Histidine oxidation in lytic polysaccharide monooxygenase

The lytic polysaccharide monooxygenases (LPMOs) comprise a super-family of copper enzymes that boost the depolymerisation of polysaccharides by oxidatively disrupting the glycosidic bonds connecting the sugar units. Industrial use of LPMOs for cellulose depolymerisation has already begun but is still far from reaching its full potential. One issue is that the LPMOs self-oxidise and thereby deactivate. The mechanism of this self-oxidation is unknown, but histidine residues coordinating to the copper atom are the most susceptible. An unusual methyl modification of the NE2 atom in one of the coordinating histidine residues has been proposed to have a protective role. Furthermore, substrate binding is also known to reduce oxidative damage. We here for the first time investigate the mechanism of histidine oxidation with combined quantum and molecular mechanical (QM/MM) calculations, with outset in intermediates previously shown to form from a reaction with peroxide and a reduced LPMO. We show that an intermediate with a [Cu–O]+ moiety is sufficiently potent to oxidise the nearest C–H bond on both histidine residues, but methylation of the NE2 atom of His-1 increases the reaction barrier of this reaction. The substrate further increases the activation barrier. We also investigate a [Cu–OH]2+ intermediate with a deprotonated tyrosine radical. This intermediate was previously proposed to have a protective role, and we also find it to have higher barriers than the corresponding a [Cu–O]+ intermediate. Graphical abstract

The boost in polysaccharide degradation stemming from LPMOs is due to an oxidative reaction on the polysaccharide substrate. Originally, the oxidation was believed to progress with O 2 as the co-substrate [1], but more recent studies have shown that H 2 O 2 leads to significantly faster reactions [15][16][17][18]. The nature of the true co-substrate in nature is still debated [19], but H 2 O 2 can be employed as co-substrate. In fact, for Lentinus similis LPMO (LsAA9), a recent study could not detect any O 2 activation in the presence of substrate, whereas H 2 O 2 led to fast scission of the glycoside bond [20].
The active site of the LPMOs consists of a copper ion ligated to three nitrogen donor atoms: one histidine ligand binds by the imidazole group, while the other histidine ligand is the N-terminal and binds both by the imidazole group and the amino-terminal -NH 2 group [3]. This has been called the histidine brace motif and it is shown in Fig. 1.
Industrial use of LPMOs for cellulose depolymerisation [22] has already begun, but we are still far from reaching the full potential. One issue is that the LPMOs self-oxidise and thereby deactivate. A significant step in the direction of better utilisation of LPMOs could be achieved by elucidation of the molecular mechanism for this deactivation. So far, most theoretical studies of the LPMO molecular mechanism have focused on either the substrate binding process [23][24][25], or the reaction with the bound substrate [26][27][28][29][30][31]. In the latter case, the theoretical investigations have shown that the reaction with the substrate requires a very strong oxidant and a copper(II)-oxyl species has been the preferred one. It has also been shown that this oxyl can be formed from H 2 O 2 [29][30][31]. Very recently, theoretical as well as experimental investigations have been carried out for the reaction of H 2 O 2 and LPMOs (with copper reduced) without substrate [32][33][34][35]. These investigations have attempted to trap and spectroscopically characterise potential intermediates relevant for substrate-or self-oxidation. Several such intermediates have been identified; one of these contains a [Cu-OH] 2+ moiety formed by extracting the hydrogen for the OH group of the coordinating tyrosine (Tyr-164 in Fig. 1), thus leaving a tyrosine radical.
