Electron transfer in polysaccharide monooxygenase catalysis

Significance

Polysaccharide monooxygenases (PMOs) are found in fungi and bacteria. PMOs act on polysaccharides such as cellulose and chitin leading to decomposition. They also function in fungal plant pathogenesis and cellular development. The reaction catalyzed by PMOs is complex. This activity requires delivery of electrons to a copper atom that is bound in the site where the chemical reaction takes place. Additionally, the enzymatic reaction requires either oxygen or hydrogen peroxide. In this work, we studied the path by which electrons are delivered to the copper site from the surface of the enzyme through studies involving specific amino acid mutations and enzyme activity. Our findings address chemical steps in catalysis and provide a framework to understand oxygen versus hydrogen peroxide reactivity.

Abstract

Polysaccharide monooxygenase (PMO) catalysis involves the chemically difficult hydroxylation of unactivated C–H bonds in carbohydrates. The reaction requires reducing equivalents and will utilize either oxygen or hydrogen peroxide as a cosubstrate. Two key mechanistic questions are addressed here: 1) How does the enzyme regulate the timely and tightly controlled electron delivery to the mononuclear copper active site, especially when bound substrate occludes the active site? and 2) How does this electron delivery differ when utilizing oxygen or hydrogen peroxide as a cosubstrate? Using a computational approach, potential paths of electron transfer (ET) to the active site copper ion were identified in a representative AA9 family PMO from Myceliophthora thermophila (MtPMO9E). When Y62, a buried residue 12 Å from the active site, is mutated to F, lower activity is observed with O₂. However, a WT-level activity is observed with H₂O₂ as a cosubstrate indicating an important role in ET for O₂ activation. To better understand the structural effects of mutations to Y62 and axial copper ligand Y168, crystal structures were solved of the wild type MtPMO9E and the variants Y62W, Y62F, and Y168F. A bioinformatic analysis revealed that position 62 is conserved as either Y or W in the AA9 family. The MtPMO9E Y62W variant has restored activity with O₂. Overall, the use of redox-active residues to supply electrons for the reaction with O₂ appears to be widespread in the AA9 family. Furthermore, the results provide a molecular framework to understand catalysis with O₂ versus H₂O₂.

Polysaccharide monooxygenases (PMOs), also known as lytic polysaccharide monooxygenases (LPMOs), are copper-dependent enzymes that hydroxylate C–H bonds in polysaccharides leading to the cleavage of glycosidic bonds. PMOs have a mononuclear active site copper and utilize either oxygen or hydrogen peroxide as a cosubstrate (1). PMOs depolymerize recalcitrant polysaccharides such as cellulose, and so their application to biofuel production has attracted considerable interest (2). Recently, more diverse roles have been shown for PMOs, from fungal virulence factors (3) and fungal allorecognition (4) to roles in bacterial pathogenesis (5). The active site of PMOs is composed of a strictly conserved site termed the “histidine brace,” which coordinates Cu through the N-terminal amine and imidazole nitrogen of H1 and the imidazole nitrogen of a second histidine about 80 residues away (6, 7). The active site contains several different secondary sphere residues conserved within different PMO families (8, 9).

A critical difference between utilizing O₂ and H₂O₂ is the number of reducing equivalents required in the respective catalytic cycles. Use of H₂O₂ necessitates only a single electron to reduce the PMO from the inactive Cu(II) form to the active Cu(I) form and no additional electrons per catalytic cycle. The use of O₂ as a cosubstrate also begins with a single electron for the initial Cu reduction but then has two proposed paths, both of which require additional electrons each cycle. The timely delivery of the second electron for the O₂ reaction has been termed the “second electron conundrum” (10). PMOs accept reducing equivalents from a variety of reductants. Cellobiose dehydrogenase (CDH), a physiologically relevant redox partner of PMOs, couples the two-electron oxidation of cellobiose to the reductive activation of PMOs (6). A variety of small molecule reductants including, ascorbate, phenols, cysteine, quinones, and lignin, can also serve as reductants for PMOs in vitro (11). Identification of which cosubstrate is the natural or physiological substrate has been a subject of debate (12, 13). The broad PMO family occupies diverse environments, and studies are increasingly pointing toward diverse functions, so it is likely that both O₂ and H₂O₂ could be used physiologically.

Several mechanisms have been proposed for copper monooxygenases. A common feature is the reduction of the active site from Cu(II) to Cu(I); herein, this step will be referred to as the “priming” electron (Scheme 1, step 1). Cu(I) then binds O₂ and converts to a Cu(II)-superoxide complex (Scheme 1, step 2). Direct H-atom abstraction (HAT) from the substrate by the Cu(II)-superoxide complex (Scheme 1, steps 1 to 4) is energetically disfavored, as calculations suggest that Cu(II)-superoxide is too weak of an oxidant to cleave carbohydrate C–H bonds (14, 15). Instead, a more potent oxidant is thought to be generated by the reduction of Cu-superoxide to form a Cu(II)-hydroperoxide complex. This electron is defined as the second electron (Scheme 1, step 5). The Cu(II)-hydroperoxide is then reduced again, followed by subsequent O–O bond cleavage to form a Cu(II)-oxyl radical (Scheme 1, step 6); this reducing equivalent is referred to as the third electron. The potent Cu(II)-oxyl radical can abstract an H-atom from the substrate forming a substrate radical and a Cu(II)-hydroxide complex (Scheme 1, step 7). The Cu(II)-hydroxide then rebounds with the substrate radical to yield the product and Cu(I) (Scheme 1, step 8). Steps 5 and 6 are specific for O₂ as a cosubstrate.

