Unique structural and ligand-binding properties of the Staphylococcus aureus serine hydrolase FphE

Significance

Staphylococcus aureus is a harmful bacterium that can form antibiotic-resistant biofilms, making infections harder to treat. An enzyme in this bacterium, called FphE, is unique to S. aureus and plays a crucial role in its pathogenicity, making it a promising target for new treatments and imaging techniques. In this work, we show that FphE has an unusual structure, forming a stable cross-subunit dimer that makes it more stable and efficient than a typical monomer. We present the structures of this dimer in complex with multiple chemically distinct compounds. These structures allow us to better understand FphE function and lay the groundwork for creating next-generation specific inhibitors and imaging tools to diagnose and treat S. aureus infections more effectively.

Abstract

Staphylococcus aureus is a human pathogen capable of forming biofilms that complicate treatment and facilitate chronic infections. A family of S. aureus serine hydrolases are important regulators of virulence and biofilm formation. Among these, FphE is highly specific to S. aureus and therefore a viable target for both imaging and therapy. Here, we present bioinformatic and structural evidence that FphE may be involved in aromatic compound metabolism. In addition, 12 distinct crystal forms reveal that FphE exists as a highly unusual but stable and flexible, cross-subunit homodimer, unique to the large alpha/beta hydrolase superfamily. Substrate engagement favors retention of the dimeric state, which is a more catalytically active form of the enzyme, and small-angle X-ray scattering confirms that the dimeric architecture occurs in solution. High-resolution cocrystal structures of FphE covalently bound to two chemically distinct ligands reveal different modes of active site engagement, supporting an atypical structural plasticity of the dimer interface. Together, these findings establish FphE as a structurally unique alpha/beta hydrolase and provide a foundation for structure-guided development of S. aureus–specific inhibitors and imaging probes.

Worldwide, Staphylococcus aureus (S. aureus) causes a variety of diseases ranging from local skin or soft tissue infections to invasive chronic infections such as bacteremia, pneumonia, or endocarditis (1, 2). Chronic infections are often linked to the ability of S. aureus to form layers of cells embedded within a biomolecular matrix, known as a biofilm (3, 4). These robust bacterial structures can form on synthetic surfaces (prosthetic implants) or heart valves, making infections more complicated and costly to monitor and treat (5). Biofilms also offer S. aureus a relatively protected environment from antimicrobial agents and the host immune response (6). As S. aureus remains a major human health threat, new methods for rapid detection and therapeutic response monitoring are needed.

Activity-based probes (ABPs), which incorporate weak electrophiles to selectively react with activated amino acid residues, enabled global mapping of active enzymes across diverse biological systems in situ (7). Our recent study applying a broad-spectrum fluorophosphonate (FP)-based ABP in live S. aureus facilitated the discovery of a previously uncharacterized family of serine hydrolases, termed fluorophosphonate-binding hydrolases (Fphs) that are involved in biofilm formation and virulence (8–18). Serine hydrolases belong to the alpha/beta hydrolase superfamily, with the serine hydrolase subfamily themselves being one of the largest enzyme families distributed across all kingdoms of life (19). They are highly druggable and execute a wide range of biological functions via hydrolytic substrate cleavage of ester, thioester, or amide bonds, allowing them to act as proteases, peptidases, lipases, esterases, and amidases. Cleavage is achieved via an active site triad made up of Ser-His-Asp/Glu containing the highly nucleophilic serine residue (20). This serine can be targeted by activity-based probes containing a reactive electrophile to label the active enzyme in vitro, in cells and even in whole organisms (7).

FphE (SAV2581 or Hy30) is a 31 kDa serine hydrolase that lacks homologs in humans or other mortality-associated bacteria aside from Listeria monocytogenes (10, 14). FphE has emerged as a promising target for the development of imaging probes that can specifically label S. aureus via covalent binding to the active site serine. We previously identified triazole urea-based covalent inhibitors for several Fphs, among them FphE (8). Optimization of an initial hit resulted in a fluorescent activity-based probe that showed FphE-specific labeling that was highest in cells growing under biofilm-forming conditions (8). In addition, FphE levels were found to be dynamic and subject to the bacterial growth state in a strain-specific manner.

FphE can be covalently targeted by β-lactam probes (21) and by a carmofur-based probe (22). Recently, screening of an oxadiazolone-based library and subsequent lead molecule optimization identified a novel antibiotic compound, named “compound 3” that is active against various S. aureus strains (23). This compound was shown to covalently bind to a set of ten S. aureus proteins, which included FphE and FphH; however, antibiotic potency was not linked to the ability of the compound to bind to the Fphs. We recently described development of ABPs for FphE using the oxadiazolone electrophile of compound 3 as a starting scaffold (14). These oxadiazolone-based ABPs were highly specific for FphE and validated it as a target for the development of imaging contrast agents for the rapid and selective detection of S. aureus infections. These studies confirmed that FphE is intracellularly located and can be targeted in live bacterial cells without causing cytotoxicity. Furthermore, we demonstrated that a fluorescently labeled version of an oxadiazolone probe can be used to noninvasively image S. aureus infection sites in mice.

X-ray crystal structures of FphE in the unbound form or with the oxadiazolone covalently bound in the active site demonstrated a highly unusual intersubunit homodimer where the catalytic triad of each monomer is composed of components of each of the two monomers (14). Interestingly, we recently found that FphE inhibitors bearing a boron warhead generate a covalent tetrahedral adduct with the catalytic serine from one monomer and the histidine from the other monomer, thereby preventing dimer dissociation of FphE (18).

Here, we present detailed structural features and the molecular basis for ligand selectivity of FphE. Our bioinformatic analysis reveals FphE is genomically linked to enzymes predicted to process aromatic substrates, providing a rationale for its preference for aromatic probes. Biochemically, we demonstrated that FphE forms a homodimer that has increased catalytic activity and is stabilized upon substrate engagement. Through the analysis of 12 distinct crystal forms including cocrystal structures with multiple ligands, we demonstrate an overall structural plasticity of this unique homodimer. We also identify two distinct binding pockets including the previously reported hydrophobic tunnel and an adjacent pocket near the catalytic site. While this unique dimer architecture enables ligand-dependent shifts in binding modes, bioinformatic analysis and our cocrystal structures reveal that selectivity is dictated not by the conserved active site, but by engagement of the less-conserved, peripheral hydrophobic tunnel. Together, these insights provide a molecular rationale for FphE selectivity and establish a framework for the rational design of S. aureus-specific probes.

