Structural and functional characterization of fosfomycin resistance conferred by FosB from Enterococcus faecium

Abstract

The Gram‐positive pathogen Enterococcus faecium is one of the leading causes of hospital‐acquired vancomycin resistant enterococci (VRE) infections. E. faecium has extensive multidrug resistance and accounts for more than two million infections in the United States each year. FosB is a fosfomycin resistance enzyme found in Gram‐positive pathogens like E. faecium. Typically, the FosB enzymes are Mn²⁺‐dependent bacillithiol (BSH) transferases that inactivate fosfomycin through nucleophilic addition of the thiol to the antibiotic. However, our kinetic analysis of FosB ^Ef shows that the enzyme does not utilize BSH as a thiol substrate, unlike the other well characterized FosB enzymes. Here we report that FosB ^Ef is a Mn²⁺‐dependent L‐cys transferase. In addition, we have determined the three‐dimensional X‐ray crystal structure of FosB ^Ef in complex with fosfomycin at a resolution of 2.0 Å. A sequence similarity network (SSN) was generated for the FosB family to investigate the unexpected substrate selectivity. Three non‐conserved residues were identified in the SSN that may contribute to the substrate selectivity differences in the family of enzymes. Our structural and functional characterization of FosB ^Ef establishes the enzyme as a potential target and may prove useful for future structure‐based development of FosB inhibitors to increase the efficacy of fosfomycin.

Short abstract

PDB Code(s): 7n7g;

1. INTRODUCTION

One of the most pressing dangers in modern medicine is antibiotic resistance. In the United States, microbial antibiotic resistance leads to more than two million infections each year, resulting in over 23,000 deaths. ¹ Organisms that display resistance are unsusceptible to treatment by certain antibiotic and antimicrobial agents, and the increase of organisms demonstrating multiple resistance is a growing issue. In 2017, the World Health Organization released its first ever list of priority pathogens. ² Many of the bacteria with the highest priority statuses are members of the antibiotic‐resistant ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) pathogens. These bacteria have collectively developed resistance to most types of commonly used antibiotics. ³ , ⁴ , ⁵ , ⁶

Vancomycin‐resistant enterococci (VRE), including E. faecium (VREfm), are opportunistic pathogens that are frequently associated with hospital‐acquired infections (HAIs). ⁷ They first emerged in the late 1980s, and by 2008, VREfm accounted for up to 80% of E. faecium clinical isolates. ⁸ , ⁹ VREfm is among the most dangerous of the ESKAPE pathogens, with outbreaks lasting up to 11 months on average and being associated with higher mortality and longer hospital stays. ¹⁰ , ¹¹ , ¹² Vancomycin is not the only antibiotic for which VREfm exhibits resistance; resistance to fosfomycin has also been observed. ¹³ , ¹⁴ , ¹⁵

Fosfomycin, known by the trade name Monurol (Figure 1), is a broad‐spectrum antibiotic that is prescribed in the United States for uncomplicated urinary tract and gastrointestinal infections. ¹⁶ Fosfomycin is effective against both Gram‐negative and Gram‐positive bacteria (including some multiple‐resistance strains like MRSA ¹⁷ , ¹⁸ ) by inhibiting bacterial cell wall biosynthesis via irreversible inactivation of the enzyme, UDP‐N‐acetylglucosamine‐3‐enolpyruvyltransferase (MurA). ¹⁹ , ²⁰ , ²¹ It is a safe antibiotic for humans, as it has few side effects and is excreted from the body in its unmetabolized form. It can be administered in a single dose, making it more convenient than many other antibiotics with the same efficacy. ²² However, the emergence of fosfomycin‐modifying enzymes has proven to be a disadvantage for the use of fosfomycin.

FIGURE 1.

Mechanism of fosfomycin resistance for FosA (Mn²⁺ and K⁺‐dependent glutathione transferase), FosB (Mn²⁺‐dependent bacillithiol transferase), and FosX (Mn²⁺‐dependent hydrase)

There are currently three known classes of fosfomycin resistance enzymes that are members of the vicinal oxygen chelate (VOC) superfamily. VOC enzymes use divalent metals to stabilize transition states that contain vicinal oxygens. All catalyze a nucleophilic addition reaction that modifies fosfomycin to open the epoxide ring (Figure 1), resulting in a product with no bactericidal properties. FosA, found in Gram‐negative bacteria, is a Mn²⁺‐ and K⁺‐dependent glutathione (GSH)‐S‐transferase that can alternatively use L‐cys as a thiol substrate as well. ²³ , ²⁴ , ²⁵ , ²⁶ FosB, from Gram‐positive bacteria, is a Mn²⁺‐dependent bacillithiol (BSH)‐S‐transferase, ²⁷ , ²⁸ and finally, FosX and FosM, are Mn²⁺‐dependent hydrases that catalyze the addition of water to fosfomycin. FosX is found in Gram‐positive and negative bacteria, whereas FosM is found in Mycobacteria. ²⁹ , ³⁰ , ³¹ In addition to the VOC fosfomycin resistance enzymes, another group of fosfomycin modifying enzymes known as FosC utilize ATP and transfer a phosphoryl group to fosfomycin. ³²

