ABSTRACT
Bacteria can utilize alternative σ factors to regulate sets of genes in response to changes in the environment. The largest and most diverse group of alternative σ factors are the extracytoplasmic function (ECF) σ factors. σP is an ECF σ factor found in Bacillus anthracis, Bacillus cereus, and Bacillus thuringiensis. Previous work showed that σP is induced by ampicillin, a β-lactam antibiotic, and required for resistance to ampicillin. However, it was not known how activation of σP is controlled or what other antibiotics may activate σP. Here, we report that activation of σP is specific to a subset of β-lactams and that σP is required for resistance to these β-lactams. We demonstrate that activation of σP is controlled by the proteolytic destruction of the anti-σ factor RsiP and that degradation of RsiP requires multiple proteases. Upon exposure to β-lactams, the extracellular domain of RsiP is cleaved by an unknown protease, which we predict cleaves at site-1. Following cleavage by the unknown protease, the N terminus of RsiP is further degraded by the site-2 intramembrane protease RasP. Our data indicate that RasP cleavage of RsiP is not the rate-limiting step in σP activation. This proteolytic cascade leads to activation of σP, which induces resistance to β-lactams likely via increased expression of β-lactamases.
IMPORTANCE The discovery of antibiotics to treat bacterial infections has had a dramatic and positive impact on human health. However, shortly after the introduction of a new antibiotic, bacteria often develop resistance. The bacterial cell envelope is essential for cell viability and is the target of many of the most commonly used antibiotics, including β-lactam antibiotics. Resistance to β-lactams is often dependent upon β-lactamases. In B. cereus, B. thuringiensis, and some B. anthracis strains, the expression of some β-lactamases is inducible. This inducible β-lactamase expression is controlled by activation of an alternative σ factor called σP. Here, we show that β-lactam antibiotics induce σP activation by degradation of the anti-σ factor RsiP.
INTRODUCTION
The bacterial cell envelope is essential for cell viability and is the target of many of the most commonly used antibiotics, including β-lactams like penicillins, penems, and cephalosporins. These are broad-spectrum antibiotics that target peptidoglycan (PG) biosynthesis by inhibiting the transpeptidase activity of penicillin-binding proteins (PBPs). This results in decreased and/or altered cross-linking of peptidoglycan, which leads to cell envelope damage and subsequent cell lysis and death (1, 2).
Members of the Bacillus cereus group, including Bacillus thuringiensis and Bacillus cereus and some strains of Bacillus anthracis, are highly resistant to β-lactam antibiotics (3–6). This resistance is due in part to expression of at least two β-lactamases (3, 5). The expression of these β-lactamases is induced by ampicillin and is dependent upon the alternative σ factor σP. σP belongs to the extracytoplasmic function (ECF) family of alternative σ factors (5).
Bacteria often utilize alternative σ factors to regulate subsets of genes required for survival under specific environmental conditions or for stress responses. ECF σ factors are the largest and most diverse group of alternative σ factors and represent the “third pillar” of bacterial signal transduction (7, 8). ECF σ factors belong to the σ70 family, but unlike the “housekeeping” σ factor, σ70, ECF σ factors contain only region 2 and region 4.2 of σ70, which recognize and bind to the −10 and −35 regions of promoter sequences, respectively (8, 9). In addition, unlike σ70, ECF σ factors are generally held inactive by anti-σ factors until bacteria encounter an inducing signal (10, 11). Upon induction, ECF σ factors are released from their cognate anti-σ factors to promote transcription of specific stress response genes.
The ECF σ factors have been subdivided into more than 40 distinct groups, with ECF01 being the best studied (reviewed in references 7, 11, and 12). σP belongs to the ECF01 family, which includes members like σE and σW from Escherichia coli and Bacillus subtilis, respectively. The activities of the ECF01 family are inhibited by their cognate transmembrane anti-σ factors (8, 13). To activate ECF01 σ factors, the anti-σ factors must be destroyed via a proteolytic cascade (14, 15). For example, the E. coli anti-σ factor RseA is degraded in response to outer membrane stress, leading to σE activation (16, 17). DegS, a serine protease, cleaves the anti-σ factor RseA at site-1 (14, 18, 19). After site-1 cleavage, the conserved site-2 protease, RseP, cleaves RseA within the membrane, leading to increased σE activity (14, 20, 21). Similarly, the σW anti-σ factor, RsiW, from B. subtilis is proteolytically degraded by site-1 and site-2 proteases. In the case of RsiW, the site-1 protease is PrsW, a metalloprotease unrelated to DegS. PrsW cleaves RsiW in response to antimicrobial peptides, vancomycin, and pH change (22–24). RsiW is further processed by the conserved site-2 protease RasP, a homolog of RseP (15).
The closely related ECF30 family member σV from B. subtilis is activated by lysozyme (25–29). Activation of σV differs from σE and σW activation in that σV is not controlled by a dedicated site-1 protease but instead utilizes signal peptidases (30, 31). Signal peptidases are essential proteases which are required to cleave substrates secreted from the general secretion or twin arginine secretion systems (32–34). The anti-σ factor RsiV binds to lysozyme, which allows signal peptidase to cleave RsiV at site-1 (30, 31). This allows the site-2 protease RasP to cleave RsiV, leading to σV activation (35).
