FosB, a Cysteine-Dependent Fosfomycin Resistance Protein under the Control of ς^W, an Extracytoplasmic-Function ς Factor in Bacillus subtilis

Abstract

We demonstrate that the Bacillus subtilis fosB(yndN) gene encodes a fosfomycin resistance protein. Expression of fosB requires ς^W, and both fosB and sigW mutants are fosfomycin sensitive. FosB is a metallothiol transferase related to the FosA class of Mn²⁺-dependent glutathione transferases but with a preference for Mg²⁺ and l-cysteine as cofactors.

Sequencing of the Bacillus subtilis genome revealed the presence of seven new ς factors, all members of the extracytoplasmic function subfamily (12, 13). We have begun to investigate their functions by mutation of each gene and the identification of target operons (8–11). In this work, we demonstrate that yndN encodes a fosfomycin resistance (Fos^r) protein that depends on ς^W for expression. We have renamed yndN as fosB, based on its similarity to the fosB gene identified on a Staphylococcus epidermidis plasmid (Fig. 1B).

FIG. 1.

The fosB(yndN) gene encodes a fosfomycin resistance protein. (A) The fosB(yndN) region of the chromosome is illustrated. The fosB gene is transcribed from a ς^W-dependent promoter similar in sequence to the sig^W autoregulatory promoter, P_w (9). (B) Multiple sequence alignment of the 144-amino-acid B. subtilis FosB protein (Bsu) with FosB from S. epidermidis (63% identity [21]) and with FosA from Tn2921 (33% identity [14]). Residues identical in all three protein sequences are shown in bold, and those residues identical between the more closely related FosB homologs are underlined in the B. subtilis sequence.

Transcription of fosB requires ς^W.

Previously, 15 ς^W-dependent operons were identified by searching the genome for sequences matching the ς^W autoregulatory site, P_w: TGAAAC N₁₆ CGTA (10). Additional candidate promoters, including one for fosB (Fig. 1A), were identified with 17-bp spacer regions (10).

To confirm the role of this predicted ς^W-dependent promoter, we generated a P_fosB-cat-lacZ operon fusion inserted ectopically in the SPβ prophage (HB8083; Table 1) and transduced the reporter fusion into wild-type, sigW, and rsiW mutant backgrounds. Promoter activity as determined in early-stationary-phase cells yielded 18.4 Miller units of β-galactosidase in the wild-type strain (HB0052), and this was reduced to background levels (∼1 unit) in the sigW mutant (HB0023). In the rsiW (anti-ς^W) mutant (HB0012), expression was elevated approximately twofold (30.5 units). This pattern is precisely that expected for a ς^W-dependent promoter.

TABLE 1.