Results

Taxonomic Distribution and Genomic Context of FphE Suggest Association with Aromatic Compound Metabolism.

To understand the biological role of FphE, we searched for homologs using a position-specific iterative BLAST (PSI-BLAST) querying FphE sequence against the National Center for Biotechnology Information nonredundant protein database (SI Appendix, Fig. S1 and Dataset S1). To ensure high structural and sequence similarity while minimizing redundancy, we excluded sequences with <40% identity and clustered the remaining 1,619 proteins at a 95% identity cutoff, yielding 119 representatives as a final set. Interestingly, all proteins are encoded within species from the phylum Bacillota, with L. monocytogenes (N = 5) being the only mortality-associated species (24), apart from S. aureus (N = 12; Dataset S2). This highly restricted distribution highlights the potential of FphE to serve as a selective imaging and diagnostic target for S. aureus infections (14). The closest homologs outside Bacillota are mainly encoded by fungi, suggesting a specialized role for this protein within the phylum (SI Appendix, Fig. S1). We found that fphE is conserved across diverse S. aureus strains under distinct gene names (e.g., SAUSA300_2518, strain USA300; NWMN_2480, strain Newman; SA2367, strain N315; SACOL2597, strain COL; SAR2661, strain MRSA252; HMPREF0769_10569, strain MN8; SAV2581, strain Mu50/ATCC 700699) and that its homologs are differentially regulated under various environmental, metabolic, and host-associated stresses (17, 25–37), although the underlying functional relevance of this regulation remains unclear (SI Appendix, Table S1 and Dataset S3). In L. monocytogenes, transcription levels of fphE were increased in a nonvirulent strain and in a virulence factor knockout (ΔprfA), indicating a potential role in pathogenicity (38–41). These findings collectively suggest that FphE is a taxonomically restricted enzyme closely related to stress-related pathways in S. aureus, supporting its potential as a species-selective biomarker.

To further explore the biological function of FphE, we examined its and its homologs’ genomic loci to identify possible conserved pathway associations (Fig. 1 A and B and SI Appendix, Fig. S2). In S. aureus USA300, fphE (SAUSA300_2518) is encoded in a seven-gene operon (SAUSA300_2512–2518) associated with biofilm regulation (Fig. 1A) (42). This operon is divergently transcribed into two mRNAs (SAUSA300_2512–2514 and SAUSA300_2515–2518), with the transcription of both controlled by a TetR-family repressor GbaA (SAUSA300_2515; glucose-induced biofilm accessory protein A) binding to the shared operator (Fig. 1A) (42). The covalent modification of GbaA by electrophiles (N-ethylmaleimide and methylglyoxal) induces these operons, and a double mutant lacking both gbaA (also known as rob) and fphE showed no additional effect on biofilm formation compared to the gbaA single mutant in S. aureus strain FK300 (42, 43). Comparative genomic analysis revealed that the latter four genes of the operon that include fphE (SAUSA300_2515–2518) are almost exclusively conserved across Staphylococcus species (23 out of 55 representative loci) with only one example outside the genus Staphylococcus (in Macrococcus equipercicus ATCC 51831; Fig. 1B and SI Appendix, Fig. S2). Interestingly, we found that fphE and an upstream gene for an amidohydrolase (SAUSA300_2517) are genomically colocalized in a majority of the representative loci across Staphylococcus and other bacterial species, suggesting possible cooperation within a biochemical pathway (80 amidohydrolases in 59 out of 119 analyzed loci; Fig. 1B and SI Appendix, Fig. S2). Moreover, in two S. aureus strains (isolate P15 and st1226), fphE and the upstream amidohydrolase gene are fused as a multidomain protein (RefSeq accessions HCY5833117.1 and CXW27440.1; SI Appendix, Fig. S2). While the crystal structure of a nearly identical amidohydrolase from a different S. aureus strain (SAV2580; 99% sequence identity) is available (PDB 3NUR), its biochemical function remains uncharacterized.

Fig. 1.

FphE is associated with aromatic compound metabolism. (A) Genomic locus of the S. aureus USA300 fphE (SAUSA300_2518). Protein names or predicted domains are given above and gene names are given below arrows. (B) A phylogenetic tree of 119 representative FphE homologs. Genomic loci are shown, and arrows are colored by their homology. Names of all species are shown in SI Appendix, Fig. S2. (C) A sequence logo for Y298 of the amidohydrolases (SAUSA300_2517) and 79 homologs encoded near genes for FphE homologs in (B). (D) Active sites of the AlphaFold2-predicted amidohydrolase (SAUSA300_2517) (Left) and Aspergillus oryzae 2,3-dihydroxybenzoate decarboxylase (2,3-DHBD) F296Y (PDB 7WKL; Right) (44, 45). (E) Inhibition potency and selectivity of JJ-OX-003 and JJ-OX-004 toward the catalytic serine (Ser103) of FphE (14).