Previous studies found that σP is induced by ampicillin (Amp) and that its activity is required for resistance to ampicillin (5). The activity of σP is inhibited by the transmembrane anti-σ factor RsiP (5, 6). However, whether σP is activated specifically by ampicillin or more generally by cell wall stress is not known. In B. subtilis, activation of σV is specific to lysozyme (26, 27), while activation of σW, σX, and σM is in response to more general cell envelope stress (9, 36, 37). Here, we show that σP is activated by a specific subset of β-lactams and that this activation occurs via regulated intramembrane proteolysis of the anti-σ factor RsiP.
RESULTS
A subset of β-lactams induces σP activation.Previously, Koehler and colleagues demonstrated that ampicillin induces expression of the β-lactamase encoded by bla1 (hd73_3490) in a σP-dependent manner in B. thuringiensis and B. cereus (5). Activation of some ECF σ factors is highly specific to an inducing signal, while others are activated by more general cell envelope stress. Thus, we sought to determine the specificity of σP activation using B. thuringiensis as a model system.
Like many ECF σ factor systems, σP is required for its own transcription (5). To monitor σP activation, we fused the σP promoter (PsigP) to the lacZ reporter gene and integrated this construct into the genome of B. thuringiensis (THE2549 thrC::PsigP-lacZ). We tested several classes of β-lactams and cell wall-targeting antibiotics for their ability to induce expression of PsigP-lacZ. We observed wide zones of PsigP-lacZ induction around cefoxitin and cefmetazole (Fig. 1). We detected fainter zones of induction in the areas around cephalothin and cephalexin (Fig. 1). Very faint zones of induction were present in the cells around ampicillin and methicillin (Fig. 1). Interestingly, we did not observe this induction surrounding the β-lactams cefoperazone and piperacillin or antibiotics that target other steps in cell wall biosynthesis, including ramoplanin, phosphomycin, nisin, bacitracin, and vancomycin (Fig. 1). We also tested compounds that do not target peptidoglycan biosynthesis, including kanamycin, polymyxin B, and erythromycin (Erm), and saw no induction of PsigP-lacZ (Fig. 1).
Expression of sigP is specifically induced by β-lactams. All the strains contained PsigP-lacZ in either a wild-type (THE2549), a ΔsigP-rsiP (EBT232), or a Δbla1 (EBT215) background. Mid-log cells were washed and diluted 1:100 in molten LB agar containing X-Gal (100 μg/ml) and poured into empty 100-mm petri dishes. Filter disks containing cefoxitin (Cef) (1 μl of 5-mg/ml cefoxitin), bacitracin (Bac) (1 μl of 50-mg/ml bacitracin), nisin (Nis) (3 μl of 100-mg/ml nisin), vancomycin (Van) (1 μl of 10-mg/ml vancomycin), cefmetazole (Cmet) (1 μl of 5-mg/ml cefmetazole), polymyxin B (Poly) (1 μl of 50-mg/ml polymyxin B), kanamycin (Kan) (1 μl of 10-mg/ml kanamycin), piperacillin (Pip) (1 μl of 5-mg/ml piperacillin), cephalothin (Cthin) (1 μl of 50-mg/ml cephalothin), ramoplanin (Ram) (1 μl of 25-mg/ml ramoplanin), cefoperazone (Cper) (1 μl of 50 mg/ml cefoperazone), phosphomycin (Phos) (1 μl of 100-mg/ml phosphomycin), Amp (2 μl of 200-mg/ml ampicillin), cephalexin (Clex) (1 μl of 50-mg/ml cephalexin), Erm (1 μl of 5-mg/ml erythromycin), and methicillin (Meth) (2 μl of 100-mg/ml methicillin) were then placed on the top agar and incubated for 16 h at 30°C.
To quantify the levels of β-lactam induction, we tested eight β-lactams for their ability to activate the PsigP-lacZ fusions using a β-galactosidase assay. Mid-log cells were incubated in the presence of various concentrations of ampicillin, cefoxitin, cefmetazole, cephalothin, methicillin, cephalexin, cefoperazone, and cefsulodin for 1 h at 37°C. We observed dose-dependent induction with a subset of these β-lactams (Fig. 2A and B). Interestingly, ampicillin, methicillin, and cephalexin showed low levels of PsigP-lacZ induction when spotted onto a lawn of cells (Fig. 1) but strongly induced PsigP-lacZ in liquid assays (Fig. 2A and B), a point we will return to later. In contrast, neither cefoperazone nor cefsulodin was able to induce on the plates or in liquid (Fig. 1 and 2B). This confirms our observation that a subset of β-lactams induces σP activation.
