An Extracytoplasmic Function Sigma Factor Controls β-Lactamase Gene Expression in Bacillus anthracis and Other Bacillus cereus Group Species

Abstract

The susceptibility of most Bacillus anthracis strains to β-lactam antibiotics is intriguing considering that the closely related species Bacillus cereus and Bacillus thuringiensis typically produce β-lactamases and the B. anthracis genome harbors two β-lactamase genes, bla1 and bla2. We show that β-lactamase activity associated with B. anthracis is affected by two genes, sigP (BA2502) and rsiP (BA2503), predicted to encode an extracytoplasmic function sigma factor and an anti-sigma factor, respectively. Deletion of the sigP-rsiP locus abolished β-lactamase activity in a naturally occurring penicillin-resistant strain and had no effect on β-lactamase activity in a prototypical penicillin-susceptible strain. Complementation with sigP and rsiP from the penicillin-resistant strain, but not with sigP and rsiP from the penicillin-susceptible strain, conferred constitutive β-lactamase activity in both mutants. These results are attributed to a nucleotide deletion near the 5′ end of rsiP in the penicillin-resistant strain that is predicted to result in a nonfunctional protein. B. cereus and B. thuringiensis sigP and rsiP homologues are required for inducible penicillin resistance in these species. Expression of the B. cereus or B. thuringiensis sigP and rsiP genes in a B. anthracis sigP-rsiP-null mutant confers inducible production of β-lactamase activity, suggesting that while B. anthracis contains the genes necessary for sensing β-lactam antibiotics, the B. anthracis sigP and rsiP gene products are not sufficient for bla induction.

Bacillus anthracis, Bacillus cereus, and Bacillus thuringiensis are the best-studied members of the B. cereus group. Extensive genomic studies, including DNA-DNA hybridization, 16S and 23S rRNA sequence comparisons, multilocus sequence typing, multilocus enzyme electrophoresis, and amplified fragment length polymorphism analysis, have revealed a high degree of phylogenetic relatedness among these organisms, leading to the proposal that B. anthracis, B. cereus, and B. thuringiensis may be viewed as a single species (7, 8, 12, 23, 28, 46-48, 51, 56). Despite the remarkably high degree of DNA sequence similarity and synteny among these species, there are notable species-specific phenotypes, many of which are attributed to plasmid content. B. anthracis, the causative agent of anthrax, typically contains two virulence plasmids, pXO1 and pXO2, which harbor genes encoding toxins and capsule biosynthetic enzymes, respectively (57, 73). Large, transmissible plasmids associated with the insect pathogen B. thuringiensis usually contain genes encoding insect toxins, which are produced as large parasporal inclusions (40, 77). Strains of B. cereus, a ubiquitous soil bacterium and opportunistic pathogen, can harbor a variety of extrachromosomal elements, including plasmids with high levels of similarity to pXO1 (44, 52).

In addition to plasmid-associated traits, some species-specific phenotypes are due to differential expression of chromosomal genes. B. cereus and B. thuringiensis have certain phospolipase, hemolysin, protease, and β-lactamase activities that are generally not associated with B. anthracis. The pleiotropic transcriptional regulator PlcR activates expression of genes associated with some of these phenotypes in B. cereus and B. thuringiensis. The B. anthracis plcR gene contains a nonsense mutation that most likely results in a protein product that cannot activate expression of target genes, leading to dramatic reductions in lecithinase, protease, and hemolytic activities associated with this species (2, 37, 38, 60, 80).

Penicillin susceptibility is a characteristic of B. anthracis that is commonly used to distinguish this species from B. cereus and B. thuringiensis, which typically exhibit inducible penicillin resistance (15, 82). The genetic basis for β-lactam resistance in this group of organisms is unknown (16). Interestingly, all sequenced strains of B. anthracis, B. cereus, and B. thuringiensis contain β-lactamase genes. Previous work in our laboratory showed that the two β-lactamase genes, bla1 and bla2, of a prototypical penicillin-susceptible strain of B. anthracis are transcriptionally silent, while the bla genes of a penicillin-resistant B. anthracis clinical isolate, strain 32, are expressed constitutively. Our studies revealed that differential expression of the bla genes is dependent on strain background and that bla1 is the major contributor to the high-level resistance to ampicillin and piperacillin in strain 32 (25).

In work presented here, we show that β-lactamase activity in B. anthracis and its close relatives B. cereus and B. thuringiensis is associated with an extracytoplasmic function (ECF) sigma factor. The ECF sigma factors are a subfamily of alternative sigma factors found in a diverse array of bacteria. The products of ECF sigma factor-regulated genes are associated with a wide range of functions, including responses to heat, osmotic and oxidative stress, and synthesis of alginate or carotenoids. ECF sigma factors have also been shown to mediate resistance to antimicrobial compounds in Bacillus subtilis (4, 5, 13, 14, 17, 18, 20-22, 72, 81). We establish that an ECF sigma factor and its cognate anti-sigma factor are associated with inducible resistance to β-lactam antibiotics in B. cereus and B. thuringiensis, but the ECF sigma factor and anti-sigma factor genes are not sufficient for β-lactamase gene expression in B. anthracis. We also show that the constitutive β-lactamase activity of an unusual penicillin-resistant clinical isolate of B. anthracis is likely due to the production of a truncated anti-sigma factor which cannot sequester its cognate ECF sigma factor. Our data reveal yet another phenotype commonly used to differentiate B. anthracis from B. cereus and B. thuringiensis that can be attributed to differences in trans-acting factors produced by these species.

MATERIALS AND METHODS

Media and growth conditions.

B. anthracis, B. cereus, and B. thuringiensis were cultured in LB medium (Difco, Detroit, MI) containing 0.5% glycerol at 30°C with shaking (200 rpm). Following overnight incubation, cultures were transferred to fresh LB medium containing 0.5% glycerol such that the starting optical density at 600 nm (OD₆₀₀) was 0.06 to 0.08. Cultures were incubated at 37°C with shaking, and samples were obtained at the late exponential growth phase (OD₆₀₀, 2.9 to 3.6). When appropriate, erythromycin (Fisher, Pittsburgh, PA) was used at a final concentration of 5 μg/ml.