B. subtilis strains, plasmids, and oligonucleotides used in this study

Strain, plasmid, or oligonucleotide	Genotype or characteristicsa	Reference, source, or derivationb
Strains
CU1065	W168 trpC2 attSPβ	Lab stock
JH642	W168 trpC2 pheA1	Lab stock
ZB307A	W168 SPβc2Δ2::Tn917::pSK10Δ6; MLSr	22
HB4247	JH642 sigW::kan	pKF90→JH642
HB0020	CU1065 sigW::kan	HB4247→CU1065
HB0010	CU1065 rsiW::kan	pXH51→CU1065
HB0008	CU1065 fosB::pAG4041 (Cmr)	pAG4041→CU1065
HB8083	ZB307A SPβ PfosB-cat-lacZ (MLSr Neor)	pAG3839→ZB307A
HB0052	CU1065 SPβPfosB-cat-lacZ (MLSr Neor)	Transduction
HB0023	HB0020 SPβ PfosB-cat-lacZ (MLSr Kanr)	Transduction
HB0012	HB0010 SPβ PfosB-cat-lacZ (MLSr Kanr)	Transduction
HB0050	CU1065 SPβ Pw-cat-lacZ (MLSr Neor)	9
HB0080	CU1065::pMC82 (PxylA-fosB)	This work
HB0081	HB0020::pMC82; sigW::kan (PxylA-fosB)	This work
HB0082	HB0008::pMC82; fosB::cat(PxylA-fosB)	This work
Plasmids
pET17b	T7 RNAP driven overexpression plasmid	Novagen
pXT	Derivative of pDG1731; allows fusion of genes to xylose-inducible xylA promoter; integrates at thrC locus	T. Msadek
pJPM122	Vector for integration of reporter fusions into SPβ (Apr Neor)	18
pJM114	Kanamycin resistance cassette vector	16
pGEM-cat-3Zf(+)	Cloning vector	20
pDG783	Kanamycin resistance cassette vector	6
pKF84	Contains sigW	Lab stock
pKF90	Contains sigW::kan	Construction analogous to sigW::MLS (reference 9)
pXH50	pGEM-cat-3Zf(+) carrying rsiW	Lab stock
pXH51	rsiW::kan in pGEM-cat-3Zf(+)	Lab stock
pAG4041	pGEM-cat-3Zf(+) carrying internal fragment (330 bp) of fosB (PCR primers 275, 276)	This work
pAG3839	pJPM122 carrying PfosB (PCR from 273, 274)	This work
pMC82	pXT carrying fosB (PCR from 470, 309)	This work
pMC50	pET17b with fosB (PCR from 308, 309)	This work
Oligonucleotidesc
275	CGGAATTCAGTTTCGCATTTGGATACA	fosB (internal; F)
276	CGTCTAGAAGCCTGTCTTGAAGGGTT	fosB (internal; R)
273	GCGAAGCTTCCGTTTTTGTTTACACTGGTA	fosB promoter F
274	GCGGATCCTAGCAAGTGATTGATTCCTTTTA	fosB promoter R
470	CGCGGATCCATTCATAATGGTCATGTT	fosB F
309	CCGGAATTCTGGTTGTGCTATCAAA	fosB R
370	CCCTTTACCAAAAGCTTTGCACC	fosB (for primer extension; R)
308	CTAGTCTAGACAGTCCGTTTTTGTT	fosB (for overproduction; F)

^aMLS, macrolides-lincosamides-streptogramin B; Neo, neomycin; Kan, Kanamycin, Ap, ampicillin; Cm, chloramphenicol. 

^bArrows indicate transformation with either plasmid or chromosomal DNA as indicated. 

^cIntroduced restriction sites used for cloning are underlined. F, forward; R, reverse (relative to fosB). 

We used reverse transcriptase primer extension mapping to identify the transcriptional start site for fosB as a G residue 10 bases downstream from the −10 region CGTA motif (Fig. 2). There were no other start sites visible in the primer extension experiment, which is consistent with the idea that ς^W is largely, if not exclusively, responsible for fosB transcription. The fosB gene is apparently monocistronic, as it is flanked on either side by genes transcribed from the complementary strand of the genome (Fig. 1A).

FIG. 2.

Expression of fosB depends on ς^W. Primer extension analysis of the fosB transcription start site is shown. RNA was isolated from logarithmically growing cells in Luria-Bertani medium of strains CU1065 (wild type), HB0020 (sigW mutant), and HB0010 (rsiW mutant). Equal amounts (100 μg) of total RNA were annealed with radiolabeled oligonucleotide no. 370 prior to analysis by reverse transcriptase primer extension. The transcription start site is the G residue shown in bold. G,A,T, and C are the sequencing ladder obtained with the same primer.

fosB and sigW mutants are sensitive to fosfomycin.