VREfm is among the high‐profile pathogens that contain the FosB enzyme, as well as Bacillus subtilis and S. aureus. ¹³ , ¹⁴ , ²⁷ , ³⁰ , ³³ In this work, we use time‐trace kinetic data and mass spectrometry to characterize the metal ion and thiol selectivity of FosB from E. faecium (FosB ^Ef ). Our results demonstrate that FosB ^Ef is a Mn²⁺‐dependent L‐cys transferase unlike other FosB enzymes from Gram‐positive pathogens. Moreover, we have obtained a crystal structure of FosB ^Ef in complex with fosfomycin. This initial characterization of FosB ^Ef may prove useful for future structure‐based development of FosB inhibitors that may increase the efficacy of fosfomycin.

2. RESULTS

2.1. Sequence alignment

FosB ^Ef is a 139 amino acid fosfomycin resistance enzyme from E. faecium that shares approximately 60% sequence similarity with two well‐characterized FosB enzymes from S. aureus and B. cereus. A sequence alignment reveals high sequence similarity between the bacterial species (Figure 2). Importantly, all of the residues that have been implicated in coordinating the divalent metal (His7, His66, and Glu115) are present. In addition, all the residues involved in binding fosfomycin are also conserved across all sequences. The residues that bind the polar end of the antibiotic are found in FosB ^Ef as Arg93, Arg124, Tyr64, and Tyr104, and the aromatic tryptophan typically found in VOC enzymes near the methyl group of fosfomycin is also present in Trp46. Together, these residues form a “cage” of hydrogen bonding interactions that stabilize fosfomycin in the active site of the FosB enzymes. ²⁷

FIGURE 2.

Sequence alignment of FosB enzymes from Enterococcus faecium, Staphylococcus aureus, and Bacillus cereus. Residues labeled with (*) are those implicated in metal binding in the active site. Residues marked with (#) are those used in fosfomycin coordination. Red blocks indicate residues that are conserved through all sequences. Blue text shows residues that are similar, but not identical

2.2. Metal specificity

The FosB enzymes were initially characterized as Mg²⁺‐dependent L‐cys transferases. ³³ However, further investigation of metal activation and substrate specificity revealed that the FosB enzymes are activated by Mn²⁺ and inhibited by Zn²⁺ with only minimal activation by Mg²⁺. ²⁷ In addition, the discovery of BSH as the low‐molecular‐weight thiol in Gram‐positive pathogens prompted characterization of FosB as a Mn²⁺‐dependent bacillithiol transferase. In general, the metal activation of FosB enzymes is Mn²⁺ ≫ Mg²⁺ > Zn²⁺.

To probe the metal activation of FosB ^Ef , we used ³¹P‐NMR. Both fosfomycin and the thiol‐fosfomycin product have a single phosphorous atom that is readily detected by ³¹P‐NMR spectroscopy. The reactions were conducted with 8 mM fosfomycin in 20 mM HEPES (pH 7.5) and 500 nM FosB ^Ef . The mixture was incubated for 5 min before the reaction was initiated by the addition of 4 mM L‐cys. We used L‐cys for these experiments for two reasons. First, all FosB enzymes characterized to date utilize L‐cys as a nucleophilic substrate, and second, L‐cys is more readily available and less expensive than BSH. Moreover, the metal activation trend is the same for either thiol substrate. The results of the ³¹P‐NMR time course kinetic analysis are shown in Figure 3. The metal activation of FosB ^Ef follows the same trend as other fosfomycin resistance enzymes, with Mn²⁺ ≫ Mg²⁺ > Zn²⁺, confirming that FosB ^Ef is a Mn²⁺‐dependent enzyme like the other fosfomycin resistance enzymes.

FIGURE 3.

