INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In the enteric bacteria, including Salmonella typhimurium and Escherichia coli, the rpoS gene encodes an alternative primary sigma (specificity) factor for RNA polymerase (42, 57). RpoS, which is also called ^S or ³⁸, has functions related to stress and stationary phase. RpoS is also a virulence factor for S. typhimurium (19). More than 50 target genes show RpoS-dependent increases in expression during various stresses, including nutrient deprivation and growth into stationary phase (for a review, see references 24 and 35). The transcriptional response to each particular stress is apparently tailored by stress- and gene-specific controls. In turn, increased RpoS abundance mediated by these diverse physiological signals is orchestrated in a complex way; evidence has been presented for control of RpoS synthesis at both the transcriptional and posttranscriptional levels (24, 35) as well as control of RpoS turnover through proteolysis mediated by the ClpXP energy-dependent protease (33, 41, 54, 62). Control of RpoS activity has also been demonstrated (47, 62).

Genetic analysis of RpoS turnover in E. coli has implicated several factors in addition to the ClpXP protease. These include the abundant DNA-binding protein H-NS (4, 61) and an orphan two-component response regulator called RssB (41) or SprE (46); the S. typhimurium ortholog of this protein is called MviA (6). The N terminus of RssB/SprE/MviA is highly similar to CheY, including an aspartate residue at position 58 that, by analogy to CheY and other response regulators, is predicted to be phosphorylated and thereby to activate MviA (26). An in vitro study of RssB phosphorylation with acetyl phosphate supports this notion (8). The C terminus of MviA is not like that of any other known protein, consistent with its unique function in proteolysis rather than in DNA binding as for most response regulators. The structure and in vitro phosphate-accepting activity of RssB/SprE/MviA suggest that one or more kinases can phosphorylate MviA in vivo, but none has yet been identified. The studies described here were initiated to search for other factors (such as a kinase) that may be required for RpoS turnover.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and construction. Strains used in this study are derived from wild-type S. typhimurium LT2 and are shown in Table 1 or Table 2. Our laboratory's standard LT2 strain was originally obtained from J. Roth. As detailed in Results, this strain is defective in the turnover of RpoS, because the mviA gene of LT2 is nonfunctional. For this reason, we used a closely related mviA⁺ S. typhimurium strain obtained from W. Benjamin (WB335 [5]) to screen for new mutants affecting the turnover pathway. WB335 was previously designated an avirulent LT2 (5); we will refer to strain WB335 as LT2A in this work.

Media and growth conditions. Bacteria were grown at 37°C in Luria-Bertani (LB) medium (55) or in minimal morpholinepropanesulfonic acid (MOPS) medium (44) as modified (7) with 0.2% of the indicated compound (glucose, glycerol, sodium pyruvate, or sodium acetate) as the carbon and energy source, with one exception noted in the text. Plates were prepared by using nutrient agar (Difco) with 5 g of NaCl per liter. Antibiotics were added to final concentrations in selective plates as follows: 100 µg of sodium ampicillin/ml, 20 µg of chloramphenicol/ml, 50 µg of kanamycin sulfate/ml, and 20 µg of tetracycline hydrochloride/ml.

Screen for mutants with a defect in RpoS turnover. For transposon mutagenesis, LT2A was transformed with the plasmid pZT344, which carries the transposon Tn10d-Cam as well as the Tn10 transposase gene at a different site in the plasmid (15). Tn10d-Cam can transpose from pZT344 to the bacterial chromosome; the resulting insertions are stable once they are transferred into a background lacking transposase. Strain TE6756, which is LT2A carrying a katE-lac operon fusion, was employed to screen for mutations affecting RpoS function. TE6756 forms Lac colonies on MacConkey lactose indicator plates after incubation for 24 h at 37°C. Pooled LT2A cells transformed with pZT344 were used as donors in a phage P22-mediated transductional cross into TE6756. Cam^r transductants were screened for a Lac⁺ phenotype. After single colony isolation, all such clones still carried ampicillin resistance derived from pZT344. Therefore, before further use, candidate Lac⁺ insertions were backcrossed into the TE6756 recipient, and purified colonies were checked to confirm that they were sensitive to ampicillin. This method results in independent insertions in the LT2A chromosome that increase expression of the RpoS-dependent reporter katE-lac [op] fusion. A useful determination of the frequency of Lac⁺ mutants could not be made, because the rate of plasmid-to-chromosome transposition was not measured; however, in other experiments the frequency of insertion mutants affecting rpoS synthesis in LT-2 is about 0.1%.