Scheme 1.

Putative oxidative mechanisms of PMOs. Brackets represent enzyme bound species.

Scheme 1 also shows a mechanism using H₂O₂ as a cosubstrate that arrives at the same Cu(II)-oxyl radical via a different path. The Cu(I) center binds H₂O₂ to form a Cu(I)-hydroperoxide species (Scheme 1, step 9), which then proceeds through O–O bond homolysis, liberating water and forming the Cu(II)-oxyl (Scheme 1, step 10). The Cu(II)-oxyl then follows steps 7 and 8 as described above. Both mechanisms are expected to leave the enzyme in the reduced Cu(I) state when the oxidized product has been released. This has been experimentally shown with H₂O₂ as a cosubstrate (16).

PMOs are part of a larger grouping of copper-dependent monooxygenases that include peptidylglycine α-hydroxylating monooxygenase (PHM), dopamine β-hydroxylase (DBH), and particulate methane monooxygenase (pMMO), all of which hydroxylate relatively high strength C–H bonds. PMOs are unique in this grouping with a single mononuclear copper binding site. DBH and PHM both contain two catalytically important copper ions, termed Cu_H and Cu_M in DBH and Cu_A and Cu_B, in PHM (17). In DBH, biophysical studies have led to the conclusion that Cu_M is the active site Cu where O₂ activation takes place, and Cu_H is involved in electron transfer to the Cu_M site (18). A very similar mechanism is generally accepted for PHM where Cu_B activates O₂ and Cu_A stores and transfers electrons to Cu_B (19). The consensus is that pMMOs also require multiple Cu atoms, though their exact roles are less well established (20, 21). Without a second Cu in PMOs, the question stands: How are additional reducing equivalents required for O₂ activation (Scheme 1, steps 5 and 6) delivered to the active site during the catalytic cycle?

In both the O₂- and H₂O₂-dependent mechanisms, the first or “priming” electron can be delivered directly to the Cu active site before substrate binding (22). Substrate affinity increases with copper reduction (23). Once bound, substrate would likely block the active site from inner-sphere delivery of additional electrons (24). Electron transfer (ET) via Tyr and Trp residues is a viable option for the delivery of additional electrons. ET during the catalytic cycle through a network of conserved Tyr and Trp residues has previously been suggested for PMOs, but not experimentally tested (25, 26). These Tyr and Trp residues have also previously been studied in the context of protecting the protein from oxidative damage via “hole-hopping,” i.e., oxidation of the residue moving a hole from the active site to the surface of the enzyme (27–30). Here, we suggest that these chains are also or perhaps predominantly functional in ET for electron delivery during the catalytic cycle. Recent work has observed that mutation of Tyr and Trp residues in an AA10 PMO, which are generally found in bacteria in contrast to the fungal AA9 PMOs, decreased the rate of peroxidase and peroxygenase activity (31). These results are described in the context of pathways that protect from oxidative damage. While this protective role may be important under high H₂O₂ or low substrate concentrations, in this work, we present evidence for a role of similarly redox-active residues in the delivery of electrons to the active site during catalysis in an AA9 PMO.

A computational approach was used to identify candidate residues that could be involved in ET. Y to F or W to F variants of the PMO from Myceliophthora thermophila, (alternatively known as Thermothelomyces thermophilus) MtPMO9E were then expressed, and activity with either O₂ or H₂O₂ as a cosubstrate was determined. Y62 was identified as a key residue for ET. Y62F and Y168F variants were found to have reduced activity relative to WT with O₂ as a cosubstrate but equally active compared to WT with H₂O₂, which has important ramifications on proposed catalytic mechanisms. Crystal structures of WT and MtPMO9E variants reveal only subtle structural perturbations. Furthermore, position 62 is conserved as either Y or W in the AA9 family, suggesting a general role at this position for electron transfer. Together, the twofold reduction in activity and the degree of conservation suggest that the residue at position 62 is involved in electron transfer, though not absolutely required. Overall, these results bring unique insight into the role of redox-active residues in ET in AA9 PMOs and clarity to the chemical steps involved in catalysis.

Results

Identification of Electron Transfer Pathways.

The computational approach employed identified potential ET residues in MtPMO9E. MtPMO9E is active on soluble substrates and thus is more amenable to biochemical studies. A structure was generated using AlphaFold2 (32) and AlphaFill (33) (for Cu placement). This structure was then used in tandem with the computation tool EHPath, which calculates ET pathways from an electron donor to an electron acceptor. The program then ranks each of the ET pathways based on their relative rates of electron transfer (34). Literature values were used for the reduction potentials for Y, W, C, and M, as is standard for EHPath. Reduction potentials of 1.0 V and 1.5 V were both used for the Cu in separate calculations. The 1.0 V potential represents the reduction of unprotonated superoxide radical anion to peroxide analogous to step 5 Scheme 1 (35) and so we first focused on this calculation. 1.5 V was additionally chosen because when protonated, the redox potential of superoxide is 1.42 to 1.71 V (36). Based on the redox couple, it is unlikely that a Y or W residue will reduce Cu(II) to Cu(I). We propose that once O₂ binds to the reduced Cu, the oxidation potential increases such that oxidation of nearby redox-active residues becomes possible.

