Differential Active Site Loop Conformations Mediate Promiscuous Activities in the Lactonase SsoPox

Enzymes are proficient catalysts that enable fast rates of Michaelis-complex formation, the chemical step and products release. These different steps may require different conformational states of the active site that have distinct binding properties. Moreover, the conformational flexibility of the active site mediates alternative, promiscuous functions. Here we focused on the lactonase SsoPox from Sulfolobus solfataricus. SsoPox is a native lactonase endowed with promiscuous phosphotriesterase activity. We identified a position in the active site loop (W263) that governs its flexibility, and thereby affects the substrate specificity of the enzyme. We isolated two different sets of substitutions at position 263 that induce two distinct conformational sampling of the active loop and characterized the structural and kinetic effects of these substitutions. These sets of mutations selectively and distinctly mediate the improvement of the promiscuous phosphotriesterase and oxo-lactonase activities of SsoPox by increasing active-site loop flexibility. These observations corroborate the idea that conformational diversity governs enzymatic promiscuity and is a key feature of protein evolvability.


Introduction
Enzymes are considered as highly efficient catalysts, so substrate recognition has long been referred to as the key-lock model with the simple view of "one sequence-one structure-one function" [1] which was subsequently refined by the induce-fit model [2]. Indeed, the protein flexibility indicates that proteins exist as a sampling of similar conformations with discrete energy levels [3,4]. Protein dynamics are essential for protein function and define its conformational landscape. Dynamics enable fast rates of Michaelis-complex formation and products release [4]. In certain cases, the transition step between conformations is the rate limiting step of enzyme catalysis [5]. Moreover, the structural diversity linked to protein flexibility constitutes a foundation of protein evolvability [4]. Indeed, the sampling of enzyme conformations allows the accommodation of different substrates within the same active site or the existence of promiscuous activities [5]. Under selective pressure, conformations allowing promiscuous binding/ activities can be selected by evolution. Promiscuity is thus considered to be one of the potential engines of enzyme evolution [1, 4,[6][7][8]. Laboratory evolution can successfully exploit promiscuous activity to generate highly efficient and specialized enzymes [9,10]. However, the molecular mechanisms underlining these specializations are still under investigation.
PTEs and PLLs belong to the amidohydrolase superfamily and, exhibit a (α/β) 8 -barrel fold (the so-called TIM-barrel) [24,26]. This fold consists of 8 β-sheets flanked by 8 α-helixes with the catalytic center localized at the C-terminus of the barrel. The catalytic center contained two metal cations coordinated by four histidines, an aspartic acid and a carboxylated lysine residue. These metals activate a bridging water molecule, which, via a nucleophilic attack onto the reactive center, allows the substrate hydrolysis. The substrate specificity is mainly governed by variations in the connecting loops of the barrel. Indeed, the major structural differences between PTEs and PLLs reside in the active site loop size and conformation [27]. A structural analysis of SsoPox, a hyperthermostable counterpart of the PLL family from Sulfolobus solfataricus [17,27,28], has revealed that loop 7 is shorter in SsoPox than in PTEs; linking the active site to the loop 8, which forms a hydrophobic channel that accommodates AHLs aliphatic chain [22]. The residue W263, located in this cavity, plays a central role in substrate binding in the SsoPox active site. Indeed, W263 is located at the start of loop 8, involved in the dimer interface of the enzyme and positions the lactone cycle of the substrate onto the bi-metallic catalytic center [22]. In this study, we deciphered position 263 contributions to the protein stability, enzymatic activity and promiscuity. We exhaustively mutated this position and selected the most improved variants for both phosphotriesterase and lactonase activities. By cross-analyzing the enzymatic, biochemical and structural properties of the selected variants, we isolated two distinct groups of variants and we propose structural features that may underline these improvements.
PLLs are believed to be natural AHL lactonases, possibly involved in quorum quenching [12,22]. Interestingly, PLLs enzymes, such as SsoPox, hydrolyze a broad spectrum of different lactones. Noteworthy, anionic detergents, which probably increase protein flexibility, strongly stimulate SsoPox promiscuous phosphotriesterase activity (up to 33-fold) [30,31], whereas they have an opposite effect on lactonase activities. In the presence of 0.1% SDS, the enzyme activity towards the best AHL substrate (3-oxo-C10 AHL (l)) is severely compromised (161-fold decrease), and, the activity towards undecanoic-γ-lactone (r) is mildly affected (1.3-fold decrease) ( Table 1). Lower SDS concentrations (0.01%) yield to milder but similar effects: the catalytic efficiency against 3-oxo-C10 AHL (l) (10-fold) is decreaed, and undecanoic-γ-lactone (r) hydrolysis is increased (2-fold) ( Table 1). Alterations in substrate specificity induced by detergent were previously observed in enzymes [32], and were mainly attributed to detergent-induced protein flexibility that enlarged the protein conformational landscape [33]. As for SsoPox, detergentinduced flexibility could promotes the increase of promiscuous phosphotriesterase activity but dramatically compromised the AHLase activity, possibly by altering the specific loop conformational sampling required for the proper binding / hydrolysis of AHLs. This result is consistent with AHLs as the native substrates of SsoPox.