Expression of PsigP-lacZ is dose dependent and dependent upon σP and RasP. (A) B. thuringiensis with transcriptional fusion PsigP-lacZ (THE2549) was grown overnight at 30°C, subcultured in LB, and grown to an OD600 of ∼0.8 before being incubated with various concentrations of β-lactams (0, 0.0625, 0.125, 0.25 0.5, 1, and 2 μg/ml) for 1 h. Cells were collected and resuspended in Z buffer. (B) B. thuringiensis with transcriptional fusion PsigP-lacZ (THE2549) was grown overnight at 30°C, subcultured in LB, and grown to an OD600 of ∼0.8 before being incubated with various concentrations of β-lactams (0, 0.0625, 0.125, 0.25 0.5, 1, and 2 μg/ml) for 1 h. Cells were collected and resuspended in Z buffer. (C) All strains contain PsigP-lacZ and the genotype and plasmid noted: wild type/Vect. (EBT169), sigP/Vect. (EBT251), ΔsigP-rsiP/pSigPRsiP (EBT238), ΔrasP/Vect. (EBT175), and rasP/pRasP (EBT176). Strains were grown to mid-log phase and then treated with 5 μg/ml cefoxitin or untreated (0) and incubated for 1 h. β-Galactosidase activity was calculated as described in Materials and Methods. These experiments were done in triplicate, and standard deviations are represented by error bars.
We found that deletion of the sigP-rsiP genes blocked expression of PsigP-lacZ in the presence of β-lactams (Fig. 1 and 2C), demonstrating that σP is required for induction of PsigP-lacZ in response to β-lactams. When we introduced a low-copy-number plasmid containing PsigP-sigP+-rsiP+ into the ΔsigP-rsiP mutant (ΔsigP-rsiP/pSigPRsiP), we restored the induction of PsigP-lacZ in response to cefoxitin (Fig. 2C). Taken together, these data suggest that a subset of β-lactam antibiotics activates σP.
σP and Bla1 are involved in resistance to some β-lactams.To determine the impact of σP on resistance to β-lactams, we measured the MICs of several β-lactams for wild-type and ΔsigP-rsiP mutant strains. We found that the wild type was greater than 100-fold more resistant to ampicillin, methicillin, and cephalothin than was the ΔsigP-rsiP mutant (Table 1). The wild type was 16- to 50-fold more resistant to cefmetazole, cefoxitin, and cephalexin than the mutant (Table 1). There was little or no difference in resistance to piperacillin, cefoperazone, and cefsulodin, which also failed to activate σP (Table 1 and Fig. 1). We also demonstrate that complementing the ΔsigP-rsiP mutant with a plasmid carrying PsigP-sigP+-rsiP+ restored resistance to ampicillin and cefoxitin (Table 2). For reasons that remain unclear, strains containing plasmids, including empty vector, have slight increases in β-lactam resistance. However, this does not impact the observation that the presence of PsigP-sigP+-rsiP+ restored resistance to ampicillin and cefoxitin.
ΔsigP-rsiP mutant is more sensitive to β-lactams than wild type
RasP is required for resistance to β-lactamsa
Since σP was shown to control expression of hd73_3490 (referred to here as bla1), which encodes a β-lactamase, we sought to determine if this gene played a role in resistance to β-lactams. We made a deletion of bla1 and determined the MIC of ampicillin and cefoxitin for this strain. The bla1 mutant was 8- to 16-fold more sensitive to ampicillin and ∼5-fold more sensitive to methicillin but no more sensitive to cefoxitin than the wild type (Table 2). This contrasts with the sigP mutant, which is greater than 1,000-fold more sensitive to ampicillin, 600-fold more sensitive to methicillin, and ∼25-fold more sensitive to cefoxitin than the wild type (Table 2). This suggests that Bla1 plays a more important role in resistance to ampicillin and methicillin than to cefoxitin. Furthermore, our data suggest that while Bla1 contributes to β-lactam resistance, additional σP-regulated genes must also contribute to β-lactam resistance.
When we tested various β-lactams for induction of PsigP-lacZ on 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) plates, we did not consistently observe a strong zone of induction surrounding ampicillin and methicillin (Fig. 1). We hypothesized that this weak induction zone was due to the wild type efficiently producing β-lactamases which degraded the inducer (ampicillin and methicillin). Thus, we were unable to observe the increased production of β-galactosidase. To test this hypothesis, we determined the effect of a Δbla1 mutant on σP activation. We found that in the Δbla1 mutant, ampicillin and methicillin produced more distinct zones of induction (Fig. 1). However, all other induction zones of the Δbla1 mutant were similar to the wild type. Thus, in the absence of Bla1, which degrades ampicillin and methicillin, we detected greater induction of PsigP-lacZ expression. Taken together, these observations suggest that the weak ampicillin induction of PsigP-lacZ on plates is in part due to the efficient degradation of the inducer by β-lactamases.