The most probable path is the most energetically coupled pathway and thus calculated to be the fastest. Herein, we describe the most probable or fastest pathways as the most coupled. For each donor-acceptor pair, the five most coupled pathways at 1.0 V were calculated and then ranked by rate. The active site Cu was chosen as the acceptor in all cases. Donors were selected by first identifying the surface-exposed Cys, Tyr, and Trp residues in MtPMO9E using the AlphaFold2 predicted structure. Eight surface residues were identified: W133, W61, W135, W111, C141, C229, Y194, and Y219. Each surface donor is expected to be accessible to either small molecule reductants or CDH, even when the substrate occupies the active site. The rates of the most coupled pathways between the acceptor Cu and each of the 8 surface donors were used to rank the pathway from each donor (SI Appendix, Table S1). The three surface donors with the most coupled pathways are listed, each with two potential pathways for ET (Fig. 1B). The paths overlaid on the structure show the three pathways of ET with the rates of the most coupled pathways (Fig. 1A). The rates of the most coupled pathways all involve W61 as the surface electron donor. EHPath calculated 14 ms for an electron to transfer from the donor W61 to the copper active site acceptor following the path: W61 > Y62 > Y168 > Cu. The second most coupled predicted ET (15 ms) connects to the same pathway but originates at Y135. The surface donors with the most coupled ET times are found on one side of the protein. These ET paths could facilitate catalytic steps 4, 5, or 6 in Scheme 1.

Fig. 1.

(A) Structure of MtPMO9E generated by AlphaFold2 (33) with Cu introduced using AlphaFill (34) and putative pathways of electron transfer with labeled residues. The fastest electron transfer pathway from each electron donor is labeled with arrows as shown in panel B. (B) Electron transfer pathway table. Electron transfer pathways and their exact mean residence time (CEMR) were calculated using EHPath (35). All surface-exposed Cys, Tyr, and Trp residues were tested as electron transfer donors and the three with the shortest CEMR were selected. Cu was chosen as the electron acceptor with a reduction potential of 1.0 V.

When a redox potential of 1.0 V is used, all of the pathways utilize residues Y62 and Y168. (SI Appendix, Table S1). As noted above, a redox potential of 1.5 V was also tested because of the range of redox potentials that may exist across the copper-oxygen reduction landscape. Interestingly when 1.5 V is used, several of the pathways do not involve residue Y168 and one pathway does not involve residue Y62 (SI Appendix, Table S1). The relevance of these computationally determined pathways was next directly tested by expression, purification, and characterization of specific variants in the predicted electron transfer pathways.

Characterization of MtPMO9E Variant Activity.

A Tyr or Trp to Phe mutation would be expected to minimally perturb enzyme structure while rendering the variant residue incapable of ET. A Tyr to Trp variant, on the other hand, should still possess redox capability. Y168 and Y62 were selected based on EHPath results. A sequence alignment was generated using the AA9 family (see below), and residues W107 and W122 were selected for use as controls. These residues are well conserved but do not appear in any of the calculated ET pathways. W122 is 14 Å from the active site, close enough to participate in electron transfer, while W107 is relatively distant at 23 Å from the active site Cu. MtPMO9E and the four variants Y62F, Y168F, W107F, and W122F were expressed in Pichia pastoris. The proteins were initially screened using a cellulose degradation assay. Endpoint assays using O₂ and phosphoric acid swollen cellulose (PASC) as the substrates showed comparable activity levels for WT, W107F, and W122F. Y168F had diminished levels of activity based on the decreased intensity of the sharp peaks between 14 and 17 min that correspond to oxidized products of the PMO reaction. These sharp features are also entirely absent in Y62F (SI Appendix, Fig. S1). Due to the lack of C4 oxidized standards, these features could not be quantified in this experiment. As a result, Y62F and Y168F variants were selected for further characterization in assays using a quantifiable soluble substrate.

Glc6 Cleavage Assay (O₂ as the Cosubstrate).

As noted above, PMOs can use either O₂ or H₂O₂ as cosubstrates, although essential details of the reaction differ (12, 13). To specifically study PMO activity with O₂ as a cosubstrate, assays were designed to minimize interference from H₂O₂ that could be formed by the reduction of O₂. At high substrate concentrations, the reduction of O₂ is almost entirely coupled to substrate hydroxylation. However, at low substrate concentrations, some H₂O₂ will be formed from the uncoupled reduction of O₂ (23). Free copper with the added reducing equivalents can also form H₂O₂ from O₂ in the reaction solution. Therefore, when testing O₂ as a cosubstrate, substoichiometric copper was used to minimize free copper, and high substrate concentrations were used to minimize the uncoupled reaction. Any H₂O₂ that might still be formed was scavenged by horseradish peroxidase (HRP) and Amplex Red supplemented into the reaction. Assays were carried out to determine the required concentration of HRP to scavenge any free H₂O₂ (15 µM) (SI Appendix, Fig. S2). Conversely, PMO activity with H₂O₂ as a cosubstrate can be studied under anaerobic conditions with the addition of known amounts of H₂O₂.

