Engineering A‐type Dye‐Decolorizing Peroxidases by Modification of a Conserved Glutamate Residue

Dye‐decolorizing peroxidases (DyPs) are recently identified microbial enzymes that have been used in several Biotechnology applications from wastewater treatment to lignin valorization. However, their properties and mechanism of action still have many open questions. Their heme‐containing active site is buried by three conserved flexible loops with a putative role in modulating substrate access and enzyme catalysis. Here, we investigated the role of a conserved glutamate residue in stabilizing interactions in loop 2 of A‐type DyPs. First, we did site saturation mutagenesis of this residue, replacing it with all possible amino acids in bacterial DyPs from Bacillus subtilis (BsDyP) and from Kitasatospora aureofaciens (KaDyP1), the latter being characterized here for the first time. We screened the resulting libraries of variants for activity towards ABTS and identified variants with increased catalytic efficiency. The selected variants were purified and characterized for activity and stability. We furthermore used Molecular Dynamics simulations to rationalize the increased catalytic efficiency and found that the main reason is the electron channeling becoming easier from surface‐exposed tryptophans. Based on our findings, we also propose that this glutamate could work as a pH switch in the wild‐type enzyme, preventing intracellular damage.


Introduction
Lignin is an aromatic heteropolymer comprising 10-30 % of plant biomass.Despite being a promising renewable resource for aromatic chemicals, it is rarely utilized by the biotechnological industry due to its resistance to depolymerization.Yet, nature has developed advanced mechanisms for this process, and there are several known organisms, such as white and brown rot fungi and soil bacteria, that can degrade lignin efficiently. [1]The enzymes responsible for lignin depolymerization originating from fungi, such as various peroxidases or laccases, are well-characterized. [2]However, the mechanism of action of bacterial enzymes involved in lignin degradation and modification still needs further investigation.These enzymes include multicopper oxidases, manganese superoxide dismutase, glutathione-dependent β-etherases, and dye-decolorizing peroxidases (DyPs). [1]ye-decolorizing peroxidases are a family of heme peroxidase enzymes [3] originating from bacteria and fungi.DyPs have been suggested to play a role in lignin degradation, but they may also play roles in anthraquinone degradation and iron metabolism. [4]Nevertheless, they are mostly investigated for their potential applications in lignin degradation and utilization of lignin-derived products.DyPs have been found to oxidize Kraft lignin, lignin-derived and lignin-mimicking phenolic model compounds, and methoxylated aromatics such as veratrylalcohol. [5]Some DyPs like DyPB from Rhodococcus jostii RHA1 or DyP2 from Amycolatopsis sp.75iv2 even show a manganese peroxidase activity in addition to their peroxidase activity, and this can enhance their reactivity towards lignin derived phenolics. [6]However, not all DyPs that have been found to react with phenolics or simple lignin models can oxidize nonphenolic compounds of more complex lignin. [7]Still, the fact that they are able to react with various phenolic and nonphenolic compounds also makes DyPs also good candidates for bioremediation of wastewaters. [8]ased on their primary or tertiary structures, DyPs were classified into three main classes. [9]B or Primitive (P) type is the smallest in size.It occasionally has a gene encoding for an encapsulin directly downstream of the DyP gene in the genome of the organism, [4] giving it a possible mechanism for secretion, increased stability, and reactivity. [10,11]A or Intermediate (I) type DyPs have an intermediate size and often include a twinarginine translocation (TAT) signal peptide [12] on their Nterminal end for relocation into the periplasm or the outside of the cell. [13]Both A and B-type DyPs originate from Bacteria.In contrast, C/D or Advanced (V) type DyPs are the largest, have the highest catalytic efficiency and substrate scope among all DyPs, and can be of bacterial or fungal origin. [9]ll structurally characterized DyPs comprise two ferredoxinlike folds of the dimeric α + β barrel superfamily.They include a heme cofactor that is located towards the C-terminus of the protein. [14]These enzymes usually exist as dimers, [14] but there are examples of different oligomeric structures, such as tetrameric [15] or hexa-and dodecameric arrangements. [16]he iron in the heme group of resting state DyPs is in the ferric state and always coordinated by a proximal histidine, while the distal site is uncoordinated.Two catalytic residues, a conserved aspartate found in a GXXDG motif present in every DyP, and an arginine can be found at the distal site of the heme. [4]The aspartate is proposed to play a role in the deprotonation of hydrogen peroxide that leads to oxidizing the iron from a Fe 3 + in the resting state to a Fe(IV)-oxo-porphyrinradical cation, also known as Compound I. The aspartate has an essential role in many DyPs; [3,18,19] however, there are examples where this residue can be replaced without significant effects on catalysis. [20]The conserved arginine is proposed to stabilize the active site architecture [14] and is essential for peroxidase activity in some cases. [18,20]s the heme group is buried and inaccessible to reducing substrates with multiple aromatic rings -a typical DyP substrate -the oxidation of these compounds is proposed to happen via a long-range electron tunneling (LRET) mechanism, in which the electron is channeled from the substrate-bound on the surface to the oxidized heme group through surface exposed tyrosine or tryptophan residues.23][24][25] Almost all DyPs have an acidic pH optimum, typically between pH 4.0-5.0. [14]However, there are also examples of Atype DyPs with a pH optimum of 7.0 [26] or 8.0. [27]This property has been attributed mostly to the distal site aspartate, although evidence suggests this is not the only residue responsible for the low pH optimum. [17]The substrate specificity of DyPs is very variable.Even members of the same group are known to oxidize azo-or anthraquinone dyes, aromatic species, or lignin model compounds and, in some cases, Mn 2 + , [14] which also supports the hypothesis that they could play an important role in bacterial lignin utilization.
BsDyP is a previously well-characterized A-type Dye-Decolorizing peroxidase originating from the Gram-positive bacterium Bacillus subtilis.This enzyme is relatively thermostable and has a broad substrate scope, [28,29] making it ideal to engineer for biotechnological applications.[32] A recent study utilizing directed evolution, aiming to increase the catalytic efficiency of BsDyP towards aromatic substrates, has found that making a loop in the second coordination sphere of the heme flexible increased catalytic efficiency towards ABTS and 2,6-dimethoxyphenol (DMP). [33]This work also proposed that modifying these loops could have great engineering potential for DyPs.
In this study, we introduce and characterize a previously unknown DyP from the Gram-positive bacterium Kitasatospora aureofaciens, called KaDyP1.Furthermore, we target a glutamate residue in Loop 2 surrounding the active site that is conserved in A-type DyPs by site saturation mutagenesis followed by screening in both the previously characterized BsDyP and KaDyP1.We express and purify variants with increased activity and characterize their catalytic efficiency towards ABTS and their thermostability.Using Molecular Dynamics (MD) simulations, we furthermore aim to explain the effect of these mutations on the catalytic efficiency at the molecular level.