To assess bla1 induction, strains were inoculated into LB medium such that the starting OD₆₀₀ was 0.1. Prior to incubation, cultures were divided into two equal portions. Water or ampicillin (0.1 μg/ml) was added. Following growth at 37°C for 2 h, samples of culture supernatants were assessed for bla1 expression.

DNA isolation and manipulation.

A Mo Bio genomic isolation kit (Mo Bio Laboratories, Solana Beach, CA) was used to obtain chromosomal DNA from B. anthracis, B. cereus, and B. thuringiensis according to the manufacturer's instructions. All PCRs were performed using the high-fidelity cloning enzymes Easy-A (Stratagene, La Jolla, CA) and Phusion (New England Biolabs, Ipswich, MA). The PCR products obtained using the Phusion enzyme were incubated for 30 min at 72°C in the presence of 1 U of Taq polymerase (New England Biolabs, Ipswich, MA) and 200 nM dATP (Invitrogen, Carlsbad, CA) before the clean-up step. The PCR products were purified using a DNA Clean and Concentrator-5 kit (Zymo Research, Orange, CA). Transformation of Escherichia coli and subsequent extraction of plasmid DNA were performed using standard procedures (9). Unmethylated plasmid DNA was electroporated into Bacillus species using a method described previously (58).

A QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used to create plasmids pUTE697 and pUTE698 (Table 1). Primer pairs CR95/CR97 and CR104/CR105 (see Table S1 in the supplemental material) annealed to opposite strands of template vectors and created new plasmids that contained the desired point mutations as described previously (75).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristic(s)a	Reference or source
B. anthracis strains
32	pXO1+ pXO2+, Apr	61
UT308	Derived from 32, pXO1− pXO2−, Apr	This study
UT335	Derived from UT308, ΔsigP(r) rsiP(r), Aps	This study
9131	pXO1− pXO2−, Aps	34
UT334	Derived from 9131, ΔsigP(s) rsiP(s), Aps	This study
B. cereus strains
569	Apr	11
UTC4	Derived from 569, ΔsigPBcPrsiPBc, Aps	This study
B. thuringiensis strains
AW43	Apr	87
UTT4	Derived from AW43, ΔsigPBtrsiPBt, Aps	This study
E. coli strains
TG1	Cloning host	76
JM109	Cloning host	76
GM2163	dam dcm	74
Plasmids
pGEM-T Easy	T/A cloning vector	Promega
pHT304	Low-copy-number vector	6
pUTE692	pHT304 carrying sigP(s) and rsiP(s) from 9131	This study
pUTE693	pHT304 carrying sigP(r) and rsiP(r) from UT308	This study
pUTE697	Derived from pUTE693, contains sigP(r) from UT308 and rsiP(s) from 9131	This study
pUTE698	Derived from pUTE692, contains sigP(s) from 9131 and rsiP(r) from UT308	This study
pUTE751	pHT304 carrying gfpmut3a	This study
pUTE705	pHT304 carrying sigP(r) from UT308 and gfpmut3a fused to first ATG of rsiP(r)	This study
pUTE707	pHT304 carrying sigP(r) from UT308 and gfpmut3a fused to second ATG of rsiP(r)	This study
pUTE708	pHT304 vector carrying sigP(s) from 9131 and gfpmut3a fused to second ATG of rsiP(s)	This study
pUTE728	pHT304 carrying sigPBc and rsiPBc from B. cereus 569	This study
pUTE729	pHT304 carrying sigPBt and rsiPBt from B. thuringiensis AW43	This study
pHT304-18Z	Low-copy-number vector containing promoterless lacZ gene	6
pUTE461	pHT304-18Z carrying bla1::lacZ transcriptional fusion	24
pUTE462	pHT304-18Z carrying bla2::lacZ transcriptional fusion	24
pUTE869	pHT304 containing sigP(s) from 9131 and rsiPBc from B. cereus 569	This study
pUTE871	pHT304 containing sigP(r) from UT308 and and rsiPBc from B. cereus 569	This study

^aAp^r, ampicillin resistant; Ap^s, ampicillin sensitive. Subscripts associated with genes indicate gene origin, as follows: r, penicillin-resistant B. anthracis; s, penicillin-susceptible B. anthracis. Bc, B. cereus; Bt, B. thuringiensis.

Strain and plasmid construction.

Strains and plasmids are listed in Table 1. Plasmids pXO1 and pXO2 were cured from B. anthracis 32 during growth in the presence of novobiocin, as described previously (41), to create strain UT308. The sigP-rsiP (BA2502-BA2503) gene pairs from the prototypical penicillin-susceptible strain B. anthracis 9131 [designated sigP(s) and rsiP(s)] and the penicillin-resistant strain UT308 [designated sigP(r) and rsiP(r)] and the homologues from B. cereus 569 and B. thuringiensis AW43 were replaced with an Ωkm cassette to create strains UT334, UT335, UTC4, and UTT4, respectively, using a protocol described previously (29). DNA sequences upstream from the sigP genes from nucleotide (nt) −809 to nt 4 relative to the translational start were amplified using CR139 (B. anthracis) or CR141 (B. cereus and B. thuringiensis) in combination with primer sigR4 (see Table S1 in the supplemental material). DNA sequences from nt −615 to nt 928 relative to the translational stop of the rsiP genes were amplified using primer CR162 in combination with sigR8 (B. anthracis) or CR143 (B. cereus and B. thuringiensis) (see Table S1 in the supplemental material). PCR products were cloned in pUTE583 such that upstream and downstream regions flanked the aph3A gene encoding kanamycin resistance. The aph3A gene was derived from pUC4, which was kindly provided by Gary Dunny. Constructs were electroporated into B. cereus 569, B. thuringiensis AW43, and B. anthracis 9131 and UT308.

Plasmids pUTE728, pUTE729, pUTE692, and pUTE693 contain the sigP-rsiP gene pairs and promoter regions from B. cereus 569, B. thuringiensis AW43, B. anthracis 9131, and B. anthracis UT308, respectively. PCR products spanning the region from 249 nt upstream of the predicted translational start site of sigP to 97 nt downstream of the apparent rsiP translational stop codon were amplified using oligonucleotide primers CR103 and rsrA2 and template DNA extracted from B. anthracis 9131 and UT308 (accession number EU979636). Similar regions were amplified from B. cereus 569 DNA (accession number EU979634) and B. thuringiensis AW43 DNA (accession number EU979635) using primers CR153 and CR155. The PCR products were purified, ligated into pGEM-T Easy (Promega, Madison, WI), and cloned into E. coli TG1 or JM109. PstI/SphI fragments containing the gene pairs were subcloned into pHT304 (6).