Metal specificity of FosB ^Ef containing Mn²⁺ (●), Mg²⁺ (■), or Zn²⁺ (▲). Each sample contains 500 nM FosB ^Ef , 8 mM fosfomycin, and 4 mM L‐cysteine

2.3. Mass spectrometry

Given the diversity of VOC fosfomycin resistance enzymes, combined with the catalytic promiscuity of FosX, capable of performing both hydrase and glutathione transferase reactions, we considered the possibility that FosB ^Ef might also be able to catalyze the addition of multiple nucleophilic substrates. To investigate nucleophile selectivity, we used mass spectrometry to analyze reaction products of FosB ^Ef in the presence of water, L‐cys, GSH, and BSH (Figures SI‐1–SI‐5). Figure SI‐1 shows the fosfomycin standard in the absence of FosB ^Ef . Figure SI‐2 shows the reaction products of fosfomycin and FosB ^Ef with the addition of BSH. A small new peak is observed at m/z of 535.0992. The new peak corresponds to the BS‐fosfomycin product and indicates FosB ^Ef can utilize BSH like other FosB enzymes of the VOC superfamily. However, this reaction required incubation overnight in order to finally observe the very small peak at m/z of 535.0992, which prompted us to reconsider whether BSH is the native preferred thiol substrate. The peak at m/z of 137.0014 corresponds to fosfomycin. Figure SI‐3 shows the mass spectrometry analysis of the reaction when FosB ^Ef is added, with the fosfomycin mass still being the only peak present. This indicates that FosB ^Ef does not utilize water as a nucleophilic substrate and is not a hydrase, FosX‐type, enzyme. Figure SI‐4 shows the reaction products of fosfomycin and FosB ^Ef with the addition of L‐cys. As anticipated, the fosfomycin peak is reduced and a new peak at m/z of 258.0169 is observed. The new peak corresponds to the L‐cys‐fosfomycin product. Finally, Figure SI‐5 shows the reaction products of fosfomycin and FosB ^Ef with the addition of GSH. No GS‐fosfomycin product was observed following the reaction, indicating the FosB ^Ef is not a glutathione‐S‐transferase like FosA.

2.4. NMR time course kinetics for thiol specificity

Because the mass spectrometry results showed only a small amount of BS‐fosfomycin product, we performed an NMR activity test using BSH and FosB ^Ef and found that no appreciable product was produced on a physiological time scale. This then led us to conduct NMR time course kinetics with L‐cys. L‐cys was chosen, once again, because all the other Gram‐positive FosB enzymes utilize L‐cys as a secondary substrate but with much less efficicency. ³⁴ The relative intensity of the product peak to the fosfomycin standard peak can be used to estimate the concentration of product at each time point. However, the peak integration is not sensitive, so it provides a rough estimation of the peak ratios and thus can only be used to qualitatively describe the products. The relative amount of L‐cys‐fos product was tracked over the span of 1.5 h as shown in Figure 4. Using the linear region of this plot, an apparent k _cat of 11.45 s⁻¹ was calculated.

FIGURE 4.

Time course plot of L‐cys‐fos product produced over a span of 90 min. Each reaction consisted of 250 nM FosB ^Ef , 8 mM fosfomycin, and 4 mM L‐cys. The slope of the linear region of this chart was used to calculate the apparent k _cat value. Trials using 2 mM BSH were tested at 20 min, however no product was detected. Due to the cost of BSH and lack of product, we decided to continue using only L‐cys. Inset shows the linear region of the plot used to calculate initial velocity

2.5. Crystal structure determination

The crystal structure of FosB ^Ef with metal and fosfomycin bound in the active site was determined to a resolution of 2.0 Å (Figure 5). The structure was solved by molecular replacement using the initial structure of FosB ^Bc ·Zn²⁺· fosfomycin (PDB Entry: 4JH3) as the search model where the Zn²⁺ was removed from the coordinate file before the phasing process. The crystal belongs to the P4₁2₁2 space group and contains 270 amino acids and two metals in the asymmetric unit. Like all other VOC fosfomycin resistance enzymes structurally characterized to date, FosB ^Ef is a homodimer with a 3D domain‐swapped arrangement of tandem βαβββ motifs where both subunits of the homodimer participate in coordination of each metal ion and formation of the active sites in the enzyme. All VOC fosfomycin resistance enzymes maintain this general structural arrangement with little variation. The overall RMSD for all atoms for FosB ^Ef and FosB ^Bc , both with Zn²⁺ and fosfomycin bound, is 0.80 Å.

FIGURE 5.