Yet, there exist currently no investigations of the actual oxidative damage. It has been noted to occur on the histidine residues coordinating copper [17], but the oxidation products and mechanism are unknown. Here, we carry out the first investigations of the initial steps of the oxidation in LsAA9. We focus on the two coordinating histidine residues. Since the substrate is known to protect from oxidative damage [17], we investigate both substrate-bound and substrate-free LsAA9. Interestingly, an investigation by Petrovic et al. [36] showed that, in addition to the substrate, a methylation of the NE2 atom [3] of the N-terminal, coordinating histidine also protects from auto-oxidative inactivation (cf. Fig. 1). This methylation is seen in many fungal LPMOs, but its function has remained unknown, since non-methylated enzymes are also catalytically active [37]. While the findings in Ref. [36] may suggest a protective role, a mechanistic explanation of the results are lacking. We have therefore included both a NE2-methylated and a non-methylated LsAA9 in this investigation.
The QM calculations for triplet states were performed with an unrestricted formalism, whereas the singlet states were obtained both with restricted and unrestricted formalisms. In the latter case, the occupation of alpha and beta orbitals was defined to be equal, which corresponds to a broken-symmetry (BS) approach [49]. The molecular orbitals from a converged triplet state were employed as starting point. The nature of the electronic states was verified and described by examining Mulliken spin populations.

QM/MM calculations
The QM/MM calculations were performed with the ComQum software [50,51]. In this approach, the protein and solvent are split into two subsystems: System 1 (the QM region) was relaxed by QM methods. System 2 contained the remaining part of the protein and the solvent, and it was kept fixed at the original coordinates (equilibrated crystal structure) to avoid the risk that different calculations end up in different local minima.
In the QM calculations, system 1 was represented by a wavefunction, whereas all the other atoms were represented by an array of partial point charges, one for each atom, taken from the MM setup. Thereby, the polarisation of the QM system by the surroundings is included in a self-consistent manner (electrostatic embedding). When there is a bond between systems 1 and 2 (a junction), the hydrogen linkatom approach was employed: The QM system was capped with hydrogen atoms (hydrogen link atoms, HL), the positions of which are linearly related to the corresponding carbon atoms (carbon link atoms, CL) in the full system [50,52]. All atoms were included in the point-charge model, except the CL atoms [53].
The total QM/MM energy in ComQum was calculated as [50,51] where E HL QM1+ptch2 is the QM energy of the QM system truncated by HL atoms and embedded in the set of point charges modelling system 2 (but excluding the self-energy of the point charges). E HL MM1,q 1 =0 is the MM energy of the QM system, still truncated by HL atoms, but without any electrostatic interactions. Finally, E CL MM12,q 1 =0 is the classical energy of all atoms in the system with CL atoms and with the charges of the QM region set to zero (to avoid double-counting of the electrostatic interactions). Thus, ComQum employs a subtractive scheme with electrostatic embedding and van der Waals link-atom corrections [54]. No cutoff is used for any of the interactions in the three energy terms in Eq. (2). The geometry optimisations were performed at the TPSS/ def2-SV(P) level of theory with a convergence criterium of 10 -6 a.u. for energies and 10 -3 a.u. for the maximum norm of the Cartesian gradients. After convergence, single-point QM energy calculations with the point-charge model were performed with the def2-TZVPD basis set and either the TPSS or B3LYP methods. Owing to convergence problems, B3LYP calculations with the substrate had to be done with the def2-SV(P) basis set and they were extrapolated to the def2-TZVPD basis set by TPSS calculations: where E QM/MM (B3/SV) is the QM/MM energy obtained at the B3LYP/def2-SV(P) level, E HL QM1+ptch2 (TP/TZ) is the QM energy of system 1 in a point-charge surrounding (first term on the right-hand-side of Eq. 1) at the TPSS/def2-TZVPD level and E HL QM1+ptch2 (TP/SV) is the corresponding term with the def2-SV(P) basis set. In the main text, only the extrapolated B3LYP-D3/def2-TZVPD results are discussed.