The formation of the soluble substrate cellotetraose (Glc4) was used to quantify the PMO-catalyzed oxidative cleavage of cellohexaose (Glc6). WT-MtPMO9E formed Glc4 at a rate of 0.37 ± 0.03 μM min⁻¹ (Fig. 2A). Y168 is referred to herein as the axial tyrosine due to proximity to the axial coordination site of the Cu center. When the axial tyrosine (Y168) is mutated to Phe, the rate decreases to 0.18 ± 0.01 μM min⁻¹ (Fig. 2A). Mutation of Y62 to Phe leads to a decrease in activity (rate = 0.22 ± 0.03 μM min⁻¹) (Fig. 2A). The decrease in activity with both Y62F and Y168F suggests their involvement in the rate-limiting step. EHPath results suggested that both Y62 and Y168 participate in electron transfer to the active site. Since Y62 is ~12 Å from the active site, it is unlikely to directly participate in active site chemistry and instead only participate in ET. The exact contribution of Y168 is more difficult to assess since this residue is in the Cu primary coordination sphere.

Fig. 2.

(A) Cellohexaose assay with oxygen as the cosubstrate: Assays contained 10 μM PMO, 9 μM CuSO₄, 1 mM cellohexaose (Glc6), atmospheric O₂, 1 mM cysteine (reductant), 15 μM HRP, and 1 mM Amplex Red. Reactions of 50 μL were carried out in 50 mM MOPS and 50 mM MES buffer (pH 6.0) at 40 °C. Aliquots of the reaction mixture (2.5 μL) were quenched into 97.5 μL of 0.2 M NaOH at 5, 15, 30, 45, and 60 min. Cellotetraose (a product of glycosidic bond cleavage) was quantified as a measure of carbohydrate hydroxylation. (B) Horseradish peroxidase (HRP)-coupled assay: Assays contained 4 μM PMO, 3.5 μM CuSO₄, 1 mM cysteine (reductant), atmospheric O₂, 1 μM HRP, and 100 μM Amplex Red. HRP-coupled assays utilized Amplex Red as a substrate for the detection of H₂O₂ over time. (C) Cellohexaose assay with H₂O₂ as the cosubstrate: Assays contained 4 μM PMO, 3.5 μM CuSO₄, 1 mM cellohexaose, 100 μM H₂O₂, and 10 μM cysteine (reductant). Reactions of 25 μL were carried out in 50 mM MOPS and 50 mM MES buffer (pH 6.0) at 40 °C. The reaction was quenched into 0.2 M NaOH as described above. The reaction was quantified by the summation of cellotetraose and cellotriose (products of glycosidic bond cleavage). The absence of PMO or H₂O₂ was used as a control in the peroxide reactions. Error bars represent the SD.

To further probe the role of Y62, two variants were expressed. First, Y62 was replaced with Trp (Y62W), which has a similar reduction potential as Tyr. Trp is highly conserved at position 62 in the AA9 family (see below). The Y62W variant catalyzed cleavage of Glc6 to Glc4 at a rate of 0.38 ± 0.01 μM min⁻¹ (Fig. 2A), comparable to that of WT-MtPMO9E. Met is also conserved at this position (see below), so a Y62M variant was also studied. Although less thoroughly investigated, Met has been reported to participate in ET reactions despite a higher potential compared to Tyr. The Y62 M variant had a rate of 0.40 ± 0.06 µM min⁻¹ in the Glc6 cleavage reaction, also comparable to WT. The observation that Trp, Tyr, and Met at position 62 result in an enzyme that can utilize O₂ as a cosubstrate suggests that the ET properties of the residue at position 62 are important in enhancing the rate of O₂-dependent catalysis.

H₂O₂ Formation Assay.

In the absence of a carbohydrate substrate, PMOs will reduce O₂ to H₂O₂. A reaction in the absence of substrate allows for observation of the first ET steps in isolation (Scheme 1, steps 1, 2, and 5). The formation of H₂O₂ was determined using the H₂O₂-dependent HRP-catalyzed oxidation of Amplex Red to resorufin (λ_max = 572 nm), which can be monitored by colorimetric readout (SI Appendix, Fig. S3). WT- and Y62W-MtPMO9E formed H₂O₂ at a high and identical rate of 0.51 ± 0.01 μM min⁻¹. A slight decrease from the WT rate was observed with the Y62M variant to 0.45 ± 0.01 μM min⁻¹. A notable decrease in H₂O₂ formation was observed with the Y62F and Y168F variants with rates of 0.23 ± 0.01 μM min⁻¹ and 0.28 ± 0.01 μM min⁻¹_, respectively. A decrease in the ability of Y62F and Y168F to use O₂ as a cosubstrate in the Glc6 assay and a decreased rate of H₂O₂ formation from O₂ suggest that Y62 and Y168F are involved in ET steps. The decrease in activity of Y168F is more substantial than Y62F, suggesting that Y168 may play additional roles in O₂ activation beyond ET. The similarity of the observed rate trends among the variants for the H₂O₂ formation and O₂-driven hydroxylation suggests that the delivery of the second or third electron is the rate-limiting step.

Glc6 Cleavage Assay (H₂O₂ as the Cosubstrate).

Using H₂O₂ as a cosubstrate negates the need for additional electrons beyond the priming electron (Scheme 1, steps 9 and 10). The rate of the H₂O₂-catalyzed reaction is several orders of magnitude faster than with O₂ (37) and also leads to rapid protein inactivation (12) so an endpoint assay was used.