Overall X‐ray crystal structure of the FosB protein from Enterococcus faecium in complex with Zn²⁺ and fosfomycin at 2.0 Å‐resolution (PDB ID: 7N7G). Final refinement had an R _work = 22.3% and R _free = 26.5%

Our crystal structure of FosB ^Ef was obtained using protein “as isolated” from purification without any removal and subsequent addition of metal to ensure consistent metal identity throughout the sample. Nevertheless, we have modelled the electron density with Zn²⁺, rather than Mn²⁺. Our reasoning is due to the “as isolated” samples being far less active than those we have intentionally stripped of all metal and substituted with Mn²⁺ as we did for the metal selectivity assays. Given the low activity and metal activation of FosB ^Ef , the inconsistency of metal binding in the “as isolated” samples is most likely due to the nutrient rich Terrific Broth used during protein expression. The same effect has previously been observed for FosB ^Bc . ²⁷

Coordination of fosfomycin by Zn²⁺ in FosB ^Ef is similar to that of FosB ^Bc and FosA. Figure 6 shows the F _o – F _c difference map, calculated before addition of fosfomycin to the coordinate file. The omit map clearly establishes the presence of fosfomycin coordinated to the metal. The geometry of metal coordination for FosB ^Ef is the same as FosB ^Bc and can be described as distorted, five‐coordinate trigonal bipyramidal where His7, His66, and a phosphonate oxygen occupy the equatorial sites, and Glu115 and the oxirane oxygen of fosfomycin occupy the axial sites. ²⁷ The trigonal bipyramidal geometry has been implicated in the FosB mechanism of fosfomycin inactivation.

FIGURE 6.

Stereo view of the active site of FosB ^Ef showing the position of fosfomycin coordinated to Zn²⁺. The electron density difference map is the F _o − F _c map calculated before addition of fosfomycin to the coordinate file (PDB ID: 7N7G) and is contoured at 3σ. The difference density map clearly establishes the position of fosfomycin in the active site

3. DISCUSSION

3.1. Thiol selectivity

An apparent k _cat value of 11.45 s⁻¹ indicates that FosB ^Ef is effective at utilizing L‐cys as a substrate. This value is higher than the apparent k _cat values for either FosB ^Bc or FosB ^Sa when L‐cys is the thiol substrate. The apparent k _cat values for FosB with L‐cys and Mn²⁺ from S. aureus and B. cereus are 0.05 and 2.0 s⁻¹, whereas the apparent k _cat values for FosB with BSH and Mn²⁺ from S. aureus and B. cereus are 5.98 and 9.3 s⁻¹, respectively. ²⁷ , ³¹ , ³⁵ Thus, the apparent k _cat value for FosB ^Ef L‐cys transferase activity is in agreement with the BSH transferase activity of FosB from S. aureus and B. cereus. This suggests that the preferred thiol substrate for FosB ^Ef is L‐cys rather than BSH.

3.2. Analysis of the E. faecium genome

The lack of BSH transferase activity and preference for L‐cys was a surprising result and prompted to us to perform a BLAST search for the genes in E. faecium that are required to synthesize BSH. Previous research identified genes of interest in the BSH biosynthesis pathway using a search based on finding homologs of mycothiol (MSH) biosynthesis genes. ³⁶ MSH is the low molecular weight thiol produced by mycobacteria that is structurally similar to BSH. The homolog search identified three genes as being responsible for BSH biosynthesis: bshA, bshB1/bshB2, and bshC. bshA acts as a glycosyl transferase and uses uridine diphosphate N‐acetylglucosamine (UDP‐GlcNAc) as well as L‐malate as substrates. bshB is the second BSH synthesis gene and catalyzes the deacetylation of GlcNAc‐Mal. bshB is usually comprised of two genes, bshB1 and bshB2, that code for two separate proteins. ³⁶ However, bshB1/bshB2 is redundant and bacteria do not need both forms to synthesize BSH. Previous research has shown that when one bshB gene is knocked out, there is still normal BSH production, however a double knock out mutant produces no BSH. ³⁶ bshC is the last of the BSH biosynthesis genes and is a homolog to mshC of the mycothiol (MSH) biosynthesis pathway. The function of bshC was assigned using EMBL‐STRING with the bshA sequence used as an input in an effort to identify other genes that co‐occur with bshA. ³⁶ When the bshC gene was knocked out, no BSH was produced, confirming that this gene is involved in BSH biosynthesis. Other Gram‐positive organisms with fosfomycin resistance proteins known to use BSH as the preferred substrate, such as FosB from S. aureus and B. subtilis, have some form of all three genes. Our BLAST search found that E. faecium only has bshA (Figure 7). This suggests that E. faecium cannot synthesize BSH and explains why BSH is a poor substrate. This supports the conclusion that L‐cys is the preferred substrate of FosB ^Ef . Figures SI‐6, SI‐7, and SI‐8 show the bsh gene locations on the complete genomic DNA of E. faecium, B. subtilis, and S. aureus, respectively.

FIGURE 7.