Both fosB (HB0008) and sigW (HB0020) mutants are fosfomycin sensitive: an MIC of 50 μg/ml for the mutants compared to 800 μg/ml for the wild type in liquid culture. Similarly, the sigW and fosB mutants have a much greater zone of growth inhibition in disk diffusion assays (∼25-mm zone for wild type versus >50 mm for the mutants). The fosB and sigW mutant strains did not display altered sensitivity to several other antibiotics, including vancomycin, cephalosporin C, penicillin G, d-cycloserine, tunicamycin, nisin, and bacitracin. Induction of fosB from a xylose-inducible promoter completely restores Fos^r to either the sigW mutant (HB0081) or, as expected, to the fosB mutant (HB0082). Thus, fosB is the only ς^W-dependent gene required for wild-type levels of Fos^r.

Expression of FosB confers fosfomycin resistance to Escherichia coli.

For mechanistic studies, we overproduced FosB in E. coli. Transformation of E. coli BL21/DE3(pLysS) with pMC50 leads to high fosfomycin resistance, even under noninducing conditions (MIC > 20 mg/ml, as judged using commercial antibiotic disks [Becton Dickinson, Cockeysville, Md.]). This is comparable to the Fos^r imparted by a similar FosA overexpression plasmid (MIC > 30 mg/ml). This suggests that an appropriate thiol cofactor for FosB is present in E. coli. To compare the relative efficacy of FosA and FosB in protecting E. coli against fosfomycin, the MIC determinations were repeated using plates containing 20 mM glucose-6-phosphate (an inducer of fosfomycin uptake). Under these conditions, FosA still supported high-level fosfomycin resistance (MIC, >30 mg/ml), while resistance of the strain expressing FosB was dramatically reduced (MIC, ∼1.25 mg/ml). This difference may relate to the lower catalytic efficiency of FosB compared to FosA (see below).

FosB: an Mg²⁺-dependent cysteine thiol transferase.

FosB was purified using modifications of the FosA procedure (3). Gel filtration chromatography indicated that FosB is a dimer in either the presence or absence of divalent metal ions. The metal ion selectivity of FosB was examined and found to be Ni²⁺ ∼ Mg²⁺ > Mn²⁺ > Fe²⁺ > Cu²⁺ > Ca²⁺ ∼ Co²⁺ > Zn²⁺ when screened with a fixed (0.5 mM) concentration of metal. Activation was almost 10-fold greater with Ni²⁺ and Mg²⁺ (Fig. 3) than with Mn²⁺. Although Ni²⁺ activates the enzyme at lower concentrations, the activation constant for Mg²⁺ (200 μM) is well below the prevailing Mg²⁺ concentration of about 1 mM. Therefore, we suggest that Mg²⁺ is the physiologically relevant metal. In contrast to FosA (2), FosB does not require a monovalent cation for optimal activity.

FIG. 3.

Activation of FosB with Ni²⁺ and Mg²⁺ using l-cysteine as the thiol substrate. The experimental data for Ni²⁺ (●) and Mg²⁺ (▪) were fit to the quadratic form of a simple binding isotherm to obtain the activation constant for the metal (K_act^M) and the maximum rate constant (k_max^M) for activation under the conditions of the assay. The solid lines are fits of the experimental data for the two metals, with K_act^Ni = 9 ± 3 μM, k_max^Ni = 1.0 ± 0.1 s⁻¹, K_act^Mg = 180 ± 30 μM, and k_max^Mg = 0.75 ± 0.04 s⁻¹. For comparison, the results for Mn²⁺ under the same conditions were K_act^Mn = 3.5 ± 0.9 μM and k_max^Mn = 0.11 ± 0.01 s⁻¹.

FosA functions as a glutathione transferase (3). Since B. subtilis, like many gram-positive bacteria, lacks detectable levels of glutathione (5, 15), it seems likely that FosB must use a different thiol. To identify the FosB thiol cofactor, the rate of appearance of conjugates of fosfomycin with various thiols was determined as described previously (3, 4). l-Cysteine and coenzyme A sulfhydryl (CoASH) are two abundant thiols in gram-positive bacteria (5, 15). Of the two, only l-cysteine supports a modest enzyme activity (Table 2). The product of the FosB-catalyzed addition of l-cysteine to fosfomycin, examined by heteronuclear multiple-bond correlation nuclear magnetic resonance spectroscopy, is (1R,2S)-1-(S-l-cysteinyl)-2-hydroxypropylphosphonate (data not shown), identical to the product produced by FosA with l-cysteine (4). Extended incubations of FosB, fosfomycin, and various divalent metal ions indicated that no degradation of the antibiotic occurred in the absence of a thiol substrate.

