Stress Triggers a Process That Limits Activation of the Bacillus subtilis Stress Transcription Factor ς^B

Abstract

Stress-induced activation of the Bacillus subtilis transcription factor ς^B is transitory. To determine whether the process that limits ς^B activation is itself triggered by stress, B. subtilis strains in which the stress pathway was artificially activated by the induced expression of a positive regulatory protein (RsbT) were exposed to ethanol stress and were monitored for the persistence of ς^B activity. Without ethanol treatment, the induced cultures displayed continuously high ς^B activity. Ethanol treatment restricted ongoing ς^B activity, but only in strains with intact rsbX and -S genes. The loss of other gene products (RsbR and Obg) known to participate in the stress activation pathway had little influence in blocking the ethanol effect. The data argue that stress upregulates the activity of the RsbX-S regulatory pair to restrict ς^B induction following stress.

ς^B is a transcription factor that controls the general stress regulon of Bacillus subtilis, a collection of genes whose products aid the bacterium in surviving any of a number of environmental traumas (10, 21, 23). Induction of the ς^B regulon occurs by the activation of ς^B itself, a process that is triggered by entry of B. subtilis into the stationary phase of growth or by the onset of environmental stress (e.g., heat, salt, acid, or ethanol) (12, 24, 25). A current model for ς^B regulation is depicted in Fig. 1. ς^B is present, but inactive, in the prestressed cell due to an association with the anti-ς^B protein RsbW. ς^B release from RsbW is effected by an additional protein (RsbV) which binds to RsbW in lieu of ς^B (4, 5, 7). The abundance of active RsbV determines the level of free ς^B (24). In unstressed cells, RsbV is largely inactive due to RsbW-catalyzed phosphorylation (7, 25). When B. subtilis enters stationary phase, unphosphorylated RsbV accumulates, likely due to the effects of an RsbV-P phosphatase (YvfP) as well as inefficient phosphorylation under the stationary-phase condition of low ATP (2, 24; K. Vijay, M. S. Brody, E. Fredlund, and C. W. Price, submitted for publication). As a result, RsbV is available to displace ς^B from the RsbW-ς^B complex and to induce the ς^B regulon. Environmental stress also activates rsbV, but does so using a separate collection of Rsb proteins (1, 6, 8, 12, 25, 27, 28). RsbT is the most upstream effector in this pathway (28). Following exposure to stress, RsbT, normally inactive and complexed to RsbS, phosphorylates RsbS and becomes free to activate the stress-specific RsbV-P phosphatase, RsbU (28). RsbU can then activate RsbV. Obg, a GTP binding protein (14, 19, 26), is also needed for stress triggering of ς^B activity; however, its explicit role in this process is unknown (17). Negative regulation is reestablished when RsbX, a RsbS-P phosphatase, dephosphorylates RsbS-P, thereby enabling RsbS to again inactivate RsbT (28).

FIG. 1.

Model of ς^B regulation. Active ς^B holoenzyme (E-ς^B) forms when the RsbV protein (V) binds to the anti-ς^B protein RsbW (W) to free ς^B (2, 7). RsbV is normally inactive (V-P) due to phosphorylation by RsbW but is reactivated by stationary-phase or stress-activated phosphatases, YvfP and RsbU (U), respectively (7, 20, 24, 28; Vijay et al., submitted). The stress phosphatase RsbU is activated by RsbT (T) (28). RsbT is normally inactive due to an association with its negative regulator RsbS (S). Upon exposure to stress, RsbT phosphorylates and inactivates RsbS (S-P) and activates RsbU (28). RsbR (R) is believed to facilitate the RsbT-RsbS interaction (1, 9). Obg, a GTP binding protein, is necessary for stress activation of RsbT, but its role is unknown (17). Negative control is resumed when RsbX (X), a phosphatase, dephosphorylates and reactivates the RsbS phosphate (28).

The genes for ς^B and seven of its regulators (RsbR, -S, -T, -U, -V, -W, and -X) are cotranscribed as an eight-gene operon from a promoter (P_A) that is recognized by the B. subtilis housekeeping ς factor (ς^A) (11, 27). An internal ς^B-dependent promoter (P_B) enhances the expression of the four downstream genes when ς^B is active (11). Thus, the levels of ς^B, its principal regulators (RsbV and -W), and the RsbX phosphatase are elevated following activation of ς^B.