/+

Screen for point mutants affecting RpoS turnover. Chemical mutagenesis was also used to isolate mutants with defective turnover of RpoS. A culture of TE6756 (LT2A katE-lac) was mutagenized with diethyl sulfate (DES) by a standard method (12) and used to grow independent cultures, which were then plated for single colonies to screen for Lac⁺ clones with elevated expression of the RpoS-dependent katE-lac [op] reporter. Elevated expression of -galactosidase was then confirmed by enzyme assay. For each independent mutant, the katE-lac reporter was removed by transduction to tetracycline resistance by using a donor phage P22 lysate grown on strain TE2929. This strain carries a Tn10d-Tet insertion 80% linked to the put locus (13). The resulting put⁺ (fusion-negative) strains were then mated with two different donors, each bearing a lac fusion on an F' plasmid marked with kanamycin resistance. One donor strain, TE6273, carries the rpoS-lac protein fusion, while the other, TE7186, carries the proV-lac [op] fusion. Assays of -galactosidase were performed with the resulting exconjugants. DES-induced mutations affecting RpoS turnover are predicted to elevate expression of proV-lac [op] but not of rpoS-lac [pr], because the latter reporter's expression reflects only changes in RpoS synthesis.

Transduction and complementation tests. All DES-induced mutations that affect RpoS turnover by the genetic test given above were assigned to genes by cotransduction and complementation tests. A new copy of the katE-lac operon fusion was first introduced into the fusion-negative version of each mutant (see above). Next, cotransduction tests were performed with P22 transducing lysates grown on each of two donor strains, one carrying a Tn10d-Tet insertion about 50% linked to mviA⁺ (TE545), the other carrying a Tn10d-Tet insertion >90% linked to clp⁺ (TE7182). With one exception, all mutants tested had defects that mapped to either the mviA or clp region by this test. The exception is briefly discussed in Results. Lesions in the clp region were further assigned by complementation tests with the plasmids pWPC9 (clpP⁺ clpX⁺), pWPC21 (clpP⁺), pWPC16 (clpX⁺), and pBR322 as a control (38). Lesions in the mviA region were assigned by a complementation test with a tandem duplication of the mviA region. This test was developed by J. Roth and colleagues (49); for a diagram and further explanation of the test, see reference 16. Briefly, strains diploid for the mviA region (and the mutation to be tested) were transduced to chloramphenicol resistance with a phage P22 donor lysate grown on strain TE6851 (mviA22::Tn10d-Cam). In the resulting Cam^r transductants, one copy of the duplicated region has inherited the mviA::Tn10d-Cam insertion of the donor, while the second copy still carries the original DES-induced mutation. If the DES-induced mutation is an allele of mviA, then both copies of mviA carry mutations, and the resulting colony phenotype will be Lac⁺. In contrast, if the DES-induced mutation is in another gene tightly linked to mviA, then most Cam^r transductants will also have repaired the DES-induced lesion in the copy of the duplication containing Tn10d-Cam, with the result that most colonies will have a Lac (wild-type) phenotype. As a control, the test was also performed with donor phage lysates grown on strain TE6826 (Tn10d-Cam insertion 50% linked to mviA⁺).

Assay of -galactosidase. Cells were centrifuged and resuspended in Z buffer (100 mM NaPO₄ [pH 7.0], 10 mM KCl, 1 mM MgSO₄) and then permeabilized by treatment with sodium dodecyl sulfate and chloroform (40). Assays were performed in Z buffer containing 50 mM -mercaptoethanol by a kinetic method with a plate reader (Molecular Dynamics). Activities (change in optical density at 420 nm [OD₄₂₀] per min) are normalized to actual cell density (OD₆₅₀) and were always compared to appropriate controls assayed at the same time. The results shown are from a single experiment; each experiment was repeated several times, with similar results (<15% variation).

Cloning of insertions and sequencing. Tn10d-Cam insertions together with flanking DNA were cloned by digestion of chromosomal DNA with BglII (for mviA insertions) or PstI (for clp) and ligation of the DNA mixture into pK184 (28) that was digested with BamHI or PstI. The desired clones were selected as Cam^r transformants, and the insertion sites were sequenced with primers specific to the cat gene (GTTTC TATCA GCTGT CCCTC CTGTT C and GACGA TATGA TCATT TATTC TGCCT C) and, in some cases with vector-specific primers.