In general, increasing the basis set from def2-SV(P) to def2-TZVPD changes the calculated energies by 10 kJ/ mol on (absolute) average, with a varying sign. Changing the functional to TPSS has a restricted effect on the triplet energies (8 kJ/mol mean absolute difference). However, TPSS systematically stabilises the singlet states, especially the closed-shell singlet, by 44 kJ/mol on average. TPSS sometimes gives a spurious broken-symmetry state, intermediate between the closed-shell and open-shell singlet, with a spin distribution similar to that of the open-shell singlet, but with 2-8 times lower magnitude. For some states, we were unable to find a proper open-shell singlet with TPSS. Therefore, we have based the discussion on the B3LYP energies and in some cases, we used geometries optimised with B3LYP/def2-SV(P).
The setup of the protein with and without the substrate has been described before and we refer to Ref. [35] for details. The total system was spherical and non-periodic with 24,186 or 24,243 atoms (without or with the substrate, respectively).
The QM system was slightly larger than in the previous study [35]. It consisted of the Cu ion, the oxy or OH ligand, the entire His-1 (coordinated to Cu by the amino-terminal group and the ND1 atom), all atoms in Thr-2, except the backbone O atom, the sidechain of His-78 (truncated at CA and coordinated to Cu by the NE2 atom), the sidechain of Tyr-164 (truncated at CA and coordinated to Cu by the OH atom), the sidechains of His-147, Glu-148 and Gln-162 (truncated at the CA atoms), as well as three water molecules. In the structure with the substrate, a disaccharide fragment was included in the QM system, whereas in the structure without the substrate, instead an additional water molecule coordinated to Cu. Structures with [CuOH] + were generated by moving the proton on Tyr-164 to the oxy group. The QM system is illustrated in Fig. 2.

Hydrogen-abstraction energies of an isolated histidine model
The two coordinating histidine residues in LPMOs display different bonding situations. To probe if this gives rise to different hydrogen-abstraction energies, we first calculated the QM energy for homolytic removal of each of the hydrogen atoms in the imidazole ring (HD2, HE1, HD1 and HE2) in three different models of the sidechain of an isolated histidine residue. The three models resemble different bonding situations in the LPMO active site (cf. Fig. 2 and the inset in Fig. 3): the HID model is protonated on the ND1 atom. This is a model of His-78 in LPMO, where NE2 binds to Cu. The HIE model is protonated on the NE2 atom, which would correspond to a non-methylated His-1 residue in LPMO (where ND1 binds to Cu). The third model (HIM) has a methyl group on NE2 (and ND1 is deprotonated). This corresponds to the methylated His-1 in LsAA9, where ND1 coordinates to Cu. The calculated hydrogen-abstraction energies are presented in Fig. 3. The effect of methylating the NE2 atom is rather small on the hydrogen-abstraction energy for the HD2 and HE1 atoms: for the HE1 atom, the hydrogen-abstraction energy increases by less than 1 kJ/mol; whereas for the HD2 atom, it increases by 12 kJ/mol for the HID model, while it decreases by 2 kJ/mol for the HIE model. The trends obtained with TPSS or a smaller basis sets are very similar (show in Table S1 in the Supporting Information). Notably, it is much easier to remove a hydrogen from the nitrogen atoms (414-418 kJ/mol) than from the carbon atoms (508-525 kJ/ mol). This may indicate that NE2 atom of the amino-terminal residue is methylated to avoid that the HE2 atom is abstracted. Still, this does not explain why the other His ligand is not methylated. Further, the hydrogen atoms bound to N in the imidazole rings are far from the copper, meaning that abstraction of this atom is an unlikely first step in the reaction. Therefore, we study the abstraction of HE1 in His-1 and His-78 as a potential first step in the next subsection (these protons were selected because they are nearest to the [CuO] + oxyl atom, cf. Fig. 2a, b).