Identification of a DyP-Encoding Gene in the Genome of K. aureofaciens
We previously identified a pyranose oxidase-encoding gene from K. aureofaciens based on sequence similarities with fungal pyranose oxidase and expressed and characterized the encoded protein, which potentially acts as an Auxiliary Activity in lignocellulose degradation by providing hydrogen peroxide as well as reducing low-molecular weight lignin degradation products. [34]We subsequently set out to identify potential peroxidases in this organism to characterize additional enzymatic components in this bacterium that may be involved in lignin degradation.The initial BLAST search using AmyDyP2 from Amycolatopsis sp.75iv2 (UniProt K7N5M8) resulted in a sequence annotated as iron uptake transporter deferrochelatase/peroxidase efeB (GeneID 33984906, Locus tag B6264_ RS10925, UniProtKB A0A1E7N504, 423 amino acids).

A Conserved Glutamate Stabilizes Loop 2 that Surrounds the Active Site in DyPs
In experimentally determined 3D structures of DyPs, the active site containing the heme group is buried and covered by three conserved loops.Loop 2 connects two continuous halves of an alpha helix (Figure 1a), and it always starts with glycine and arginine; the amino acid ending the loop and starting the second part of the alpha helix is, in all cases, the proximal histidine coordinating the heme group.The size of loop 2 correlates with the enzyme size, with B-type DyPs having the shortest (16-20 amino acids), A-type DyPs having a medium length (23-39 amino acids), and C/D-type DyPs having the longest loop 2 (40-55 amino acids) (Table S1).The amino acid composition of loop 2 is variable, even among members of the same type of DyP (Figure 1b).However, in A-type DyPs, a glutamate residue appears in primary and tertiary structures (Figure 1b and c) at specific positions.This amino acidcorresponding to E309 or E312 in KaDyP1 and BsDyP, respectively -forms a hydrogen bonding interaction with a threonine (T235 or T242) in loop 1 with its carboxyl group, and with the arginine at the start of loop 2 (R293 or R299) with its backbone (Figure 1c).Also, the indicated threonine in loop 1 is separated only by a conserved glycine from the catalytic aspartate on the distal side of the heme (D233 or D240).The arginine at the start of loop 2 shows a repulsive interaction with the catalytic arginine (R347 or R339).Here, we investigate the effect of modifying this stabilizing interaction network by the variation of the conserved glutamate using two A-type DyPs: the previously well-characterized B. subtilis BsDyP, [33] and KaDyP1 from K. aureofaciens, described here for the first time (Figure 1c).7PKX [33] ), B-type KpDyP from Klebsiella pneumoniae (yellow, PDB: 6FKS) [19] and C/D-type DyP2 from Amycolatopsis sp.75iv2 (gray, PDB: 4G2 C [35] ).Heme groups are shown as sticks.(b) Multiple sequence alignment of Loop 2 of known A-type DyPs.][38][39][40][41][42] (c) The conserved glutamate in loop 2 makes a hydrogen bond with threonine in loop 1 in the homology model of KaDyP1 (blue) and the x-ray structure of BsDyP (salmon, PDB: 7PKX [33] ).

Replacing the Conserved Glutamate in Loop 2 Increases Activity in two A-Type DyPs
Based on the catalytic properties of BsDyP variants [33] and our findings in MD simulations, we assumed that destabilizing Loop 2 could modulate substrate accessibility and oxidation.As the conserved glutamate in the newly characterized KaDyP1 E309 and BsDyP E312 is involved in stabilizing interactions, we decided to target this residue, using saturation mutagenesis to find amino acid substitutions that result in altered catalytic properties while simultaneously contributing to improving our knowledge on the structure-function relationships of DyP enzymes.Site saturation mutagenesis is a random mutagenesis technique used in protein engineering in which a single codon or set of codons results in all possible amino acids at the position.Therefore, libraries of variants were constructed and screened for activity and stability using ABTS and hydrogen peroxide as substrates and compared to the values of the wildtype enzyme (Figure S1 and Figure 2).
After constructing the KaDyP1 E309 saturation library, ~150 variants were screened for ABTS activity (Figure S1a).To test the efficacy of the engineering approach, DNA from 72 variants was sent to sequencing, selected among those that showed higher (41), similar (9), and lower (22) activity than wild-type.The sequencing results revealed that 18 substitutions (including the wild-type, glutamate) were present in variants at position E309, out of the 20 possible; the only two amino acids not found at that position were tyrosine and tryptophan (Figure S1b).All variants with known genotypes were rescreened (six replicates each) by assessing not only their enzymatic activity but also their thermal stability (Figure 2).The results show that variants separated into two groups: those exhibiting comparable activity and stability as the wild-type and those showing (2.5 to 4-fold higher) increased activity to wild-type and slightly decreased stability, variants E309L, E309I, E309V, E309P, and E309Q.Except for glutamine, all variants of this group do not establish hydrogen bonds.Leucine, isoleucine, and valine show similar properties, being hydrophobic with branched side chains and incapable of making hydrogen bonds.Proline is also hydrophobic, lacks a backbone amide hydrogen, and is, therefore, even less capable of making polar interactions.Glutamine was a somewhat unexpected finding, highly similar to glutamate, capable of hydrogen bonding interactions but carrying no charge.
In the saturation library of E312 BsDyP, 84 variants were initially screened for activity (Figure S1c); the seven variants showing increased (2.5-fold) activity than wild-type were rescreened and sequenced.Two variants show a replacement of E312 phenylalanine and two for tryptophan; the remaining three showed a wild-type genotype (Figure S1d and Figure 2b).Variants E312W and E312F show 2 to 4-fold higher activity than the wild-type and comparable thermal stability.Therefore, we decided to use spectroscopic, biochemical, and kinetic approaches to characterize variants KaDyP1 E309L and E309Q, and BsDyP E312W and E312F variants.