Cloning of mviA and construction of pTE695. The mviA⁺ gene was isolated from LT2A by PCR with the primers CGCGA ATTCC ATATG ACGCA GCCAT TGGTC GGAAA AC and CGCGG ATCCT TATTC TGCAG ACAAC ATCAA GCGCA GTCGA C. These primers are derived from the E. coli sequence and introduce some changes to the S. typhimurium gene that are silent with respect to the amino acid sequence. The mviA gene was first cloned into pK184 and subsequently inserted as an NdeI-BamHI fragment into pTE571 (9), which is a derivative of pBAD18 (21). Site-directed mutants of the mviA D58 codon were constructed as described previously (10). The sequence of the mviA gene from LT2 was determined from the appropriate PCR product (Biotech Core, Palo Alto, Calif.).

Western (immunoblot) and immunoprecipitation analysis. Cultures were grown as described in the text to an OD₆₀₀ of 0.4. Electrophoresis and transfer were as described previously (9), except that detection was with tissue culture supernatant containing an anti-RpoS monoclonal antibody (R12), which is of the 2a isotype. The secondary reagent was biotinylated goat anti-mouse immunoglobulin (Ig) (Southern Biotechnology); subsequent chemiluminescence detection steps were as described previously (9). For immunoprecipitation, cells were grown to an OD₆₀₀ of 0.4 and then labeled for 1 min with 100 µCi of Tran³⁵S-label L-[³⁵S]methionine and L-[³⁵S]cysteine (ICN), and labeled proteins were prepared as described previously (3, 27). Immunoprecipitation was performed with the R12 monoclonal antibody (MAb) and protein A-Sepharose beads (Sigma).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of mutants with defects in RpoS turnover. A katE-lac [op] fusion was used as the reporter to screen for new mutations affecting RpoS turnover in S. typhimurium. Mutants with elevated RpoS show increased transcription of katE-lac and can be visualized as Lac⁺ clones on MacConkey lactose indicator plates. Both transposon and chemical mutagenesis were employed, as described in more detail in Materials and Methods. The screens were complicated by the finding that mutations which increase RpoS synthesis are common and affect multiple genes (data not shown). Therefore, in the transposon mutagenesis, we first screened for elevated expression of katE-lac in the LT2A background and then eliminated insertions that also elevated expression of katE-lac after transduction into our standard LT2 background. As discussed below, others have reported (6), and we have confirmed, that LT2 is an mviA mutant and defective for RpoS turnover. Mutants affecting RpoS turnover are predicted to have the parental Lac phenotype in LT2 bearing katE-lac, since the turnover pathway is already defective. In contrast, mutants causing increased RpoS synthesis should show increased katE-lac expression in both LT2 and LT2A backgrounds. A total of 90 mutants arising by insertion of Tn10d-Cam and Mud-Cam were tested; 12 of these passed the second screen and were studied further (Table 2). DNA sequencing and PCR tests showed that 3 are alleles of clpX and 9 are alleles of mviA. No new genes were identified in this experiment.

Mutations known to affect expression of katE-lac in E. coli. We also tested mutations in the S. typhimurium counterparts to genes previously shown to affect RpoS turnover in E. coli. Derivatives of both LT2 and LT2A were constructed that carry the RpoS-dependent katE-lac reporter as well as insertion mutations disrupting the mviA, clpP, and hns genes. Expression of katE-lac was not changed by any of these mutations in the standard wild-type LT2 strain background during growth to stationary phase in rich medium (Fig. 1, left panel). In contrast, when any one of these mutations was present in LT2A, a three- to fourfold increase in expression of the reporter was observed under the same conditions. For the mviA and clpP mutants, introduction of an rpoS null allele reduced katE-lac expression to <3% of the level seen in the rpoS⁺ parent; for the hns mutant, the rpoS-independent expression was about 10% (data not shown). We conclude that our laboratory strain of LT2 is defective for the RpoS turnover pathway and that LT2A contains a functional pathway. We mapped the LT2 defect to the mviA region by cotransduction tests; the mutation was confirmed to be the V102G allele described for a closely related LT2 strain, WB600 (5, 6). Because the hns and mviA genes are tightly linked, it was important to show that strain TE7512 (LT2A hns::Kan) is mviA⁺.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Testing mutations predicted to affect RpoS turnover for their effect on expression of a katE-lac [op] reporter. Cultures were grown overnight to stationary phase in LB medium (left panel) or minimal MOPS medium, with 0.2% glucose as the source of carbon and energy (right panel). The activity of -galactosidase was assayed as described in Materials and Methods and is reported in arbitrary units. Strains have LT2 or LT2A backgrounds as indicated and are either wild type (white bar), mviA::Kan (dark gray bar), clpP::mini-Tn5 (light gray bar), or hns::Kan (black bar).