The cloning strategy described above was used to create plasmids pUTE705, pUTE707, pUTE708, and pUTE751. Briefly, primers CR103 and CR134 and DNA from UT308 were used to amplify DNA containing sigP(r), the first 9 nt of rsiP(r), and the upstream promoter region from UT308 DNA. This fragment, digested with PstI and SphI, was simultaneously ligated into pHT304 with an SphI/HindIII fragment containing the gfpmut3a gene to create plasmid pUTE705 (84). Similarly, primer CR103 was used in combination with CR106 to create a PCR fragment containing sigP and 53 nt of rsiP amplified from UT308 and 9131 DNA to create plasmids pUTE707 and pUTE708, respectively. Finally, plasmid pUTE651 was created by introducing the same gfpmut3a gene, amplified from plasmid pAD123 (33) using primers CR126 and CR127, into pHT304 previously digested with SphI and HindIII.

Plasmids pUTE864 [sigP(s) rsiP_Bc] and pUTE866 [sigP(r) rsiP_Bc] were created using overlap extension PCR. Briefly, DNA corresponding to the sigP genes in B. anthracis 9131 and UT308 were amplified using primers CR103 and CR176 (see Table S1 in the supplemental material), while DNA corresponding to the rsiP gene of B. cereus was amplified using primers CR175 and CR155 (see Table S1 in the supplemental material). PCR products corresponding to sigP from either B. anthracis strain were mixed with the PCR product corresponding to the B. cereus rsiP gene without additional primers in a PCR for two cycles, during which the homologous regions of primers CR176 and CR175 were allowed to anneal. Finally, primers CR103 and CR155 were added to the PCR mixture for an additional 25 cycles to create PCR products containing sigP from the B. anthracis penicillin-susceptible or -resistant strain in combination with the rsiP gene from B. cereus 569. These PCR products were cloned into pHT304 using the procedure described above.

The DNA associated with recombinant loci in mutant isolates and all cloned DNA were sequenced to confirm the fidelity of the constructs.

5′ end mapping.

RNA was extracted from UT308 (Ap^r) cells grown to the late exponential phase as described previously (43). The 5′ ends of the bla1 and sigP transcripts were determined using the 5′ rapid amplification of cDNA ends system (version 2.0; Invitrogen, Carlsbad, CA) according to the manufacturer's instructions, with the following modifications. The bla1 and sigP primers (CR129 and CR132, respectively [see Table S1 in the supplemental material]) were incubated at 70°C for 10 min and then hybridized to 1 to 2 μg of RNA overnight at 30°C in S1 hybridization buffer. Following ethanol precipitation of the RNA-primer complex, the Invitrogen protocol was followed.

β-Lactamase assays.

Supernatants from cultures in the late exponential growth phase were tested for β-lactamase activity using the chromogenic substrate nitrocefin (Clabiochem, San Diego, CA, and BD, San Jose, CA). Supernatant samples were diluted 10-fold. Ten microliters of diluted supernatant was added to a 100 mM nitrocefin solution according to the manufacturer's instructions. Following incubation for 30 min at 37°C, the absorbance at 486 nm was assessed using a Thermo Electron Multiskan Spectrum and the SkanIt 2.2 software. The β-lactamase activity of the parent strain, UT308, was defined as 100 arbitrary units. Samples from three independent cultures were tested.

Fluorescence microscopy.

Samples from cultures in the late exponential growth phase were placed on a glass slide, covered with a coverslip, and visualized using a fluorescein isothiocyanate filter with a Nikon Eclipse E600 microscope.

β-Galactosidase assays.

Samples were collected from B. cereus and B. thuringiensis cultures during late exponential growth. The 1-ml samples were pelleted, resuspended in 1 ml of Z-buffer, and homogenized using a Mini-bead Beater 8 (Biospec Products, Bartlesville, OK). The beads were removed by centrifugation (15,000 × g for 3 min), and an aliquot of the remaining sample was assayed for β-galactosidase activity as described by Miller (67). Three independent cultures were assessed for β-galactosidase activity. For B. anthracis, cultures were grown in the presence or absence of ampicillin and 1-ml samples were collected after 4 h of growth at 37°C. β-Galactosidase activity was assessed using the procedure described above.

Determination of MICs of ampicillin.

B. anthracis, B. cereus, and B. thuringiensis spores were streaked onto Mueller-Hinton (Fluka-Biochemica, Buchs, Switzerland) agar plates and incubated overnight at 37°C. Colonies were used to inoculate phosphate-buffered saline to obtain turbidity equivalent to 0.5 McFarland standard. Per the instructions of the Etest strip manufacturer (AB-Biodisk, Piscataway, NJ), cell suspensions were applied to 1-day-old 90-mm plates containing Mueller-Hinton agar at a depth of 4.0 ± 0.5 mm and erythromycin at a final concentration of 5 μg/ml. Sterile swabs were used to inoculate the plates for confluent growth, and Etest strips were placed individually onto the freshly swabbed plates. The MICs were determined after incubation at 37°C for 20 h.

RESULTS

bla gene expression by a penicillin-resistant B. anthracis strain is associated with an ECF sigma factor.

We reported previously that prototypical penicillin-susceptible B. anthracis strains harbor two β-lactamase genes, bla1 and bla2, which are transcriptionally silent even in the presence of subinhibitory levels of β-lactam antibiotics. We also demonstrated that the β-lactamase activity produced by the atypical clinical isolate B. anthracis strain 32 is due to high-level, constitutive expression of bla1 and bla2. Although sequence differences immediately upstream of both bla genes are apparent, they are not responsible for the differences in bla gene expression in the penicillin-susceptible and -resistant strains (25).