Selected Variants with Replaced Glutamate Show Increased Catalytic Efficiency
We produced and purified for the first time the wild-type KaDyP1.The UV-visible spectrum shows the typical Soret band at 408 nm (Figure S2); the calculated Reinheitszahl (Rz) value (A 404/408 /A 280 ratio), which reflects the purity of hemoproteins is 2, indicating a good purity yield.The heme b content was estimated to be 0.6 mol per mole of protein, indicating that the enzyme preparation is partially heme-depleted (Table S2).The enzyme is most active in the acidic pH range of pH 4.0-5.0,typical for DyPs (Figure S3).The apparent kinetic parameters for ABTS and hydrogen peroxide were determined and are comparable to other characterized DyPs (Table 1, Figure S4).
The properties of purified variants KaDyP1 E309Q and E309L, the BsDyP E312W, and E312F variants were compared to the wild-type enzymes.In all cases, the UV-visible spectra fingerprints, the Reinheitszahl (Rz) values, and the heme content of the variants did not change significantly compared to the wild-type enzymes (Table S2, Figure S2), indicating the loop flexibility did not influence the heme incorporation and thus; presumably, it did not change the first coordination sphere of the heme group.Similar is true for the pH optimum (Figure S3), although, for the KaDyP1 E309Q and E309L, there is a slight shift of the optimal pH from 4.5 in the wild type to 4.9 (E309Q) and 5.2 (E309L).Both variants of BsDyP, E312W, and E312F show a similar optimal pH to the wild-type (pH 4.3).Interestingly, the selected variants showed increased catalytic efficiency for ABTS due to increased turnover numbers (up to 15-fold higher) and decreased K m ; for example, variant KaDyP1 E309Q and BsDyP E312F show a 2-and 25-fold higher catalytic efficiency for ABTS, respectively (Table 1 and Figure S4).Generally, the K m for hydrogen peroxide is higher for variants (2-10 fold) except for BsDyP E312F, which is comparable to wild-type, showing that the variants show a decreased affinity for hydrogen peroxide.Furthermore, the data obtained indicated that, in general, the enzyme variants show a 2-4-fold decreased K i for hydrogen peroxide (i.e., they are more sensitive to this oxidant) except for variant KaDyP1 E309L, which shows a 2-fold higher K i .
The thermodynamic stability of KaDyP1 and variants was studied by probing the tertiary structure (fluorescence intensity) at increasing temperatures.The unfolding induced by temperature is described by a two-state process where the folded and unfolded states seem to be the only states that accumulate at significant amounts (Figure S5).KaDyP1 is a moderately thermostable enzyme with a melting temperature (T m , where 50 % of the protein molecules were denatured) of 46 °C (Table 2).The kinetic thermal denaturation profiles indicate that the enzyme shows at T 50 at 37 °C, at which the enzyme activity was reduced by 50 % after incubation for 30 min (Figure S5).KaDyP1 E309Q and E309L variants show a minor decrease in thermodynamic and kinetic stability (Table 2, Figure S5).In the case of BsDyP E312W and E312F variants, a more significant drop in stability, with a 10 °C decrease in melting temperatures and T 50 values (Table 2, Figure S5), possibly related to the more notorious increase in the k cat values towards ABTS 5-15 fold (Table 1, Figure S4), resulting in a clear activity-stability trade-off.This behavior was previously observed in BsDyP variants evolved in  : k cat and K M estimated by Michaelis-Menten fits up to 1 mM a , 1.5 mM b or 0.6 mM c of substrate concentration.d : Ref. [33]  the laboratory using directed evolution approaches showing increased activity for lignin phenolic substrates. [33]