Sequence alignment of the amidohydrolase SAUSA300_2517 and its 79 homologs identified across 50 of the 119 loci revealed a highly conserved tyrosine residue (Y298), along with putative metal-binding residues (E7, H171, and D295; Fig. 1C and SI Appendix, Fig. S3A). Structural homologs (PDB: 7WKL, 7WMB, 6JQX, and 7BPC) of the AlphaFold model of SAUSA300_2517 utilize corresponding aromatic residues (tyrosine or phenylalanine) to engage π–π interactions for recognizing the benzene ring of their aromatic substrates, such as catechol, salicylic acid, and 2,5-dihydroxybenzoic acid (Fig. 1D, SI Appendix, Fig. S3 B–E, and Dataset S4) (44, 46, 47). Consistent with this, in our previous work, we observed that FphE was efficiently labeled by probes that contain an aromatic ring (14). In particular, the probe JJ-OX-004 that contains an aromatic ring between the lipid chain and oxadiazolone warhead showed selectivity toward the catalytic serine (Ser103) of FphE with dramatically improved potency compared to JJ-OX-003, which lacks this aromatic feature (Fig. 1E). We previously demonstrated that FphE has carboxylesterase activity against fluorogenic substrates, with preference for 4-methylumbelliferyl caprylate (4-MUC) (14). In the present study, we further show that FphE lacks protease activity (SI Appendix, Table S2). Taken together, these observations suggest that FphE as an esterase and the upstream amidohydrolase may participate in a shared biochemical pathway for modifying aromatic substrates.

FphE Forms a Homodimer in Solution.

FphE has a predicted molecular weight of 31 kDa and SDS-PAGE analysis of probe labeled S. aureus cell lysates, as well as recombinantly produced FphE, confirm that the protein resolves at this predicted molecular weight (8, 14, 17). However, when recombinant FphE was subjected to size exclusion column chromatography, two distinct but relatively equal peaks were observed that resolve as the same molecular weight (~31 kDa) when analyzed by SDS-PAGE analysis (Fig. 2A). We subjected the purified protein peak 1 and peak 2 to native-PAGE under nonreducing conditions and observed two distinct protein bands, a dimer corresponding to peak 1 and a monomer corresponding to peak 2 (Fig. 2B). We observed no interconversion between the monomer and dimer upon changing the concentration of either form. Although each form contained a small amount of the other as a contaminant, no change in equilibrium was observed when the samples were progressively diluted to lower protein concentrations and directly analyzed by native-PAGE. We tested whether labeling by our FphE-selective probe JJ-OX-007 containing a BDP-TMR fluorophore (14) affects the oligomeric state of FphE (Fig. 2C), as well as examined whether the oligomeric status could be modulated by addition of the lipid ester substrate, 4-MUC. We incubated the dimeric and monomeric form of FphE at room temperature for 30 min in the presence or absence of substrate, and the samples were either treated with JJ-OX-007 or left untreated and incubated for an additional hour at 37 °C (Fig. 2D). Interestingly, we observed spontaneous conversion to the monomeric form (Fig. 2E), that was reduced by the coincubation with the 4-MUC substrate. (Fig. 2E, lanes 1 and 2, Coomassie stain). A similar trend was observed upon additional treatment with the JJ-OX-007 fluorescent probe that enabled more sensitive detection of both monomeric and dimeric forms compared to Coomassie staining (Fig. 2 E and F and SI Appendix, Table S3). To further assess whether engagement of the active site could influence the oligomeric equilibrium of FphE, we examined the effect of the boronic acid-based FphE inhibitor Z27, which is reported to form a covalent adduct with the catalytic serine of one monomer while simultaneously engaging the catalytic histidine of the other monomer that forms the active-site triad across the dimer interface (18). Incubation of dimeric FphE with increasing concentration of Z27 at 37 °C resulted in a dose-dependent increase in the relative abundance of the dimeric form (SI Appendix, Fig. S4 A and B). We also evaluated the effects of temperature on dimer-to-monomer conversion (SI Appendix, Fig. S4C). To minimize effects during substrate incubation time, we transferred the samples to the target temperatures after substrate and probe addition. As expected, the presence of the substrate correlated with a reduction in dimer-to-monomer conversion at 37 °C (SI Appendix, Fig. S4 C, Top row). However, at lower temperatures such as room temperature or 4 °C, no change in oligomeric status was observed even after 1 h incubation. Fluorescence from probe labeling decreased for both forms as the temperature decreased, which is likely due to overall reduced enzyme activity at lower temperatures (SI Appendix, Fig. S4 C, Left column). Protein from both size exclusion peaks generated the same crystal forms (see below). However, the dimer showed higher cleavage activities toward 4-MUC model substrate than the monomer (SI Appendix, Fig. S5).

Fig. 2.

Biochemical characterization of FphE oligomeric states and probe labeling. (A) Size exclusion chromatography of recombinant FphE revealing two major peaks corresponding to dimeric (peak 1) and monomeric (peak 2) forms. (B) Native-PAGE analysis of increasing protein concentration from each peak. (C) Chemical structure of FphE covalent fluorescent probe JJ-OX-007. (D) Schematic overview of fluorescent probe-based detection of monomeric and dimeric FphE. (E) Native-PAGE analysis of oligomeric status and probe labeling of FphE monomer and dimer. (F) Quantification of the native-PAGE in Fig. 2E. Values are shown as mean ± SD (n = 2) from independent experiments. 4-MUC, 4-methylumbelliferyl caprylate; ns, not significant.

FphE Crystals Assemble into an Unusual Homodimer Architecture in 12 Unique Crystal Forms.

FphE crystals grew in various conditions into a range of different crystal forms (SI Appendix, Table S4). FphE crystallized in a pH range from 5.5 to 8.5 in the presence of 15 to 30% polyethylene glycol compounds or in high salt conditions (SI Appendix, Tables S4–S6). Crystal growth was accelerated by the presence of magnesium, potassium, and calcium salts, but crystals also appeared in the absence of any additional salts. Processing revealed 12 distinct crystal forms of FphE. We arranged them roughly by their highest diffracting crystal. Crystal forms 1 and 12 were the most frequently appearing crystals, with form 1 always diffracting to high resolution (as high as 1.38 Å) and form 12 crystals never diffracting to better than 2.5 Å and often worse than ~5 Å. The form 12 crystal could not be phased using molecular replacement due to severe tNCS (translational noncrystallographic symmetry). Interestingly, these 1 and 12 crystal forms sometimes appeared in the same crystallization drop, and they could not be distinguished based on optical presentation. The remaining 10 crystal forms appeared far less frequently and were sometimes also found in the same crystal drop as forms 1 and 12. Occasionally, their occurrence appeared to be correlated with the presence of specific compounds in the reservoir mixture (SI Appendix, Table S4) or when specific ligands were cocrystallized, but no concrete pattern could be established. Crystal form 10 could also not be processed due to potential twinning issues causing complications at molecular replacement. Crystal form 11 suffered from poor diffraction patterns that prevented processing beyond merging and scaling steps. The remaining 9 crystal forms of FphE were fully processed, refined, and deposited to the Protein Data Bank (48) (SI Appendix, Tables S7 and S8).

