [41,42]. Note that the elution times vary slightly between different runs. Double (C1/C4) oxidized products (peaks eluting after~40 min) dominate since CDH oxidizes the reducing end of the C4-oxidized products formed by the LPMO. Note that C4-oxidized products are unstable, which explains why some peaks become smaller over time in panel (D). that the increased ability of CDH oxy+ to support LPMO reactions is not caused by changes in the ability to reduce the LPMO but to increase production of H 2 O 2 .

Activation of LPMO by ChDH and DH oxy+
To gain further insight into the CDH-LPMO interplay, we replaced the full-length enzymes in conversion experiments with DH oxy+ and ChDH at concentrations (31.25 and 750 nM, respectively) expected to produce the same amount of H 2 O 2 . Unexpectedly, in the absence of an additional reductant like ascorbic acid (AscA), both dehydrogenase domains were capable of driving the LPMO reaction to some extent, despite the lack of an electron transferring cytochrome domain (Fig. 6). The DH domain of ChCDH was previously shown to be incapable of direct electron transfer to LPMO [21], but activation of LPMO by some flavoenzymes without cytochrome domains under steady-state conditions has been reported [34]. Considering that all the investigated flavoenzymes bind FAD noncovalently and that in our experiments ChDH (at 750 nM) was much more effective than DH oxy+ (at 31.25 nM) at initiating LPMO activity (Fig. 6A), we considered FAD itself as a potential redox mediator between LPMO and DH. Thus, as a control, we carried out reactions with added FAD and compared it with reactions supplemented with the commonly used reductant AscA. Adding 0.25 mM FAD or 0.1 mM AscA to reactions containing 750 nM ChDH led to a slight increase in the amount of formed LPMO reaction products (Fig. 6B). A doubling of LPMO activity was observed upon adding FAD to reactions containing 31.25 nM DH oxy+ , whereas in this case, the addition of 0.1 mM   AscA led to a much higher, 55-fold, increase in the LPMO product concentration after 24 h. These observations show that FAD can act as a redox mediator between the DH domain and the LPMO. The strong effect of AscA on the reactions with DH oxy+ demonstrates that at low DH concentrations the concentration of dissociated FAD, and, thus, reduction of the LPMO becomes rate limiting. In this case, the addition of another reductant (AscA) alleviates the limitation and enables the LPMO to convert the available H 2 O 2 .
Titrations of FADH 2 with oxidized SmLPMO10A and NcLPMO9C (Fig. 7) confirmed that FADH 2 indeed reduces both enzymes with the expected 1 : 2 stoichiometry (1 electron per LPMO active site copper), underpinning the potential of FADH 2 /FAD redox cycling between DH and LPMO. Whether dissociated FADH 2 is of relevance as a natural redox mediator system is debatable, since the DH concentration in the experiments was higher than those usually observed in vivo.
We also considered that superoxide, generated via a one-electron reduction of O 2 at the cytochrome domain, could act as a potential reductant for LPMO. Such a reduction was previously shown to occur at high superoxide concentrations, albeit with low efficiency [15]. Using a fluorescence-based assay [35] we observed that under anaerobic conditions ChCDH reduced LPMO completely, while under aerobic conditions only 45% of the LPMO was reduced due to simultaneous reoxidation of CDH by O 2 (Fig. 6C). The fact that both ChDH and DH oxy+ reduced LPMO under anaerobic conditions (Fig. 6D,E) indicates that LPMO reduction by these truncated enzymes does not involve superoxide and is rather due to FADH 2 /FAD redox cycling, as discussed above.