Molecular Dynamics Simulations of KaDyP1 and BsDyP Wild-Type Enzymes
To explain at the molecular level how the above mutations modify the activity of these enzymes, we have run 100 ns long Molecular Dynamics (MD) simulations on both wild-type enzymes.As the pH optimum of these enzymes (pH 4.5-5.0) is close to the side-chain pK a of the targeted glutamate (pK a 4.25 [43] ), we also included simulations of the wild-type enzymes where the sidechain of E309 in KaDyP1 or E312 in BsDyP is protonated.We have considered enzyme dimers and included two independently run simulations, resulting in four trajectories of monomers that we analyzed for each variant.In these simulations, the parts of the protein with defined secondary structural elements show a quite stable atompositional root-mean-square derivation (RMSD) for their backbone atoms, ranging from 0.04-0.34nm in KaDyP1 and 0.04-0.38nm in BsDyP wild type, throughout the simulations (Table S3, Figure S6).The unstructured loop regions have increased mobility based on the calculated atom-positional root-mean-square fluctuations (rmsf) (Figure 3a and c).The mobility of the loop regions surrounding the buried heme group causes it to become exposed to the solvent, which is evident by the strong fluctuation of the solvent-accessible surface area (SASA) of the heme group throughout the trajectory (Figure 3 b and d).
However, we noticed that the secondary structure of the two parts of the α-helix separated by loop 2 is relatively stable throughout the simulations based on an analysis of the secondary structure elements, [44] except residues 294-296 of BsDYP -in the first α-helical part right before loop 2 beginswhich show the reduced formation of the α-helix (Table S4).Another indicator of the stability of these helices is the distance of the two hydrogen bonds keeping the two α-helices together.These are the distances of the carbonyl oxygens and amide hydrogens of residues V290-H331 and I291-I332 for KaDyP1 and F296-V326 and T295-H325 of BsDyP (Figure S7).If we compare the simulations that include KaDyP1 E309 or BsDyP E312 in a  deprotonated glutamate or a protonated glutamic acid form, a more stable formation of the continuous helix for both enzymes is present in simulations involving the protonated form.Overall, all simulations of wild-type BsDyP show slightly more diversity in these distance distributions than KaDyP1 simulations.The increased diversity in the sampled conformation agrees with the secondary structure propensities in Table S4.
The accepted catalytic mechanism towards bulky substrates of DyPs involves electron tunneling (LRET) from the substrate via a surface-exposed tyrosine or tryptophane to the oxidized heme group in the substrate oxidation step.KaDyP1 and BsDyP have 5 and 10 Tyr and 4 and 5 Trp residues, respectively.The residues that correspond to W263 in TcDyP from Thermomonospora curvata, which was previously identified as the main site for catalysis [23] in this A-type enzyme, are W279 for KaDyP1 and W284 for BsDyP (Figure S9).Another study found that the electron tunneling from the surface-exposed tryptophan is also influenced by nearby tyrosine or tryptophan residues, and an aromatic dyad motif can stabilize the radical being formed on the surface. [24]BsDyP has such a dyad with W284 and Y387, while the only aromatic residue near W279 in KaDyP1 is F395.However, for TcDyP when W376 was mutated to F376equivalent to F395 in KaDyP1 -catalytic efficiency improved, suggesting W376 offers a radical sink for off-pathway electron transfer. [23]While the exact requirements for Trp residues to be involved in the electron pathways in A-type DyPs are still unclear, we decided to consider W279 for KaDyP1 and W284 in BsDyP for further analysis.This decision is also supported by our initial calculations of decay factors (Table S5).In Ka-DyP1 W279 offers the shortest electron channeling path, while in BsDyP W284 and Y343 -a conserved Y in every DyP -are the ones with the lowest decay factors.The experimental study of TcDyP concluded that the catalytic function of the residue corresponding to Y343 is minimal, most likely due to its small surface-exposed area. [23]Neither W279 in KaDyP nor W284 in BsDyP are near the mutated glutamate residue.However, they are connected to the catalytic aspartate via a network of residues as also observed in TcDyP.The residues involved in this network are M229, N227 and Q231 for KaDyP1 or F237, F235 and N233 for BsDyP.As KaDyP1 E309 and BsDyP E312 connect to the catalytic aspartate via a hydrogen bonding network (see next section), the modification of the Glu residue could lead to allosteric effects that influence the position of this Trp residue.In case of KaDyP1, Q231, while for BsDyP F235 and F237 are part of the electron channeling network from their respective tryptophans (Figure S10).
A study involving the B-type KpDyP utilized the GROMOS + + program epath [45,46] to identify which tyrosine or tryptophan residues could play a role in the LRET process of these enzymes. [25]This program finds electron tunneling pathways in proteins by calculating the highest possible product of decay factors k, corresponding to the "shortest path" an electron can take between two atoms.The higher this value is, the more likely it will be to channel the electrons between two atoms.In the catalytic mechanism of DyPs, electron channeling is proposed to happen between the NE1 atom of the surfaceexposed to tryptophan and the iron center of the heme group.Thus, we calculated the logarithm of these decay factors for the snapshots of our 100 ns long wild-type simulations, with both deprotonated glutamate and protonated glutamic acid forms of KaDyP1 E309 or BsDyP E312, to determine if the protonation status of the glutamate could influence the available electron channeling tunnels (Figure 4, Figure S10).
For the deprotonated state, the logarithm of decay factors is mostly between À 12 and À 10 for KaDyP1 and between À 14 and À 12 for BsDyP, with some snapshots also sampling higher values.These agree with the ones calculated for the corresponding residue of the A-type DyP EfeB from Escherichia coli (W240, log k = À 11.31 [25] ).With the protonation of the targeted glutamate to glutamic acid, the calculated log k shifts towards higher values for both enzymes: In the case of KaDyP1 the peak around log k = À 11 decreases, and a new peak emerges near log k = À 8 (i.e., an increase of the total decay factor by three orders of magnitude); while for BsDyP the main peak shifts towards log k = À 12 and also the peak towards log k = À 10 becomes more pronounced.This could influence the catalytic efficiency of these enzymes towards bulky substrates, especially considering that this modification is on the logarithmic scale, and it means orders of magnitude of improvement in decay factors.Another important factor that can affect the catalytic activity of these enzymes is the position of the catalytic arginine and aspartate in the heme group.This region is responsible for the binding and deprotonation of hydrogen-peroxide and its oxidation and formation of Compound I. [47] If we analyze these positions, we see that if E309 or E312 are deprotonated, the catalytic aspartate is in a closer position, while the arginine is farther away, compared to the conserved glutamate being protonated (Figure 4 and S8).To summarize, we have found that the protonation state of E309 or E312 in the wild-type enzymes influences catalytically important properties, including the positions of catalytic residues and the possibilities from electron channeling from their respective surface exposed tryptophans.