The summation of Glc3 and Glc4 reveals that approximately 100 µM of cleavage products are formed by WT-MtPMO9E in the presence of 100 µM H₂O₂ WT-MtPMO9E, and all tested variants were similarly active for the hydroxylation reactions of Glc6 with H₂O₂ as cosubstrate. This finding stands in stark contrast to the lower activity of the Y62F variant with O₂ as the cosubstrate. This difference strengthens the hypothesis that Y62 plays a role in the second or third electron transfer steps. Indeed, all variants tested utilized H₂O₂ as a cosubstrate similarly. This result also suggests that all variants tested are well-folded proteins with active sites capable of binding Glc6.

Crystal Structures of Variant MtPMO9E Maintain Similar Side-Chain Geometry.

To better understand potential structural effects of mutations to Y62 and Y168, crystal structures were solved of WT-MtPMO9E and the variants. Both wild-type and variant crystals formed under similar conditions and in the same space group (C 2 2 2₁). Crystal structures were solved at 1.8 to 3 Å resolution. Crystal statistics are summarized in SI Appendix, Table S2. Each crystal was found to have two monomers in the asymmetric unit with similar overall structure (RMSD between chains = 0.163 to 0.305, SI Appendix, Table S3). The WT-MtPMO9E structure was found to very closely resemble the AlphaFold2-predicted (32) model initially used for electron transfer predictions (RMSD = 0.515). WT-MtPMO9E also has a similar fold to other previously crystallized PMOs, e.g., NcPMO PDB 5TKH (38), with an RMSD of conserved regions of 0.740. Like this structure, the catalytic site has clear copper density coordinated by His 1 and His 83 (Fig. 3A). Based on the histidine brace coordination geometry (SI Appendix, Table S4), the copper center is likely to be of mixed Cu(II) and Cu(I) valence (SI Appendix, Table S4). The mixed valence is likely the result of X-ray photoreduction as reported by Tandrup et al. (39). There is also density for a ligand equatorial to the Cu in both chains that is too large to be a water molecule (Fig. 3A). We putatively modeled O₂ into this site, but at this resolution, we cannot differentiate dioxygen from peroxide or hydroperoxide species. Further, based on imperfect map-model fit, it is likely that the crystal contains multiple ligand species coordinated at different distances and orientations (Fig. 3A and SI Appendix, Fig. S4A). This is common in PMO crystals and may also reflect partial photoreduction of the Cu center (39, 40).

Fig. 3.

Crystal structures of wild-type (cyan), Y168F (orange), Y62W (purple), and Y62F (green) variant Mt PMO9E. (A) Electron density of WT (Top) histidine brace is shown compared to Y168F (Bottom). An overlay of the aligned WT and Y168F structures is shown in SI Appendix, Fig. S5C. The side chains overlay closely. (B) Electron densities of WT Y62 (Top) is compared with Y62W (Middle) and Y62F (Bottom). Blue densities are 2F_O-F_C maps displayed at 2, 1.5, 1.5, and 1.3 σ for WT, Y168F, Y62W, and Y62F, respectively. Green densities are F_O-F_C maps displayed at 3 σ for WT, Y168F, and Y62F and 2.3 σ for Y62F. In all cases, chain A of the asymmetric unit is shown. Chain B is shown in SI Appendix, Fig. S5.

Crystal structures of Y168F, Y62W, and Y62F variants were solved to 2.2 Å, 2.0 Å, and 3.0 Å, respectively. For all variants, the histidine brace region was similar to that of the wild type, including a Cu center and a putative oxygen molecule coordinated equatorially (SI Appendix, Fig. S4A). Importantly, density in the variants is clearly resolved for comparison with WT-MtPMO9E. In each case, the plane of the aromatic variant side chain matches that of the wild type (Fig. 3). In the Y168F crystal structure, Phe 168 overlays very similarly with wild-type Tyr with the conspicuous absence of the hydroxyl-coordinating Cu (Fig. 3A). Without the hydroxyl group of Y168, the Cu center is shifted an average of 0.3 Å toward C4 of the phenyl ring. No other significant structural changes occur (SI Appendix, Table S4). From these comparisons, we conclude that the Y168F mutation does not significantly alter the spatial arrangement of the catalytic site.

The Y62W structure has clear density for Trp at position 62 (Fig. 3B). It overlays closely with wild-type Tyr despite the added bulk of tryptophan, retaining a similar planarity (SI Appendix, Fig. S4C). This is likely due to backbone interactions with residues G132 and E63, though the tyrosine hydroxyl can additionally hydrogen bond with the backbones of Y168 and V151 (SI Appendix, Fig. S4). The Y62F structure was solved to a lower resolution, but density for Phe is nonetheless visible in a similar conformation as wild-type Tyr. As with the Y168F structure, no other significant residue changes are observed near the catalytic site for either Y62 variant structures.

The crystal structures also enabled a more precise look at the distances of ET traversed between Y168, Y62, and Y61 in the most coupled electron transfer pathways as predicted by EHPath (SI Appendix, Table S5). In the wild-type structure, Y168 is 10.4 Å from Y62 and Y62 is 10.5 Å from Y61. If Y62 does not participate in the electron transfer (as was calculated to be possible albeit a less probable pathway at a higher redox potential of 1.5 V), the distance between Y168 and Y61 is 16.5 Å.

Bioinformatics of Position 62 in the AA9 Family.