Conclusions
Taken together, these experiments support the importance of H 2 O 2 -driven LPMO catalysis. While ChCDH and CDH oxy+ transfer electrons to the LPMO with similar efficiencies, they produce different amounts of H 2 O 2 , which translates into different LPMO activities. In accordance with the correlation between H 2 O 2 production by the CDH and LPMO activity, it has been noted earlier that the rate of ChCDH-driven LPMO activity is similar to the rate of H 2 O 2 production by ChCDH in reactions with O 2 as the only electron acceptor [33]. It has been suggested that H 2 O 2 -driven LPMO activity implies that, when once reduced, an LPMO could carry out multiple oxidative cleavages [15]. The apparent stoichiometry of the reactions that may be derived from the data displayed in Fig. 3 [15,36].
The cytochrome domain allows for specific intermolecular electron transfer from CDH to LPMO, but LPMO reduction can also be accomplished by a small molecule reductant like AscA or, interestingly, also by FADH 2 when the flavodehydrogenase is applied at high concentrations. The exact mechanism of LPMO activation is currently disputed [15,18]. In light of this debate, it is worth noting the AscA experiments depicted in Fig. 6B: the amount of oxidized products generated after 24 h in a reaction containing 0.1 mM AscA and 31.25 nM DH oxy+ was 72 times higher when compared to a reaction containing only 0.1 mM AscA. This difference supports the hypothesis that under commonly employed reaction conditions in LPMO experiments the production of H 2 O 2 is rate limiting.
Aside from providing novel insight into H 2 O 2 -driven LPMO catalysis and providing a newly engineered enzyme, CDH oxy+ , with potential applications in industrial biomass processing, the present study sheds light on a rarely investigated biological role of CDH. The very low O 2 turnover of CDHs might provide a slow but steady H 2 O 2 supply for LPMOs in a physiological setting, thereby avoiding oxidative damage of LPMO.

Generation of ChCDH and CDH oxy+
The codon-optimized expression plasmid pPIC-MtSopt encoding the full-length wild-type CDH from the thermophilic fungus C. hotsonii (syn. M. thermophilum; ChCDH) under the control of the methanol-inducible AOX promoter was used as a parental plasmid for the generation of CDH oxy+ . In a previous study, we already used this plasmid as a template for a semirational approach resulting in variant N700S with a moderately improved H 2 O 2 production rate [28]. In the present study, the plasmid pPIC-MtCDH N700S encoding the above-mentioned variant was used as the starting point for an additional round of protein engineering by site-saturation mutagenesis. A cdh library containing all possible codons at the targeted position N748 was prepared using the forward primer: TACCACTNNSCCAACTTCTTACATTGTCG and the reverse primer: AAGTTGGSNNAGTGGTAGGAA-CACC. A library of 368 clones was screened by a differential screening using the peroxidase-coupled ABTS assay for detection of H 2 O 2 production and the 2,6-dichloroindophenol assay to normalize the expression level of the enzyme variants [28]. In this screening, five variants showed an increased H 2 O 2 production compared to the parental enzyme N700S. The sequencing results showed that all of them carried the additional mutation N748G. With respect to its increased oxygen reactivity, the identified ChCDH variant N700S/N748G was named CDH oxy+ .

Generation of ChDH and DH oxy+
To obtain CDH and its oxygen-converting variant without the electron-transferring cytochrome domain, two truncated sequences (expression plasmids pPIC-wtDH and pPIC-DH oxy+ ) were synthesized by a commercial provider (ATG: biosynthetics, Merzhausen, Germany). In both plasmids, the sequence encoding for the native signal sequence (nucleotides 1-63) was directly followed by the dehydrogenase domain encoding sequence [nucleotides 750-2481 (F230-L806)]. The 687 nucleotides encoding for the cytochrome domain and the interdomain linker (N1-S229) were omitted in pPIC-wtDH. The expression plasmid pPIC-DH oxy+ additionally carried the two mutations N700S and N748G.

Heterologous production and purification
All four enzymes were recombinantly produced in P. pastoris X-33 cells and purified as previously reported [37]. Table 1 shows the activities of the CDH and DH variants which were measured with 2,6-dichloroindophenol as a chromogenic electron acceptor and 30 mM lactose as a substrate in 50 mM potassium phosphate buffer, pH 6.0. Data are expressed as mean values AE SD from three measurements. The calculated molar absorption coefficients at 280 nm given in Table 1 were used to calculate the protein concentrations and the specific activities.
The LPMO 10A from S. marcescens (SmLPMO10A, also known as CBP21) was recombinantly produced in Escherichia coli and purified by chitin affinity chromatography as previously described [38]. A threefold molar surplus of Cu(II) SO 4 was added to the pure protein to ensure that each LPMO molecule has a copper cofactor. Excess copper was removed by applying the LPMO-copper solution on a PD Midi-Trap G-25 column (GE Healthcare, Vienna, Austria) equilibrated with 50 mM potassium phosphate buffer, pH 6.0. The protein concentration was determined by measuring A280 using the calculated molar absorption coefficient of 35.20 mM À1 Ácm À1 .
The LPMO 9C from Neurospora crassa (NcLPMO9C, previously termed PMO-02916) was recombinantly produced in P. pastoris in a bioreactor and purified by a two-step chromatographic procedure as previously described [39]. The purified enzyme solution was repeatedly diafiltered against 50 mM potassium phosphate buffer, pH 6.0. The protein concentration was determined by measuring A280 using the calculated molar absorption coefficient of 46.91 mM À1 Ácm À1 .