RsiP is degraded in response to cefoxitin in a dose-dependent manner.The anti-σ factors of other ECF01 family members are degraded, which leads to the activation of their cognate σ factors (7, 14, 15). We sought to determine if β-lactams activate σP by inducing degradation of RsiP. To investigate this, we constructed a strain with an anhydrotetracycline (ATc)-inducible copy of green fluorescent protein (GFP) fused to the N terminus of RsiP (GFP-RsiP). The inducible promoter allows us to uncouple expression of RsiP from induction of σP. The GFP-RsiP fusion allows us to follow the fate of the cytoplasmic portion of RsiP. Expression of GFP-RsiP complements an rsiP null mutation (see Fig. S1 in the supplemental material) and localizes to the membrane (Fig. S2). We then induced the synthesis of GFP-RsiP in exponential-phase cells and monitored its processing before and after treatment with cefoxitin. We chose to utilize cefoxitin for these experiments because cefoxitin induces σP activation over a wide concentration range and the ΔsigP-rsiP mutant strain grows at most of these concentrations (Fig. 2A and Table 1). Cell pellets were then lysed by sonication, and Western blot analyses were performed using anti-RsiP antisera against the extracellular portion of RsiP or anti-GFP antisera, which detect GFP fused to the intracellular portion of RsiP.
FIG S1
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S2
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
When cells producing GFP-RsiP were grown in the absence of cefoxitin, we detected full-length GFP-RsiP at the expected size of ∼60 kDa using anti-RsiP antisera. This band was absent in the empty-vector control (Fig. 3A). When cells were incubated with cefoxitin (5 μg/ml) for various times, we found that the level of full-length GFP-RsiP decreased over time (Fig. 3A and Fig. S3A). We observed loss of GFP-RsiP by 30 min to 1 h after exposure to cefoxitin (Fig. 3A and Fig. S3A). This suggests that GFP-RsiP is likely degraded in the presence of cefoxitin.
RsiP levels decrease in the presence of cefoxitin. B. thuringiensis expressing tetracycline-inducible gfp-rsiP (EBT360) or empty vector (EV; EBT169) was subcultured 1:50 into LB supplemented with ATc (50 ng/ml). At mid-log phase, cells were incubated with 5 μg/ml of cefoxitin for various times (0, 15, 30, 60, 120, or 180 min) (A) or increasing concentrations of cefoxitin (0, 0.05, 0.5, 5, 50, or 500 μg/ml) for 1 h (B). The immunoblot was probed with antisera against RsiP (α-RsiP76–275). Streptavidin IR680LT was used to detect HD73_4231 (PycA homolog), which served as a loading control (62, 63). The color blot showing both anti-RsiP and streptavidin on a single gel is shown in Fig. S3. Numbers at right indicate molecular masses in kilodaltons.
FIG S3
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
We also tested the effect of cefoxitin concentration on GFP-RsiP levels by incubating cells with a range of cefoxitin concentrations (0 to 500 μg/ml) for 1 h. We found that increasing concentrations of cefoxitin resulted in a greater decrease of full-length GFP-RsiP (Fig. 3B and Fig. S3B). We obtained comparable results when we performed blotting assays for the N-terminal domain using anti-GFP antisera (Fig. S4). These data suggest that activation of σP occurs via loss of RsiP in a cefoxitin dose-dependent manner.
FIG S4
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
RasP is necessary for σP activation.Both σE and σW are activated by regulated intramembrane proteolysis of their cognate anti-σ factors. Proteolysis of these anti-σ factors requires multiple proteases, including the highly conserved site-2 proteases RseP and RasP, respectively (14, 15). We hypothesize that activation of σP requires multiple proteases, including the conserved site-2 protease RasP to degrade RsiP. To test this, we used BLAST to identify a putative membrane-embedded metalloprotease, HD73_4103, which is 76% similar and 60% identical to B. subtilis RasP and is here referred to as RasP (Fig. S5) (38–43). To determine if RasP was required for σP activation, we generated a strain containing a deletion of rasP and the PsigP-lacZ reporter. In the absence of RasP, we did not detect increased expression of PsigP-lacZ reporter in response to cefoxitin (Fig. 2C). In MIC experiments, we found that, similarly to the ΔsigP-rsiP mutant, the ΔrasP mutant was more sensitive to ampicillin and cefoxitin (Table 2). We found that both resistance to β-lactams and induction of PsigP-lacZ could be complemented when a plasmid expressing rasP+ was introduced into the ΔrasP mutant (Fig. 2C and Table 2). These data suggest that RasP is required for σP activation.
FIG S5
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
RasP is required for degradation of RsiP.To determine if RasP is required for degradation of RsiP, we expressed the GFP-RsiP fusion in both the wild type and a ΔrasP mutant. We treated cells with 5 μg/ml cefoxitin for various lengths of time from 0 to 180 min (Fig. 4 and Fig. S6). In the wild type, we observed loss of full-length RsiP over time (Fig. 4 and Fig. S6). In contrast, we observed loss of full-length GFP-RsiP and the accumulation of a smaller ∼35-kDa band in the ΔrasP mutant (Fig. 4 and Fig. S6). This suggests that RasP is required for complete degradation of RsiP. Since a truncated product accumulates in the ΔrasP mutant, RasP is likely required for site-2 cleavage and an unidentified protease is required for cleavage at site-1.