Although hundreds of FphE crystals were processed, the monomeric form of FphE was never observed in the crystalline state with all crystal forms being a homodimer (Fig. 3). In all cases the homodimer is formed in a very unusual way where the N-terminal residues 1 to 136 of one FphE monomer are assembled with the C-terminal residues of 168 to 276 of the second FphE monomer to create two, connected “full-length” FphE molecules that may resemble the monomeric state. These monomers are connected via the remaining residues 137 to 167 which form a long bent helix that runs antiparallel to the corresponding helix from the other FphE monomer (Fig. 3A). The active site still contains the conserved catalytical triad for serine hydrolases, in this case formed by Ser103, His257, and Glu127, with Ser and Glu originating from one FphE monomer and His from the other. The hydrophobic active site pocket and access tunnel (Fig. 3 B–D) are therefore also formed by residues from both FphE monomers. Specifically, the connecting helix ~155 to 167, as well as helices 170 to 178 and 183 to 207, contain residues from both monomers. These helices form a so-called “lid” in hydrolases of a similar fold (49), including FphH (11) or FphI (15). However, the FphE conformation is rather unique in that the helices form a distinct entrance to the active site rather than a dynamic lid that may cover the active site after ligand binding. Our previous prediction of the monomer of FphE (9), as well as predictions using Alphafold2 (50) for this study, suggest that FphE has two helices in place of the long connecting dimer helices 136 to 154 and 156 to 167 (Fig. 3D). It appears that the dimer is formed when the C-terminus starting at residue 155 bends away from the core protein with residues 156 to 167 forming a long helix instead of a helix that bends backward toward the FphE monomer. However, the bend at the middle of the long connecting dimer-helix is still retained to some degree in all crystal forms and appears to be the main differentiating factor of the crystal forms. Cα alignment of the first ~1 to 136 residues of one FphE monomer demonstrates how the connecting helix differs significantly between the crystal forms (Fig. 4 A and B and SI Appendix, Fig. S6). Alignment of the N terminus results in the ends of the connecting helices (distance of Cα atom of residue 137 and 167) being between 13.1 Å and 19.2 Å across the different crystal forms. This is due to the extent of bending in the connecting helices which is characterized by a shift of the central residue 155 of the helices. Due to the differences in the bend of this linker region (SI Appendix, Table S9), the positions of the two assembled FphE monomers shift relative to each other. Alignment of one FphE copy shifts the other one by up to 25 Å in one axis and by up to 13 Å in another axis across the different crystal forms. The shifted positions of the two FphE monomers then facilitate variable crystal interfaces which results in the multiple crystal forms. These unique folds for the various helices can be influenced by ligands covalently bound to the active site serine. Some crystal forms were only observed in the presence of a specific ligand: form 2 only with oxadiazolone compound 3 (see below), forms 4 and 6 only with borolane-based compound Q41, and form 7 only with boronic acid-based compound W41 (18). The different crystal forms also influence the accessibility of the active site (SI Appendix, Fig. S7), with all crystal forms showing a similar open hydrophobic tunnel leading to the active site but with slight variations in the dimensions and arrangement of the tunnel.

Fig. 3.

FphE homodimer. (A) Ribbon representation of FphE homodimer in crystal form 1 (PDB ID 8T87, 1.62 Å) (14), FphE chain A in light cyan and chain B in pink, with active site triad Ser-His-Glu displayed in brown. N and C termini annotated; dimer interfacing helix residue numbers indicated. (B) Active site pockets illustrated as a blue surface model, identified via the FPocketWeb browser (51). (C) Protein chains and active site pockets surface representation. (D) Secondary structure schematic of predicted monomer in lavender compared to FphE homodimer in crystal form 1 (8T87) with FphE chains in blue and pink. N and C termini annotated with capital letters. Active site triad residue location noted with a black dot. For the homodimer chain A residue numbers for each helix and strand are indicated. Hydrophobic tunnel helices and loops marked with thick black lines. Helix dimer interface illustrated via dotted lines.

Fig. 4.

FphE crystal forms. (A) Ribbon representation of alignment of chain A (or C in two cases) Cα atoms of N-terminal residues 1 to 136 of each crystal form. Overlay of all crystal forms and individual pictures for each crystal form in different ribbon colors. Form 1 chain A+B dimer cyan (PDB ID 8T87, dataset resolution 1.62 Å) (14); form 2 A+A* crystal symmetry dimer blue (8G49, 1.60 Å); form 3 AA* pink (8G48, 1.95 Å); form 4 AB gray (8UWM, 1.97 Å) (18); form 5 AB magenta, CC* yellow (9COM, 2.04 Å); form 6 AB light blue, CD purple (9EBF, 2.50 Å); form 7 AB ochre (8UIX, 2.39 Å) (22); form 8 AA* green (9D87, 2.30 Å); and form 9 AB dark purple (9EDJ, 2.60 Å). Largest shift of other dimer ends between crystal forms 4 and 8 of 25 Å (C-terminus residue 276 Cα atom) indicated in red. (B) Same alignment turned by ~90 degrees. Largest shift of other dimer end between crystal forms 2 and 6 of 13 Å (residue 247 Cα atom) indicated. (C) SAXS result of FphE homodimer. Ab initio envelope (beige surface model) of FphE in solution overlaid via ATSAS-CIFSUP with FphE crystal form 1 (8T87). (D) SAXS result of FphF homodimer. Ab initio envelope of FphF in solution overlayed via ATSAS-CIFSUP with FphF crystal (6VH9, chain A in red and chain D in orange).