Oxygen reactivity of CDH variants
The oxygen turnover of ChCDH and CDH variants was measured by incubating either 1.5 µM of the CDH and DH with low oxygen turnover (ChCDH and ChDH) or 30 nM of the CDH and DH variants with high oxygen turnover (CDH oxy+ and DH oxy+ ) with 5 mM cellobiose or lactose. Reactions had a total volume of 500 µL and were incubated at 40°C in 50 mM sodium phosphate buffer, pH 6.0, at a constant agitation of 800 r.p.m. Samples were withdrawn regularly, incubated at 95°C for 10 min to stop the reaction and measured by HPLC for aldonic acid formation.

Standard LPMO/CDH reactions
Standard reactions were set up with 1 µM SmLPMO10A and 10 gÁL À1 b-chitin, 15 mM lactose and the different CDH variants (i.e., ChCDH, CDH oxy+ , ChDH, or DH oxy+ ) with concentrations varying from 7.8 to 3000 nM. When using the truncated DH variants, lower concentrations were used as these variants produce more H 2 O 2 compared to the full-length enzymes. All reactions were carried out in 50 mM potassium phosphate buffer, pH 6.0, in a Thermomixer set to 30°C and 800 r.p.m. At regular intervals, samples were taken from the reactions, and the LPMO activity was stopped by directly separating the soluble fractions from the insoluble chitin particles by filtration using a 96-well filter plate (Millipore, Burlington, MA, USA) operated with a vacuum manifold. For samples where both aldonic acids from the LPMO and CDH (i.e., chitobionic acid and lactobionic acid) were subjected to quantitative analysis, the enzyme reaction was first stopped by 15 min heat inactivation at 100°C followed by filtration (0.22 µm). SmLPMO10A-generated chito-oligosaccharides were further degraded by incubation with 1 µM SmGH20 chitobiase as previously described [40] to produce N-acetyl-glucosamine and chitobionic acid, where the latter product represents the sum of all solubilized oxidized sites. To determine the total oxidized products, the LPMO reactions were stopped by heat inactivation (100°C, 15 min) followed by 20 h of incubation with 2 µM of the S. marcescens exo-chitinase A (SmChi18A) and 2 µM endo-chitinase (SmChi18C) as well as 1 µM SmGH20 chitobiase.

Quantitation of aldonic acids
Aldonic acids generated by CDH were quantified following two previously published HPLC methods [41,42]. Quantitation of GlcNAcGlcNAc1A (i.e., chitobionic acid) from reactions containing SmLPMO10A was accomplished using an RSLC system (Dionex, Sunnyvale, CA, USA) equipped with a 100 9 7.8-mm Rezex RFQ-Fast Acid H+ (8 %) (Phenomenex, Torrance, CA, USA) column operated at 85°C. Samples of 8 µL were injected into the column and solutes were eluted isocratically using 5 mM sulfuric acid as mobile phase with a flow rate of 1 mLÁmin À1 . The elution of products was monitored and recorded at 194 nm.
In-house made chitobionic acid solutions with known concentrations were used to generate a standard curve of oxidized products. Briefly, (GlcNAc) 2 was dissolved to a final concentration of 5.0 mM in 50 mM Tris/HCl pH 8.0 and incubated overnight with 0.12 mgÁmL À1 Fusarium graminearum chito-oligosaccharide oxidase [43] at 20°C as previously described [40] to generate chitobionic acid (i.e., GlcNAcGlcNAc1A).