ς^B-dependent transcription is only transiently activated by stress (18, 25), declining by 20 to 30 min after its initial induction (Fig. 2). The observation that RsbX, the most upstream negative regulator of the stress pathway, is expressed at higher levels when ς^B becomes active suggested that the transience of the ς^B stress response could be attributed to an effect of elevated RsbX protein levels on the phosphorylation state of RsbS (28). Although the persistence of ς^B activity following stress induction of mutant strains lacking RsbX indicated that RsbX had a role in this process, manipulating RsbX levels by its expression from inducible promoters failed to show a credible correlation between the absolute levels of RsbX and the degree of ς^B activation in stressed cells (22). These results suggested that the RsbX protein was necessary, but not sufficient, to limit the induction of ς^B following stress.

FIG. 2.

Ethanol induction of ς^B in wild-type B. subtilis. BSA46 (ctc::lacZ) was grown in Luria broth (LB) (16) at 37°C. At an optical density at 540 nm (OD₅₄₀) of 0.15, ethanol (4% vol/vol) was added to half of the culture (0 time). Samples were taken at 15-min intervals and were analyzed for β-galactosidase (13). The data is given in Miller units (15).

To further investigate the mechanism responsible for the transience of the stress induction of ς^B, we sought to separate the effect of stress in triggering the pathway from its possible effect in limiting the duration of the response. To accomplish this, we took advantage of the finding that the enhanced synthesis of RsbT, relative to its negative regulator, RsbS, is sufficient to induce the ς^B stress pathway in the absence of stress (17, 28). This allowed us to artificially activate the pathway and then test the effects of stress and the need for particular rsb gene products on the duration of the response.

We used a B. subtilis strain (BSA419), in which a P_spac::rsbT fusion plasmid (pHV501T) had entered the chromosome by a single-site recombination event at rsbT (Table 1). BSA419 contains a sigB operon in which rsbR, -S, and -T are expressed from the P_A promoter and a second copy of rsbT and the remaining downstream sigB genes, separated from P_A by the plasmid sequences, are expressed under the control of the inducible spac promoter (17). When P_spac is not induced, only RsbR and -S are evident in Western blots (Fig. 3, lane 1). rsbT is also expressed, but is difficult to detect in unstressed cells by Western blotting (8). Induction of P_spac with isopropyl-β-d-thiogalactopyranoside (IPTG) yields the anticipated increase in the products of the six genes that are downstream of P_spac (Fig. 3, lane 2).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genotype	Constructiona or source
Bacillus strains
PY22	trpC2	P. Youngman
BSA46	trpC2 SPβ ctc::lacZ	3
XS352	trpC2 aph3′5"/sigB rsbST	18
BSA419	trpC2 Pspac::rsbT SPβ ctc::lacZ	17
BSJ13	trpC2 Pspac::rsbT Pxyl::obg SPβ ctc::lacZ	17
BSJ38	trpC2 Pspac::rsbT rsbX::spec SPβ ctc::lacZ	BSA625 → BSA419
BSJ39	trpC2 aph3′5"/sigB ΔrsbST	XS352 → PY22
BSJ40	trpC2 aph3′5"/sigB ΔrsbST Pspac::rsbT SPβ ctc::lacZ	pHV501T → BSJ39
BSJ41	trpC2 aph3′5"/sigB ΔrsbST Pspac::rsbT SPβ ctc::lacZ	BSA46 → BSJ40
BSJ42	trpC2 aph3′5"/sigB rsbRΔ5 Pspac::rsbT SPβ ctc::lacZ	pJM49 → BSA419
Plasmids
pHV501T	Apr Emr Pspac::rsbT	17
pRS11	Apr KanrPA::rsbR rsbS	18
pJM49	Apr KanrPA::rsbRΔ5 rsbS	This study
pML7/X::spec	Apr Cmr SprPBrsbVW sigB rsbX::spec	4

^aTransformations were performed as described by Yasbin et al. (29). 

FIG. 3.

Western blot analysis of BSA419 after treatment with ethanol. Cells were grown as described in the legend to Fig. 4, with samples harvested 30 min after induction by pouring over ice chips. Following centrifugation, the cells were resuspended in buffer (50 mM Tris-HCl [pH 8.0], 0.1 mM ETDA, 0.03% phenylmethylsulfonyl fluoride) and were disrupted by passage through a French press. The extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, were transferred to nitrocellulose, and were probed by Western blotting by using monoclonal antibodies raised against RsbV, -W, -X, -R, -S, -T, and -U and ς^B (8). The anti-RsbX antibody detects doublet bands of unknown significance (24). Lane 1, cells immediately before addition of IPTG; lane 2, 30 min after IPTG induction; lane 3, 90 min after addition of IPTG, without ethanol treatment; lane 4, 90 min after addition of IPTG, with ethanol treatment (60 min).