In some cases, the two monomers of FphE that form the homodimer are identical, resulting in a crystal form with only one chain A with an identical crystal symmetry mate A* forming the homodimer (SI Appendix, Table S9, forms 2, 3, and 8). In other cases, the homodimer is formed via chain A and chain B of the asymmetric unit (forms 1, 4, 7, and 9). In one case, the asymmetric unit contains two homodimers of chain A with chain B and a second one formed by chain C and chain D (form 6). And the final form contains a homodimer formed by chain A with chain B with a second homodimer formed by chain C and its crystal symmetry mate C* (form 5). It was especially challenging to obtain the correct molecular replacement solution for this crystal form without prior knowledge of the correct assembly.

The long antiparallel connecting helices of residues 136 to 167 set up a domain in which hydrophobic residues (Ile147, Val148, Ile151, Leu152, Leu156, Met160, and Phe163) are involved in the dimer interface, whereas charged residues (Thr138, Lys141, Asp142, Asp145, Asp146, His149, Thr153, Glu154, Glu157 and Lys161) face the opposite solvent-exposed direction. This hydrophobic interfacing and hydrophilic solvent exposed arrangement is a potential key driver of the unusual helix conformation of the dimer. Our sequence conservation analysis also indicates that the hydrophobic interfacing residues are highly conserved in the 119 FphE homolog sequences, with the noninterfacing residues showing less conservation, while mostly retaining their charged state (SI Appendix, Fig. S8 and Dataset S5). Ile151 is the least conserved among the hydrophobic interface residues, but it is frequently replaced with either valine or glycine (Dataset S5). It is plausible that other alpha/beta hydrolase enzymes with a similar conservation pattern at this region of the fold may also be able to adopt the FphE dimer configuration.

The FphE Homodimer Crystal Architecture Is Present in Solution.

While we were able to confirm that FphE crystallizes as a dimer and also resolves in solution as a dimer by size exclusion chromatography, we wanted to determine how the structure of the dimer in solution relates to the observed unique crystal architectures. Therefore, we determined the in-solution structure of FphE using Small-angle X-ray scattering (SAXS, SI Appendix, Table S10). FphE SAXS data (Fig. 4C and SI Appendix, Figs. S9A and S10) showed that the dimer peak clearly contained a dimer structure of FphE in-solution that closely matched the crystal structures of FphE. Crystal form 1 was a good match to the SAXS data with the connecting helix clearly visible. Minor deviations of the data from crystal form 1 (SI Appendix, Fig. S9A) further confirm the flexibility of the FphE homodimer suggested by the multiple crystal forms. Creation of an updated SAXS data-based model of crystal form 1 (SI Appendix, Fig. S9B) emphasized this flexibility as the overall fold remains a close match (SI Appendix, Fig. S9C) with only the middle of the connecting helices altered to better match the SAXS data. In addition, we determined the SAXS in-solution structure of FphF. Both are serine hydrolases of similar size, with both adopting a traditional alpha/beta hydrolase fold, yet they do not appear to be functionally related with only 23% sequence identity. We previously described FphF as a homodimer in solution; however, the crystal structure showed a tetramer (9). We proposed that the FphF homodimer is formed via antiparallel association of its first β-strand. Fitting such a homodimer was the best fit for the SAXS data, confirming our hypothesis (Fig. 4D and SI Appendix, Figs. S11A and S12). Creation of an updated model of the FphF crystal structure to fit the SAXS data did not result in significant changes, suggesting a less flexible and more uniform structure of FphF compared to FphE (SI Appendix, Fig. S11 B and C). While the N-terminal homodimer interface of FphF is uncommon for alpha/beta hydrolases, the FphE interface appears to be completely unique for this enzyme.

The FphE Homodimer Is Unique within the Large Alpha/Beta Hydrolase Superfamily.

FphE has a general alpha/beta hydrolase fold (Inter Pro IPR029058) (52), making it related to a group of over 3,000 proteins containing this fold in the protein data bank (PDB) (48). This overall set of structures is made up of subfamilies AB_hydrolase_1 (934 experimental determined structures), AB hydrolase superfamily (191 structures), and Pfam Abhydrolase_6 (PF12697, 179 structures). These groups are very broad and thus do not provide insights into a putative biological function of FphE. We used the FphE Alphafold 2 (50) predicted monomer structure (SI Appendix, Table S11) and dimer crystal structure (SI Appendix, Tables S12 and S13) to screen for 3D matches in the PDB (48) as well as in the predicted proteome of S. aureus using the DALI webserver (53). For the monomer, the top hit was Zearalenone lactonase (54) which structurally aligns well in the β-sheet region and some helices (129 Cα pairs at 1.2 Å rmsd) but the active site loops and helices, including lid helices differ significantly and the proteins have only 18% sequence identity. In fact, all of the top hits for the predicted FphE monomer structure have similarly low sequence identity and align structurally in a similar way with decreasing statistics of the overall alignments. Searching for structure matches to FphE in the dimer conformation only resulted in hits that loosely match the first ~140 residues with statistics indicating that these are only general alpha/beta fold hits. Thus, the unusual FphE dimer conformation appears to lack any matches to proteins in the PDB (48) as well as to predicted structures of the S. aureus proteome found in AlphaFoldDB. In addition, AlphaFold 3 (55) was also unable to predict the FphE dimer that we observed in our determined crystal structures.

The Structure of FphE with a Covalent Fluorophosphonate Probe Reveals Shared Binding Features with JJ-OX-004 Despite Distinct Electrophile Chemotypes.