In an effort to identify trans-acting regulators of the bla genes, the predicted functions of neighboring open reading frame (ORFs) were investigated. Located approximately 5 kb upstream from the bla1 gene are two overlapping ORFs predicted to encode an ECF sigma factor (BA2502) and an anti-sigma factor (BA2503) (Fig. 1). As is true for other ECF sigma factors, the predicted amino acid sequence of BA2502 contains well-conserved domains, region 2 and region 4, that interact with the β′ and β subunits of RNA polymerase and recognize the −10 and −35 promoter elements, respectively (19, 71, 85, 89). In addition, the apparent BA2502 amino acid sequence lacks region 3, a domain that is present in other subclasses of sigma factors but is not present in ECF sigma factors (62). The location of the BA2503 ORF and its predicted cotranscription and cotranslation with BA2502 make it an ideal candidate for an anti-sigma factor-encoding gene. Furthermore, the predicted amino acid sequence appears to contain cytoplasmic, membrane-spanning, and extracytoplasmic domains, consistent with the sequences of other anti-sigma factors (31, 69, 70).

FIG. 1.

Schematic representation of the bla1 locus. The sigP (BA2502) and rsiP (BA2503) genes are located approximately 5 kb upstream of the bla1 gene. Asterisks indicate the locations of the sequence differences between the sigP-rsiP gene pairs from the penicillin-resistant strain B. anthracis 32 and the prototypical penicillin-susceptible strain 9131.

To determine if the BA2502-BA2503 locus affects bla gene expression, plasmids harboring bla1- and bla2-lacZ transcriptional fusions were introduced individually into UT308, a pXO1⁻ pXO2⁻ derivative of the penicillin-resistant strain B. anthracis 32, and UT335, an isogenic mutant lacking both ORFs. Only the UT308 strains containing the bla1 or bla2 transcriptional fusions exhibited β-galactosidase activity, demonstrating that the locus is required for bla gene expression in the penicillin-resistant strain background (Table 2). We also determined that deletion of the two ORFs abolished β-lactamase activity (Fig. 2) and resistance to the β-lactam antibiotic ampicillin (Table 3).

TABLE 2.

Expression of the B. anthracis bla1 and bla2 promoters in the penicillin-resistant strain UT308 and the isogenic sigP-rsiP-null strain UT335^a

Strain	Expression (Miller units)
	bla1::lacZ (pUTE461)	bla2::lacZ (pUTE462)	lacZ (pHT304-18Z)
UT308	510.36 ± 116.84	728.16 ± 36.37	1.64 ± 0.46
UT335	1.19 ± 0.19	1.68 ± 0.58	1.87 ± 0.37

^aThe plasmids contained bla1 and bla2 promoters fused to a promoterless lacZ gene as indicated. The values are means ± standard deviations.

FIG. 2.

β-Lactamase production by B. anthracis strains in the presence and absence of sigP-rsiP gene pairs. β-Lactamase activities were determined using culture supernatants from parent strain 9131 (Pen^s) and the corresponding ΔsigP ΔrsiP mutant UT334, parent strain UT308 (Pen^r) and the corresponding ΔsigP ΔrsiP mutant UT335, and ΔsigP ΔrsiP mutants complemented with pUTE692 (sigP and rsiP from the Pen^s strain), pUTE693 (sigP and rsiP from the Pen^r strain), and pHT304 (empty vector). Black bars indicate a Pen^s strain (9131) background, and gray bars indicate a Pen^r strain (UT308) background. The error bars represent standard deviations. Bkgrd, background.

TABLE 3.

MICs of ampicillin for B. anthracis, B. cereus, and B. thuringiensis parent and sigP-rsiP-null strains

Strain	MICs of ampicillin (μg/ml) with the following sources of sigP and rsiP:
	Vector only (pHT304)	B. anthracis 9131(pUTE692)	B. anthracis UT308(pUTE693)	B. cereus 569(pUTE728)	B. thuringiensis AW43(pUTE729)
B. anthracis 9131 (parent)	0.032-0.047
B. anthracis UT334 (ΔsigP ΔrsiP)	0.023-0.047	0.032-0.047	192-256	1.5-6	1-3
B. anthracis UT308 (parent)	192->256
B. anthracis UT335 (ΔsigP ΔrsiP)	0.032	0.032-0.047	128->256	0.023-0.047	0.023-0.047
B. cereus 569 (parent)	32-48
B. cereus UTC4 (ΔsigP ΔrsiP)	0.047-0.064	0.047-0.064	64-128	48-96	16
B. thuringiensis AW43 (parent)	32-96
B. thuringiensis UTT4 (ΔsigP ΔrsiP)	0.047-0.064	0.25-2	64-128	32-48	3-8

To establish the effects of the BA2502-BA2503 locus on bla gene expression in a prototypical B. anthracis strain, the tandem ORFs were deleted from B. anthracis 9131 (34), a prototypical penicillin-susceptible strain lacking pXO1 and pXO2, to create UT334. We compared the β-lactamase activities of equal volumes of culture supernatants from the parent strains 9131 (Pen^s) and UT308 (Pen^r) and the BA2502-BA2503 deletion mutants UT334 and UT335, respectively, using the chromogenic substrate nitrocefin. As indicated above, deletion of the gene pair from the penicillin-resistant strain abolished β-lactamase activity (Fig. 2), whereas the same mutation in the penicillin-susceptible strain did not alter β-lactamase activity; β-lactamase activity remained undetectable in culture supernatants. Furthermore, complementation of UT334 and UT335 with pUTE693, a low-copy-number vector containing BA2502 and BA2503 from penicillin-resistant strain UT308, conferred β-lactamase activity similar to that of UT308 in both recombinant strains. Complementation of UT334 and UT335 with pUTE692, which contains the corresponding ORFs from penicillin-susceptible strain 9131, did not result in enzyme activity (Fig. 2). These data indicate that the products of BA2502 and/or BA2503 in the penicillin-susceptible and penicillin-resistant strains 9131 and UT308 are functionally different.

Considering the availability of sigma factor gene (sig) designations in Bacillus gene nomenclature and our data indicating that the presence of BA2502 and BA2503 affects resistance to the β-lactam antibiotic penicillin, we designated the two genes sigP and rsiP (repressor of sigP), respectively.

Conserved sequences 5′ of the bla and sigP transcriptional start sites.

Genes transcribed by specific sigma factors have conserved DNA recognition sequences 5′ of their transcription start sites. Since ECF sigma factor genes are typically autoregulated, the specific recognition sequences are also in the promoter regions of the ECF sigma factor genes (49). We compared the DNA sequences 5′ of the bla1, bla2, and sigP coding regions and discovered two common sequences, as shown in Fig. 3. A conserved 10-bp sequence, 5′-ATGGAACAAA-3′, includes an upstream G region and an AAC motif, features typically found in the −35 binding elements recognized by ECF sigma factors (59). A common 7-bp consensus sequence, 5′-TTGTCTA-3′, is located 13 bp from the putative −35 binding element, consistent with a potential −10 element.

