The Bacteriophage–Phage-Inducible Chromosomal Island Arms Race Designs an Interkingdom Inhibitor of dUTPases

ABSTRACT Stl, the master repressor of the Staphylococcus aureus pathogenicity islands (SaPIs), targets phage-encoded proteins to derepress and synchronize the SaPI and the helper phage life cycles. To activate their cycle, some SaPI Stls target both phage dimeric and phage trimeric dUTPases (Duts) as antirepressors, which are structurally unrelated proteins that perform identical functions for the phage. This intimate link between the SaPI’s repressor and the phage inducer has imposed an evolutionary optimization of Stl that allows the interaction with Duts from unrelated organisms. In this work, we structurally characterize this sophisticated mechanism of specialization by solving the structure of the prototypical SaPIbov1 Stl in complex with a prokaryotic and a eukaryotic trimeric Dut. The heterocomplexes with Mycobacterium tuberculosis and Homo sapiens Duts show the molecular strategy of Stl to target trimeric Duts from different kingdoms. Our structural results confirm the participation of the five catalytic motifs of trimeric Duts in Stl binding, including the C-terminal flexible motif V that increases the affinity by embracing Stl. In silico and in vitro analyses with a monomeric Dut support the capacity of Stl to recognize this third family of Duts, confirming this protein as a universal Dut inhibitor in the different kingdoms of life. IMPORTANCE Stl, the Staphylococcus aureus pathogenicity island (SaPI) master repressor, targets phage-encoded proteins to derepress and synchronize the SaPI and the helper phage life cycles. This fascinating phage-SaPI arms race is exemplified by the Stl from SaPIbov1 which targets phage dimeric and trimeric dUTPases (Duts), structurally unrelated proteins with identical functions in the phages. By solving the structure of the Stl in complex with a prokaryotic (M. tuberculosis) and a eukaryotic (human) trimeric Dut, we showed that Stl has developed a sophisticated substrate mimicry strategy to target trimeric Duts. Since all these Duts present identical catalytic mechanisms, Stl is able to interact with Duts from different kingdoms. In addition, in silico modeling with monomeric Dut supports the capacity of Stl to recognize this third family of Duts, confirming this protein as a universal Dut inhibitor.

similar molecular mechanism of recognition, some discrepancies have been observed-for example, while equimolecular interactions (one Dut trimer with three Stls) similar to the f 11Dut-Stl structure have been reported, alternative stoichiometries (one Dut trimer interacting with one Stl dimer or 2 Stl monomers) have also been proposed (15).
In this work, we take a step forward in the study of the molecular basis of this fascinating mechanism of targeting biological processes used by the main SaPIbov1 repressor. Here, we solve the structure of the Stl N-terminal portion in complex with prokaryotic (M. tuberculosis) and eukaryotic (H. sapiens) trimeric Duts. The structures confirm the exquisite molecular mimicry mechanism used for the Stl to recognize Duts from different kingdoms of life. Moreover, the structure shows how motif V of the human Dut, a P-loop that covers the active center once the substrate has been bound, also participates in the interactions. Therefore, Stl targets all the catalytic motifs of the trimeric Duts and thereby reduces its escape capacity. Moreover, our in silico and in vitro studies show that Stl would also be able to recognize monomeric Duts, thus confirming this protein as a universal Dut inhibitor.

RESULTS
The N-terminal portion of Stl binds to Mycobacterium tuberculosis and human Duts. To better understand the molecular basis of the broad-spectrum inhibitory capacity on trimeric Duts shown by the Stl repressor from SaPIbov1 (13,15,16), we undertook the structural characterization of this repressor in complex with one representative prokaryotic and one representative eukaryotic Dut. We selected the Duts from M. tuberculosis (mDut) and humans (hDut) since their interaction with Stl had been previously demonstrated (14,15,17). Functional and structural analyses have shown that Stl is a modular protein with a highly flexible region between its middle and C-terminal domains which, respectively, mediate the recognition and interaction with trimeric and dimeric Duts from S. aureus phages (11). This flexibility has hampered obtaining the three-dimensional structure of the full-length Stl, both alone and in complex with its target Duts. Therefore, for our structural studies we decided to use a construct that included the Stl N-terminal DNA-binding domain (DBD) and the middle domain (Stl N-ter ; residues 1 to 156), which had already been used to obtain the structure in complex with the trimeric Dut of S. aureus phage f 11 (f 11Dut) (11). As a first step, we analyzed whether this portion of Stl is equally responsible for recognizing prokaryotic and eukaryotic trimeric Duts (6,14,15). Native PAGE assays showed that for both mDut and hDut, a band corresponding to the Dut-Stl N-ter complex appears concomitantly with the disappearance of the bands corresponding to each individual protein (see Fig. S2 in the supplemental material). Titration assays showed that at a molar ratio of 1:1, the formation of the Dut-Stl N-ter complex is maximal with a minimal residual amount of each of the participating proteins, a pattern identical to that observed for the control protein f 11Dut. These assays support the interaction model observed in the f 11Dut-Stl N-ter complex structure, in which a trimer of f 11Dut binds three independent Stl N-ter monomers (11). This equimolar stoichiometry has also been proposed for the complex of full-length Stl with mDut and hDut using alternative experimental approaches (14,15). As a second step, we analyzed the binding kinetics of Stl N-ter for these Duts using biolayer interferometry (BLI). Our analysis showed that the two Duts have almost identical affinities for Stl N-ter (K D [equilibrium dissociation constant], 34.44 and 39.25 nM for hDut and mDut, respectively), and that this is only 5 times lower than that shown for f 11Dut (7.94 nM), one of its biological targets (Table 1; Fig. S3). Similar differences in affinity between f 11Dut and hDut for the full-length Stl were reported previously using the isothermal titration calorimetry (ITC) technique (0.23 and 0.10 mM for hDut and f 11Dut, respectively) (15,18). These differences in affinity in relation to f 11Dut are mainly due to a higher dissociation rate constant for hDut and mDut (around 10-and 5-times-higher K off , respectively), since both Duts showed a similar (mDut) or even higher (hDut) association rate constant (Table 1; Fig. S3). Furthermore, hDut and mDut showed affinities for Stl N-ter similar to that of the Dut from S. aureus phage 80a (40 nM [19]). These results support that this portion of Stl has selectively evolved to recognize trimeric enzymes that catalyze dUTP hydrolysis.