TABLE 2.

Steady-state kinetic parameters of the thiol substrate and metal-ion selectivity of FosB

Thiol substrate	Metal ion	kcat (s−1)	kcat/Kmthiol (M−1 s−1)	Kmthiol (mM)
l-Cys	Mg2+	6.3 ± 0.3	180 ± 20	35 ± 3
GSH	Mg2+	0.027 ± 0.002	1.8 ± 0.2	15 ± 2
CoASH	Mg2+	—a	0.40 ± 0.03	>50
l-Cys	Ni2+	7.8 ± 0.5	190 ± 30	41 ± 3
GSH	Ni2+	0.066 ± 0.005	9.1 ± 1.7	7 ± 1
l-Cys	Mn2+	—a	6.9 ± 0.2	>200
GSH	Mn2+	—a	0.093 ± 0.001	>50
CoASH	Mn2+	—a	0.0009 ± 0.0001	>100

^aA linear dependence on the thiol substrate concentration was observed, precluding a determination of k_cat. Therefore, a conservative lower limit for K_m^thiol was estimated from the substrate concentration range used. 

Although with glutathione the catalytic efficiency of FosB is significantly less than that of FosA (k_cat/K_m = [1.7 ± 0.3] × 10⁵ M⁻¹ s⁻¹), they are about equally active with l-cysteine (k_cat/K_m = 410 ± 40 M⁻¹ s⁻¹). These rate constants are still at least 10⁶ to 10⁸ greater than those reported for the spontaneous reactions (2). The optimal catalytic efficiency of FosB (with l-cysteine) is far lower than FosA efficiency (with glutathione), which may reflect an intrinsic difference in catalytic efficiency or could indicate that a physiologically relevant cofactor for FosB has not yet been identified. To verify that l-cysteine is the thiol donor in vivo, it will be necessary to characterize the product of fosfomycin inactivation from intact cells.

Summary.

Fosfomycin is a clinically important, broad-spectrum antibiotic that irreversibly inactivates MurA, which catalyzes the first committed step in peptidoglycan biosynthesis (17). Resistance arises predominantly via mutations in the chromosomally encoded transport pathways (7) or by resistance genes found on transmissible plasmids. At least two classes of plasmid-borne Fos^r determinants have been described (reviewed in reference 19). The best characterized, fosA, encodes a Mn²⁺-dependent glutathione transferase (1–3). A related resistance gene, fosB, is from an S. epidermidis plasmid. FosB is 38% identical to FosA, suggesting a similar mechanism of action (21), as confirmed in this study. Indeed, the plasmid-borne fosA resistance determinants may have arisen from chromosomal genes, like fosB, that serve a defensive role within soil microorganisms such as B. subtilis.

There are several significant mechanistic differences between FosA and FosB. FosB, unlike FosA, does not function efficiently with glutathione and instead appears to use l-cysteine as the physiological thiol donor. Analyses of S. epidermidis FosB have also indicated that glutathione is not involved in detoxification (R. Asano and J. Davies, personal communication). A second mechanistic difference is that FosB uses Mg²⁺ rather than Mn²⁺ as metal cofactor. The final notable difference between FosA and FosB is catalytic efficiency. FosA exhibits a very high catalytic efficiency (k_cat/K_m), perhaps in response to selection pressures imposed by the clinical use of fosfomycin.

The assignment of fosB to the ς^W regulon further validates the “consensus search” approach for defining alternative ς factor regulons (9–11), and it supports the emerging picture of ς^W as a regulator of a broad, “antibiosis” regulon involved in both the production of, and defense against, antimicrobial compounds.