FphE is covalently inhibited by electrophilic chemotypes including fluorophosphonates (FPs) (17), triazole ureas (8), oxadiazolones (14, 23), β-lactams (21), carmofur (22), and boronic acids (18). Members of the Fph protein family have been systematically profiled using the activity-based probe FP-tetramethylrhodamine (FP-TMR). To enable further structural analysis of FphE, we designed and synthesized an alkyne version of the FP probe, JB101 (SI Appendix, Scheme S1), and then determined the structure of FphE bound to this molecule. As expected, JB101 forms an irreversible covalent bond to the active site Ser103 residue from one FphE dimer copy (residues 1 to 136), with the fluorine atom of JB101 being replaced with the oxygen side chain of the serine residue (Fig. 5 A–C). The core aliphatic backbone of the ligand is located in the hydrophobic active site tunnel that we described previously in detail (Fig. 5 D and E) (14). The phosphoryl oxygen atom of JB101 occupies the well-described oxyanion hole (56, 57) of alpha/beta hydrolases, formed here via interactions with main chain amide groups of Ser104 and nearby Ala28 from the same chain as the active site serine (Fig. 5 B and F). In the case of FphE, this atom is further stabilized via hydrogen bonds to two side chain orientations of Ser104. There are many additional hydrophobic interactions along the JB101 ligand backbone. Carbon atoms near the phosphorus atoms are stabilized by backbone carbons of Gly27 and the Cβ atom of Asn29. All other interactions are formed with the other dimer copy (residues 168 to 276) of FphE which forms the active site tunnel via the end of the dimer connecting helix 156 to 166 and the following 170 to 178 and 183 to 207 helices, as well as the loop containing the active site triad residue His257 (~256 to 261). The electron density quality of JB101 atoms of the alkyl linker decreases with distance from the covalent link to the active site serine, suggesting that the ligand may have alternate conformations in the tunnel. This is a similar ligand-binding orientation to what we previously reported for the JJ-OX-004 ligand which contains an oxadiazolone electrophile (See below: Fig. 6 H and I) (14). In addition, based on our structure, the terminal alkyne group of JB101 is likely able to reach the protein surface, which explains how this probe can easily be modified via click chemistry without losing specificity for FphE.

Fig. 5.

FphE covalently bound by FP-JB101. (A) Crystal structure of FphE in complex with fluorophosphonate compound JB101 (crystal form 1, PDB ID 8SBQ, at 1.50 Å). Ribbon representation with carbon atoms of FphE chain A in light cyan, chain B in pink (helix 171.B-178.B in the foreground transparent), JB101 carbon atoms in gray, hydrogen bonds in red. Side chains of key residues displayed and labeled. The 2Fo−Fc electron density map for JB101 and Ser103 of chain A is shown as a blue mesh at 1 σ. (B) Close-up of indicated window in (A). (C) JB101 chemical structure with nucleophilic attack of active site Ser103 indicated. (D) View from above in either ribbon or (E) surface representation. (F) LigPlot 2D illustration of JB101 interactions with FphE residues.

Fig. 6.

FphE covalently bound by compound 3. (A) Crystal structure of FphE in complex with oxadiazolone compound 3 (crystal form 2, PDB ID 8G49, at 1.60 Å). Ribbon (transparent) representation with carbon atoms of FphE chain A in light cyan, chain B in pink, compound 3 carbon atoms in gray. Side chains of key residues displayed and labeled. The 2Fo−Fc electron density map for compound 3 and Ser103 of chain A is shown as a blue mesh at 1 σ. (B) Hydrogen bonding network of compound 3 in the same orientation as in (A) depicted with water molecules as red spheres and hydrogen bonds as red lines. (C) Compound 3 chemical structure and reaction with active site Ser103 indicated. (D) Further rotated view of A, view from above in either ribbon or (E) surface representation. (F) LigPlot 2D illustration of compound 3 interactions with FphE residues. (G) Heatmap of sequence consensus (%) for key residues forming the binding pocket or interacting with JJ-OX-004 and compound 3. Boxes with a cross indicate residues without direct interactions. (H−K) Structural comparison of ligand binding to FphE. Transparent gray ribbon representation of FphE, only side chains of active site triad residues Ser103 and His267 displayed. Covalent bound ligands in two orientations as balls and sticks with carbon atoms in black. (H) FP-JB101 (PDB ID 8SBQ, at 1.50 Å). (I) Oxadiazolone-JJ-OX-004 (8T88, 1.54 Å) (14). (J) Oxadiazolone-compound 3 (8G49, 1.60 Å). (K) Boronic acid-Z27 (8UGM, 1.65 Å) (18).

The Oxadiazolone Compound 3 Occupies a Unique Pocket Compared to Other Ligands.

We next determined the crystal structure of FphE with an oxadiazolone-based molecule, compound 3, that was recently discovered to have antibiotic potency against methicillin-resistant S. aureus (MRSA) (23). This inhibitor covalently binds to ten proteins in MRSA bacteria, including FphC, FphE, and FphH. Further studies suggested that combined inhibition of FabH (an essential β-Ketoacyl-Acyl Carrier Protein Synthase), FphC, FphE, and AdhE (an aldehyde−alcohol dehydrogenase) contributes to the antimicrobial activity of compound 3. We previously described the inhibition of FphH by compound 3 (11), but no structure of this molecule has been reported. Cocrystallization of FphE with compound 3 resulted in crystal form 2 that was only observed in the presence of this molecule. In this form, two homodimer copies of FphE are perfect crystal symmetry mates A and A* (Fig. 6 A and B). As expected, compound 3 is covalently bound to the active site Ser103 with the oxadiazolone ring opened to form a stable carbamate linkage as was proposed previously (Fig. 6C) (23). Consistent with the crystal structure of FphE bound to the related oxadiazolone probe, JJ-OX-004, (14) the free oxygen atom of the oxadiazolone group is located near the oxyanion hole (56, 57) but not directly in the hole (Fig. 6A). The oxygen still forms a hydrogen bond with the amide group of Ala28 but in contrast to JB101 (2.9 Å O-N distance), the atom is shifted to no longer be able to form a hydrogen bond with the Ser104 backbone nitrogen atom (4.2 Å O-N distance). Beyond the oxadiazolone group, compound 3 does not fully occupy the hydrophobic tunnel and does not reach the surface of FphE as described for JB101 or JJ-OX-004 (Fig. 6 H–J). In contrast, compound 3 occupies a nearby pocket formed by residues Ala28 and Asn29 together with Arg193, Thr194, and Trp197 from the other dimer copy (Fig. 6 A and D). In addition, a hydrogen bonding network, with several water molecules, links the ligand to residues Ser102, Asn144 from one FphE dimer copy and Thr194, His257, Tyr260 from the other copy (Fig. 6B). This pocket is covered by helix 156 to 166 and the following 170 to 178 helix that connect the monomers in the dimer (Fig. 6E). Hydrophobic interactions are observed with Met160, Phe163, Leu167, Ile169, Met177, and Trp197 (Fig. 6F).