Structures of Stl N-ter bound to mDut and hDut. Once we had confirmed that both prokaryotic and eukaryotic Duts showed high affinity for Stl N-ter , the structures of its complex with mDut (mDut-Stl N-ter ) and with hDut (hDut-Stl N-ter ) were solved at 2.75-and 1.94-Å resolution, respectively, by molecular replacement using the individual components as searching models (Stl N-ter , PDB identifier [ID] 6H49 [11]; mDut, PDB ID 1MQ7 [20]; hDut, PDB ID 1Q5U [21]) ( Table 2). The mDut-Stl N-ter complex was crystallized in the space group P2 1 , and the crystal asymmetric unit contains two mDut trimers and six Stl N-ter monomers, forming two mDut-Stl N-ter complexes with a 1:1 stoichiometry in which each mDut trimer interacts with three independent Stl N-ter protomers ( Fig. 1A and Fig. S4). The hDut-Stl N-ter complex was crystallized in the space group P2 1 2 1 2 1 , and its asymmetric unit contains one trimer of hDut and three Stl N-ter monomers, with a 1:1 assembly identical to both that which was observed in mDut-Stl N-ter and that which was previously reported for the f 11Dut-Stl N-ter complex (11) (Fig. 1B).
The structures of the complexes of mDut and hDut with Stl N-ter confirm not only the 1:1 binding stoichiometry proposed by the biochemical assays (see above [11]) but also the mechanism of trimeric Dut recognition by Stl observed in the complex of Stl N-ter with f 11Dut, its natural target (11), as well as the mechanism recently described between Stl and Dut from the crustacean Litopenaeus vannamei (lvDut-Stl N-ter [22]). For all these Dut- Stl N-ter complexes, the Dut maintains its trimeric state and binds three independent Stl monomers, each of which uses predominantly its middle domain to interact with individual active centers of trimeric Dut.
In the mDut-Stl N-ter structure, the two independent complexes are largely identical with a root mean square deviation (RMSD) of 1.1 Å for the superimposition of 642 Ca atoms (Fig. S4) corresponding to the Dut trimer and the three Stl N-ter protomers. In both complexes, clear and traceable electron density is present for the entire Dut protomers, except for the eight C-terminal residues corresponding to the catalytic P-loop motif V, which covers the active center of the enzyme once the nucleotide has been bound. In the complex, the Dut active center is occupied by the Stl N-ter molecule, forcing the projection of this Ploop into the solvent. On the other hand, for all the Stl molecules, the eight N-terminal and the three C-terminal residues are not visible, supporting their nonparticipation in the interactions with the Dut that allows them a high flexibility. In addition, two of the Stl protomers, one for each of the complexes, show regions of the N-terminal DBD (residues 33 to 53 and 59 to 62 in one protomer and residues 38 to 39 and 47 to 61 in the other) where electron density is so weak that it prevents the structure from being traced (Fig. S5), indicating that this domain of the repressor has a greater freedom of movement in the complex due to its weak participation in the interaction with the Dut.
In the hDut-Stl N-ter complex, the three Stl protomers are well defined with the exception of the 8 to 10 N-terminal residues for which, as in the case of the mDut-Stl N-ter complex, no electron density is observable. Surprisingly, in the case of the hDut, the structure allows not only the trace of the main body of each Dut protomer but also the C-terminal motif V ( Fig. 1B and Fig. S1). This P-loop, unable to position over the active center occupied by the repressor, is projected toward the Stl and introduces its C-terminal end between the middle domain and the DBD of Stl. This structure gives molecular insight into the contribution of this catalytic motif in Stl binding, which was previously proposed by biochemical and genetic data on the interaction of Duts from S. aureus phages with Stl (7,11,16) and recently observed in the structure of the lvDut-Stl N-ter complex (22).