Hydrogen abstraction from histidine in the protein
The oxidation of the two histidine residues binding to Cu in the enzyme (His-1 and His-78 as is shown in Fig. 1) was investigated employing QM/MM, as described in the Methods section. We focus on [CuO] + as the reactive species, since this moiety can be formed from reaction with H 2 O 2 , and H 2 O 2 is known to be responsible for oxidative damage [30,36]. We compare the results to the same reactions   2+ , we have studied the abstraction of the HE1 proton in His-1, as well as in His-78. We have also studied the abstraction of HE1 from His-1 without any methylation of NE2 of His-1. The calculations are performed for three spin states: the closed-shell singlet, the triplet and the open-shell singlet states. Since also the substrate is known to protect the enzyme against oxidative damage [15,19,55], we started from two different (previously optimised [35]) structures, one with and one without a bound trisaccharide substrate. Note that when the substrate is bound, the water ligand of the Cu ion trans to Tyr-164 is displaced (cf. Fig. 2a, b). Consequently, we have 3 × 2 × 2 × 3 = 36 separate reaction profiles that are described in the following. The results are collected in Table S2 and summarised in Table 1.
We start by a discussion of the electronic structures of the [CuO] + reactant state (R, shown in Fig. 2a, b). As discussed previously, based on calculations with smaller QM systems [26,30,31], the [CuO] + reactant can best be described as Cu 2+ (d 9 ) coupled to a Oradical. Thus, it has two unpaired electrons, which can either be ferromagnetically coupled to   (Fig. 4c), the spin on Cu in the triplet state is still 0.54-0.67 e, whereas that on O in the formed OH group has decreased to 0.16-0.47 e, and the reacting His group has acquired almost one electron (0.99-1.14 e for His-1, but 0.78-0.86 e on His-78, mainly on the C atom that lost its proton, ~ 0.7 e). In the transition state (TS1), the situation is  Fig. 2a). The substrate displaces the water ligand, leading to a five-coordinate Cu ion and slightly shorter bonds to all ligands (0.005-0.05 Å, but 0.15-0.16 Å for Tyr-164; Fig. 2b). In the intermediate, I, the Cu-O distance to the OH group has increased by 0.11-0.19 Å, reflecting its protonation ( Fig. 4c; but only 0.03 Å for two complexes for which I is an open-shell singlet that gives shorter distances to Cu). The other ligands show rather small changes. The transition state is intermediate.
In the reactant state (R), the O atom of [CuO] + forms a hydrogen bond to a water molecule (W1), which also receives a hydrogen bond from one of the HE2 atoms of Gln-162 (Fig. 2a). W1 also donates a hydrogen bond to another water molecule (W2), which is hydrogen bonded to one of the carboxylate atoms of Glu-148. For structures with the substrate, W1 also receives a hydrogen bond from HE2 of His-147 (Fig. 2b). The OE1 atom of Gln-162 receives a hydrogen bond from the OH group of Tyr-164. All these interactions are kept also in the reaction intermediate I.
We now turn to the reactant state with a [CuOH] 2+ moiety (Figs. 2c, 4d-f). For this state, the triplet is still most stable, but only 1-2 kJ/mol more stable than the open-shell singlet without the substrate (7-13 kJ/mol with the substrate). The closed-shell singlet is 39-43 kJ/mol less stable. The [CuOH] 2+ reactant state has a complicated electronic structure with ~ 0.62 e on Cu, only ~ 0.22 e on OH, but 0.78 (without) or 0.91 e (with substrate) on the deprotonated Tyr ligand in the triplet state. A similar spin distribution was observed also in a previous study (with a smaller QM region) and it fits experimental observations [35]. For the open-shell singlet, the spin populations are ~ 0.50 e on Cu, ~ 0.15 e on OH (aligned with Cu) and 0.81-0.88 e on Tyr (with the opposite sign). Thus, it is best described as Cu 2+ -OH --TyrO ⋅ , but with some radical character on OH -.