Isolation of W263 variants with improved phosphotriesterase and lactonase activities
Phosphotriesterase activity screening. All selected variants of the saturation site of position 263 have been produced and partially purified in a heating step (see methods). This method might induce a bias resulting from the different expression of variants, but has the merit of enabling fast selection and comparison of the activities of the various expressed proteins. The ability of each variant to hydrolyze paraoxon has been evaluated with 1 mM (Figure 2A) and 100 µM ( Figure S3A). Compared to wild-type SsoPox, variants W263L, W263M and W263F are the most efficient at improving specific activities, ranging from 30-50 and 20-35-fold at 1 mM and 100 µM, respectively. These variants constitute the Phosphotriesterase Selected Variants group (PteSV). Notably, all 19 substitutions of position 263 increase paraoxonase activity (Figure 2A). These variants are also the most proficient at hydrolyzing CMP-coumarin (an organophosphate nerve agent analogue; Figure S1-II) with specific activity improvements ranging from 4-to 11-fold compared to wild-type enzyme ( Figure S3B).
Lactonase activity screening. The lactonase activity of each variant has been screened using a genetically modified Pseudomonas aeruginosa-based bioluminescence screening method ( Figure S4). Variants W263I, W263T and W263V were selected ( Figure 2B) and constitute the Lactonase Selected Variant group (LacSV).

Phosphotriesterase kinetic characterization of selected variants
Catalytic efficiencies of paraoxon hydrolysis have been determined for the selected variants (PteSV group) ( Table 2). Variant W263F is the most proficient (k cat /K M = 1.21 Roman numbers indicate the chemical structures of indicated molecules presented in Figure S1. r corresponds to racemic solution and l at the pure levorotatory enantiomer. Data obtained with cobalt as cofactor. ND corresponds to an undetermined value. When V max could not be reached, the linear part of the MM plot was fitted to a linear regression and corresponded to the catalytic efficiency.
SsoPox clearly preferred the levorotatory enantiomer of the AHLs; kinetics performed for the racemic mix resulted in characterization for the levorotatory enantiomer.

Lactonase kinetic characterization of selected variants
Selected variants have been characterized with the best, possibly natural, substrate of wild-type SsoPox (3-oxo-C10 AHL), the worse AHL substrate (3-oxo-C12) and promiscuous oxo-lactones with long aliphatic chains (undecanoic-δ-lactones and undecanoic-γ-lactones) ( Table 2). The kinetic experiments reveal a mild decrease in the 3-oxo-C10 AHLase activity for all variants (ranging from 2-to 3-fold for the W263F variant to undetectable activities for W263L-M variants) ( Figure S5A). On the contrary, the moderate 3-oxo-C12 AHLase activity of wildtype SsoPox was improved by 1.3-to 55-fold (W263F and W263V, respectively), but no activity was detected for variants W263L-M ( Figure S5A). Regarding the other lactones, all selected variants possess higher catalytic efficiencies than the wild-type enzyme, ranging from 1.4-(W263V) to 148-fold for undecanoic-γ-lactone (W263T) ( Figure S5) and from 3.3-(W263L) to > 74-fold for undecanoic-δ-lactones (W263I) (Figure 2C & Figure S5AC). Notably, contrary to the wild-type enzyme, all selected variants (with the exception of W263F) exhibit a substrate inhibition for undecanoic-δ-lactones with K I values ranging between 789 ± 186 µM and 7 400 ± 2 475 µM, respectively, for the W263V and W263M variants. This substrate inhibition has not been observed for undecanoic-γlactones and AHLs. Because these mutants dramatically increase the δ-and γ-lactonase activities (up to 148-fold), but have a less important effect on AHLs (up to 55-fold improvement for 3-oxo-C12 AHL) or even decrease this activity (with 3-oxo-C10 AHL), these classes of lactones may utilize a different binding mode to the active site and/or a different active site conformation. Differential Flexibility Governs SsoPox Activities PLOS ONE | www.plosone.org