Dut trimer is the target of Stl. The binding of Stl N-ter to f 11Dut showed negligible structural changes in the Dut trimer (11). To test whether prokaryotic and eukaryotic Dut trimers are also the conformational targets, we compared the structures of these Duts in complex with Stl N-ter with those in their corresponding free forms (apo and nucleotide bound). The six mDut monomers from the two mDut-Stl N-ter complexes in our structure showed an RMSD below 0.9 Å compared with the free mDut monomers from both the apo (PDB ID 1MQ7) and dUTP-bound (PDB ID 1SJN) forms ( Fig. 2A). Similar results were observed with hDut and the hDut-Stl N-ter complex, which showed RMSDs lower than 0.6 Å compared to hDut in apo (PDB ID 1Q5U) or dUDP-bound (PDB ID 1Q5H) forms ( Fig. 2A). For the latter comparison, the C-terminal motif V was not included since, although it is visible in both structures, the motifs are projected in different directions due to the presence of Stl or dUDP. Not only do the individual protomers in the complexes possess identical conformations, the trimers in their free and Stl-bound forms also have the same conformation ( Fig. 2A). The structural comparison of mDut trimers in apo and dUTP-bound forms (PDB IDs 1MQ7 and 1SJN, respectively) with the Stl N-ter complex showed RMSDs as low as 0.6 Å and, in the case of the hDut trimer, RMSDs lower than 0.5 Å with respect to its apo or dUDP-bound forms (PDB IDs 1Q5U and 1Q5H, respectively), supporting that the target of Stl is the Dut trimer.
Interestingly, a detailed view of this structural comparison shows that the trimer of mDut in complex with Stl is more similar to the nucleotide-bound form than the apo form. The differences are mainly concentrated in the unique and short a helix (mDut residues 65 to 70) from the catalytic motif III (Fig. S1). In the Apo mDut, this helix is distorted, losing its helical topology, whereas in the nucleotide-and Stl-bound forms it folds as identical a helices (Fig. S6). This observation indicates that Stl recognizes the Dut active center in its competent conformation for nucleotide binding, in agreement with the capacity of Stl to inhibit the enzymatic activity of Dut (6,14).
Similarly, we analyzed the conformational changes induced in Stl by the recognition and binding to the different trimeric Duts. With the exception of two Stl monomers in the mDut-Stl N-ter complex, which presented a greater flexibility in their DBDs that prevented the tracing of some areas (Fig. S5), the remaining Stl protomers showed an almost identical conformation regardless of whether they were bound to a phage, bacterial, or human Dut (Fig. 2B), yielding RMSDs between 0.3 and 1.1 Å (superimposition of 142 to 146 residues). In some cases, the differences were greater between Stl protomers within the same complex than between complexes from different species (Fig. S5). Marginally larger differences were observed when comparing Stls in complex with the Stl free form (PDB ID 6H49) (Fig. 2C), although in any case these differences exceeded RMSDs of 1.6 Å, indicating that the N-terminal portion of the Stl protomer has evolved to acquire a conformation competent to recognize the active center of trimeric Duts, which is highly conserved from phages to human.
Stl mimics the nucleotide to interact with trimeric Duts. The analysis of the structures of mDut-Stl N-ter and the hDut-Stl N-ter complexes shows that Stl is using a similar strategy to interact with prokaryotic and eukaryotic trimeric Duts. In both complexes, the repressor inserts the middle domain, specifically the helix a8 and its linkers with helices a7 and a9, into the catalytic site of Dut, in the same way that has been shown previously in f 11Dut-Stl N-ter ( Fig. 3) (11). From these Stl structural elements, the residues Y112 and Y113 interact with the conserved Asp and Tyr residues of trimeric Dut motif III (mDut D83 and Y86, hDut D79 and Y82, f 11Dut D81 and Y84), taking up the place of the substrate ribose, pyrimidine ring, and catalytic water ( Fig. 3 and Fig. S7). Meanwhile, Y113 together with Y105 and Y116 also mimics the nucleotide by interacting with conserved Dut catalytic residues, which mediate contacts by interacting with the phosphates of the nucleotide such as the Arg and Ser of motif II (mDut R64 and S65, hDut R62 and S63, f 11Dut R64 and S65) and the Tyr of motif III (mDut Y86, hDut Y82, and f 11Dut Y84) (Fig. 1C, Fig. 3, Fig. S1 and S7, and Tables S2 and S3). The nucleotide mimicry mechanism of binding carried out by the Stl helix a8 represents the main anchor point to the Dut, which is reinforced with additional interactions that recognize partially conserved residues in Duts. This is the case for Stl Y106, which interacts with conserved or partially conserved residues from motifs I and III, or Stl S114 and D117, which interact with positively charged residues (Arg or Lys) located in Dut motifs II and IV (Fig. 3, Fig. S1 and S7, and Tables S2 and S3). Our previous work with S. aureus phage Duts has shown that the presence of a basic residue of Arg or Lys in motif IV implied changes in affinity for Stl (16). Given the variability in this residue among the