RsiP degradation is dependent upon the site-2 protease RasP. B. thuringiensis wild type (EBT360) or ΔrasP (EBT366) containing a tetracycline-inducible copy of gfp-rsiP was subcultured 1:50 into LB supplemented with ATc (50 ng/ml). At mid-log phase, cultures were incubated with cefoxitin (5 μg/ml) for the time indicated at 37°C. The immunoblot was probed with anti-GFP antisera. EV is wild type with pAH9 (EBT169), and GFP is wild type with pAH13 (UM20). Streptavidin IR680LT was used to detect HD73_4231 (PycA homolog), which served as a loading control (62, 63). The color blot showing both anti-GFP and streptavidin on a single gel is shown in Fig. S6. Numbers at right are molecular masses in kilodaltons.
FIG S6
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
Mutations in rsiP result in constitutive sigP expression.To further characterize the σP signal transduction system, we isolated mutants which resulted in constitutive expression of PsigP-lacZ. We selected for mutants with increased resistance to cefoxitin by plating cultures of the wild-type PsigP-lacZ strain (THE2549) on LB-cefoxitin (200 μg/ml) agar. At this concentration of cefoxitin, wild-type B. thuringiensis fails to grow. These strains were tested for PsigP-lacZ expression in the absence of cefoxitin by streaking on LB–X-Gal. We isolated 8 independent mutants with increased resistance to cefoxitin that have constitutive PsigP-lacZ expression. We hypothesized that these strains harbored mutations in rsiP. We PCR amplified and sequenced the sigP and rsiP genes from the constitutive mutants. The 8 constitutive mutants contained mutations in different regions of the rsiP gene that resulted in C-terminal truncations of RsiP (Fig. S7). We selected four rsiP mutants for further study. We found that each mutant strain showed increased PsigP-lacZ expression even in the absence of β-lactams (Fig. 5). When a wild-type copy of rsiP (pSigPRsiP) was introduced to each of these mutants, PsigP-lacZ expression was no longer constitutive but was induced in the presence of cefoxitin (Fig. S8). This indicates that the rsiP mutations were responsible for the increased PsigP-lacZ expression.
Truncations of RsiP lead to constitutive σP activation. To determine if RasP was required for ampicillin-inducible PsigP-lacZ expression, we assayed β-galactosidase activity of B. thuringiensis with transcriptional fusion PsigP-lacZ and different rsiP truncation mutants (WT, THE2549; RsiP1–220, THE2602; RsiP1–80, THE2628; RsiP1–61, THE2637; RsiP1–16, THE2642) and a ΔrasP deletion (WT, EBT140; RsiP1–220, EBT116; RsiP1–80, EBT148; RsiP1–61, EBT133; RsiP1–16, THE2605). Cells were grown overnight at 30°C, subcultured in LB, and grown to an OD600 of ∼0.8 before being incubated with cefoxitin (5 μg/ml) for 1 h. The experiment was performed in triplicate, and standard deviations are represented by error bars.
FIG S7
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S8
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
In the σV and σW systems, RasP cleaves the anti-σ factors RsiW and RsiV within the transmembrane domain to activate the cognate σ factors (15, 35). The RsiP transmembrane is predicted to be residues 54 to 71 based on TMHMM (44). Two of the four RsiP truncations produce proteins with the transmembrane domain intact, while the remaining RsiP truncations lack the transmembrane domain. Since RasP is known to cleave proteins within the transmembrane domain, we hypothesized that those truncations which still contain a transmembrane domain would require RasP in order to activate σP. To test this, we introduced the ΔrasP mutation into each of the rsiP mutants. In the absence of RasP, strains containing truncations which have a transmembrane domain (RsiP1–220 and RsiP1–80) (Fig. 4 and Fig. S7) no longer constitutively activate σP (Fig. 5). However, the strains with the rsiP truncation lacking the transmembrane domain (RsiP1–16 and RsiP1–61) constitutively activate σP even in the absence of RasP (RsiP1–16 and RsiP1–61) (Fig. 4 and Fig. S5). Thus, RasP is required for σP activation when the transmembrane domain of RsiP is intact, consistent with the role of RasP as a site-2 protease.