FIG. 3.

Conserved sequences in the bla1, bla2, and sigP promoter regions. The putative −35 and −10 sequences are indicated by boldface type. The conserved G residue and AAC motif of the putative −35 region are underlined in the consensus sequence. Putative transcriptional start sites of bla1 and sigP, as determined from the results of 5′-end mapping, are indicated by boldface type and underlining.

We attempted to map the 5′ ends of bla1, bla2, and sigP gene transcripts using RNA from the β-lactamase-positive penicillin-resistant strain B. anthracis 32. We detected transcripts with 5′ ends mapping 8 to 9 nt from the putative −10 element of the sigP promoter and 8 and 15 nt from the putative −10 element of the bla1 promoter (Fig. 3 and data not shown). Despite repeated attempts, we were unable to determine the 5′ end(s) of bla2 transcripts. Nevertheless, the locations of the putative transcriptional start sites relative to the consensus sequences indicate that the conserved DNA sequences are likely sigma factor-specific −35 and −10 binding elements.

The penicillin-resistant B. anthracis strain produces a truncated anti-sigma factor.

To further investigate the different β-lactamase phenotypes conferred by the sigP and rsiP genes of the penicillin-susceptible and penicillin-resistant strains, we compared the nucleic acid sequences of the gene pairs and the amino acid sequences of the predicted gene products. DNA sequencing results and subsequent alignments of the genes from the penicillin-susceptible and penicillin-resistant strains revealed two nucleotide sequence differences (Fig. 1; see Fig. S1 in the supplemental material). One difference occurs within sigP and is predicted to result in a single amino acid difference; amino acid 24 of σ^P from the penicillin-susceptible strain is aspartic acid, while amino acid 24 of σ^P from the penicillin-resistant strain is glycine (see Fig. S2 in the supplemental material). This amino acid difference could affect protein activity since it occurs within region 2, a conserved domain in all sigma factors that contains residues important for interactions with the β′ subunit of polymerase, as well as the −10 promoter element (42, 53, 54, 71, 79, 85, 89). For clarity, we designated the sigP gene from the penicillin-susceptible strain “sigP(s)” and the sigP gene from the penicillin-resistant strain “sigP(r).”

The other nucleic acid sequence difference is located 17 nt downstream from a predicted translational start codon of rsiP (Fig. 1; see Fig. S1 in the supplemental material). In the penicillin-resistant strain, a nucleotide deletion results in a frameshift mutation predicted to result in a truncated, 12-amino-acid protein (see Fig. S3 in the supplemental material). Only the first 5 amino acids of the 12-amino-acid protein would resemble the protein produced by the prototypical penicillin-susceptible strain initiating translation from the same position. However, in sequences from these strains we also noted a methionine codon located 51 and 50 nt downstream from the first predicted translational start codon in the penicillin-susceptible and penicillin-resistant strains, respectively. The presence of a ribosome binding site upstream from either ATG codon was not apparent. In both strains, translation initiation from the second ATG, which is in frame with the first ATG, would result in an amino-terminally truncated protein lacking the first 17 amino acids. Typically, the amino-terminal domains of this class of anti-sigma factors reside in the cytoplasm, where they bind and sequester the cognate ECF sigma factor under normal growth conditions (31, 69). Moreover, expression of the cytoplasmic N terminus alone can prevent the ECF sigma factor from interacting with polymerase. In the case of σ^E and RseA, an ECF sigma factor and anti-sigma factor pair found in E. coli, only the first 97 amino acids of the anti-sigma factor are required to inhibit the interaction of the sigma factor and RNA polymerase (69).

To determine the translation initiation codon for rsiP, we created translational fusions in which the green fluorescent protein (GFP) gene gfpmut3 lacking its native start codon was cloned downstream of the first ATG (pUTE705) and downstream of the second ATG using sequences from the penicillin-susceptible strain (pUTE708) and from the penicillin-resistant strain (pUTE707). Since each translational fusion construct contains the sigP coding region, GFP synthesis was assessed in the sigP-rsiP-null strains UT334 (derived from the Pen^s parent) and UT335 (derived from the Pen^r parent), using fluorescence microscopy and Western hybridization. In both mutants, the presence of pUTE705 or pUTE708 resulted in GFP production, while the presence of pUTE707 did not appear to result in synthesis of GFP (Fig. 4). A translational fusion similar to the one in pUTE705 containing sigP from the penicillin-susceptible strain instead of the resistant strain was also created, but we were unable to introduce it into the UT335 strain. Polyclonal anti-GFP serum reacted with cell extracts from strains that exhibited fluorescence, but it did not react with extracts from nonfluorescent cells, even when increasing amounts of protein extract were used (data not shown). These results indicate that translation initiation occurs at the first ATG in both strain backgrounds and predict that a full-length and functional anti-sigma factor is present in the prototypical penicillin-susceptible B. anthracis strain and a 12-amino-acid peptide is present in the penicillin-resistant strain. For clarity, we designated the rsiP gene from the penicillin-susceptible strain “rsiP(s)” and the rsiP gene from the penicillin-resistant strain “rsiP(r).”

FIG. 4.

Fluorescence microscopy of strains UT334 (ΔsigP ΔrsiP) and UT335 (ΔsigP ΔrsiP) harboring GFP translational fusions, observed using light and a fluorescein isothiocyanate filter. The fusions contained sigP genes from a Pen^s or Pen^r (asterisk) background and the 5′ end of the rsiP gene from Pen^s and Pen^r (asterisk) backgrounds. Plus and minus signs indicate that the strains are β-lactamase positive and β-lactamase negative, respectively.

Finally, we used site-directed mutagenesis to create plasmids containing hybrid sigP-rsiP gene pairs. As indicated in Table 4, pUTE697 harbors sequences corresponding to sigP(r) and rsiP(s), while pUTE698 harbors sigP(s) and rsiP(r). The plasmids were introduced into UT334 [sigP(s)-rsiP(s)-null strain], and β-lactamase activity was measured. Several attempts were made to introduce these constructs into the UT335 strain, but a UT335 derivative containing sigP from the penicillin-susceptible strain could not be obtained. As expected, the recombinant strain containing rsiP(r) exhibited β-lactamase activity, while strains harboring rsiP(s) did not exhibit β-lactamase activity (Table 4). This result is consistent with the synthesis of a truncated nonfunctional anti-sigma factor by the penicillin-resistant strain.