Thermal stability and thermophilicity of SsoPox variants
SsoPox is an extremely thermostable enzyme (melting temperatures (T m ) = 106 °C) [27]. We here show that substitutions at position 263 are destabilizing (Figure 3; Table  S1). Indeed, all variants exhibit a lower T m compared to the wild-type enzyme (between 92.0 ± 2.1 °C for SsoPox-W263L to 84.1 ± 1.6 °C for SsoPox-W263V). In addition to SsoPox-W263M, PteSV variants possess a higher thermal stability than LacSV variants. Contrary to the wild-type enzyme that did not lose any paraoxonase activity with an increase in temperature [17], the selected variants possess different profiles (Figure 3). Interestingly, the loss of paraoxonase activity for the variants PteSV occurs before the loss of structure (corresponding to the T m value). On the contrary, the paraoxonase activity of LacSV variants exhibit a behavior that is similar to that of the wild-type enzyme, i.e., the activity keeps increasing with temperature (or remains higher than ambient temperature activity for W236T around its T m ). The temperature-induced behaviors of LacSV and PteSV confirm their respective distinct behaviors observed in enzyme kinetics.

Structural characterization
In order to understand the structural consequences of variations at position 263, the crystal structures of all selected variants have been solved. No major differences in the wildtype enzyme structure are observed. Position 263 is located at the dimer interface [22]; and it modulates the relative orientation of its contacting residues, F104, in the second monomer ( Figure S6A). Consequently, substitutions of W263 affect the relative orientation of both monomers; for each mutant, the monomers are closer to each other compared to the wild-type enzyme (with the exception of W263L). The dimer reorientation yields significant movements that, ranging from 3.4 Å (W263T) to 4.7 Å (W263I) compared to wild-type SsoPox ( Figure S6C; Table S1).
The active sites of variant structures superimpose well with the wild-type enzyme structure ( Figure S6AB). However, the different residues selected at position 263 modulate the size of the active site cavity, which is increased as compared to wildtype enzyme. The active site cavity is globally larger for LacSV than for PteSV (Figure S7). The HTL-bound structure ( Figure  S8A) of the lactonase variant W263I exhibits the same binding mode than that observed in the wild-type SsoPox structure [22]: both structures are well superimposed (Figure S8B). This feature suggests that the observed catalytic enhancements are related to substrate binding, rather than to catalysis improvement. The loss of the interaction of the lactone ring with W263 is compensated by an interaction with a polyethylene glycol molecule that fills the cavity created by the W263I mutation ( Figure S8A).
The main difference, albeit subtle, between all structures relate to the conformation of the active site loop 8, which carries the position 263 [22]. Indeed, all selected variants present a slightly altered loop 8 conformation compared to wildtype enzyme (Figure 4ACE). All variants possess loop 8 conformations that are similar (Figure 4E) with the exception of the W263L variant ( Figures 4C & S9A). Moreover, the Roman numbers indicate the chemical structures of indicated molecules presented in Figure S1. normalized B-factor confirms that the substitution of W263 is highly destabilizing, because the normalized B-factor of the loop is increased in all variants, compared to the wild-type structure (Figure 4BDF & S9B). The increased flexibility (or increased normalized B-factor) is even more pronounced in PteSV variants than in LacSV variants ( Figure 4BD).