Two Distinct Binding Sites in FphE Underlie Pocket-Specific Ligand Selectivity.

To assess whether ligand binding correlates with evolutionary conservation, we generated sequence logos for 119 FphE homologs using WebLogo (58) and calculated residue-wise consensus values for positions interacting with both JJ-OX-004 and compound 3 (Fig. 6G, SI Appendix, Fig. S13, and Dataset S5). Both compounds share the same oxadiazolone warhead and an aromatic ring but differ in their target selectivity. JJ-OX-004 is a selective inhibitor of FphE with only minor cross-reactivity with FphB in Staphylococcus bacteria, whereas compound 3 is a more general antimicrobial agent with other targets including FabH which is not a serine hydrolase (14, 23). Interactions near the active site, including the GSSSG motif at the catalytic serine, are similar and highly conserved (SI Appendix, Fig. S13; Gly27, Ala28, and Pro129). The unique buried pocket occupied by compound 3 is formed by residues that are also highly conserved (Fig. 6G and SI Appendix, Fig. S13; Asn29, Ser102, Asn144, Met160, Phe163, Ile169, Met177, Trp197, and Tyr260). In contrast, the tail of JJ-OX-004 does not enter this pocket but reaches the surface via a hydrophobic opening that is formed by the side chain of moderately conserved Leu156 and the backbone of less conserved residues (Fig. 6G and SI Appendix, Fig. S13; Glu157, Ser178, and Tyr191). JB101 occupies the same distal hydrophobic tunnel as JJ-OX-004 but bears a fluorophosphonate (FP) warhead, which has much higher reactivity and thus broader serine hydrolase labeling properties compared to the oxadiazolones (59). This increased reactivity likely explains the reduced target selectivity of the FP probe despite the similar pocket engagement profile to JJ-OX-004. These observations indicate that both the location of binding within FphE and the intrinsic reactivity profile of the warhead contribute to selectivity outcomes. Our findings suggest that the presence of an optimal electrophilic warhead, an aromatic ring, and a hydrophobic pocket-occupying substituent produces molecules with a high degree of selectivity for FphE.

Discussion

Although serine hydrolases play crucial roles in bacterial physiology and pathogenesis, systemic characterization of this family of enzymes in S. aureus has only recently been initiated. The discovery of the Fph family has provided new insights into biofilm formation and virulence in S. aureus (11, 17, 18), as well as identified new targets for the development of species-specific molecular imaging tools (14). Among this serine hydrolase family, FphE stands out as a structurally unique and pathogen-selective target. This taxonomic exclusivity positions it as a potential selective biomarker for S. aureus infections. The ability to discriminate S. aureus from other related bacterial infections is critical in clinical settings where rapid and efficient identification of a specific bacteria affects treatment choices. Beyond the diagnostic potential of FphE, its exact functional role remains unclear. Broad variability in fphE transcriptional responses to external stimuli (17, 25–37) suggests that it may participate in adaptive processes that enable S. aureus to respond to environmental and host-associated conditions, rather than serving a single dedicated metabolic function inside the bacterial cell. Transcriptomic analyses indicate that fphE (SA2367, SACOL2597) expression is controlled by the alternative sigma factor, σ^B regulon, with expression patterns varying across stress conditions (25–27). fphE induction has been observed in response to nitroxyl donor Angeli’s salt (NWMN_2480 and SA2367) (28), mupirocin (SACOL2597) (29), tea tree oil (SACOL2597) (30), and rhodomyrtone (SAR2661) (31), while repression occurs with diclofenac (SACOL2597) (27), menD mutation (SACOL2597) (60), or by interspecies competition (SAUSA300_2518) (32). Expression is further modulated across growth states (strain UAMS-1) (34) and within vancomycin-intermediate resistant (strain SA137/93G) and susceptible (strain SA1450/94) strains (33), suggesting its role in adaptive responses. RST4–1 represents a major human genotype of S. aureus, but its occasional isolation from bovine mastitis cases raised the possibility of pseudogenization of fphE (61) that suggests it may contribute to host-specific adaptation. Functional studies provide additional evidence that FphE contributes to the adaptive response, as fphE disruption reduced human endothelial cell damage (35), and we reported that fphE transposon mutant displayed reduced liver bacterial load without affecting heart or kidney colonization (17). Several proteomic studies (62–67) have confirmed the presence of the FphE protein in various S. aureus strains and growth conditions without directly linking FphE function to a specific phenotype but giving further support to the hypothesis that FphE has a general role in adaptive responses to environmental stresses. Notably, one study showed that FphE could hydrolyze nitrophenyl acetate, demonstrating esterase activity toward an aromatic ester (67).

Our previous study demonstrated that JJ-OX-004, which bears an aromatic ring on the scaffold, exhibited greater selectivity for the active site serine on FphE than JJ-OX-003 (14). While the initial probe design was guided by an existing scaffold that contains both an aromatic ring and an oxadiazolone electrophile rather than by direct structural information about the target (23), adding the benzene ring to the JJ-OX-003 scaffold proved essential for labeling FphE. Although the endogenous substrates of FphE remain to be identified, our combined bioinformatic and structural analyses provide insights into molecular features that may be relevant to its physiological function. Consistent with this structure–activity relationship, our bioinformatic analysis suggests that FphE may be associated with the processing of aromatic compounds, as supported by the genomic colocalization of fphE with enzymes implicated in aromatic compound metabolism and its phenotypic association with biofilm formation (18). Furthermore, our structural analysis reveals that the FphE homodimer forms an extended hydrophobic tunnel, which is compatible with engagement of substrates containing hydrophobic features. Taken together, these observations illustrate how integrating bioinformatic context with structural analysis can help define key molecular features relevant to endogenous substrate identification, while simultaneously guiding the development of highly selective activity-based probes.