TABLE 4.

β-Lactamase activity of UT334 (sigP-rsp-null strain) containing sigP and rsiP alleles^a

Background	Plasmid	Cloned sigP allele	Cloned rsiP allele	β-Lactamase activity
UT308 (Penr)	pHT304			+
9131 (Pens)	pHT304			−
UT334	pUTE692	sigP(s)	rsiP(s)	−
UT334	pUTE693	sigP(r)	rsiP(r)	+
UT334	pUTE697	sigP(r)	rsiP(s)	−
UT334	pUTE698	sigP(s)	rsiP(r)	+

^aThe sequences of the cloned sigP and rsiP alleles are identical to the sequences in the penicillin-resistant [sigP(r) or rsiP(r)] or penicillin-susceptible [sigP(s) or rsiP(s)] strains. β-Lactamase activity in culture supernatants is positive or negative.

bla gene induction and expression in B. cereus and B. thuringiensis require homologous sigP and rsiP loci.

The penicillin susceptibility of prototypical B. anthracis and constitutive β-lactamase synthesis by the unusual penicillin-resistant B. anthracis strain are interesting, considering that the closely related species B. cereus and B. thuringiensis commonly exhibit inducible β-lactamase activity. A protein BLAST search (BlastP) revealed that most sequenced B. cereus and B. thuringiensis strains contain a gene pair that is nearly identical in sequence and location to the B. anthracis sigP and rsiP genes. We cloned and sequenced the sigP and rsiP loci of the common laboratory strains B. cereus 569 and B. thuringiensis AW43 (see Fig. S2 and S3 in the supplemental material). To determine if these homologues affect β-lactamase expression, we replaced the B. cereus 569 and B. thuringiensis AW43 sigP-rsiP gene pairs with an aph3 gene, encoding kanamycin resistance, to create strains UTC4 and UTT4, respectively. These mutants were unable to grow on media containing a β-lactam antibiotic, a phenotype that was complemented by introducing the native B. cereus 569 and B. thuringiensis AW43 sigP and rsiP genes in trans (Table 3).

To establish whether B. anthracis bla gene expression could be induced in these species, we introduced B. anthracis bla1- and bla2-lacZ transcriptional fusions into B. cereus 569, B. thuringiensis AW43, and the sigP-rsiP-null strains UTC4 and UTT4. The presence of the B. anthracis bla1- and bla2-lacZ transcriptional fusions in the parent strains, but not their presence in the mutant strains, resulted in significantly increased β-galactosidase activity following the addition of sublytic concentrations of a β-lactam antibiotic (Table 5), indicating that sigP and rsiP play roles in bla gene induction and expression in B. cereus and B. thuringiensis.

TABLE 5.

Expression of the B. anthracis bla1 and bla2 promoters in B. cereus and B. thuringiensis grown in the presence and absence of ampicillin (1 μg/ml)^a

Organism	Expression (Miller units)
	bla1::lacZ (pUTE461)		bla2::lacZ (pUTE462)		lacZ (pHT304-18Z)
	Without ampicillin	With ampicillin	Without ampicillin	With ampicillin	Without ampicillin	With ampicillin
B. cereus	0.43 ± 0.61	46.03 ± 18.42	0.30 ± 0.12	134.19 ± 53.92	0.11 ± 0.15	0.12 ± 0.04
UTC4	0.10 ± 0.07	0.08 ± 0.07	0.02 ± 0.03	0.18 ± 0.17	0.03 ± 0.04	0.18 ± 0.21
B. thuringiensis	0.34 ± 0.12	13.70 ± 7.46	0.75 ± 0.04	60.43 ± 17.27	0.58 ± 0.10	0.60 ± 0.16
UTT4	0.26 ± 0.02	0.28 ± 0.09	0.35 ± 0.11	0.39 ± 0.19	0.50 ± 0.09	0.53 ± 0.29

^aThe plasmids contained bla1 and bla2 promoters fused to a promoterless lacZ gene as indicated. The values are means ± standard deviations.

The B. cereus and B. thuringiensis sigP and rsiP homologues are sufficient for penicillin resistance in prototypical B. anthracis.

Given that the B. anthracis bla genes could be induced in B. cereus and B. thuringiensis strains harboring sigP and rsiP homologues, we questioned the lack of β-lactamase induction in prototypical B. anthracis strains during growth in the presence of β-lactam antibiotics. We considered three possible reasons for the lack of induction: (i) B. anthracis cannot sense the presence of a β-lactam antibiotic in the environment, (ii) B. anthracis lacks the ability to transduce the signal generated by the presence of a β-lactam antibiotic to the anti-sigma factor, and (iii) in B. anthracis the protein product of rsiP cannot respond to the upstream signal.

To investigate these hypotheses, we complemented B. anthracis UT334 [sigP(s)-rsiP(s)-null strain] with pUTE728 and pUTE729, plasmids that contained sigP and rsiP from B. cereus and B. thuringiensis, respectively. The recombinant strains exhibited increased MICs of ampicillin compared to the penicillin-susceptible parent strain 9131 and UT334 containing only the vector (Table 3). Interestingly, introduction of the B. cereus and B. thuringiensis genes into B. anthracis UT335 [sigP(r)-rsiP(r)-null strain] did not result in increased resistance to ampicillin, illustrating further differences between the two B. anthracis strains. To determine if the increased β-lactam resistance of the B. anthracis sigP-rsiP-null strains containing the B. cereus and B. thuringiensis sigP and rsiP genes was a result of bla induction, we compared the β-lactamase activities of recombinant strains grown in the presence and absence of ampicillin. The β-lactamase activities of culture supernatants from B. anthracis sigP-rsiP-null strain UT334 containing the B. cereus and B. thuringiensis sigP and rsiP genes were elevated following the addition of ampicillin (Fig. 5). As expected, the β-lactamase activity of UT334 complemented with the B. anthracis sigP and rsiP genes from the resistant or susceptible strains was not affected by the addition of a β-lactam antibiotic. These results indicate that the increased MICs of ampicillin for the B. anthracis sigP-rsiP-null strain (UT334) complemented with the B. cereus and B. thuringiensis sigP and rsiP genes results from bla induction.