solved complexes (mDut R110, hDut R105, and f 11Dut K133), we looked in greater detail at the structures. In all the structures, the basic residue is salt bridged with Stl Interkingdom dUTPase Inhibition by the Stl Repressor Microbiology Spectrum D117 and makes hydrophobic and polar interactions with Y116, but this second interaction differs depending on the Dut recognized by Stl. In the original f 11Dut-Stl N-ter complex, Y116 is completely inserted into the Dut core and interacts with residues from catalytic motifs II (R64) and IV (K133) and the phage-specific motif VI (I110) (11). The Y116 side chain conformation in the hDut-Stl N-ter is identical to that in the f 11Dut complex, although its placement is not restricted, because this Dut lacks motif VI. In contrast, the absence of motif VI is exploited in the complex with mDut where Y116 presents, in the different Stl subunits, rotameric conformations alternative to that observed in the f 11Dut and hDut complexes. As in phage and human complexes, Y116 from some Stl protomers interacts with mDut D109 and R110, but in other Interkingdom dUTPase Inhibition by the Stl Repressor Microbiology Spectrum protomers, the rotamer is identical to that observed in the Stl free form, losing these interactions and forming new ones with L67 and R70 in motif II (Fig. 3C). Intermediate conformations between the two rotamers are also observed in some protomers. While these changes confirm the versatility of Stl in mimicking nucleotide interactions, they could also explain the Stl affinity differences shown for the different Duts. Moreover, differences in affinity could also be explained by peripheral interactions provided by the Stl residues in the DBD and helix a5, which vary according to the recognized Dut (Tables S2 and S3).
Motif V is an active element in Stl binding. The participation of motif V of trimeric Duts in the recognition and binding to Stl has been controversial (7,16,23). Our previous work with trimeric Duts from S. aureus phages showed that the level of participation of motif V varies between Duts (16). The recently reported lvDut-Stl N-ter complex shows the first example of how the Dut C-terminal end would recognize Stl (22). In the two lvDut-Stl N-ter complexes present in the crystal structure, the C-terminal tail is visible in 4 of the 6 Dut subunits, projected from the Dut core to reach the Stl and introducing a Phe residue from motif V into a hydrophobic pocket located between the DBD and middle Stl domain. In the hDut-Stl N-ter complex, all the Dut C-terminal tails are well ordered, showing a similar mechanism of Stl recognition and complex stabilization by introducing a conserved Phe from Dut motif V in the Stl interdomain hydrophobic pocket ( Fig. 4 and Fig. S8). However, C-terminal Dut tail binding is a consequence of the interaction of more than one residue, since the Stl repressor exploits residues from loop a4-a5 (G66, I67, and P68) and the helix a7 (Y98, S99, N102, K103, and N107) to interact with the main and side chains of different Dut motif V residues (D127, R130, G133, G134, and G136) ( Fig. 4 and Table S3). On binding to Stl, the Dut C-terminal tail shows a totally different placement from that observed for nucleotide recognition. In order to allow this drastic change of the C-terminal tail direction, the flexibility provided by several Gly residues and the pivoting interactions of Arg130 at the beginning of the tail are fundamental. Therefore, Stl recognition and binding are brought about by multiple conserved residues from motif V in different ways, anchoring the C-terminal tail by the Phe, providing flexibility by the Gly residues, and being a hinge through the Arg residue (Fig. 4).
In contrast, no density attributable to the C-terminal tail was found in the mDut-Stl N-ter complex. The main difference between hDut and mDut in this region corresponds to the replacement of the anchoring Phe by a His (H135), a polar residue whose accommodation in the Stl hydrophobic pocket is unfavorable energetically. Therefore, we hypothesized that the Phe-His change eliminated the main anchor point of the C-terminal tail in Stl, preventing the participation of this structural element in receptor binding. In order to confirm our proposal, we generated an mDut mutant where H135 was replaced by a Phe (mDut H135F ) and analyzed its binding to Stl. It is worth noting that this position in motif V is occupied by an aromatic residue, typically Phe, which stacks over the uracil ring, and that previous replacement of the mDut His135 by another aromatic residue (Trp) had minimal effect on the dUTPase activity (18). Using the BLI technique, a K D of 12.45 nM was calculated for the mDut H145F binding to Stl N-ter , representing a 3-fold improvement in affinity relative to wild-type mDut (Table 1 and Fig. S3). This gain-of-function supports the participation of motif V in the binding once the anchoring Phe is introduced. Therefore, these results put an end to the controversy about the involvement of the C-terminal Dut domain (motif V) in the recognition and binding to Stl, highlighting the conserved Phe in motif V as a key actor in this process.