BSA419 contains a lacZ reporter gene fused to a ς^B-dependent promoter (ctc::lacZ). Concomittant with induction of P_spac, there was a rapid rise in ς^B-dependent transcription, which remained high throughout the duration of the experiment (Fig. 4A). When the induced culture was exposed to ethanol stress 30 min after IPTG induction (Fig. 4A), reporter gene activity showed a small increase, followed by a decline in β-galactosidase activity that resembled the decline seen when ς^B is induced by stress in wild-type B. subtilis (Fig. 2). This difference in the activity of ς^B in stressed and unstressed cultures was also evident in the accumulation of the sigB operon products (Fig. 3). The sigB genes (rsbV, rsbW, sigB, and rsbX), controlled from P_B, continued to generate products in the absence, but not in the presence, of ethanol stress (Fig. 3, lane 3 versus lane 4). Ethanol treatment thus curtails the activity of ς^B, even when the activation of ς^B is independent of stress. The culture which was not IPTG treated did not show ethanol induction. This is likely due to the restricted expression of rsbU, which is downstream of the integrated plasmid. The uninduced culture did, however, display a modest increase in ς^B activity upon entry into stationary phase. Presumably, this occurred when ς^B, present at low levels in this strain, became active and triggered its further expression from P_B. The IPTG- and ethanol-treated cultures were growth impaired and did not enter stationary phase during the course of the experiment.

FIG. 4.

Effect of ethanol on ς^B induction in P_spac::rsbT strains. B. subtilis strains carrying Spβ ctc::lacZ were grown at 37°C in LB to an OD₅₄₀ of 0.1. The cultures were diluted 1:10 into fresh LB and were incubated further. When growth had recovered (OD₅₄₀ of 0.05), portions of the culture were either left untreated (■) or were treated with 1 mM IPTG (▴). Thirty minutes later, as indicated by the arrows, ethanol (4%, vol/vol) was added to a portion of each of the cultures (open symbols). Samples from each of the cultures were taken every 15 min and were analyzed for β-galactosidase. Results are the averages of two experiments. The Miller unit values (15) were normalized to 1 by using the highest respective value of each strain. (A) BSA419 (P_spac::rsbT), 1 = 113 Miller units; (B) BSJ38 (P_spac::rsbT rsbX::spec), 1 = 128 Miller units; (C) BSJ41 (P_spac::rsbT ΔrsbST), 1 = 151 Miller units; (D) BSJ42 (P_spac::rsbT rsbRTΔ5), 1 = 163 Miller units.

In previous studies, we noted that stress-induced ς^B activity did not decline in B. subtilis strains lacking RsbX (18, 22). We therefore tested whether the fall in ς^B activity, which occurred when the IPTG-induced culture was ethanol treated, also required RsbX. BSJ38 (Table 1) is a strain containing the P_spac::rsbT integration present in BSA419 plus a disruption of rsbX (rsbX::spec) (Fig. 5, lane 2). A culture of BSJ38 was induced with IPTG and a portion was exposed to ethanol stress. As was the case with the RsbX⁺ strain, IPTG induction resulted in ς^B activation; however, unlike the RsbX⁺ strain, ethanol treatment did not lead to a decline in ς^B reporter gene activity (Fig. 4B). Thus, the ethanol-dependent drop in ς^B activity requires functional RsbX.

FIG. 5.

Rsb profiles of BSA419 and mutant strains. B. subtilis strains were grown at 37°C in LB to an OD₅₄₀ of 0.1, were treated with IPTG (1 mM) to induce Pspac upstream of rsbT, and were harvested 30 min after induction and processed as described in the legend to Fig. 3. Lane 1, BSA419 (P_spac::rsbT); lane 2, BSJ38 (P_spac::rsbT rsbX::spec); lane 3, BSJ41 (P_spac::rsbT ΔrsbST); lane 4, BSJ42 (P_spac::rsbT rsbRΔ5).