FIG. 5.

β-Lactamase activities of B. anthracis strains containing the sigP-rsiP gene pairs from B. anthracis, B. cereus, and B. thuringiensis grown in the presence and absence of ampicillin. The β-lactamase activities of the sigP-rsiP-null strain UT334 complemented in trans with the sigP-rsiP gene pairs from penicillin-susceptible B. anthracis (pUTE692), penicillin-resistant B. anthracis (pUTE693), B. cereus (pUTE728), and B. thuringiensis (pUTE729) and with the empty vector (pHT304) were compared to the β-lactamase activity of the parent penicillin-resistant strain, UT308. Supernatants from the indicated strains were sampled 4 h after induction with 0.1 μg ampicillin/ml (gray bars) or water (black bars). The error bars represent standard deviations. Bkgrd, background; Ba, B. anthracis; Bc, B. cereus; Bt, B. thuringiensis.

We also compared the ampicillin resistance of the sigP-rsiP-null mutants B. cereus UTC4 and B. thuringiensis UTT4 complemented with sigP-rsiP gene pairs from the three species (Table 3). We found that the gene pairs from B. cereus and B. thuringiensis could restore some level of resistance to ampicillin in both species backgrounds. However, pUTE692 [harboring B. anthracis sigP(s) and rsiP(s)] was able to increase resistance to ampicillin only in the UTT4 background. Taken together, these data indicate that protypical Pen^s B. anthracis strains like 9131 contain the proteins necessary to sense the presence of a β-lactam antibiotic and transduce the signal to the anti-sigma factor, as evidenced by the increased resistance and β-lactamase activity of B. anthracis UT334 (sigP-rsiP-null strain) harboring the B. cereus or B. thuringiensis sigP and rsiP homologues. However, the product of the rsiP(s) gene from B. anthracis, when produced in B. anthracis 9131 or B. cereus 569, does not respond to an upstream signal, resulting in continued sequestration of the ECF sigma factor.

Functional differences in the B. anthracis sigP(s) and sigP(r) alleles.

We were unable to introduce the sigP(s) gene alone (in the absence of an rsiP allele) into UT335 [sigP(r)-rsiP(r)-null strain]. Moreover, the presence of sigP(s) (without rsiP) in UT334 [sigP(s)-rsiP(s)-null strain] yielded a recombinant strain that exhibited a growth defect (data not shown). Notably, the sigP(r) gene was readily electroporated into either strain background, and the recombinant strains exhibited growth rates similar to those of the parent strains. To circumvent the problems associated with introducing sigP(s) into B. anthracis without its cognate negative regulator, we created hybrid gene pairs consisting of the B. anthracis sigP(s) or sigP(r) gene fused to the B. cereus rsiP gene (rsiP_Bc) and introduced them into the B. cereus and B. anthracis sigP-rsiP-null strains UTC4 and UT334, respectively. The recombinant B. anthracis and B. cereus strains containing pUTE869 [sigP(s) rsiP_Bc] exhibited MICs of ampicillin that were higher than those exhibited by strains containing pUTE871 [sigP(r) rsiP_Bc] (Table 6). These results strongly suggest that the σ^P produced by the penicillin-susceptible strain exhibits greater activity than the σ^P produced by the penicillin-resistant strain, a difference most likely attributed to the amino acid change (D24G) in region 2.

TABLE 6.

MICs of ampicillin for B. anthracis and B. cereus sigP-rsiP-null strains containing hybrid sigP-rsiP gene pairs

Strain	MICs of ampicillin (μg/ml)
	sigP(s) rsiPBc (pUTE869)	sigP(r) rsiPBc (pUTE871)	Empty vector (pHT304)
B. anthracis UT334	2-12	0.047-1	0.023-0.047
B. cereus UTC4	16-24	2-6	0.047-0.064

DISCUSSION

The susceptibility of B. anthracis to β-lactam antibiotics is well established, and as a result, penicillin has historically been the drug of choice for anthrax treatment worldwide (83; M. N. Jones, R. J. Beedham, P. C. B. Turnbull, and R. J. Manchee, presented at the International Workshop on Anthrax, Winchester, England, 1995). However, treatment of anthrax with penicillin is not always successful, and the use of penicillin for prophylaxis and treatment of anthrax in experimental animals has produced various outcomes (36, 39, 65; A. M. Friedlander, presented at the Third International Conference on Anthrax, Plymouth, England, 1998). B. cereus and B. thuringiensis, two Bacillus species that are closely related to B. anthracis, commonly exhibit inducible penicillin resistance, and some strains of B. cereus and B. thuringiensis have been associated with human disease. B. thuringiensis has been implicated in burn wound infections (30). Reports of B. cereus infections have included reports of necrotic enteritis, liver failure, endocarditis, meningitis, bacteremia, and lethal pneumonia (10, 26, 27, 50, 52, 63, 64, 68). Investigations of the genetic basis for differential expression of the β-lactamase genes in these species should contribute to the taxonomy of these bacteria.

In this work we demonstrated that constitutive expression of the bla genes in a naturally penicillin-resistant clinical isolate of B. anthracis is associated with the ECF sigma factor-anti-sigma factor gene pair, sigP-rsiP. In our model, a nucleotide deletion in the 5′ end of the rsiP gene most likely results in a nonfunctional, 12-amino-acid peptide that is unable to sequester its cognate ECF sigma factor, σ^P. As a result, σ^P is available to interact with RNA polymerase, promoting expression of bla1 and bla2. In a prototypical B. anthracis strain, σ^P is sequestered by a full-length anti-sigma factor, RsiP, resulting in the familiar penicillin-susceptible phenotype. bla gene expression by the prototypical strain is not induced by the presence of a β-lactam antibiotic. Our data also suggest that the single predicted amino acid difference in region 2 of σ^P from the penicillin-resistant strain is associated with decreased σ^P activity since the MICs of ampicillin for B. cereus and B. anthracis sigP-rsiP-null strains containing sigP(s) in combination with B. cereus rsiP were higher than those of the same strains containing sigP(r) paired with the B. cereus rsiP gene.