RasP cleaves within the transmembrane domain of RsiP and is not the regulated step in σP activation.In the case of σW and σV, the rate-limiting step in σ factor activation is site-1 cleavage (15, 35). Since the identity of the site-1 protease is not currently known, we sought to determine if RasP cleavage of RsiP is a rate-limiting step in σP activation. To test this, we constructed truncations of GFP-RsiP that lack the extracellular portion of RsiP. One truncation includes the transmembrane domain (gfp-rsiP1–72), and one truncation lacks the transmembrane domain (gfp-rsiP1–53). We expressed the truncated GFP-RsiP proteins in wild-type and ΔrasP backgrounds and exposed these strains to cefoxitin (5 μg/ml). In wild-type strains, we found that both GFP-RsiP1–72 and GFP-RsiP1–53 were degraded (Fig. 6 and Fig. S9). However, in the ΔrasP mutant GFP-RsiP1–72 accumulated, while GFP-RsiP1–53 was degraded (Fig. 6 and Fig. S9). These data indicate that GFP-RsiP1–72 requires RasP for degradation while GFP-RsiP1–53 does not. One possible interpretation is that GFP-RsiP1–72 is not produced or localized properly to the membrane. Thus, we confirmed that GFP-RsiP1–72 localizes to the membrane by fluorescence microscopy (Fig. S2). This suggests that the RasP cleavage site of RsiP occurs within the transmembrane domain between amino acids 53 and 72. The presence or absence of cefoxitin had no effect on the degradation (Fig. 6 and Fig. S9). Since GFP-RsiP1–72 is constitutively degraded, we conclude that GFP-RsiP1–72 mimics the site-1 cleavage product and that RasP activity is not induced by cefoxitin. This suggests that RasP cleavage of RsiP is not the regulated step in σP activation and that site-1 cleavage is the step that is controlled by the presence of β-lactams.
Truncation of RsiP results in constitutive degradation in a RasP-dependent manner. B. thuringiensis containing a tetracycline-inducible copy of gfp-rsiP, gfp-rsiP1–72 (rsiP without the extracellular domain), or gfp-rsiP1–53 (rsiP without the transmembrane and extracellular domains) was constructed in either the wild type (rasP+) or a ΔrasP mutant strain (GFP-RsiP wild type [rasP+], EBT360; GFP-RsiP ΔrasP, EBT366; GFP-RsiP1–53 wild type [rasP+], EBT518; GFP-RsiP1–53 ΔrasP, EBT510; GFP-RsiP1–72 wild type [rasP+], EBT516; GFP-RsiP1–72 ΔrasP, EBT533). Strains were subcultured 1:50 into LB supplemented with ATc (100 ng/ml), grown to mid-log phase, and then incubated for 2 h without (−) or with (+) cefoxitin treatment (5 μg/ml) at 37°C. The immunoblot was probed with anti-GFP antisera. Streptavidin IR680LT was used to detect HD73_4231 (PycA homolog), which served as a loading control (62, 63). The color blot showing both anti-GFP and streptavidin on a single gel is shown in Fig. S9. Numbers at right are molecular masses in kilodaltons.
FIG S9
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
DISCUSSION
Many ECF σ factors are induced in response to extracytoplasmic stressors and initiate transcription of a subset of genes to modulate the cell’s response to these stresses. ECF σ factors can respond to signals such as misfolded periplasmic protein, antimicrobial peptides, or lysozyme. The ECF σ factors encoded in highly related organisms can vary widely. For example, B. subtilis encodes 7 ECF σ factors, while B. thuringiensis encodes 15 predicted ECF σ factors. The only ECF σ factor that these organisms share is σM (45). Thus, there is a variability in how bacteria utilize ECF σ factors to respond to stress. Ross et al. demonstrated that the novel ECF σ factor σP is induced in the presence of ampicillin and initiates transcription of β-lactamases (5). Here, we demonstrated that σP responds specifically to a subset of β-lactams, while other β-lactams and cell wall-targeting antibiotics fail to induce σP activation. We also showed that σP confers various degrees of resistance to these β-lactam antibiotics. We found that σP was not required for resistance to other cell wall antibiotics, including vancomycin, nisin, and bacitracin, suggesting specificity in resistance to β-lactams and not a general cell envelope stress response.
For ECF σ factors to be activated, their cognate anti-σ factors must be inactivated. This can be accomplished via various mechanisms, including a conformational change of the anti-σ factor; partner switching, where an anti-anti-σ factor frees the σ factor from the anti-σ factor; or proteolytic destruction of the anti-σ factor (9, 11). The anti-σ factors RseA in E. coli and RsiW and RsiV in B. subtilis are degraded sequentially by regulated intramembrane proteolysis. Each of these anti-σ factors requires a different family of proteases to cleave the anti-σ factor at site-1 (14, 22, 30, 46, 47), while site-2 cleavage is carried out by the conserved site-2 protease (14, 15, 35). We hypothesize that σP is activated in a similar manner. Our data indicate that σP is released from RsiP by proteolytic degradation when β-lactams are present. We found that RasP is required for activation of σP. We also observe that an RsiP degradation product approximately the size of our predicted RasP substrate accumulates in a ΔrasP mutant. This indicates that RasP is required for degradation of RsiP. Our data also suggest, similarly to other anti-σ factors, that site-2 cleavage of RsiP is not the rate-limiting step, since the C-terminal RsiP truncations are constitutively degraded and lead to constitutive σP activation in the absence of β-lactams. Thus, we hypothesize that RasP is required for site-2 cleavage of RsiP and that an as-yet-unidentified protease is required to initiate degradation of RsiP by cleaving RsiP at site-1. We hypothesize that, like other ECF σ factors activated by regulated intramembrane proteolysis, site-1 cleavage of RsiP is likely the rate-limiting step in σP activation.