W263 substitutions increase SsoPox promiscuous activities
A single mutation of W263 can significantly increase the paraoxonase activity of the enzyme (by 23-fold with the W263F substitution), the oxo-lactonase activity (by 148-fold with the W263T substitution) and the activity toward the poor AHL Roman numbers indicate the chemical structures of indicated molecules presented in Figure S1.   substrate (3-oxo-C12 AHL; by 55-fold (W263V)). The impact of this substitution is such that SsoPox-W263F hydrolyzes nerve agent analogs with remarkable efficiency (k cat /K M is 8.85 (±2.99)x10 4 M -1 s -1 against IMP-coumarin), whereas the wildtype SsoPox is unable to hydrolyze this compound [19]. Noteworthy, subtle changes, such as W263L or W263I, yield to a large tradeoff in catalytic activities. The W263 substitution seems to have similar effect as SDS on SsoPox: promiscuous phosphotriesterase activity is increased in both cases. Interestingly, SDS has a small stimulatory effect on SsoPox-W263F catalytic efficiency (1.7-6.1-fold increase), whereas it strongly stimulates the wildtype enzyme (with paraoxon and CMP-coumarin as substrates; 6.4-to 12.4-fold increase) [19]. Thus, the detergent inducedflexibility and substitutions of W263 may have overlapping structural effects on the enzyme. However, whereas detergentinduced flexibility has no effect on the lactonase activity, we isolated variants of W263 that exhibit increased lactonase activities (W263T/V/I). These variants exhibit dramatic improvement of their oxo-lactonase catalytic efficiencies (148fold for variant W263T), while the activity for the best AHL substrate is compromised (3-oxo-C10 AHL; for the wild-type enzyme).
Residue W263 of SsoPox is a key position that interacts with the substrate, is located in loop 8 and is involved in the dimer interface [22] (Figure S6). The effects of W263 substitutions on the enzyme structure are numerous, because the W263 mutation provokes a reshaping of the active site cavity ( Figure  S7) and, for some variants a slight re-orientation of the protein homodimer ( Figure S6). These structural features may contribute to the observed variation of catalytic activities in the W263 variants, compared to the wild-type enzyme. However, the modulation of the enzyme dimer are not systematic for all variants, and therefore cannot explain the systematic effect of W263 on promiscuous activities. Additionally, our study demonstrates that all 19 possible substitutions of the W263 residue increase the SsoPox paraoxonase activity. Such a global effect suggests a non-specific effect of the substitutions on the catalytic cycle, and eliminates a specific role of W263 mutants in the promiscuous substrates accommodation. Moreover, the substitutions of W263 have another effect on the enzyme: all substitutions are consistently destabilizing ( Figure  3) and dramatically increase the active site loop 8 flexibility (Figure 4).

A given promiscuous activity requires a given conformational subset of the active site loop
Interestingly, the selected mutations for increased phosphotriesterase and lactonase activities (PteSV and LacSV) are overall mutually exclusive and constitute two distinct groups. Regarding the PteSV group, while these variants exhibit higher phosphotriesterase activity, they demonstrate mild improvements for the oxo-lactones and no or wild-type-like AHLase activity (W263L-M and W263F, respectively). Conversely, the LacSV group dramatically increases oxolactonase activity and 3-oxo-C12 AHLase activities, while the phosphotriesterase activity is mildly modulated and 3-oxo-C10 AHLase is decreased. This dichotomy between the two groups of substitutions is further illustrated by different active-site loop conformational properties. The residues selected in the LacSV group (T, V, I) are overall smaller than those selected in the PteSV group (F, L, M) with the exception of isoleucine (I). Mutations W263V-T therefore yield enlarged active site cavities for the lactone variants. The analysis of the normalized Bfactors of the variants clearly reveal that the substitution W263 is destabilizing (as confirmed by the T m of the variants), but also that the active site loops of PteSV are significantly more disordered than those of LacSV. This feature is nicely illustrated by the thermophilicity profile of these variants. PteSV lose their activity before the loss of the global structure (T m ), whereas LacSV do not, and exhibit a similar thermophilicity profile to that of the wild-type enzyme [17]. These data strongly suggest that the increased disorder in loop 8 caused by the PteSV substitutions collapses the loop as the temperature increase, whereas it does not occur with the LacSV.
The consequences of the W263 substitution are a decrease in the overall protein stability, a reshaping of the active-site cavity, a slight re-orientation of the enzyme homodimer, a very important increase in loop 8 flexibility and concomitantly an increase in the enzyme's promiscuous activities. The different group of W263 substitutions yield to different conformational loop samplings that possess different structural and physical properties, and are distinct from wild-type behavior. Moreover, because the activity profiles of LacSV and PteSV show little overlap, the promiscuous activities may then require not only different loop conformations, but also different subsets of the conformational loop landscape. The enzyme would then use a given conformational subset of the active-site to process a given molecule. More precisely, this conformational subset would produce one specific loop conformation that allows the productive binding of the given substrate and another that permits an efficient release of corresponding products. The fact that all selected mutations, for both group of activities, dramatically lower the K M of the enzyme is consistent with the idea that the conformational flexibility required for enhancing promiscuous activities mainly improves the substrate binding (or reduces non-productive bindings), and/or products release but not catalysis (i.e. the chemical step).
Similarly to the case of paraoxonase 1, the promiscuous activities utilize the catalytic machinery in different combinations [6][7][8]. This different usage of the active site enables the proper alignment of the promiscuous substrates and the catalytic residues [1, 7,8]. Position W263 thus exemplifies a point mutation that can dramatically alter enzyme specificity without the complete loss of the native function (weak tradeoff) [4]. Such a mutation may then provide an evolutionary advantage to the organism, and give birth to a specialized novel enzyme after gene duplication. Flexible loops in the enzyme support the notion of fold polarity [35], whereby a part of the active site (the loop) is weakly connected to the protein scaffold and thus provides the potential, with little mutational events, for evolving new functions.