Simulations of Variants
To find a rational explanation for the behavior of the variants identified in our site-saturation mutagenesis experiments, we have also performed MD simulations using the KaDyP1 E309Q and E309L and the BsDyP E312F variants, similar to our wild-type simulations.As for the wild-type, we collected four 100 ns long trajectories of monomers for each considered variant we analyzed.
As mentioned above, a significant loss of kinetic and thermodynamic stability was observed in BsDyP variants and little to no difference in KaDyP1 (as compared to the wild-type).The biggest change in stability in our simulations for the BsDyP E312F variant is at the α-helices, separated by loop 2 described previously (Figure 1 and Figure S7, Table S4).The secondary structure analyses using the GROMOS + + dssp program show that in the first part of the helix, near loop 2, the helix partly loses its secondary structure and samples conformations where those amino acids are already in a loop.Also, in the BsDyP E312F variant, the hydrogen bonds keeping the two α-helices together break up, and the two parts show larger distances (Figure S7).This effect also occurs to some extent for the KaDyP1 variants; however, it is much less pronounced than in BsDyP E312F, where these distances do not sample the expected hydrogen bonding distance of ~2.
If we consider the solvent-accessible surface area (SASA) of the heme of the variants, we can see that it generally decreases in the variants (Table S6, Figure S11).This property was variable in our wild-type simulations (Figure 3 b and d).These results were somewhat surprising, as with these mutations, we abolished a stabilizing hydrogen-bond interaction between loop 1 and loop 2. However, the heme becoming less accessible to the solvent and any substrate can benefit the catalytic efficiency: this closure can protect the heme from decoupling reactions.The closure of the heme pocket is also in line with the increased apparent K M values for hydrogen-peroxide we could measure for the variants (Table 1), as there are fewer channels for this substrate to access the distal site.
We have also examined the possible electron channeling pathways between the surface-exposed tryptophans and the heme group for the variant enzymes.We observed that the resulting distributions of the logarithm of the decay factor k also sample higher values more often and resemble the distributions observed for the wild-type enzyme with protonated E309 or E312 (Figure 4).The increase of these values is in line with the observed increase in turnover numbers and catalytic efficiency towards ABTS in our experiments, as this property influences the oxidation efficiency of substrates of larger size.
The emerging peaks of the E309 or E312 variants towards log k = À 9 and above compared to the deprotonated wild-type enzyme are already in the range of the surface exposed tryptophan that plays a role in the catalytic mechanism of AauDyP (W377, log k = À 9.49 [25] ).Even if only a portion of the enzyme is found in this beneficial conformation, resulting in similar activity as the much more active AauDyP, we could expect an increase of catalytic parameters, as is the case for KaDyP1 and BsDyP discussed here.These higher values also approach the decay factor calculated for the tyrosine of lignin peroxidase (TceLiP, log k = À 8.38 [25] ); however, they stay well below the calculated values for tryptophan in cytochrome c peroxidase (CcP, log k = À 5.62 [25] ).
To show that the introduced mutations do not hinder the LRET ability of KaDyP1 or BsDyP, all variants were incubated with hydrogen-peroxide or ABTS or both, similarly to a previous study done with TcDyP, [23] and then analyzed on SDS-PAGE gels.If the enzymes indeed have the ability for long-range electron transfer, the surface exposed radicals that form in the presence of hydrogen-peroxide without the presence of a reducing substrate can cause a portion of the protein fraction to form covalent dimers or multimers.In the presence of a reducing substrate (ABTS), this multimerization should happen to a lesser extent, as the protein radicals are reduced at the surface exposed sites.Furthermore, reduced multimerization proves that the enzymes can oxidize the substrates at the surface exposed radical sites.Indeed, both the KaDyP1 and BsDyP variants have the ability for LRET as we observed clear dimerization and multimerization bands (Figure S12) when reacting with hydrogen-peroxide.Furthermore, in the presence of ABTS, the multimerization happens to a lesser extent, proving that this substrate is indeed being oxidized at the surface exposed radical sites in BsDyP, KaDyP and their variants.
As mentioned earlier, the positions of the catalytic aspartate and arginine on the distal side of the heme group can also influence reactivity with hydrogen peroxide.In the variants, we see that the distance of the catalytic aspartate to the heme iron decreases, and the distance of the arginine to the heme iron increases compared to what we can observe if the target Glu residues in KaDyP1 or BsDyP are protonated and resemble more the distributions obtained for the deprotonated form of these residues (Figure S8).We have seen in our variants that the apparent K M values towards hydrogen-peroxide vary as compared to the wild-type, as well as the inhibition constant, indicating that hydrogen bonding becomes less favorable, for example, in the case of variant KaDyP1 E309L and BsDyP E312W.
Based on the experimental and simulated properties of the KaDyP1 E309 or BsDyP E312 variants, we propose that the glutamate residue works as a pH switch (Figure 5).The decay factors (log k) between the surface exposed tryptophan residues and the heme iron shift towards higher values in the simulations in which the glutamate is protonated (Figure 4), suggesting a more likely electron channeling.In the deprotonated form of this glutamate, the electron channeling seems less likely, compared to the protonated form or the variants investigated here.Experimentally, we see in the variants that we have increased turnover numbers and catalytic efficiency compared to the wild type, which aligns with this observation of more likely electron channeling.It also indicates that if the glutamate in the loop is protonated in the wild type, oxidation of bulky substrates becomes easier, as the electron channeling process governs it.
With the distal site residues, we see an opposite effect.The aspartate is closer to the heme in the variants.In our distribution graphs of the distance between the CB atom of the catalytic aspartate and the heme iron (Figure S8a) we see two peaks for the wild-type in KaDyP1 at around 0.8 nm and 0.9 nm, and one main peak for BsDyP at 0.75 nm (Figure S8b).The distribution shifts towards the higher values in the protonated glutamic acid form: in KaDyP1 to 1 nm, while in BsDyP to 0.9 nm.In the glutamate variants we see a shift towards lower distances of the aspartate, to 0.8 nm and 0.55 nm in KaDyP and BsDyP respectively.The catalytic arginine is further away in the variants, at 0.7-0.8nm, while it is a bit closer around 0.7 nm if the glutamate is protonated, in KaDyP1 (Figure S8c).This effect is much clearer for BsDyP, where the distance of the arginine is Figure 5. Schematic representation of the positions of the distal aspartate (dark red, Catalytic Asp) and arginine (dark blue, Catalytic Arg) to the heme group, and the electron channeling pathways (dark red arrows) from the surface exposed tryptophans (purple, Surface Trp), depending on the protonation state of the glutamate, or its replacement by other residues (green, Loop Glu).The straight, full arrows represent generally higher decay factor (log k, higher likelihood of a successful electron channeling) values, while the dashed, curvy arrows represent lower decay factor values observed in our MD simiulations.An equilibrium of the protonated and deprotonated states of the Glu is expected in the wild-type enzyme at the optimal pH (4.3-4.5) for these enzymes.The variants show a combination of these two states, indicating that protonation could be beneficial for the activity of these enzymes.Chemical structures were drawn using ChemSketch. [61]round 0.7 nm if the glutamate is deprotonated or if it is mutated to phenylalanine, while it is around 0.5 nm if the glutamate is deprotonated (Figure S8d).Altogether, the behavior of the distal site catalytic residues resembles the behavior of the deprotonated wild-type variant.In our variants, this generally causes an increased K M towards hydrogen peroxide, which is likely not advantageous for catalysis at low substrate concentrations.All this indicates that if the glutamate found in the loop is deprotonated, oxidation of substrates of larger size becomes more difficult: both the electron channeling pathways and the position of the distal catalytic residues are more favorable in the protonated form.This model suggests that the conserved glutamate in loop 2 could have a very important role in nature and in the design of tailored biocatalysts for specific applications.Conversely, BsDyP and KaDyP1 have an N-terminal Tat signal peptide for secretion via the Tat system.This system mostly handles cofactor-containing enzymes folded into a functional conformation in the cytoplasm.[12] A deprotonated glutamate in the cytoplasm would push the enzyme towards a less active form, preventing unwanted oxidation and accidental cellular damage.As the soil bacteria where these enzymes originate from often live in acidic soil environments and prefer acidic pH environments, [48,49] the excreted enzyme is more likely to contain a protonated glutamic acid, thereby switching to a more active form, which can degrade lignin or other oxidizable species in the extracellular space.