Interestingly, hydrogen abstraction from His-1 (both with and without the methyl group) failed for all systems with [CuOH] 2+ -a potential-energy scan of the abstraction of the HE1 atom to the Cu-bound OH group was uphill by 108-159 kJ/mol and releasing the H-O bond distance restraint led to that the hydrogen atom returned to His-1. However, with His-78, a transition state was found (TS1), as well as a stable intermediate (I) with the HE1 hydrogen atom abstracted from histidine to a water bound to Cu 2+ (illustrated in Fig. 4d-f without   This gives rather small changes in the Cu-N distances (Fig. 2). In the [CuOH] 2+ state, the proton on Tyr-164 has moved to the OH ligand, but it keeps the hydrogen bond to Gln-162 (Fig. 2c) Table S2 and summarised in Table 2.
The reaction starts from the intermediate (I) and goes via a transition state (TS2) to a product (P), which contains a hydroxylated histidine ring with OH on either His-78 or His-1: the C-O bond length is 1.31-1.35 Å in the product (P) compared to 2.27-3.35 Å in the starting intermediate (I)-structures of the intermediate (I), transition state (TS2) and product (P) are displayed in Fig. 5. The product is always most stable in the closed-shell singlet state, i.e. with the Cu ion in the + I oxidation state. When the ). The second column shows whether the model involves the substrate or not. The third column shows whether the hydrogen atom was abstracted from His-1 with or without the methyl group (1 M   proton goes to Tyr-164): a and d intermediates (I); b and e transition states (TS2) and c and f products (P). The figures also show selected distances in Å. All energies are in kJ/mol reaction is started from [CuO] + , the triplet state for the product is 245-362 kJ/mol less stable than the closed-shell singlet and the open-shell singlet state is not found. When the reaction is started from [CuOH] 2+ , the intermediate has a copper-bound water molecule, which cannot directly rebind to the histidine radical. Instead, one of the protons needs to move to another group, which complicates the reaction. The most stable product is the closed-shell singlet with a hydroxylated histidine and the proton moved back to Tyr-164 (i.e. the same state as for the reactions started from [CuO] + ; cf. Fig. 5f). However, other products can also be found, especially on the triplet surface, e.g. with C=O or CH(OH) groups on the histidine or with the proton on Gln-162, which is the closest proton acceptor. Yet, these products are always appreciably less stable, by 158-221 kJ/mol for the triplet structures and by 24-43 kJ/mol for the alternative protonation states.
In the product state (P) without the substrate, the hydroxyl group on His-1 forms a hydrogen bond to W1 (Fig. 5c). With the substrate, it instead receives a hydrogen bond from W1 and donates a hydrogen bond to W3. The hydroxyl group on His-78 donates a hydrogen bond to W1 (without the substrate; Fig. 5f) or to the substrate.
In all cases without the substrate, the Cu ion remains five-coordinated in the product state, but the bonds to the water molecule and to Tyr-164 are long, 2.32-2. 36  The transition state for the rebound reaction is always found on the triplet surface at a C-O distance of 1.6-1.9 Å (but 1.4 Å when Tyr is deprotonated; cf.

Discussion
In Fig. 6, we summarise the results for both reactions (hydrogen abstraction and rebound) for the lowest spin state for the six reactions started from [CuO] + and putting all reactants at the same energy level (with the reactant state, R, as the reference). Both with and without the substrate, the methyl group on His-1 increases the highest of the two reaction barriers by 10 kJ/mol (from 91 to 101 kJ/mol without and from 152 to 162 kJ/mol with the substrate; the rate-limiting step also changes from hydrogen abstraction to rebound. This shows that the methyl group in fact protects His-1 from selfoxidation. The difference may seem small, but it corresponds to an increase in the estimated turnover time (k cat -1 ) from 0.3 to 18 h (estimated with classical transition-state theory [56] and a pre-exponential constant of k B T/h = 6.2 × 10 12 /s). Moreover, the oxidation of unmethylated His-1 has a lower maximum barrier (91 kJ/mol) than that of His-78 (100 kJ/ mol), but the oxidation of methylated His-1 has essentially the same barrier (101 kJ/mol), explaining why there is no need to methylate also His-78. With the substrate, the maximum barrier for hydroxylation of His-78 is smaller than that The colours indicate whether the hydrogen atom was abstracted from His-1 with or without the methyl group (blue and red, respectively) or from His-78 (green). The energies for the hydrogen-abstraction from the C4 atom of the substrate are shown in grey. All energies are given in kJ/mol. Results in regular style were obtained in the triplet state, those in italics were obtained for the closed-shell singlet and those in bold face were obtained for the open-shell singlet for His-1 both with and without methylation, but all barriers are so large that no reaction is expected (k cat < 5 × 10 -12 /s). Likewise, the [CuOH] 2+ state leads to barriers that are too high for any reaction to occur, both with and without substrate (269-290 kJ/mol).