Our biochemical analysis of recombinant FphE suggests that this enzyme forms a stable dimer that resists dissociation even when diluted. This robustness of FphE dimerization suggests a functional or structural necessity of the cross-monomeric assembly. The FphE fluorescent probe JJ-OX-007 efficiently labels both monomeric and dimeric forms without inducing interconversion, confirming that catalytic accessibility is maintained in the dimeric state. Also, this result suggests that JJ-OX-007 minimally perturbs the oligomeric equilibrium of FphE and may serve as a valuable tool for the biochemical study of FphE. However, the higher hydrolytic activity of the dimer toward the 4-MUC substrate suggests subtle structural differences in the active site pocket that enhance catalytic efficiency in the dimeric conformation. Interestingly, coincubation with the substrate 4-MUC was associated with a reduced rate of dimer-to-monomer conversion at 37 °C, implying the active-site engagement may stabilize the dimeric assembly. Together, these observations suggest that the FphE monomer appears to be the more stable form at higher temperature in the absence of substrate, whereas substrate binding and lower temperature stabilize the dimeric, more catalytically active form. This behavior potentially indicates a dynamic regulation of FphE based on substrate availability and temperature conditions. The lack of a monomeric crystal form may be due to the fact that crystallography was always performed at 16 °C, or that the dimer introduces crystal interfaces via the connecting helices that are not present in the monomer. Temperature dependence of the dimeric status further suggests that dynamic equilibrium may be influenced by environmental factors, which may have physiological relevance under stress conditions where oligomeric transitions could modulate enzyme activity.

Our results indicate that there are general alpha/beta hydrolase homologs of the FphE monomer in the PDB but none that are close enough matches to provide structure–function information. While FphE is a member of a family of enzymes with a conserved and well-characterized fold, the homodimer we describe here represents a unique architecture among alpha/beta hydrolase members and the dimer architecture is not easily predicted using current modeling methods. SAXS data confirm that the observed crystal architecture is highly similar to the in-solution confirmation and clearly differs from other alpha/beta hydrolase homodimer interfaces, as seen for related hydrolases such as FphF. The FphE homodimer crystal forms suggest a level of flexibility of the overall fold that is not dependent on a traditional lid arrangement, presenting another unique feature of this architecture. All these aspects have implications for the binding of substrates or inhibitors.

The structure of JB101 bound in the active site of FphE provides structural insights into covalent FP interactions within the S. aureus FphA-J proteins, even though an FP probe was used to identify this class of enzymes. Our structure can be used to describe FphA-J protein interactions with FPs in general. Such FP-alkynes are used frequently to target active site serine residues and a crystal structure of a computationally designed serine hydrolase with the FP-ligand covalently bound has been reported (68). Several other studies have also reported on FP ligands that are covalently bound to serine active site residues, including FAS thioesterase (69) and phospholipases A (70). The structure of JB101 bound in the active site of FphE, when combined with the JJ-OX-004 structure (14) offers a framework for further structure-guided probe design and optimization.

The compound 3 bound structure, on the other hand, reveals an alternative interaction site within the unique homodimer architecture. This molecule binds such that it occupies multiple binding regions in and near the active site. This includes the active site triad, which is bound in manner similar to all other covalent bound ligands for this class of hydrolases (Fig. 6 H–K). In addition, the first phenyl ring of compound 3 occupies a space pointing toward the protein surface, similar to JB101, JJ-OX-004 (14) and a boronic acid ligand Z27, previously described by us (Fig. 6 H–K) (18). Finally, the final two phenyl rings bend away from the surface to occupy a space only observed for compound 3, suggesting a potentially high level of plasticity in inhibitor/substrate binding.

The high conservation of residues that interact with the bound inhibitors provides a structural rationale for the contrasting selectivity of JJ-OX-004 and compound 3, which share the same oxadiazolone electrophile and adjacent aromatic ring but differ in the substituent attached via an amide linkage. Compound 3 interacts with highly conserved regions of the active site and a neighboring pocket, whereas JJ-OX-004 engages less conserved peripheral residues within a hydrophobic channel unique to FphE. This interaction network not only partially explains the selectivity of JJ-OX-004 for FphE but also provides a generalizable design principle for targeting nonconserved hydrophobic pockets outside the catalytic triad to achieve FphE specificity. These insights show that combining structural data with conservation analysis from bioinformatics can effectively guide future rational design of selective probes and inhibitors. Taken together, these findings advance our understanding of FphE from a poorly characterized serine hydrolase to a structurally distinct enzyme with a potentially conserved metabolic role. It also establishes a rationale for further development of potential S. aureus–specific probes and inhibitors.

Materials and Methods

Bioinformatic Analysis.

FphE homologs were identified using sequence- and structure-based bioinformatic approaches. Detailed procedures are provided in SI Appendix.

Chemicals and Protein Preparation.

Small-molecule probes and inhibitors were obtained commercially or synthesized. Recombinant FphE was expressed and purified using standard methods. Full details are provided in SI Appendix.

Biochemical and Activity Assays.

Oligomeric state, probe labeling, and enzymatic activity of FphE were assessed using established gel-based and fluorogenic assays. Experimental details are described in SI Appendix.

Structural Analysis.

X-ray crystallography and small-angle X-ray scattering experiments were performed using standard data collection and analysis pipelines. Data processing and refinement procedures are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Data, Materials, and Software Availability

The crystal structure datasets generated and analyzed during the current study are available in the worldwide Protein Data Bank under PDB IDs 8SBQ (71), 8G49 (72), 8G48 (73), 9COM (74), 9EBF (75), 9D87 (76), and 9EDJ (77). SAXS datasets are available in the Small Angle Scattering Biological Data Bank SASBDB IDs SASDX64 (78) and SASDX74 (79). All other data are included in the manuscript and/or supporting information.