Stl is a universal Dut binder. Monomeric Duts originated from trimeric Duts by gene duplication and in tandem fusion, and because of that, both families exhibit highly conserved active centers and catalytic mechanisms (24,25). We therefore considered whether Stl would also recognize monomeric Duts, thus being a universal binder of Duts. For a deeper insight into this hypothesis, we followed an in silico approach to produce a model of an Stl N-ter -monomeric Dut complex by taking advantage of the structural information available in the PDB and that provided here. Since Stl mimics the nucleotide to interact with trimeric Duts, we exploited the conservation in the catalytic mechanism between monomeric and trimeric Duts to generate the model by superimposing the nucleotides present in the active centers of both Dut types. We selected the monomeric Dut from Epstein-Barr virus ( EBV Dut; PDB ID 2BT1 [25]), and the trimeric Duts from M. tuberculosis (PDB ID 1SIX [20]) and humans (PDB ID 2HQU [26]), since their complexes with Stl are presented in this work and are representative of prokaryotic and eukaryotic Duts, respectively. Once the active centers of monomeric and trimeric Duts were aligned by superimposing their nucleotides, confirming their structural relationship (Fig. S9), Stl was positioned on the monomeric EBV Dut by superimposing the Dut component of the corresponding trimeric Dut-Stl N-ter complexes. Regardless of the trimeric Dut used (mDut or hDut) to position Stl on the monomeric Dut, the results obtained were similar and no steric problems were observed in any of the complexes (Fig. 5A). The models show that Stl inserts its a8 helix into the EBV Dut active center and that Stl could also exploit its Tyr (Y105, Y112, Y113, and Y116) residues to recognize conserved catalytic residues of the monomeric Dut, highlighting the interactions of Y112 and Y113 with the catalytic Asp (D76) and the uracil ring recognition Tyr (Y73) residues of EBV Dut ( Fig. 5A and Table S4). Likewise, the EBV Dut-Stl N-ter models show that other Stl interactions with conserved residues of motifs II and IV in trimeric Duts are also replicated for monomeric Dut (Table S5). In this way, Stl D117 mediates a salt bridge with EBV Dut R280 equivalent to that observed with Lys/Arg from motif IV in trimeric Duts, and EBV Dut R171 and S172 also replicate the interactions with Stl provided for identical residues (mDut R64/S65 and hDut R62/S63) in trimeric Dut motif II. Although some conserved Stl-trimeric Dut interactions are not replicated in the EBV Dut-Stl N-ter complex (e.g., the Stl S114 interaction with a positively charged residue in trimeric motif II), the majority of interactions are replicated (Table S5), supporting that Stl can bind and inhibit monomeric Dut. Indeed, the Stl positioning is appropriate for the EBV Dut C-terminal P-loop, which is highly flexible as deduced from its lack of density in the PDB available structures, to recognize Stl in a manner similar to that observed in the hDut-Stl N-ter complex. To experimentally check this inhibitory capacity, we produced and purified EBV Dut and analyzed its dUTPase activity in the presence and absence of Stl N-ter . The enzymatic assays showed a decrease in dUTPase activity of about 60% in the presence of Stl N-ter (Fig. 5B). A comparable inhibition (40 to 80%) of dUTPase activity induced by Stl on mDut, hDut, or D. melanogaster Dut was previously reported (13)(14)(15), supporting a monomeric-Stl interaction similar to that observed with these trimeric Duts.
Finally, in order to evaluate the possible universality of Stl as a Dut binder, we performed a sequence analysis of trimeric and monomeric Duts, which showed in most cases that the positions in the Dut motifs mediating interactions with Stl are highly conserved, with the exception of motif I, whose contribution to binding is minimal (Fig. 6). It should be noted that the inclusion of monomeric Duts, which have a reshuffled sequence relative to the trimeric Duts (25), means that only a part of the catalytic motifs can be correctly aligned, decreasing the degree of conservation. The conservation is especially high in residues from motifs II and III, with more variation in motif IV residues, which, as our previous in vitro and in vivo results have shown (7,11,16), could modulate the affinity between Duts and Stl. This high conservation supports that Stl can recognize and interact with a wide range of trimeric and monomeric Duts, proposing this protein as a universal Dut binder. Given the inhibitory nature of this interaction, the results shown here support the proposed use of Stl as a proteinaceous inhibitor of Duts (15), which could be shared by Stls from other SaPIs that have high sequence homology in their middle and N-terminal domains (6).

DISCUSSION
We had previously demonstrated that different phage dUTPases, which are structurally unrelated, are able to interact with the SaPIbov1 Stl repressor and had deciphered the particularities of the molecular mechanism of recognition for each of them (6,9,11). However, this capacity is not limited to Duts from phages, since it has been shown that other trimeric Duts from distant organisms in the "tree of life" are also able to interact with Stl (13)(14)(15). Given the exceptional capacity of Stl to interact with proteins that maintain identical enzymatic activity only but differ in their structure, we wondered if the molecular mechanism used to recognize trimeric Duts from prokaryotes and eukaryotes would be similar to that used to recognize Duts from phages.