The O₂ cosubstrate results with the Y62 variant in MtPMO9E inspired an investigation into the conservation of Y62 in the AA9 family. Using Pfam03443 as the input, the 7,007 sequence AA9 family list was generated using the EFI-EST website (41). A random selection of the 7,007 sequences was used to generate the representative set of 4,000 sequences (Fig. 4B). The least similar sequence in the representative set has a 38% sequence identity with MtPMO9E. For this analysis, position 62 in the AA9 family is defined as the residue that aligns with position 62 in MtPMO9E. A sequence alignment of the AA9 family reveals that 64% of the residues at position 62 could be redox-active (Tyr, Trp, or Met) (Fig. 4A).

Fig. 4.

Conservation of residues at position 62. (A) Sequence logo generated from a multiple sequence alignment of the 240 sequences in the cluster of AA9 enzymes containing MtPMO9E using Weblogo 3 (42). Y, W, and M are labeled blue, all other residues labeled black. (B) Sequence logo generated from a multiple sequence alignment of 4,000 sequences of AA9 family using the Weblogo 3 website. Y, W, and M are labeled blue, all others residues labeled black. (C) SSN of AA9 family, Pfam 03443 was used as an input for the EFI-EST (41) website to generate an SSN containing 7,007 sequences. Sequences with an alignment score of 80 or greater will cluster. The clusters of sequences were visualized using Cytoscape (43). Each cluster is numbered based on the relative number of sequences. Cluster 7 is the MtPMO9E-containing cluster. (D) Percent occurrence of residues at position 62 in a 270-sequence cluster containing MtPMO9E of AA9 enzymes.

An SSN of the AA9 family was generated to further investigate the conservation at position 62 as a function of clustering (SI Appendix, Fig. S6). Starting with the AA9 family and using the EFI-EST website, the clusters were generated based on a clustering cutoff of 80% sequence identity (41). Six out of the 9 largest clusters in the AA9 family have redox-active residues at position 62 in over 70% of all sequences (SI Appendix, Table S6). Further, a sequence alignment of the 240-sequence cluster containing MtPMO9E (Fig. 4C, cluster 7) showed 83% conservation of Trp and 11% conservation of Tyr at position 62. Thus, 94% of the enzymes in this cluster have redox-active residues at position 62. These results, taken together with the function of a redox-active residue at position 62 in MtPMO9E, suggest that the majority of AA9 enzymes undergo evolutionary pressure to maintain a redox-active residue at this position.

Interestingly, 7% of the residues in position 62 are Phe in the AA9 family and 6% are Phe in the MtPMO9E containing cluster 7 (Fig. 4D). Clusters 2 and 9 are 63% and 52% Phe at position 62, respectively. Based on the data reported here, sequences with Phe at position 62 are expected to be poor at using O₂ as a cosubstrate and thus could only hydroxylate polysaccharides with H₂O₂ as a cosubstrate. Perhaps these PMOs have a different biological role, such as those that participate in cell remodeling in N. crassa (4). Another possibility would be that these enzymes activate O₂ using another redox-active residue that is located in a different position, though this position would have distance constraints.

Only 11% of the cluster 8 residues have a Tyr, Phe, Met, or Trp at position 62; a significant deviation from the distribution observed in the other 8 clusters. It is possible, as noted above, that enzymes in this cluster use a different residue in place of Y62. Cluster 3 is intriguing in that the residue in position 62 is Trp (40%) or Met (39%); this cluster accounts for the highest amount of Met at position 62 in the AA9 family. Additionally, cluster 3 is subdivided into three dominant subclusters, with nearly all the sequences containing Met at position 62 in one subcluster. We have shown here that Met has the same activity as WT (Tyr) or variant Trp in position 62. Several experimental and computational studies show Met can directly participate in ET (44–46). However, the increased redox potential and the relative scarcity of literature examples of Met as a redox-active residue when compared to Tyr and Trp suggest a role that requires further investigation.

Discussion

The timely and tightly controlled delivery of electrons to transition metal active sites in oxygen-utilizing enzymes is essential to catalysis. In PMOs, delivery of the additional electrons beyond the priming electron is especially difficult to envision given that the active site is a unique mononuclear copper site, and after the one-electron reduction to Cu(I), substrate occupancy would block further inner-sphere delivery from the substrate-binding face of the enzyme. The results reported here bear directly on this important mechanistic question of electron delivery to the PMO active site.

With MtPMO9E as a representative example, we provide evidence that Y62 delivers the additional electrons required to activate O₂ but is not required when H₂O₂ is the cosubstrate. Thus, we hypothesize that Y62 can reduce Cu-superoxide to Cu-hydroperoxide (Scheme 2). The mechanistic pathway involving O₂ would lead to oxidation of Y62 to form a tyrosyl radical. This radical is reduced by an ET mechanism that generates a radical on the surface of the protein, which is reduced by a reductant like ascorbate, cysteine, or by CDH. The Cu-peroxide species can be reduced to form an active oxidant that hydroxylates the polysaccharide substrate. The Y62F variant can still utilize H₂O₂ as a cosubstrate, suggesting that Y62 is not required for steps 7, 8, 9, or 10 (Scheme 1).

Scheme 2.