Pictorial representation of the bsh genes from Enterococcus faecium, Bacillus subtilis, and Staphylococcus aureus. E. faecium only has bshA. B. subtilis and S. aureus both have all three bsh biosynthesis genes. Shown are the number of base pairs between each gene

3.3. Crystal structure

One possible explanation for why FosB ^Ef does not utilize BSH can be observed in the overall structure of the protein. Figure 8 shows a surface visualization of the proposed thiol substrate binding site of FosB ^Ef and a FosB homolog from Bacillus anthracis (FosB ^Ba , PDB ID: 4JD1) that is known to utilize BSH. Fosfomycin is positioned at the bottom of a narrow channel poised for nucleophilic attack. While the FosB enzymes have never been crystallized with BSH, previous studies have implicated Arg35 and Lys36 for BSH recognition in FosB ^Ba and Lys35 and Lys36 in FosB ^Sa . ³⁷ These positively charged residues are positioned appropriately to bind the negatively charged malate domain of BSH. ³⁶ However, in those positions, FosB ^Ef instead has Glu35 and Thr36. Due to the inherent difference in the charge of these amino acids, it is likely that it is not electrostatically possible for FosB ^Ef to bind and utilize BSH as a thiol substrate.

FIGURE 8.

Surface visualization of the active site and proposed thiol substrate binding residues of FosB ^Ef (orange) and FosB ^Ba (blue). The difference in electrostatics at residues 35 and 36 provide insight into why FosB ^EF cannot utilize BSH while FosB ^Ba can. The positively charged residues in FosB ^Ba are positioned appropriately to bind the negatively charged malate domain of BSH, whereas the negatively charged residues in FosB ^Ef would not favor BSH

3.4. Sequence similarity network

A sequence similarity network was generated for the FosB family of the VOC superfamily (IPR022858) to identify potential structure/function relationships that could account for the unexpected substrate selectivity of FosB from E. faecium. The full network consisted of 1,179 sequences. Within the Cytoscape visualization software, the stringency for edges was increased by filtering on the alignment score metric. At a value of 10⁻⁶⁰, the nodes containing sequences for FosB from Enterococcaceae species were isolated in a cluster of sequences from Staphlococcaceae (Figure 9a). This cluster was extracted into a new network and the stringency for drawing edges was increased to create separate clusters. At a value of 10⁻⁶⁵, a cluster containing the FosB ^Ef sequence separated from the cluster containing FosB ^Sa (Figure 9b). The FosB ^Ef ‐containing cluster contains sequences from three other Enterococcaceae species as well as FosB sequences identified in plasmids from S. aureus clinical isolates.

FIGURE 9.

Sequence similarity networks for the FosB family (IPR022858). (a) Full network view consisting of 1,179 nodes and 129,314 edges drawn at a cutoff value of 10⁻⁶⁰ alignment score (average alignment length = 138 residues, average percent identity = 83%). Nodes are colored by taxonomic family (Bacillaceae, blue; Staphylococcaceae, magenta; Enterococcaceae, green; Paenibacillaceae, orange; other, red). Diamond‐shaped nodes indicate the sequence has been annotated by Swiss‐Prot. Circle‐shaped nodes are from the TrEMBL database. (b) Extracted view of the boxed cluster in panel (a). This network contains 117 nodes and 1,466 edges drawn at a cutoff value of 10⁻⁶⁵ alignment score (average alignment length = 139 residues, average percent identity = 90%)

A sequence alignment was generated using nodes from the FosB ^Ef and FosB ^Sa clusters that identified ~45 positions that are differentially conserved between the two clusters (Figure SI‐9). Three positions were identified that may contribute to the substrate selectivity differences between the two enzymes: residues 9, 35, and 36. Residue 9 (Thr9 in E. faecium and Cys9 in S. aureus) is located directly in the active site while residues 35 and 36 (Glu35/Thr36 in E. faecium and Lys35/Lys36 in S. aureus) are located on the edge of the thiol‐binding pocket. A second sequence alignment was generated using the FosB ^Ef , FosB ^Sa , and randomly selected sequences from all major clusters in Figure 9a to identify residues that are globally essential for catalysis in FosB enzymes (Figure SI‐10). Conservation at position 9 is either cysteine or leucine for most of the FosB family with Thr only being found in the E. faecium cluster. Positions 35/36 are strongly conserved as two positive residues, most often as Lys35/Lys36 (as seen in the FosB ^Sa sequence). FosB ^Ef is the only sequence to lack a positively charged residue in either position. These two residues have been proposed to interact with the dicarboxylate malate‐moiety of bacillithiol in FosB enzymes. ³⁷ Thus, the SSN analysis confirms the structural analysis, and it is likely that loss of positive charge in these two positions prevents FosB ^Ef from effectively binding bacillithiol, but with sufficient catalytic machinery to utilize L‐cys as a substrate.