The complexes of Stl with the mycobacterial and human Duts show that Stl recognizes and binds to the Duts by exploiting interactions with almost all the conserved catalytic residues of these enzymes, and thus, Stl is a universal Dut repressor in multiple organisms along different directions of the tree of life. The human-Stl complex confirms our previously proposed and recently observed implication of the C-terminal motif V in trimeric Dut binding. Contradictory results derived from analyses performed with different Duts from different S. aureus phages, 80a and f 11, led to disagreements about the involvement of motif V in the recognition and binding to Stl. The structural data presented here and those recently described for the Stl in complex with Dut from L. vannamei (22) confirm and provide visualization of our previously proposed model. For this purpose, Stl exploits interactions with one of the most conserved residues of trimeric Dut motif V, a phenylalanine that is crucial in the catalytic mechanism by positioning itself over the uracil ring once the nucleotide is bound, stabilizing the P-loop. This phenylalanine is inserted into a hydrophobic pocket of Stl generated by the interface between its N-terminal DNA-binding domain and the middle domain, which recognizes the Dut nucleotide binding pocket. Although this Phe is highly conserved among trimeric Duts, which strongly supports Stl as a possible universal inhibitor of these enzymes, some variations in this position have been observed, as is exemplified by the Dut of M. tuberculosis. In this case that residue is a histidine, which could penalize its affinity for Stl, since its positioning in the hydrophobic pocket is energetically unfavorable. We confirmed this hypothesis by changing the His of M. tuberculosis Dut in this position to a Phe, and it showed that the mutant Dut has 3-times-higher affinity for Stl than the wild-type Dut. This result definitively confirms the participation of motif V in the binding of trimeric Dut to Stl and is further evidence of the high level of selection that Stl has followed to recognize most of the catalytic residues of these Duts in such a way as to prevent the generation of Dut escape mutants while retaining enzyme activity. Therefore, this sophisticated mechanism of Stl confirms the evolutionarily favorable strategy of binding by mimicking the target substrate and not by the recognition of a specific domain or sequence, which guarantees the success in SaPI derepression and dissemination.
At this point, we contemplated another relevant aspect. Of the three families of Duts (monomeric, dimeric, and trimeric), the binding strategy of Stl had been proven for both dimeric and trimeric forms (11), and given that trimeric and monomeric Dut catalytic motifs are highly conserved, we considered the possible spreading of the mimicry strategy of Stl to monomeric Duts. These Duts are specific to herpesviruses, which have only vertebrate hosts; therefore, Stl could not have been evolutionarily selected to recognize these Duts, as happened with the trimeric and dimeric forms. However, the structural and mechanistic relationship between trimeric and monomeric Duts (25), as well as the versatility shown by Stl in binding structurally unrelated proteins, opened up the possibility for Stl-monomeric Dut interaction. Our in silico study generating a structural model of the Stl-monomeric Dut complex, together with deep sequence analysis, supported the versatility shown by this repressor. Our model shows no steric hindrance in the binding of Stl to monomeric Duts and a similar mimicry mechanism, exploiting interactions with catalytic residues in the Dut active center. Four Stl tyrosines (Y105, Y112, Y113, and Y116) mimic the dUTP substrate interacting with conserved residues in monomeric Duts. Indeed, analysis of more than 10,500 sequences of Duts revealed that catalytic residues from four of the five motifs (Arg and Ser from motif II; Val, Ile, Asp, Tyr, and Arg from motif III; Asp and Arg from motif IV; Ile, Gly, and Phe from motif V) are highly conserved in both monomeric and trimeric Duts. Finally, we experimentally confirmed that the monomeric Dut for Epstein-Barr virus is enzymatically inhibited by Stl in a manner similar to that of other prokaryotic and eukaryotic trimeric Duts, supporting the universality of Stl as a direct competitor of dUTP and a natural inhibitor of Duts.

MATERIALS AND METHODS
Gene cloning. General DNA manipulations were performed using standard procedures. The gene encoding dUTPase from M. tuberculosis was amplified from genomic DNA (strain H37Rv) using the primers mDut-Fw and mDut-Rv (see Table S1 in the supplemental material). The PCR product was cloned in pETNKI 1.1 vector (NKI Protein Facility LIC vector system) previously digested with KpnI (Fermentas) and treated with T4 DNA polymerase (New England Biolabs [NEB]). The mutant mDut H145F was designed and made using the Q5 site-directed mutagenesis kit (NEB) and the primers mDutH145A-Fw and mDutH145F-Rv (Table S1). The synthetic gene encoding the common region of nuclear and mitochondrial isoforms of the human dUTPase (UniProtKB access code H0YNW5, Fig. S10) with the overhangs hDut-FW and hDut-RV was manufactured by the IDT Company and was subsequently cloned into a pET28a vector using the NEBuilder HiFi DNA assembly cloning kit (NEB) and the primers pET28A-FW and pET28a-RV (Table S1).