Y62-dependent O₂ activation mechanism. The MtPMO9E mechanism is depicted in the absence of substrate for clarity. The Cu(II) active site is first reduced by the priming electron followed by O₂ binding to form the Cu(II)-superoxide complex. The Cu(II)-superoxide species then oxidizes Y62 to form the Cu(II)-hydroperoxide complex and a neutral radical at Y62. The radical at Y62 is then quenched via residues in the electron transfer pathway resulting in a radical at the enzyme surface that will readily be quenched by a small molecule reductant or CDH.

As noted, mutating Y62 does not lead to complete abolition of activity. While in some reports, complete inactivation of the enzyme is observed when the ET pathway is mutated, these cases may represent enzymes in which a singular electron pathway exists. Our computational results indicate that in MtPMO9E, there is ET redundancy so that while Y62 is utilized along the most coupled pathway, alternative pathways can still be used when Y62 is mutated to a non-ET residue. These experiments cannot deduce if step 5 or step 6 is the rate-determining step because Y62 and Y168 may play roles in either or both of these steps. Thus, Y62 and Y168 may either play a role in a preequilibrium step before the rate-determining step or the rate-determining step itself.

Our bioinformatic analysis suggests that the use of a redox-active residue at position 62 is generalized throughout the AA9 family since 64% of all AA9 family members have either Y, W, or M in position 62. The high level of conservation of Y, W, or M at position 62 relative to the background level suggests an evolutionary pressure to maintain a redox-active residue at this position. The lack of absolute conservation may indicate that alternative ET pathways are viable in some PMOs.

Restoration of activity of the Y62W variants is intriguing, as Tyr and Trp are both capable of ET though they have distinct PCET properties. The relative abundance of Trp at position 62 and His at position 60 in both the MtPMO9E cluster 7 and the AA9 family suggest a conserved and well-tuned mechanism of ET. Although the proton acceptor for Y62 is still unknown, a water molecule is a reasonable hypothesis. In our MtPMO9E crystal structures, residue 62 appears relatively solvent accessible, with a pocket of ordered water molecules directly adjacent.

As noted above, it is challenging to deconvolute the ET contributions of Y168 from the active site geometry and electronic structure contributions. However, a complete description of the role of Y168 is likely to be more complex than that of Y62. Y168 could, and likely does, provide a direct contribution to the overall electronics of the active site. Further spectroscopic and computational studies will help illuminate the role of Y168 in ET, proton transfer (8) reactive intermediate energetics (47) and/or protection under oxidative conditions (27).

Mutations at position 62 and position 168 showed negligible perturbations to the respective structures. Of note, mutation of position 62 to Phe or Trp had no major effect on the active site geometry, with each aromatic side chain observed in a similar conformation despite changes to steric bulk and polarity. These structures provide further evidence that the effects on activity of mutation at positions 62 and 168 can be ascribed to changes in electron transfer properties rather than any structural perturbations. Interestingly, the maintenance of the planar orientation at position 62 in each variant suggests that the local peptide environment plays a major role in preserving this orientation.

In previous studies, Trp and Tyr radicals have been detected and characterized when prereduced PMOs are treated with H₂O₂ in the absence of substrate. The observed tyrosyl radical was assigned to the axial tyrosine in the active site, which is conserved throughout the AA9 family (Y168 in MtPMO9E numbering) (27, 28). The observed tryptophanyl radical, W62 in NcLPMO9C and W79 in HjLPMO, is ~6 Å from the Cu active site and in the same position as Y62 in MtPMO9E (16, 28). These studies concluded that the above radicals form as part of an evolved protective pathway. These radicals are proposed to reduce reactive oxidative intermediates that form at the active site in the absence of substrate. In this protective pathway paradigm, Y168 and Y62 would become oxidized after quenching an oxidative species (analogous to Scheme 1, step 7). In the catalytic mechanism described in this work, we propose a direct role for Y168 and Y62 in catalysis through the reduction of Cu(II)-superoxide to form Cu(II)-peroxide.

The oxidation of a Tyr or Trp by either a Cu(II)-oxyl or a Cu(II)-superoxide are thermodynamically similar, and it is likely that both are possible depending on the concentrations of polysaccharide substrate, reductant, O₂, and H₂O₂. While we can design in vitro experiments that favor one event over the other, the use of either the catalytic or protective mechanisms will depend on the native, often variable ecological environment of the organism. However, a catalytic role of these residues offers a more complete understanding of PMOs.

In recent work, redox-active residues in AA10 PMOs were shown to contribute to the stability and durability of PMOs under oxidative conditions that favor peroxidase and peroxygenase activity (31). These results support the potential role of redox-active residues in protection against oxidative damage (29, 30). The results presented here support a paradigm in which functional electron transfer occurs during the catalytic cycle with O₂ as a cosubstrate and utilizes conserved redox-active amino acids. However, these two roles for these residues need not be exclusive, and the potential dual-use of aromatic residues in different PMOs or under varying activating conditions may lead to the observed strong conservation of the residues of interest. As PMO characterization continues, roles in biology beyond that of polysaccharide degradation will likely continue to be uncovered. The diversity of environs PMO producing organisms inhabit likely influences mechanisms of catalysis, including the use of O₂ or H₂O₂. In addition to adding key mechanistic detail to PMO catalysis, our findings provide a clear framework to understand reactivity with O₂ versus H₂O₂.

Materials and Methods

Expression and Purification of MtPMO9E.