4. CONCLUSION

In this paper we have expressed and characterized the fosfomycin resistance enzyme FosB from E. faecium. We found that FosB ^Ef is activated by Mn²⁺ and has reduced activity with Zn²⁺ and Mg²⁺, similar to the other FosB enzymes. However, unlike the other FosB enzymes, we were surprised to find that FosB ^Ef is an L‐cys transferase rather than a BSH transferase. Furthermore, we obtained the 3D crystal structure of FosB ^Ef in complex with fosfomycin. Given the rise of multidrug‐resistance and extensively drug‐resistant bacteria like E. faecium, many researchers have turned to the revitalization of approved, safe antibiotics like fosfomycin. Our characterization of FosB ^Ef provides a starting platform for future work regarding structure‐based development of FosB inhibitors that may eventually play a role in combatting fosfomycin resistance.

5. EXPERIMENTAL

5.1. General materials

All buffer salts were obtained from VWR. Metals were purchased as their chloride salts; manganese (II) chloride was from ACROS Organics, magnesium chloride hexahydrate and zinc chloride were purchased from Fisher Scientific. L‐cysteine was obtained from Fisher Chemical, and glutathione was purchased from VWR. Fosfomycin disodium salt was from MP Biomedicals, LLC. BSH was synthesized as bacillithiol disulfide (BSSB) by the Vanderbilt Chemical Synthesis Core and reduced to BSH prior to use with a Pierce Immobilized Reductant Column from ThermoFisher Scientific.

5.2. Protein expression and purification

A pET‐20b plasmid containing the gene encoding non‐His tagged wild type (WT) FosB from E. faecium was transformed into Escherichia coli BL21 (DE3) cells. The cells were plated on LB agar containing 100 μg/ml of ampicillin and 75 μg/ml of fosfomycin and incubated for approximately 16 h at 37°C. A singular colony was isolated from the LB‐agar plates and used to inoculate 2 ml of LB (Fisher Bioreagents) starter culture containing 75 μg/ml of ampicillin. After approximately 8 h of incubation at 37°C with shaking, 2 ml of the starter growth was used to inoculate 75 ml of LB containing 75 μg/ml of ampicillin. This small starter culture was grown at 37°C at 225 rpm for 12–16 h, after which 10 ml of culture was added to 1 L of Terrific Broth containing 80 μg/ml of ampicillin (6 L total). The 1 L cultures were grown at 37°C with shaking until the OD₆₀₀ reached ~1 and were then induced with 0.5 mM IPTG. The temperature was then reduced to 18°C and the cultures were grown for 18–20 h. The cells were then harvested by centrifugation at 8,000g for 10 min. The cell pellets were stored at −80°C.

The E. coli cell pellet containing overexpressed FosB ^Ef was resuspended in 2 ml of lysis buffer (20 mM Tris HCl, pH 7.5) per gram of pellet. Lysozyme was then added to the slurry at 1 mg/ml and this mixture was stirred at 4°C for 30 min. 5 mg of DNase was then added, and the slurry was stirred for an additional 30 min at 4°C. The mixture was sonicated to ensure the complete lysing of cells, and the lysate was cleared by centrifugation at 20,000g for 20 min.

An ammonium sulfate precipitation was performed on the cleared lysate solution. Fractions were precipitated at 20, 30, 40, 50, 60, and 70% ammonium sulfate. An SDS‐PAGE analysis found that the 40–60% fraction contained the highest ratio of FosB ^Ef to other proteins. This fraction was dialyzed into 20 mM HEPES (pH 7.5) overnight for further purification via anion‐exchange chromatography.

After dialysis, the protein was loaded onto a GE Healthcare HiPrep DEAE FF 16/10 anion exchange column that was equilibrated with 20 mM HEPES (pH 7.5) using a BioRad NGC FPLC equipped with a separate sample load pump. FosB ^Ef , having an estimated PI of 5.21, adheres to the DEAE material and turns a pink color that is visible on the column. The protein was eluted over 20 column volumes using a gradient of 0–50% NaCl in 20 mM HEPES (pH 7.5). The salt level was held constant for 2 column volumes at 12, 18, 27, and 50% NaCl. Fractions were then analyzed by SDS‐PAGE, with the purest fractions being combined and dialyzed into 50 mM HEPES and 150 mM NaCl pH 7.5 overnight for further purification via gel filtration chromatography.

After dialysis, the protein was concentrated and loaded onto a GE Healthcare HiPrep 26/60 sephacryl S‐100 HR gel filtration column equilibrated with the same buffer using a BioRad NGC FPLC. Collected fractions were then analyzed by SDS‐PAGE and the purest fractions were combined and dialyzed overnight into 20 mM HEPES pH 7.5.