Expression and protein production. mDut and its mutant mDut H145F were produced in Escherichia coli BL21(DE3) in LB medium supplemented with kanamycin at 33 mg/mL. hDut was produced in E. coli Rosetta in LB medium supplemented with kanamycin and chloramphenicol, both at 33 mg/mL. Cultures were grown at 37°C up to an optical density at 600 nm (OD 600 ) of ;0.5 when the protein expression was induced with 0. When required, His tag from purified Duts was removed by digestion using glutathione S-transferase (GST)-tagged PreScission protease at a molar ratio of 1:50 (protease/Dut) at 4°C for 16 h and then loaded onto tandem His-Trap and GST-Trap columns to remove the undigested proteins and the GST-protease from the sample. Purified Duts were concentrated through an Amicon Ultra system (30-kDa cutoff) and further purified by size exclusion chromatography. The proteins were loaded onto a preequilibrated Superdex S75 column (GE Healthcare) with buffer C (0.1 M Tris-HCl, pH 8, 0.4 M NaCl for mDut and mDut H145F and 0.02 M Tris, pH 8, 0.125 M NaCl, 1 mM MgCl 2 for hDut). After an isocratic elution, fractions were analyzed by SDS-PAGE, and those with the highest purity were concentrated, flash frozen in liquid nitrogen, and stored at 280°C.
The proteins f 11Dut and Stl N-ter (residues 1 to 156) were expressed and purified as previously described (11).
Native PAGE. A fixed concentration of Stl N-ter (11 mM) was mixed with increasing concentrations of Dut from a molar ratio of 0.5:1 to that of 4:1 (Dut/Stl N-ter ) in buffer containing 75 mM HEPES, pH 7.5, 250 mM NaCl, and 5 mM MgCl 2 and incubated overnight at 4°C. Samples were loaded in 8% polyacrylamide native gels. Gels were prerun in 25 mM Tris-HCl, pH 8.3, 1.44% (wt/vol) glycine buffer, at 4°C, 150 V for 1 h. Soon after, samples were loaded and the electrophoresis was performed in the same buffer and at the same temperature and voltage for 135 min for f 11Dut and 210 min for hDut and mDut.
Biolayer interferometry assays. Binding studies were performed on the Octet system (Sartorius). All kinetic assays were performed in freshly prepared and filtered buffer containing 75 mM HEPES (pH 7.5), 250 mM NaCl, 5 mM MgCl 2 , 1% bovine serum albumin (BSA), 0.005% Tween 20, and 10 mM imidazole at 28°C. The samples were dispensed in dark 96-well polypropylene plates and placed on the shaker at 1,000 rpm. Sequentially, His-tagged Duts (at 218.7 nM) were captured on nickel-nitrilotriacetic acid (Ni-NTA) biosensors for 280 s followed by one step of 200 s in buffer to remove Dut excess and 60 s to establish the baseline. For the association, the Dut captured was exposed to decreasing concentrations of untagged Stl N-ter , from 218.7 nM to 3.4 nM in serial one-half dilutions, for 120 s. Dissociation was performed in buffer for 120 s. Data were analyzed using the Octet Data Analysis HT program (Sartorius) and fitted to a 1:1 kinetic model. dUTPase activity assay. A Malachite Green phosphate assay was used to analyze the inhibitory capacity of Stl N-ter on the dUTPase activity by quantifying P i released (27,28). Assays were carried out in 50 mL of reaction buffer containing 20 mM HEPES (pH 7.5), 250 mM NaCl, 10 mM MgCl 2 , 10 mM imidazole, and 0.0025 U of inorganic pyrophosphatase (Thermo Scientific). The dUTPase activity of 0.3 mg of EBV Dut was measured, and 4 mg of Stl N-ter (molar ratio of 1:20, dUTPase/repressor) was included to test the inhibitory activity. In the inhibition assays, Stl and Dut were incubated for 15 min at room temperature before the reaction. The reactions were started by the addition of 15 mM dUTP, samples at different time points (0 to 5 min) were taken, and reactions were stopped by adding 200 mL of acidic Malachite Green solution. After 20 min of incubation at room temperature, the absorbance at 630 nm was measured and the P i production was calculated based on a standard curve of P i included in the assay.
Crystallization and data collection. Crystals from both complexes were grown as sitting drops at 21°C with a vapor-diffusion approach. Initial crystallization attempts were set up in the crystallogenesis service of the IBV-CSIC using commercial screening assays JBS I and II (Jena Biosciences) and JCSG1 (Molecular Dimensions) in 96-well plates. Complexes were formed by mixing untagged mDut or His-tagged hDut with untagged Stl N-ter , in a 1-to-1 molar ratio (calculated for the monomer), and mixtures were incubated at 4°C overnight prior to the crystallization plate setup. Crystallization drops were generated by mixing equal volumes (0.3 mL) of sample and the corresponding reservoir solution and were equilibrated against 100 mL reservoir solution. The mDut-Stl N-ter complex was crystallized at 17 mg/mL in a reservoir solution consisting of 20% polyethylene glycol (PEG) 10000 and 0.1 M Na-HEPES, pH 7.5. Crystals were cryoprotected with a solution consisting of reservoir solution increased up to 35% PEG 10000 and supplemented with 10% ethylene glycol. The hDut-BovI-Stl N-ter complex was crystallized at 3 mg/mL in a reservoir solution of 1.6 M (NH 4 ) 2 SO 4 and 1 M Li 2 SO 4 , and crystals were cryoprotected with reservoir solution supplemented with 5% glycerol and 10% ethylene glycol.