The role of RsbX in the stress-induction pathway is thought to involve reactivation of RsbS, a negative regulator of RsbT (28). Given that the activation of ς^B in our artificial system was caused by the induced expression of RsbT rather than by a putative stress-triggered inactivation of RsbS by RsbT, we asked whether the fall in ς^B activity following ethanol treatment required RsbS. The RsbS⁻ strain was constructed by transforming the P_spac::rsbT plasmid into BSJ39, a strain containing a deletion in the rsbS and -T region of the sigB operon (Table 1). The resulting strain (BSJ41) has an inducible source of RsbT but lacks RsbS (Fig. 5, lane 3). As was also observed with the RsbX⁻ strain, the strain lacking RsbS failed to restrict ς^B activity after stress (Fig. 4C). This result is consistent with the notion that gratuitous expression of rsbT results in an inactivation of RsbS, which can be at least partially reactivated by RsbX in stressed B. subtilis but not in unstressed cells.

Recently, Gaidenko et al. found that RsbR could influence the ability of RsbT to phosphorylate RsbS (9). They proposed that RsbR modulated the inactivation of RsbS by RsbT, either in response to environmental signals or as part of a feedback mechanism to prevent continued stress signaling (9). This result prompted us to ask whether RsbR played a role in the stress-dependent restriction of ς^B activity which we observed in our present experiments. A RsbR⁻ mutation was constructed by deleting a 500-bp EcoRI fragment from the interior of rsbR on the plasmid pRS11 (18). The resulting plasmid (pUM49) was then linearized with ScaI and was transformed into BSA419 to generate BSA42 (rsbRΔ5 P_spac::rsbT) (Fig. 5, lane 4). When BSA42 was induced with IPTG and treated with ethanol, its ς^B activity profile (Fig. 4D) resembled that of the parent strain (Fig. 4A). There was a small reproducible difference (15% lower) in the degree to which ς^B activity fell in the RsbR⁻ strain compared to the decline in the RsbR⁺ strain; however, given that the principal pattern of decline was still evident, we conclude that RsbR is not an important component of this process. Thus, RsbX and -S, but not RsbR, are essential for the stress-activated drop in ς^B activity. Presumably, stress influences the activation state of the RsbX phosphatase and its ability to reactivate RsbS-P.

In earlier studies, we discovered that an essential GTP binding protein of B. subtilis, Obg, is needed for ς^B activation by stress (17). Obg was also found to interact with RsbT, -W, and -X in the yeast dihybrid system (17). Given the possible interaction of Obg with RsbX, we tested whether the stress-dependent stimulation of RsbX is affected by Obg. B. subtilis BSJ13 (Table 1), which carries the P_spac::rsbT construction within rsbT, as well as a second inducible promoter (P_xyl) driving the expression of obg, was used for this experiment. By withholding xylose, we can deplete Obg from the culture. This depletion of Obg causes a slowing of growth and a failure of stress to induce ς^B (17). After culturing BSJ13 in a medium without xylose to a point where growth had slowed and ς^B could no longer be activated by stress, we induced the stress pathway with IPTG and examined the ability of ethanol to restrict ς^B activity in these Obg-depleted cells. As is seen in Fig. 6, ethanol treatment could still curtail ς^B activity in the absence of Obg. Thus, the putative stress activation of RsbX appears to be independent of Obg.

FIG. 6.

Effect of ethanol on ς^B induction in Obg-depleted cells. BSJ-13 (P_spac::rsbT P_xyl::obg) was grown in LB without xylose in order to deplete Obg. When growth slowed (time 0), IPTG (1 mM) was added to a portion of the culture (triangles) to induce P_spac upstream of rsbT, while the remaining portion was left untreated (squares). Thirty minutes later, as indicated by the arrows, ethanol (4%, vol/vol) was added to a portion of each of the cultures (open symbols). Samples from each of the cultures were taken every 15 min and were analyzed for β-galactosidase (13). Results are the averages of two experiments. The Miller unit values were normalized to 1 by using the highest value for the strain (1 = 132 Miller units).

The data presented herein argue that, aside from inducing ς^B activity, ethanol stress activates a process that limits this induction. Although ethanol treatment was the only stress examined in the present study, other stresses (e.g., acid shock and salt stress) also induce ς^B transiently and likely engage in a similar process. The ethanol-responsive process requires RsbX and RsbS and presumably involves the ability of RsbX to dephosphorylate and reactivate RsbS-P. The limiting factor in this reaction is not the RsbX protein, but rather is its activation. RsbX was present at higher levels in the culture that was not ethanol treated than in the ethanol-treated culture (Fig. 3, lane 3 versus lane 4) and yet was relatively ineffective in curtailing ς^B activity. We conclude that either stress activates RsbX directly or there are additional stress-responsive factors which modulate the activity of RsbX.