Screening methods
Sample preparation for the screening steps. A sitesaturation of position W263 of SsoPox was ordered to a service provider (GeneArt, Invitrogen, Germany). Each variant were checked by sequencing and stored as Escherichia coli DH5α cell glycerol stocks. The 20 plasmids (pET22b-SsoPox-W263X [30]) have been purified from E. coli DH5α cells and transformed into the BL21(DE 3 )-pLysS strain for protein production. Protein production was performed in 3 mL of ZYP medium [36] (100 µg/ml ampicillin, 34 µg/ml chloramphenicol) as previously described [19,20,30]. Cells were harvested by centrifugation (3 000 ×g, 4 °C, 10 min), re-suspended in 500 µL of lysis buffer (50 mM HEPES pH 8, 150 mM NaCl, 0.2 mM CoCl 2 , 0.25 mg/ml lysozyme, 0.1 mM PMSF and 10 µg/ml DNAseI) and stored at -80°C. Suspended frozen cells were thawed and disrupted by three steps of 45 seconds of sonication (Ultrasonic processor xl; power 5). Cell debris were removed by centrifugation (13 000 ×g, 25 °C, 30 min). Partial purification of the protein was performed by 15 minutes incubation at 70 °C, which exploited SsoPox extreme thermal stability [17,27,28]. Aggregated proteins were harvested by centrifugation (13 000 ×g, 25 °C, 30 min), and the estimated purity was evaluated as > 70%. The total protein quantity of samples was evaluated using a nanospectrophotometer (Nanodrop, Thermofisher Scientific; France). These protein amounts were used to calculate specific activities. This method enables fast comparison of variants' activities, even if it might introduce a bias due to potential differences in expression/ purification yields of the variants.

Purification of SsoPox and its variants for kinetic measurements and crystallographic studies
Protein production was performed using the E. coli strain BL21(DE 3 )-pGro7/GroEL (Takara Bio). Productions have been performed in 500 mL of ZYP medium [36] (100 µg/ml ampicillin, 34 µg/ml chloramphenicol) as previously explained [19,20,30], but 0.2% (w/v) arabinose (Sigma-Aldrich, France) was added to induce the expression of the chaperones GroEL/ES. Purification was performed as previously explained [30]. Briefly, a single step of 30 minutes incubation at 70 °C was performed, followed by differential ammonium sulfate precipitation, dialysis and exclusion size chromatography. Proteins were quantified using a nanospectrophotometer (Nanodrop, Thermofisher Scientific, France) and a protein molar extinction coefficient generated with the protein primary sequence in PROT-PARAM (Expasy Tool software) [40].

Enzyme kinetics
Experiments were performed in triplicate at 25 °C and recorded using a microplate reader (Synergy HT, BioTek, USA) and the Gen5.1 software in a 6.2 mm path length cell for a 200 µL reaction in a 96-well plate as previously explained [19]. Catalytic parameters were obtained by fitting the data to the Michaelis-Menten (MM) equation [41] using the Graph-Pad Prism 5 software. When the V max could not be reached in the experiments, the catalytic efficiency (k cat /K M ) was obtained by fitting the linear part of MM plot to a linear regression using Graph-Pad Prism 5 software.
Lactonase kinetics. Lactonase kinetics were performed using a previously described protocol [11,19]. The time course hydrolysis of lactones were performed in lac buffer (2.5 mM Bicine pH 8.3, 150 mM NaCl, 0.2 mM CoCl 2 , 0.25 mM Cresol purple and 0.5% DMSO) over a concentration range 0-2 mM for AHLs and 0-5 mM for δ/γ-lactones. Time course hydrolysis of undecanoic-γ-lactone (d, l) (Figure S1-XVI) and 3-oxo-C10 AHLs (l) (Figure S1-XI) in presence of 0.1 and 0.01% SDS has also been performed in lac buffer. Duplicate kinetics of SsoPox Differential Flexibility Governs SsoPox Activities PLOS ONE | www.plosone.org with 250 µM of racemic (d, l) and enantiopure (l) 3-oxo-C8 AHLs (Figure S1-X) have been performed to determine the enantioselectivity of SsoPox. Cresol purple (pK a 8.3 at 25 °C) is a pH indicator used to follow lactone ring hydrolysis by acidification of the medium. Its molar coefficient extinction (ε 577 nm = 2 923 M -1 cm -1 ) was evaluated by recording the absorbance of the buffer over a range of acetic acid concentrations (0-0.35 mM). For some SsoPox variants, the MM plots have been fitted to the substrate inhibition equation [41] using the Graph-Pad Prism 5 software to determine a K I for undecanoic-δ-lactone. Consequently, the calculated catalytic efficiencies in these conditions are valid only at low substrate concentrations.