Purified FosB ^Ef was prepared with Mn²⁺, Mg²⁺, or Zn²⁺ by dialyzing into 50 mM Bis‐Tris (pH 6.0) with 5 mM EDTA, 2 mM 1,10‐phenanthroline, and 10 mM Chelex for 4 h to remove all bound metals. The protein was then dialyzed into 20 mM HEPES (pH 7.5) with 200 μM of the target metal. After dialysis, the protein was then concentrated, and the concentration was determined using a Thermo Scientific NanoDrop One.

5.3. Sequence alignment

Sequences of the FosB enzymes from E. faecium, S. aureus, and B. cereus were aligned using T‐Coffee, a multiple sequence alignment server package. ³⁸ Final alignment shading and image creation were done with the pyBoxshade desktop program.

5.4. Continuous ³¹P NMR assays

To investigate metal specificity, FosB ^Ef was purified and prepared with 200 μM Mn²⁺, Mg²⁺, or Zn²⁺ bound. According to published protocol, ³¹ saturation conditions of 8 mM fosfomycin prepared in 20 mM HEPES (pH 7.5) and 500 nM FosB ^Ef were allowed to incubate for 5 min before the reaction was initiated by the addition of 4 mM L‐cys. The sample was then transferred to an NMR tube and allowed to continue reacting at 25° C. The ratio of the concentration of fosfomycin to the concentration of the product was monitored continuously by obtaining spectra at various time points up to 80 min using a Bruker Avance 500 MHz NMR. The data were analyzed using the MestReNova‐LITE software.

5.5. Mass spectrometry

According to previously published protocol, ³¹ all samples were prepared in water with 8 mM fosfomycin and 500 or 250 nM FosB ^Ef with the thiol samples additionally containing 4 mM L‐cys, 4 mM GSH, or 2 mM BSH. Each reaction was initiated by the addition of FosB ^Ef and allowed to continue at 25°C overnight to ensure full product formation. The reactions were then quenched with 500 μl chloroform to precipitate the enzyme out of solution. Samples were centrifuged to ensure full separation, and the aqueous layer was analyzed using a Waters Xevo G2‐XS QToF Mass Spectrometer.

5.6. Protein crystallization

Initial crystals of FosB ^Ef were grown using the sitting‐drop vapor‐diffusion method at 298 K by mixing 1 μl of protein solution (20 mg/ml in 20 mM HEPES buffer, pH 7.5) and 1 μl of reservoir solution (Hampton Research Index 55; 100 mM HEPES pH 7.5, 0.5 M magnesium chloride, and 30% PEG MME 550 [w/v]) in a Hampton Research MRC 2 well crystallization plate. Larger crystals were grown using the hanging‐drop vapor‐diffusion method at 293 K by mixing 3 μl of protein solution (20 mg/ml in 20 mM HEPES buffer, pH 7.5) and 3 μl of reservoir solution (Hampton Research Index 55, 0.5 M magnesium chloride and 30% PEG MME 550 [wt/vol]) in a Hampton Research VDX plate. Crystallization hits were obtained in several different conditions, but those grown in HR Index 55 yielded the best diffraction. The final optimized conditions were obtained after mixing equal volumes (3 μl) of the protein solution (30 mg/mL in 20 mM HEPES buffer, pH 7.5) and reservoir solution (100 mM HEPES pH 7.5, 0.15 M magnesium chloride, and 32% PEG MME 550 [wt/vol]). All crystals were cryoprotected in the mother solution and 15% glycerol prior to freezing in liquid nitrogen and data collection.

5.7. Data collection and refinement

Diffraction data were collected at the University of Alabama using a Rigaku Synergy DW rotating anode X‐ray generator equipped with a Rigaku Hypix detector. Crystals were maintained at 100 K using an Oxford Cryosystems cryostat. The collected diffraction data sets were processed with CrysAlis^Pro. ³⁹ Phasing of diffraction data was done by molecular replacement PHASER ⁴⁰ using PDB Entry: 4JH3. Manual model building for each structure was performed using Coot model building software. ⁴¹ Water molecules were placed with the Coot routine, Find Waters. The final models were obtained by iterative cycles of model building in Coot and structure refinement using Refmac5 in the CCP4 suite of programs (Collaborative Computational Project, 1994). ⁴² All protein figures were prepared with Chimera. ⁴³ Data collection and refinement statistics are given in Table 1.

TABLE 1.