Diffraction data were collected from single crystals at 100 K on the ALBA (Barcelona, Spain) and DLS (Didcot, UK) synchrotrons. Processing and reduction were done with iMosflm (29) and Aimless (30) programs (CCP4 suite [31]). Data collection statistics are shown in the crystallography table (Table 2).
Model building. Crystallographic phases for mDut-Stl N-ter and hDut-Stl N-ter complexes were obtained by molecular replacement using PHASER (32) and the structures of mDut (PDB ID 1MQ7 [20]), hDut (PDB ID 1Q5U [21]), and Stl N-ter (PDB ID 6H49 [11]) as searching models. Final models were generated by iterative cycles of refinement using Refmac (33) and manual rebuilding and optimization with Coot (34). Maximum likelihood was applied in each cycle of restrained refinement, using automatic weighting and experimental sigmas to weight X-ray terms and excluding 5% data for R free calculation. The temperature factors were refined as isotropics. TLS refinement was applied in the last cycles of hDut-Stl N-ter complex refinement. Data refinement statistics are given in the crystallography table (Table 2). Atomic coordinates and structure factors have been deposited in the Protein Data Bank with identification codes 7PWX for mDut-BovI-Stl N-ter and 7PWJ for hDut-BovI-Stl N-ter .
Structural analysis. The CCP4 suite (31) and PISA server (35) were used to analyze interactions in the complexes and to superimpose the structures. The omit map from domain V on the hDut-Stl N-ter complex was calculated using the Phenix suite (36). The residues from domain V (R130 to G136) were the subject of omit selection specifying 3 Å as the solvent exclusion radius.
Structural modeling of the EBV Dut-Stl N-ter complex. The structural model EBV Dut in complex with Stl N-ter was generated by the following steps. As a representative monomeric Dut, we used the structure of EBV Dut in complex with the dUTP analogous nucleotide a-b-imino dUTP (PDB ID 2BT1). The Coot program (34) was used to superimpose the nucleotide in this structure with the nucleotides inside one of the active centers of the structures of mDut (PDB ID 1SIX) or hDut (PDB ID 2HQU). The conformation of nucleotides in all the structures is almost identical. In this way, the structure of the monomeric EBV Dut was aligned according to its substrate recognition with both trimeric Duts. Then, each one of these trimers of Dut was superimposed with the corresponding trimer of Dut in complex with Stl N-ter (mDut-BovI-Stl N-ter or hDut-BovI-Stl N-ter ), so that Stl was positioned on the active site of EBV Dut with the supposed recognition conformation. At this point, the EBV Dut-Stl N-ter models were finished by eliminating the structures of the corresponding trimeric Duts, creating one EBV Dut-Stl N-ter complex for the Stl N-ter from mDut-Stl N-ter and another for that from hDut-Stl N-ter . Energy minimization of both models was then performed using the Yasara energy minimization server, using the default Yasara force field and minimization values (37). Since the two EBV Dut-Stl N-ter models are very similar, the one generated with Yasara with the Stl N-ter from the mDut-Stl N-ter complex structure was used to produce the figures presented in this study.
Sequence alignment and logo diagram. In order to determine the amino acid profile and sequence variability of monomeric and trimeric Dut sequences, we performed the following analysis. Dut sequences from the Pfam database under ID PF00692 (14,635 sequences), which include monomeric and trimeric Duts, were downloaded and computationally distributed based on their length. Sequences showed a size mainly between 125 and 225 residues, so those sequences outside this range (generally corresponding to Dut fragments or Dut domains fused to other proteins) were eliminated. The size analysis also showed a bimodal distribution with peaks centered on 150 and 190 residues. Inspection of representative sequences of these peaks indicated that the first one corresponds to trimeric Duts and the second to monomeric Duts plus trimeric Duts carrying insertions (e.g., signaling peptides). Finally, to avoid overrepresentation of certain families, homologs were clustered using a 95% identity coverage with MMseqs2 version 12.113e3 (38). One representative homolog per cluster was selected, resulting in a final data set of 10,616 sequences that were aligned using the MAFFT program version 7.475 (39), and the resulting alignment was visualized using the SeaView program version 5.0.4 (40). The high level of diversity in sequences as a consequence of the large number of proteins included involved empty columns across the alignment. For adding reliability to the resultant logo of the alignment, these empty columns were removed using the "gappyout" parameter of the trimAl tool version 1.4. rev14 (41), alignment size-dependent guideline. This sequence alignment was uploaded to the Skylign webserver (42) to obtain the corresponding logo, which is a graphical representation of sequence conservation where the frequency of each amino acid residue at the specific position is proportional to the height of its symbol. In the graphic, the presence of a single letter indicates that the residue is invariable, whereas the presence of different letters corresponds to the most common amino acids.
Data availability. Atomic coordinates and structure factors have been deposited at the RCSB Protein Data Bank (PDB) (PDB code 7PWX for mDUT-Stl N-ter and 7PWJ for hDUT-Stl N-ter ).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 3.5 MB.