Initially, we tried to compare the σ^P proteins produced by the penicillin-susceptible and -resistant strains by introducing the sigP genes into the B. anthracis sigP-rsiP-null mutants on a low-copy-number vector. However, we were unable to obtain a UT335 strain containing sigP(s), and a UT334 strain complemented with sigP(s) exhibited growth defects and almost no β-lactamase activity. Surprisingly, addition of the 5′ end of rsiP(s) to constructs containing sigP(s) resulted in strains with increased β-lactamase activity. For example, complementing UT334 [sigP(s)-rsiP(s)-null strain] with a construct containing sigP(s) and the 5′ end of rsiP, corresponding to the first 5 amino acids of RsiP(s), increased the β-lactamase activity from 0 to approximately 45 arbitrary units. If the 5′ end of rsiP(s) encoding the first 17 amino acids of RsiP(s) was included with sigP(s), the recombinant strain exhibited β-lactamase activity of approximately 100 arbitrary units, which is similar to the activity of the parent Pen^r strain UT308 (data not shown). Notably, a similar construct, pUTE708, was used to assess the translational start of RsiP (Fig. 4). Recombinant strain UT335(pUTE708) had significant growth defects (data not shown), suggesting that constitutive expression of σ^P from the penicillin-susceptible strain is detrimental to B. anthracis.

Given that all sequenced B. anthracis strains contain bla1 and bla2 genes and the sigP-rsiP gene pair, we wondered why β-lactamase activity is not induced when B. anthracis is grown in the presence of a β-lactam antibiotic. Using commonly used laboratory strains of B. cereus and B. thuringiensis, we demonstrated that the inducible β-lactamase expression exhibited by these species is dependent on the presence of their native sigP and rsiP genes. Furthermore, complementation of B. anthracis UT334 [sigP(s)-rsiP(s)-null strain] with the B. cereus and B. thuringiensis sigP-rsiP gene pairs resulted in MICs of ampicillin ranging from 1 to 6 μg/ml. This is significantly higher than the MICs of ampicilin for UT334 complemented with its native sigP-rsiP gene pair (0.032 to 0.047 μg/ml). In addition, these strains exhibited increased β-lactamase activity when they were grown in the presence of a β-lactam antibiotic. These results indicate that prototypical B. anthracis strains are able to sense the presence of a β-lactam antibiotic and transduce the signal to the B. cereus and B. thuringiensis anti-sigma factors, ultimately resulting in expression of the bla genes.

In well-studied systems of other bacteria, sequestration of ECF sigma factors is alleviated when the cognate anti-sigma factor is degraded through a process called regulated intramembrane proteolysis. In E. coli, the ECF sigma factor, σ^E, is regulated by the anti-sigma factor, RseA. Under normal growth conditions, RseA tethers σ^E to the cytoplasmic membrane. In response to unfolded proteins in the periplasmic space, DegS, a periplasmic serine protease, carries out site 1 cleavage of RseA. The cleaved RseA is subject to site 2 cleavage by the protease YaeL/RseP (1, 3, 55, 86, 88). The remaining amino-terminal portion of RseA is then subject to further degradation by ClpXP, ultimately resulting in free σ^E (35). The B. subtilis ECF sigma factor, σ^W, is regulated by the anti-sigma factor RsiW, which is also subject to site 1 and site 2 cleavage by PrsW and YluC (also referred to as RasP), respectively (32, 45, 78). After the remaining RsiW fragment is degraded by the cytoplasmic proteases ClpXP, σ^w associates with core RNA polymerase and directs transcription of target genes (90).

Based on our data, we considered the following models for regulation of bla gene expression in B. cereus, B. thuringiensis, and B. anthracis. We propose that prototypical B. anthracis, B. cereus, and B. thuringiensis strains can sense the presence of a β-lactam antibiotic in the environment. The B. cereus and B. thuringiensis RsiP protein responds to this stress, and σ^P is released, resulting in transcription of the bla genes. However, in prototypical penicillin-susceptible B. anthracis strains, RsiP does not respond to the signal, possibly because the anti-sigma factor is not recognized and/or is degraded by proteases or because RsiP is defective in the ability to receive a signal from an upstream factor. In either case, σ^P remains sequestered, and the bla genes are not expressed.

Overall, our studies of SigP-RsiP function in the B. cereus group support the emerging idea that phenotypic differences between B. anthracis, B. cereus, and B. thuringiensis are in large part due to altered gene expression rather than to the gain or loss of coding regions. The sigP and rsiP gene products can be added to the growing list of trans-acting factors that differentially affect transcription of genes common to the B. cereus group species. The best example of this is the PlcR regulon (2, 37, 60). In B. cereus and B. thuringiensis, PlcR controls expression of multiple genes, some of which are associated with pathogenesis. Although many of these genes are present in the B. anthracis genome, the B. anthracis plcR gene contains a nonsense mutation resulting in a nonfunctional protein product. Multiple phenotypic differences between B. anthracis and the other two species are attributed to low-level expression of the PlcR regulon in B. anthracis. Another trans-acting regulator, AtxA, is located on a B. anthracis plasmid and controls expression of B. anthracis-specific virulence genes and some chromosomal genes common to the B. cereus group species. B. cereus and B. thuringiensis strains do not carry the plasmid harboring atxA and therefore exhibit differential expression of the AtxA-controlled chromosomal genes. Interestingly, expression of a B. thuringiensis plcR gene in a B. anthracis strain containing atxA resulted in a significant decrease in sporulation, a phenotype that was rescued by deletion of atxA (66). These results suggest that the plcR and atxA regulons in B. anthracis are not compatible and that the nonsense mutation in the B. anthracis plcR gene provided a selective advantage.

What, if any, is the selective advantage of an apparently noninducible SigP-RsiP system in B. anthracis? We have not found culture conditions that allow induction of the B. anthracis bla genes, and our data suggest that constitutive expression of sigP is detrimental to B. anthracis growth. In the case of a penicillin-resistant strain, this toxicity appears to be alleviated by an amino acid change in σ^P, which reduces the activity of the sigma factor. It is possible that the sigP-rsiP system could be induced by a non-β-lactam signal in the host, adversely affecting growth. Regulatory relationships between σ^P and other genes involved in pathogenesis and development of B. anthracis are unknown. Future experiments will define the σ^P regulon.