Conclusions
In this work, we have investigated the role of a conserved glutamate residue found in loop 2 of A-type DyPs, BsDyP, and KaDyP1 presented here for the first time.We did site saturation mutagenesis experiments followed by the screening of the resulting variants, and we found that variants KaDyP1 E309Q and E309L, BsDyP E312F and E312W showed increased catalytic efficiency compared to wild-type, while showing comparable (KaDyP1) or lower (BsDyP variants) thermal stability.We used molecular dynamics simulations to rationalize the changed properties of the variants.We have found that the increased catalytic efficiency of the variants is caused by the changes in electron tunneling pathways, making the tunneling more likely in simulated variants.There is also a clear change in the positions of the catalytic arginine and aspartate on the distal side of the heme group.This explains changes in the apparent increased K M values towards hydrogen-peroxide and decreased inhibition constant of the variants.Our MD simulations also included protonated and deprotonated variants of the conserved glutamate.These, together with the data collected on the E309 or E312 variants, indicate that protonation shifts the enzyme towards a more active form than the deprotonated form.As the pH optimum of these enzymes is in the pH 4.0-5.0range, and the bacteria these enzymes originate from prefer acidic soil environments, we propose the loop glutamate could act as a pH switch.Intracellularly, the enzymes would be pushed towards a less active, deprotonated form, but after secretion, they would be protonated and activated in the environment.All the above findings help us to understand the mechanism of these enzymes more in depth, and to identify simple ways to engineer them for various biotechnological applications.

Identifying KaDyP1 in the Genome of K. aureofaciens
Starting with a BLAST search using the characterized Amycolatopsis sp.75iv2 (ATCC 39116) DyP2 dye-decolorizing peroxidase (UniProt K7 N5 M8 [35] ) as a template we selected candidate genes from various strains of K. aureofaciens and compared these with entries in the NCBI database (Identical Protein Groups) and genome annotations.We limited our search to two genome assemblies, ASM208260v1 (K.aureofaciens DM1, reference genome) and ASM118895v3 (K.aureofaciens ATCC 10762), which gave identical results.

Model Building and Molecular Dynamics Simulations
We built a homology model for wild-type KaDyP1 using SWISS-Model [50] by searching for templates using the full length of the amino acid sequence of KaDyP1 (Uniprot: A0 A1E7 N504).The template with the highest sequence identity was a dye-decolourizing-type peroxidase A (DtpA) from Streptomyces lividans (59 % sequence identity; PDB code: 5MAP [51] ), which was used to build the homology model, as a homodimer (Figure S13).The heme groups were modeled in PyMol [52] by aligning the homology model with the template structure and transferring the heme from the template into the model.The homology model agrees closely with the predicted AlphaFold2 structure now available in Uniprot [53,54] with a backbone atom-positional root-mean-square-deviation of 0.558 Å upon alignment using PyMol.For BsDyP, the wild-type Xray structure (PDB-code: 7PKX [33] ) was used for the MD simulations.The missing loop was modelled using the loop modeling functionality of SWISS-Model [50] by using its x-ray structure as the template.We created the models of KaDyP1 variants E309Q and E309L and BsDyP variant E312F from the wild-type enzyme structures using the mutagenesis wizard in PyMol.All enzymes were modelled and simulated as homodimers.
We performed molecular dynamics (MD) simulations using GROMOS11 [55] with the GROMOS forcefield 54 A8. [56] For the simulations where we considered E309 or E312 to be protonated, the protonation state of the glutamate was defined in the topology.The molecules were energy minimized using the steepest descent algorithm and solvated in simple point charge (SPC) [57] water boxes with a minimum solute-to-wall distance of 0.9 nm.We added sodium or chloride counter ions by replacing random water molecules to achieve a neutral net charge and additional ion pairs to provide an ionic strength of 0.1 M. The systems were equilibrated to 298 K in six 20 ps steps, increasing temperature by 50 K each step and gradually releasing initial position restraints on the protein atoms from a force constant of 2.5 .10 4 kJ/mol/nm 2 to zero, followed by 20 ps of unrestrained simulation at 298 K.This was followed by 100 ns long production runs at 298 K and constant pressure of 1 atm.The temperature was kept constant using Nosé-Hoover chains, with a relaxation time of 0.1 ps, while the barostat had a relaxation time of 0.5 ps.The bond lengths were kept constant using the SHAKE algorithm, and a timestep of 2 fs was used.Electrostatic interactions were described with the reaction field method.All simulations were done in duplicates of homodimers, resulting in four trajectories per monomer for each variant.