This fits well with the recent suggestion that the [CuOH] 2+ state may protect the active site from histidine oxidation, based on experimental and theoretical characterisation of a tyrosine radical [32][33][34][35]. Further, an intermediate with a characteristic signature of a histidine radical was very recently observed and characterised by HERFD-XAS and UV-Vis spectroscopy [57]. It was suggested that this intermediate decays to an intermediate with a tyrosine radical, which was found to be inactive with respect to substrate oxidation. These result fit well with the mechanism proposed here, in which the histidine radical intermediate I is formed from the reactive [CuO] + species. The formation of I is uphill, and I was indeed found to be short lived [57].
Finally, we also note that a lower barrier (57 kJ/mol) is calculated for the abstraction of the hydrogen at C4 from the substrate by [CuO] + (grey path in Fig. 6b; further results are given in the Supporting Information). This shows that the substrate is much more prone to be oxidised than the His ligands. We have not investigated whether a methylation of His-1 affects the reaction energy of the substrate C-H abstraction, but previous QM-cluster results and our calculations on the isolated histidine residues suggest that this is not the case [26].
We can, thus, conclude that a self-oxidation mechanism involving [CuO] + abstracting a hydrogen from His-1 is consistent with the observation that methylation of His-1 makes the deactivation less likely. It is known that histidine residues susceptible to oxidative damage in some other metalloenzymes are typically converted to 2-oxo histidine [58,59]. The product P contains 2-hydroxy-histidine and it is possible that it may further react to 2-oxo histidine. Our results for isolated histidine highlights that the most susceptible bonds in the imidazole ring are the N-H bonds. Thus, any further reaction to 2-oxo histidine may involve breaking of the NE2-H bond in His-1, and we are currently investigating this possibility.

Conclusions
We have studied oxidation of the histidine ligands in LPMO, investigating the hypothesis that methylation of the His-1 residue, observed in many families of LPMO, may protect against such oxidation. We have considered oxidation of both His-1 and His-78, with both [CuO] + and [CuOH] 2+ as the reactive state, as well as reactions for the enzyme both with and without a bound substrate. We make several interesting observations. • The reaction consists of an initial hydrogen abstraction of the HE1 proton, followed by rebound of the formed OH or OH 2 group. •  Our results point out two tentative reasons why His-1 is methylated. First, the simple model calculations indicate that the methylation avoids hydrogen abstraction of the HE2 atom (which is replaced by the methyl group); the bonddissociation energy of N-bound hydrogen atoms is ~ 100 kJ/ mol lower than that of C-bound hydrogen atoms. However, this atom is too far from the Cu-bound O group (4.7-5.0 Å) to allow a direct reaction, studied by the current methods. Second, methylation increases the maximum barrier for oxidation of His-1 by 10 kJ/mol, viz. to the same level as for His-78 and increasing the lifetime of the enzyme 60-fold. We believe that the second reason is more important and also explains why only His-1 and not His-78 is methylated. This nicely illustrates how QM calculations may explain the design of enzymes.
Funding Open access funding provided by Lund University.
Data availability statement Raw data from the calculations are available from the corresponding author upon request.

Conflict of interest The authors declare no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.