Biophysical studies
Melting temperature determination. Circular Dichroism spectra were recorded as previously explained [19] using a Jasco J-810 spectropolarimeter equipped with a Pelletier type temperature control system (Jasco PTC-4235) in a 1 mm thick quartz cell and using the Spectra Manager software. Briefly, measurements were performed in 10 mM sodium phosphate buffer at pH 8 with a protein concentration of 0.1 mg/mL. Denaturation was recorded at 222 nm by increasing the temperature from 20 to 95 °C (at 5 °C/min) in 10 mM sodium phosphate buffer at pH 8 containing increasing concentrations (1.5-4 M) of guanidinium chloride. The theoretical T m without guanidinium chloride was extrapolated by a linear fit using the GraphPadPrism 5 software.
Thermophilicity analysis. The temperature dependence of the paraoxonase activity of SsoPox variants were studied over the range of temperatures 25-85 °C with a 10 °C increment. The paraoxon hydrolysis (50 µM) (ε 405 nm = 17 000 M -1 cm -1 ) was monitored in 500 µL with 1-cm path length cell and a Cary WinUV spectophotometer (Varian, Australia) using the Cary WinUV software in pte buffer pH adjusted with NaOH to pH 8 at each temperature. Measurements were performed in triplicate.

Structural analysis
Crystallization. Crystallization assays were performed as previously described [30,42] using enzymes concentrated at 6 mg/mL -1 . Co-crystallization assays with C10-HTL were performed using the same protocol [22] but adding 4 µL of a 60 mM C10-HTL solution (in Ethyl acetate:DMSO; 1:1) to 120 µL of the protein solution. Crystallization was performed using the hanging drop vapor diffusion method in 96 well plates (Greiner Microplate, 96 well, PS, F-bottom) on ViewDrop II seals (TPP Labtech). Equal volumes (0.5 µL) of protein and reservoir solutions were mixed using a HoneyBee X-8 (Cartesian) crystallization instrument, and the resulting drops were equilibrated against a 150 µL reservoir solution containing 20-30% (w/v) PEG 8000 and 50 mM Tris-HCl buffer (pH 8). Thin crystals appeared after few days at 277 K.
Data collection and structure determination. Crystals were first transferred to a cryoprotectant solution composed of the reservoir solution and 20% (v/v) glycerol, a 1/30 (v/v) ratio of a 60 mM C10 HTL solution was added to the cryo-protectant solution for co-crystallization trials. Crystals were then flashcooled in liquid nitrogen. X-ray diffraction data were collected for W263L, W263M, W263I, and W263V crystals at 100 K using synchrotron radiation at the ID23-1 beam line (ESRF, Grenoble, France) and an ADSC Q315r detector. The diffraction data for W263F and W263I-C10 HTL crystals were collected at the Proxima-1 beam line (SOLEIL, Gif-sur-Yvette, France) using a PILATUS-6M detector. X-ray diffraction data were integrated and scaled with the XDS package [43] ( Table  4). The phases were obtained using the native structure of SsoPox (PDB code 2vc5) as a starting model, performing a molecular replacement with MOLREP [44] or PHASER [45]. The models were built with Coot [46] and refined using REFMAC [47]. Structure illustrations were performed using PyMOL [48]. B-averages of structures were evaluated using Baverage software from CCP4 suite [49]. RMSD were evaluated comparing all structures to wild-type structure (2vc5) using Swiss Prot PDB Viewer software [50].