Our data suggest that a subset of β-lactams induce σP activation. We found that, in addition to ampicillin, σP is activated by cefoxitin, cefmetazole, cephalothin, cephalexin, and methicillin but not by piperacillin, cefoperazone, cefsulodin, or antibiotics that target other steps in peptidoglycan biosynthesis. This raises the question of what the signal is for σP activation. The β-lactams could be sensed directly or indirectly. For example, RsiV directly senses lysozyme and degradation of RsiV is rapid (31). In contrast, activation of σE is indirect and due to buildup of products that occur when the outer membrane is damaged (31, 48). Our data suggest that RsiP degradation is a relatively slow process. One possible interpretation of this is that β-lactam-induced peptidoglycan (PG) damage must accumulate to induce RsiP degradation. We hypothesize that the β-lactams that we tested have different affinities for penicillin-binding proteins (PBPs) and that this affinity may explain why some β-lactams induce σP while others do not. In other organisms, including Streptococcus pneumoniae, B. subtilis, and E. coli, β-lactams can differentially target PBPs (49–51). This raises the possibility that activation of σP could be the result of inhibition of specific PBPs. Unfortunately, at this time we do not know which PBPs are targeted by the different β-lactams in B. thuringiensis. Thus, the precise mechanism and signal responsible for σP activation remain to be clearly defined.
MATERIALS AND METHODS
Media and growth conditions.All B. thuringiensis strains are isogenic derivatives of AW43, a derivative of Bacillus thuringiensis subsp. kurstaki strain HD73 (52). All strains and genotypes can be found in Table 3. All B. thuringiensis strains were grown in or on LB medium at 30°C unless otherwise specified. Cultures of B. thuringiensis were grown with agitation in a roller drum. Strains containing episomal plasmids were grown in LB containing chloramphenicol (Cam; 10 μg/ml) or erythromycin (Erm; 10 μg/ml). E. coli strains were grown at 37°C using LB-ampicillin (Amp; 100 μg/ml) or LB-Cam (10 μg/ml) medium. To screen for threonine auxotrophy, B. thuringiensis strains were patched on minimal medium plates without or with threonine (50 μg/ml) (53, 54). The β-galactosidase chromogenic indicator 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) was used at a concentration of 100 μg/ml. Anhydrotetracycline (ATc; Sigma) was used at a concentration of 100 ng/ml.
Strains
Strain and plasmid construction.All plasmids are listed in Table 4, which includes information relevant to plasmid assembly. Plasmids were constructed by isothermal assembly (55). Regions of plasmids constructed using PCR were verified by DNA sequencing. The oligonucleotide primers used in this work were synthesized by Integrated DNA Technologies (Coralville, IA) and are listed in Table S1 in the supplemental material. All plasmids were propagated using OmniMax 2-T1R as the cloning host and passaged through the nonmethylating E. coli strain INV110 before being transformed into a B. thuringiensis recipient strain.
Plasmids
TABLE S1
Copyright © 2019 Ho et al.This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
To construct deletion mutants, we cloned DNA 1 kb upstream and 1 kb downstream of the site of desired deletion using primers listed in Table S1 onto the temperature-sensitive pMAD plasmid (erythromycin resistant) between the BglII and EcoRI sites (56).
Complementation constructs were constructed in pAH9, which is an E. coli–Gram-positive bacterial shuttle vector with a pE194 origin of replication (57). Chromosomal DNA including the promoter sequence was cloned for PsigP-sigP+-rsiP+ and cloned into pAH9 digested with EcoRI and HindIII, while rasP was cloned downstream of the PsarA promoter from Staphylococcus aureus by digesting with EcoRI and KpnI. In B. thuringiensis, PsarA has moderate constitutive expression.
To generate strains containing the sigP promoter fused to the lacZ reporter integrated into the chromosome, we constructed a number of intermediate vectors. To switch the antibiotic resistance of the temperature-sensitive pMAD vector, we constructed pTHE946, which contains the E. coli origin (ColE1 ori) of replication, an Erm resistance gene (for selection in Gram-positive bacteria), an Amp resistance gene (for selection in E. coli strains), and the temperature-sensitive origin (pE194 ori) from pMAD (7.3-kb StuI and BamHI fragment) as well as the conjugation origin of transfer and the Cam resistance gene from pRPF185 (SmaI and BamHI fragment). The thrC (primers 2917 and 2918) and thrB (primers 2919 and 2920) genes were cloned into the ScaI- and SalI-digested pTHE946 plasmid (lacking Ermr and Ampr genes) to generate a vector (pTHE948) which can integrate into the thrC operon. A promoterless lacZ fragment (primers 2922 and 2923) was added between the thrC and thrB genes of pTHE948 (XhoI and SbfI) to generate pTHE950. This plasmid (XhoI and NotI digested) was used to clone the sigP promoter (primers TE2929 and 2930) to generate the PsigP-lacZ promoter fusion (pTHE949).