Site Saturation Mutagenesis
We constructed the plasmids expressing the KaDyP1 E309 saturation variants by amplifying the KaDyP1 gene by PCR in two fragments with codons NNS at the E309 site (Table S7, reactions 3 and 4).The fragments were assembled with the pGEX-6P-1 plasmid backbone using NEBuilder HiFi DNA Assembly Master Mix, using the standard reaction protocol.We transformed electrocompetent BL21* cells directly with the assembled mixture.All the above plasmids contain an N-terminal GST-tag.The PCR reactions were performed with Q5®-High-Fidelity DNA Polymerase (New England Biolabs), using the recommended setup including 5X Q5 High GC Enhancer: initial denaturation at 98 °C for 2 min, followed by 35 cycles of 10 s denaturation of 98 °C, 30 s of annealing at the temperature included in Table S7, and elongation for a time noted in Table S7 at 72 °C, followed by a final extension of 2 min at 72 °C, and cooling to 4 °C.Saturation mutagenesis of residue E312 of BsDyP was performed using the QuickChangeTM Site-Directed Mutagenesis (QCM) protocol with primers forward 5'-GGGCAGAAAAAGNNSACAGATCCCGTGAAGC -3' and reverse 5'-GCTTCACGGGATCTGTSNNCTTTTTCTGCCC -3' (where N designates a 1: 1 mixture of all four nucleotides and S designates a 1: 1 mixture of G and C).The PCR reaction was performed in a 50 μL volume.The initial denaturation step was performed by pre-heating the mixture to 94 °C for 2 min, followed by 28 denaturation cycles at 95 °C for 2 min, annealing at 60 °C for 1 min, and extension at 72 °C for 8 min; a final extension step at 72 °C for 10 min.PCR products were digested with DpnI (Thermo Fisher Scientific) for 6 h at 37 °C and purified using the GFX PCR DNA Purification Kit (GE Healthcare).

Screening Saturation Libraries for Activity and Stability
To screen both saturation mutagenesis libraries, individual colonies from the transformed BL21* cells were picked to 96-well plates, each containing individual KaDyP1 and BsDyP plasmids modified at the E309 and E312 sites, respectively.Pre-cultures of variants were grown in 200 μL of media in 96-well plates overnight, 750 rpm at 37 °C.Ten μL of the pre-culture were transferred to 170 μL fresh media and incubated for ~5 h shaking at 750 rpm at 37 °C, after which 10 μL of 1.5 mM hemin and 10 μL of 2 mM IPTG were added and incubated at 25 °C overnight, 750 rpm.Cells were harvested by centrifugation at 4 °C, 4000 rpm, 10 min, supernatants discarded, and cell pellets resuspended in a solution of 30 % B-PER™ Bacterial Protein Extraction Reagent (Thermo Fischer Scientific, Rockford, USA) in 20 mM Tris pH 7.6, for cell disruption.After centrifugation at 4000 rpm for 30 min, we harvested supernatants (crude extracts).We measured the activity using 20 μL of the supernatant in a reaction mixture containing 1 mM 2,2'-azinobis(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS), 0.1 mM H 2 O 2 in 100 mM Sodium Acetate buffer pH 4.5, the final reaction volume was 200 μL.The activity was measured by following the progress of the reaction overtime at 420 nm (ɛ 420, ABTS = 36 mM À 1 * cm À 1 ) and 25 °C using a Synergy2 microplate reader (BioTek, Winooski, VT, USA).Activities were compared to the average of wild-type activities.For the stability screening, crude extracts were incubated at 40 °C for 15 min.The residual activity (ratio of the activity after 15 min at 40 °C to the initial activity of the variant (v), normalized to the parent type (p) -(Ar/Ai)v/(Ar/Ai)p) was determined.One unit (U) of enzymatic activity was defined as the enzyme required to reduce 1 μmol of substrate per minute.