A detailed description of the expression and purification protocol can be found in supplemental information. In brief, the full-length gene of WT-MtPMO9E and variants were cloned into a pPICZa vector and expressed with secretion signal sequence in Pichia Pastoris. The protein was purified without a tag from the Pichia pastoris secretome. MtPMO9E was then collected, concentrated, and desalted into 50 mM MOPS and 50 mM MES buffer (pH 6.0). Protein purity was verified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using precast stain-gel gets (Bio-Rad) (SI Appendix, Fig. S7), and protein identity was verified using whole protein mass spectrometry (SI Appendix, Fig. S8).

Electron Transfer Pathway Calculations Using EHPath.

A structure was generated using AlphaFold2, and AlphaFill was used to position Cu in the active site. Using this structure, a truncated file was generated of only Met, Tyr, Trp, Cys, and Cu coordinates; this file is termed the bridge. The Cu center is defined as the donor, and individual files were generated with W133, W61, W135, W111, C141, C229, Y194, and Y219 as acceptors. The Cu center was defined with E = 1.0 V or 1.5 V and λ = 1.0. EHPath was run selecting the Cu donor and one acceptor, “electron” transfer is selected and α = 1, the cutoff number was 5, and printed results were 5. All results are listed in SI Appendix, Table S1.

Bioinformatic Analysis of MtPMO9A and the AA9 Family.

A more detailed description of the bioinformatics analysis protocol can be found in supplemental information. In brief, Pfam 03443 was used as an input for the EFI-EST website to generate an SSN containing 7,007 sequences. An alignment score of 80 was used, and a sequence alignment was conducted on all sequences representing the 9 largest clusters (SI Appendix, Fig. S8) using Clustal Omega (48).

Aerobic Cellohexaose Assay.

Assays contained 10 μM PMO, 9 μM CuSO₄ (Sigma), 1 mM cellohexaose (Glc6) (Megazyme), atmospheric O₂, 1 mM cysteine (Fischer Scientific), 15 μM HRP (Sigma), and 1 mM Amplex Red (Cayman Chemical). The PMO, CuSO₄, and Glc6 were incubated at 40 °C for 10 min before adding HRP, Amplex Red, and cysteine. The addition of cysteine initiated the reactions. Reactions of 50 μL were carried out in 50 mM MOPS and 50 mM MES buffer (pH 6.0) at 40 °C. An aliquot of the reaction mixture (2.5 μL) was quenched in 97.5 μL of 0.2 M NaOH (Sigma) at 5, 15, 30, 45, and 60 min. The quenched samples were analyzed using HPAEC coupled with an electrochemical detector (Dionex). Traces were background subtracted using a water trace and analyzed within the MATLAB 2022b software suit. Cellotetraose was quantified using the integrated area of the corresponding peaks compared to the standard curve made with known concentrations of cellotetraose (Megazyme).

Anaerobic H₂O₂-Driven Cellohexaose Assay.

Anaerobic activity assays with H₂O₂ as a cosubstrate were performed at room temperature in an anaerobic glove box ([O₂] < 0.6 ppm, MBRAUN). The PMO, CuSO₄, and Glc6 were allowed to equilibrate in the glovebox at 4 °C overnight before use. Cysteine was brought in fresh and equilibrated for 1 h before use. A 1/50 dilution of H₂O₂ was prepared fresh from a 9.8 M stock (Fisher Scientific) and allowed to equilibrate for 1 h in the box before use. Assays contained 4 μM PMO, 3.5 μM CuSO₄, 1 mM cellohexaose, 100 μM H₂O₂, 10 μM cysteine. PMO, CuSO₄, Glc6, and cysteine were incubated at room temperature for 10 min before adding H₂O₂. Reactions of 25 μL were carried out in 50 mM MOPS and 50 mM MES buffer (pH 6.0) at room temperature. After 5 min, the reaction was quenched in 0.2 M NaOH inside the glovebox. The quenched samples were analyzed using HPAEC coupled with an electrochemical detector. Traces were analyzed analogously to the anaerobic cellohexaose assay above. The reaction was quantified by a summation of cellotetraose and cellotriose. In the absence of PMO or peroxide reactions, the absence of H₂O₂ was used as a control.

H₂O₂ Formation by the PMO in the Absence of Polysaccharide Substrate.

MtPMO9E (4 μM),1.3 μM HPR, 3.5 μM CuSO₄, 100 μM Amplex Red, and 1 mM cysteine were added to 50 mM MES and 50 mM MOPS at pH 6.0 in a 96-well plate. Resorufin formation was monitored at 572 nm at 15-s intervals for 30 min using a SpectraMax340 spectrophotometer (Molecular Devices). Negative controls omitted cysteine. All assays were performed in triplicate. The rate of H₂O₂ formation was determined from a linear regression of the linear region, typically corresponding to the 0 to 30 min range.

Crystallography of MtPMO9A.

Crystals of all variants were grown using vapor diffusion and appeared as thin plates in space group C 2 2 2₁. Data collection on crystals was performed at the Advanced Light Source (ALS) beamlines 5.0.1, 5.0.2, and 5.0.3. Further details on crystallization, data collection, data processing, and model refinement can be found in SI Appendix.

Other Methods.

Further methods on AlphaFold2 predictions, aerobic PASC assay, HPAEC-PAD analysis, and intact protein mass spectrometry can be found in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB) with accession codes 9BJQ (49), 9BJR (50), 9BJS (51), and 9BJT (52). All other data are included in the manuscript and/or SI Appendix.