B. thuringiensis DNA transformation.Plasmids were introduced into B. thuringiensis by electroporation (58, 59). Briefly, recipient cells were grown to late log phase at 37°C. For each transformation, cells (1.5 ml) were pelleted by centrifugation (9,000 × g) and washed twice in room-temperature sterile water. After careful removal of all residual water, 100 μl of sterile 40% polyethylene glycol (PEG) 6000 (Sigma) was used to gently resuspend cells. Approximately 2 to 10 μl of unmethylated DNA (>50 ng/μl) was added to cells and transferred to an 0.4-cm-gap electroporation cuvette (Bio-Rad). Cells were exposed to 2.5 kV for 4 to 6 ms. LB was immediately added, and cells were incubated at 30°C for 1 to 2 h prior to plating on selective media.
Construction of deletions or promoter-lacZ fusions in B. thuringiensis.To generate unmarked mutants and thrC::PsigP-lacZ strains, we used plasmid vectors containing the temperature-sensitive origin of replication (pE194 ori) from the pMAD plasmid (56). At permissive temperatures (30°C), pMAD replicates episomally as a plasmid. At nonpermissive temperatures (42°C), pMAD must integrate into the chromosome via homologous recombination; otherwise, the plasmid will be lost to segregation and the strain will become sensitive to erythromycin. Plasmids were transformed into a B. thuringiensis recipient strain and selected for on LB-Erm agar at 30°C. To select for the integration of the deletion plasmid into the recipient strain genome, plasmid-containing bacteria were grown at 42°C on LB-Erm plates. The plasmid-integrated strain was then struck on LB agar at 30°C twice. Individual colonies were patched on LB and LB-Erm agar to identify the Erm-sensitive bacteria which had lost the deletion plasmid by segregation. To verify each deletion, genomic DNA was isolated from each strain candidate and PCR was used to verify the deletion. Integration of the PsigP-lacZ fusion into the thrC operon results in threonine auxotrophy and can be identified by lack of growth on minimal medium plates without threonine.
Zones of inhibition and zones of induction.To determine the zones of inhibition and induction by various antibiotics, we first washed mid-logarithmically grown cells in fresh LB. Washed cells were diluted 1:100 in molten LB agar containing X-Gal (100 μg/ml) and poured into empty 100-mm petri dishes. Sterile cellulose disks (8 mm) were saturated with different antibiotics and allowed to dry for longer than 10 min. After each antibiotic disk was placed on the solidified agar, plates were incubated at 30°C overnight before observation.
β-Galactosidase assays.To quantify expression from the sigP promoter, we measured the β-galactosidase activity of cells containing a PsigP-lacZ promoter fusion. Overnight cultures were diluted 1:50 in fresh LB medium and incubated for 3 h at 30°C. One milliliter of each subculture was pelleted (9,000 × g), washed (in LB broth), and resuspended in 1 ml LB broth lacking or including specified antibiotics. After 1 h of incubation at 37°C, 1 ml of each sample was pelleted and resuspended in 200 μl of Z buffer. Cells were permeabilized by mixing with 16 μl chloroform and 16 μl 2% Sarkosyl (26, 60). Permeabilized cells (100 μl) were mixed with 10 mg/ml ortho-nitrophenyl-β-galactoside (ONPG; Research Products International; 50 μl), and optical density at 405 nm (OD405) was measured over time using an Infinite M200 Pro plate reader (Tecan). β-Galactosidase activity units (μmol of ONP formed min−1) × 103/(OD600 × ml of cell suspension) were calculated as previously described (61). Experiments were performed in triplicate with the mean and standard deviation being shown.
MIC assay.To determine the MICs of various antibiotics, we diluted overnight cultures of bacteria (washed in LB) 1:1,000 in medium containing 2-fold dilutions of each antibiotic. All MIC experiments were performed in round-bottom 96-well plates. Each experiment was performed in triplicate, and the plates were allowed to incubate for 24 h at 37°C before observation of growth or no growth.
Immunoblot analysis.Samples were electrophoresed on a 15% SDS-polyacrylamide gel, and proteins were then blotted onto a nitrocellulose membrane (GE Healthcare, Amersham). Nitrocellulose was blocked with 5% bovine serum albumin (BSA), and proteins were detected with either 1:10,000 anti-GFP or 1:5,000 anti-RsiP76–275 antiserum. Streptavidin IR680LT (1:10,000) was used to detect two biotin-containing proteins, PycA (HD73_4231) and AccB (HD73_4487), which served as loading controls (62, 63). To detect primary antibodies, the blots were incubated with 1:10,000 goat anti-rabbit IR800CW (Li-Cor) and imaged on an Odyssey CLx scanner (Li-Cor) or an Azure Sapphire imager (Azure Biosystems). All immunoblot assays were performed a minimum of three times with a representative example being shown.
ACKNOWLEDGMENTS
This work was supported by the Department of Microbiology and Immunology at the University of Iowa and by NIH grant R21AI146769 to C.D.E.
We thank Theresa Koehler for strains and advice. We also thank Leyla Slamti for protocols and members of the Ellermeier and Weiss labs for helpful comments.
FOOTNOTES
- Received July 17, 2019.
- Accepted July 18, 2019.
- Copyright © 2019 Ho et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