Protein Production and Purification
The enzymes KaDyP1 and BsDyP and their respective variants were produced in Tuner E. coli cells transformed with the plasmids harboring the respective genes.KaDyP and variants have an Nterminal GST-Tag cleavable by HRV 3 C protease, while BsDyP and variants have an N-terminal His-Tag.Recombinant strains were cultivated in 2 L of LB medium at 37 °C shaking 140 rpm, and 0.1 mM IPTG and 75 μM hemin were added when OD 600 = 0.6-0.8, and temperature was reduced to 16 °C, in the case of KaDyP1, and 25 °C for BsDyP, overnight.Cells were harvested after centrifugation at 8000 rpm (11325×g) for 10 min (Beckman Coulter, JA-10 rotor).Cells were suspended in 50 mM Tris-HCl, pH 7.3 with 150 mM NaCl Buffer for KaDyP1, and 20 mM Tris-HCl buffer, pH 7.6, with 200 mM NaCl and 20 mM Imidazole for BsDyP, containing DNase I (10 μg mL À 1 extract), MgCl 2 (5 mM), and a mixture of protease inhibitors, antipain, and leupeptin (2 μg mL À 1 extract) was added before disruption , using a French press.Cell debris was removed after centrifugation at 18000 rpm, 4 °C for 2 h (Beckman Coulter JA-25.50 rotor).Supernatants were filtered and loaded onto a 5 mL GSTrap column (GE Healthcare) equilibrated with 50 mM Tris-HCl + 150 mM NaCl pH 7.3.The column was washed, and the bound protein was eluted with a buffer containing 50 mM Tris-HCl, 150 mM NaCl, and 10 mM reduced glutathione at pH 8.0.The eluted protein was buffer exchanged to 50 mM Tris-HCl supplemented with 150 mM NaCl, pH 7.3, and concentrated to a final volume of 5 mL.One mg of HRV 3 C protease was added to cleavage the GST-tag and incubated at 4 °C overnight.The cleaved sample was passed through a GSTrap column using the standard syringe protocol and collected in the flow-through.The sample was concentrated by ultrafiltration and loaded onto a Superdex 75 column equilibrated with 20 mM Tris-HCl with 200 mM NaCl, pH 7.3.We checked purity using SDS-PAGE using self-cast gels with 12 % acrylamide content, run with a Mini-PROTEAN Tetra Vertical Electrophoresis Cell (BioRad), and pooled pure, active fractions (Figure S15).The purified samples were kept at 4 °C to avoid activity loss upon freeze-thaw cycles until further analysis.In the case of BsDyP, cell crude extracts were loaded onto a 5 mL HisTrap column (GE Healthcare) pre-equilibrated with 20 mM Tris-HCl buffer, pH 7.6 (supplemented with 0.2 M NaCl and 20 mM imidazole).The proteins were eluted by applying a 0-100 % gradient of 1 M Imidazole in 20 min with a 5 mL/min flow rate.The red and active protein fractions were collected and pooled.Imidazole was removed using a PD-10 desalting column (GE Healthcare) that changed the buffer to 20 mM Tris-HCl, pH 7.6 with 0.2 M NaCl.Aliquots of enzyme preparation were stored at À 20 °C.

Enzyme Activity Assays
The pH profile was determined in reactions containing 0.6 mM or 1 mM ABTS for KaDyP1 and BsDyP, respectively, in the presence of 0.1 mM of H 2 O 2 , in Britton-Robinson buffer (100 mM phosphoric acid, 100 mM boric acid and 100 mM acetic acid) mixed with 1 M of NaOH to the desired pH in the range 2 to 10 at 25 °C.Reactions were monitored at 420 nm and 25 °C (ABTS, � 420nm = 36,000 M À 1 cm À 1 ) using the Synergy2 microplate reader (BioTek, Vermont, USA).Reactions to estimate the kinetic parameters for ABTS (0.1-4 mM in the case of KaDyP1, and 0.01-4 mM for BsDyP) were performed with 0.05 mM H 2 O 2 for KaDyp1 wild-type, 0.08 mM for E309Q variant, and 0.1 mM for E309L variant, and 0.2 mM H 2 O 2 for BsDyP and variants, in 100 mM sodium acetate buffer at the appropriate pH.Reactions to estimate kinetic parameters of H 2 O 2 (2.5 μM-2 mM for KaDyP1 and 0.01-4 mM) were performed in 0.6 mM ABTS for KaDyP1 wild-type, 1.5 mM for E309Q and 0.5 mM for E309L, and 1 mM ABTS was used for BsDyP wild-type and variants, in 100 mM sodium acetate buffer at the appropriate pH.The kinetic data were fitted directly using the Michaelis-Menten equation or the equation for the non-linear curve that fits enzyme kinetics affected by substrate inhibition (v = Vmax[S]/ (Km + [S](1 + [S]/Ki))) (Origin software or qtGrace software).Measurements were performed at least in triplicate.For more details see text S1 in the supporting information.

Protein Stability
The thermodynamic stability was assessed using 10 μM of purified enzyme preparations in 20 mM Tris-HCl buffer, pH 7.6, with 200 mM NaCl.The temperature was increased at 1 °C/min from 20 °C to 100 °C using a thermostatically controlled thermal block.Fluorescence emission was recorded every 0.5 °C, with an excitation wavelength of 296 nm and emission wavelength of 330 nm (Varian Cary Eclipse Fluorescence Spectrophotometer).We calculated the folded fractions at every temperature point, and the melting temperature (T m ) was determined at the point where half of the protein fraction unfolded.We measured thermal inactivation by incubating all enzymes from 25 to 50 °C for 30 min.The incubation was stopped by placing enzymes on ice, and the residual activity was measured in the presence of 1 mM ABTS and 0.1 mM H 2 O 2 in 100 mM Sodium acetate buffer at the appropriate pH (Table S8).T 50 values -the temperature where the enzyme loses half its activitywere estimated using a fit to the sigmoidal curve.

Figure 2 .
Figure 2. Rescreening for activity and stability of KaDyP1 variants library saturated at the E309 site (a) and BsDyP library saturated at the E312 site (b).The name of amino acids indicates the known substitutions in the obtained variants.

Figure 3 .
Figure 3. Examples of Root-mean-square fluctuation (rmsf) of the α-carbon atoms of each amino acid and the fluctuation of the solvent accessible surface area (SASA) of the heme group in KaDyP1 (a and b respectively) and BsDyP (c and d respectively) throughout a 100 ns long MD simulation.Regions in αhelical and β-sheet secondary structures are marked with pink trapezes and orange rectangles, respectively (a, c).

Figure 4 .
Figure 4. Normalized distributions of the logarithm of decay factor k for the electron channeling from the NE1 atom of W279 in KaDyP or W284 in BsDyP to the heme iron including both monomers in both independent replicate simulations for all variants.Values are calculated with the GROMOS + + program epath.

Table 1 .
Apparent steady-state kinetic parameters towards ABTS and H 2 O 2 for KaDyP1 and BsDyP and their variants. mM)

Table 2 .
Thermodynamic and kinetic stability of KaDyP1 and BsDyP.The melting temperature T m is at which half of the protein is unfolded.The T 50 is the temperature at which the enzyme activity was reduced by 50 % after incubation for 30 min.