Crystallographic data collection and refinement statistics

PDB entry	7N7G
Wavelength (Å)	1.54
Resolution range (Å)	23.78–2.0 (2.072–2.0)
Space group	P 41 21 2
Unit cell (Å)	79.9, 79.9, 95.6
Total reflections	3,91,729 (27,027)
Unique reflections	21,509 (2,084)
Multiplicity	18.2 (13.0)
Completeness (%)	99.66 (99.95)
Mean I/sigma(I)	22.16 (1.76)
Wilson B‐factor	30.78
R‐merge	0.1081 (1.342)
R‐meas	0.1112 (1.4)
R‐pim	0.02514 (0.3878)
CC1/2	0.999 (0.709)
CC*	1 (0.911)
Reflections used in refinement	21,457 (2,084)
Reflections used for R‐free	1078 (107)
R‐work	0.223 (0.284)
R‐free	0.265 (0.394)
CC(work)	0.945 (0.762)
CC(free)	0.915 (0.656)
Number of non‐hydrogen atoms	2,360
Macromolecules	2,222
Ligands	42
Solvent	96
Protein residues	269
RMS (bonds)	0.015
RMS (angles)	1.85
Ramachandran favored (%)	97.34
Ramachandran allowed (%)	2.66
Ramachandran outliers (%)	0
Rotamer outliers (%)	4.64
Clashscore	2.72
Average B‐factor	34.97
Macromolecules	34.46
Ligands	50.92
Solvent	39.91

5.8. Sequence similarity network generation

A sequence similarity network ⁴⁴ (SSN) was generated for the IPR022858 family in the InterPro database (version 86.0) using the Enzyme Function Initiative – Enzyme Similarity Tool ⁴⁵ (EFI‐EST) maintained by the Enzyme Function Initiative (www.enzymefunction.org). All 1,179 sequences were used to generate the SSN. The resulting network was downloaded as a Cytoscape ⁴⁶ readable xgmml file for visualization. Multiple‐sequence alignments were created using MAFFT and visualized with Jalview.

CONFLICT OF INTEREST

There are no conflicts of interest to declare.

AUTHOR CONTRIBUTIONS

Vanessa Wiltsie: Data curation (equal); formal analysis (equal); writing – original draft (equal); writing – review and editing (equal). Skye Travis: Data curation (equal); formal analysis (equal); writing – original draft (equal); writing – review and editing (equal). Madeline R. Shay: Data curation (equal); formal analysis (equal); writing – original draft (equal). Zachary Simmons: Data curation (supporting). Patrick Frantom: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); methodology (equal); project administration (equal); resources (equal); software (equal); validation (equal); visualization (equal); writing – original draft (equal); writing – review and editing (equal). Matthew Thompson: Conceptualization (lead); data curation (lead); formal analysis (lead); funding acquisition (lead); investigation (lead); methodology (lead); project administration (lead); resources (lead); software (lead); supervision (lead); validation (lead); visualization (lead); writing – original draft (lead); writing – review and editing (lead).

Supporting information

Data S1 Figure SI‐1 Mass spectrometry analysis of 8 mM fosfomycin carried out in water at 25°C.

Figure SI‐25. Mass spectrometry analysis of 250 nM Fos BEfF, 8 mM fosfomycin, and 2 mM BSH, carried out in water at 25°C.

Figure SI‐32. Mass spectrometry analysis of 0.5 μM Fos BEfF and 8 mM fosfomycin carried out in water at 25°C.

Figure SI‐43. Mass spectrometry analysis of 0.5 μM Fos BEfF, 8 mM fosfomycin, and 4 mM  l ‐cys, carried out in water at 25°C.

Figure SI‐54. Mass spectrometry analysis of 0.5 μM Fos BEfF, 8 mM fosfomycin, and 4 mM GSH, carried out in water at 25°C.

Figure SI‐5. Mass spectrometry analysis of 250 nM Fos BEF, 8 mM fosfomycin, and 2 mM BSH, carried out in water at 25°C.

Figure SI‐6. Genome of Enterococcus faecium. The gene for bshA is highlighted in its locus.

Figure SI‐7. The genome for Bacillus subtilis is shown. Highlighted are the genes for bshA, bshB1, bshB2, and bshC in their respective loci.

Figure SI‐8. The genome of Staphylococcus aureus is shown with the genes for bshA, bshB2, and bshC highlighted in their respective loci.

Figure SI‐9. Resulting sequence alignments from the nodes of FosBEfF and FosBSaA clusters from the SSN. Highlighted in blue are areas of differentially conserved residues between the two clusters. Of interest are residues 9, 35, and 36.

Figure SI‐10. Results of a second sequence alignment using FosBEfF, FosBSaA, and randomly selected sequences from all of the previously identified major clusters to identify residues that are implicated as being essential for catalysis universally for FosB enzymes.

Wiltsie V, Travis S, Shay MR, Simmons Z, Frantom P, Thompson MK. Structural and functional characterization of fosfomycin resistance conferred by FosB from Enterococcus faecium . Protein Science. 2022;31:580–590. 10.1002/pro.4253