Rapid kinetic studies of cellobiose dehydrogenase
Rapid kinetic experiments were performed with an SX20 stopped-flow spectrophotometer (Applied Photophysics, Leatherhead, UK) using a photomultiplier tube (PMT) detector. All experiments were performed in 50 mM sodium phosphate buffer, pH 6.0, at 30°C. Substrate and enzyme solutions were degassed prior to all measurements unless stated otherwise. The cellobiose-dependent reduction of the CDH cytochrome domain was measured in single mixing mode by following the absorbance at 563 nm (heme a-band). For the determination of the interdomain FAD-to-heme b electron transfer, CDH and cellobiose were used at final concentrations of 1.5 µM and 0.4-25 mM, respectively. Electron transfer from the reduced heme b to SmLPMO10A was studied in sequential mixing mode. Initially, CDH (20 µM) was mixed with 5 mM of cellobiose in an aging loop until the full reduction of the heme b cofactor was observed (320 s). The reduced enzyme was rapidly mixed with SmLPMO10A (15-60 µM) or buffer. These experiments were performed aerobically at a dissolved oxygen concentration of 250 µM. Approximately the same concentration of H 2 O 2 was accumulated during prereduction of CDH with cellobiose in the aging loop. The same experiments were therefore performed in the presence of 500 UÁmL À1 catalase from Corynebacterium glutamicum (Sigma Aldrich, St. Louis, MO, USA) in the aging loop to remove H 2 O 2 before reacting the enzyme with LPMO. The required aging time in these experiments was 360 s. Observed rates (k obs ) were determined by fitting the initial slope of at least three experimental repeats to an exponential function using the PRO DATA SX software (Applied Photophysics).
Titration of SmLPMO10A and NcLPMO9C with FADH 2 All experiments were performed in 50 mM sodium phosphate buffer, pH 6.0, in an anaerobic glove box (Whitley DG250; Don Whitley Scientific, Shipley, UK), which was continuously flushed with a nitrogen/hydrogen mixture (90/ 10). Residual oxygen was removed with a palladium catalyst and the generated water vapor was absorbed by silica gel. To probe the interaction of FADH 2 with LPMO, oxidized FAD (60 µM) was approximately 70% reduced with sodium dithionite to avoid an excess of reductant. FADH 2 was reoxidized with LPMO by adding aliquots of 3 µL Cu (II)-loaded NcLPMO9C from a 515 µM stock solution or 1.5 µL Cu(II)-loaded SmLPMO10A from a 906 µM stock solution to the cuvette. Spectra were recorded after mixing. Blank runs containing FADH 2 in buffer or LPMO in buffer were subtracted from these spectra. Electronic absorption spectra were recorded with an Agilent 8453 UV-visible spectrophotometer (Santa Clara, CA, USA) equipped with a photodiode array detector. All reagents and enzyme solutions used in the glove box were extensively degassed by applying alternate cycles of vacuum and nitrogen pressure. All experiments were carried out in quartz cuvettes (200 µL total volume) at a constant temperature of 25°C.
Monitoring the redox state of SmLPMO10A by fluorimetry All reactions were carried out in 50 mM potassium phosphate buffer, pH 6.0 at 25°C, prepared under aerobic or anaerobic conditions, and contained 2 µM SmLPMO10A mixed with either ChCDH (0.25 µM) or ChDH (0.25 µM) or DH oxy+ (31.25 nM). Enzyme mixtures were transferred to a sealable fluorescence quartz cuvette equipped with screw cap and septum (Hellma, M€ ullheim, Germany), which was inserted into a Cary Eclipse Fluorescence spectrophotometer (Agilent Technologies, Santa Clara, CA, USA). Excitation and emission wavelengths were set to 280 and 340 nm, respectively, and the PMT detector voltage set to 550 V, as previously described [35]. Reactions were initiated by addition of lactose (1 mM final concentration) using a Hamilton syringe. In a control reaction, carried out in anaerobic conditions, the lactose/CDH system was replaced by AscA (10 µM final concentration) to reduce the LPMO, providing a reference fluorescence signal corresponding to 100% reduction of SmLPMO10A. All fluorescence signals were expressed as the variation in fluorescence relative to the maximum increase observed in this control reaction.

Author contributions
DK planned, performed, and interpreted oxygen conversion, rapid kinetic and anaerobic titration experiments, and wrote the original draft of the manuscript; ZF planned, performed, and interpreted conversion and electron transfer experiments and contributed to the original draft; BB planned, performed, and interpreted oxidation studies and fluorescence experiments; SG supported conversion and oxidation studies; MP and CS generated, produced, and purified the CDH variants and characterized them; VGHE and RL equally planned the study, acquired funding and administrated the project, supported the evaluation and interpretation of experiments, and wrote the final version of the manuscript.