Role of lrhA in RpoS turnover. Very recently, evidence has been obtained with E. coli for the involvement of lrhA in RpoS turnover (20). This gene lies immediately upstream of and is transcribed toward the nuo operon encoding the 14-subunit energy-conserving NADH dehydrogenase. (The next characterized open reading frames [ORFs] comprise the ack-pta operon, which lies about 6 kb upstream of lrhA and is divergently transcribed.) Gibson et al. found that an lrhA null mutation was as effective as an sprE/rssB null mutation in elevating RpoS levels in E. coli. They also found that constitutive and, likely, overexpression of lrhA due to a transposon insertion 349 bp upstream of lrhA resulted in hyperactive RpoS turnover, a phenotype that requires both SprE/RssB and ClpXP function. We tested the role of lrhA in S. typhimurium, using several transposon and interposon insertions isolated in a previous study (2), including an lrhA::-Cm insertion at codon 100 of this 313-codon gene. No effect was seen for this presumed null allele of lrhA, either with katE-lac (Fig. 2) or with any of three other RpoS-dependent reporter fusions (data not shown). However, a threefold decrease in expression of the RpoS-dependent katE-lac reporter was seen for the zeg-6816::Tn10d-Cam insertion, which lies 379 bp upstream of the lrhA ATG codon. This is consistent with the hyperturnover phenotype observed for the similarly positioned insertion in E. coli (20). An mviA null mutation was epistatic to zeg-6816::Tn10d-Cam (data not shown). We conclude that loss of LrhA does not affect RpoS turnover in S. typhimurium during growth in LB medium but that its overexpression may do so.

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Expression of katE-lac [op] in strains carrying mutations in the lrhA region. All strains are LT2A containing katE-lac [op]: TE6756 (wild type) (white bar), TE7637 (zeg-6816::Tn10d-Cam) (speckled bar), TE7640 (lrhA::-Cm [codon 100]) (light gray bar), TE7638 (lrhA::-Cm [codon 297]) (dark gray bar), and TE7639 (nuo p14::-Cm) (black bar). Cultures were grown to stationary phase in LB medium, and -galactosidase activity was assayed as described in Materials and Methods. The map (bottom) shows the position of each insertion relative to the lrhA and nuoA genes.

Role of phosphorylation in activation of MviA. The aspartate residue at position 58 of MviA (D58) is predicted to be the phosphorylation site based on experiments with CheY (50) and other response regulators, such as NtrC (30, 51). Studies with these proteins also lead us to predict that the mutant MviA D58N protein should be nonphosphorylated, and consequently this protein will function poorly or not at all. We constructed derivatives of the plasmid pBAD18 (21) that express wild-type S. typhimurium mviA or its D58N mutant derivative and tested the function of these genes as described in the legend to Fig. 3. In the absence of the inducer arabinose, the P_BAD promoter is partially active (e.g., reference 22). At this level of expression, the wild-type MviA protein gives wild-type (i.e., inhibited) levels of expression of the katE-lac reporter, whereas the D58N mutant protein is defective. Since MviA D58N functions poorly, D58 phosphorylation is likely required for wild-type activity. When arabinose was added to increase transcription from the P_BAD promoter, both versions of MviA showed increased activity, although D58N did not function as well as the wild type.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   Complementation of an mviA insertion mutant by plasmids carrying wild-type mviA or a mutant (D58N) affecting the predicted phosphorylation site. Strains analyzed were derivatives of TE7199 (S. typhimurium LT2A putPA1303::Kanr-katE-lac [op] mviA22::Tn10d-Cam recA1) containing the following plasmids: vector (pTE570), wild-type mviA (pTE695), and mviA D58N (pTE695 D58N). Cultures were grown to an OD600 of 0.3 in minimal MOPS medium with 0.2% glucose, 50 µg of ampicillin/ml, and the indicated amount of arabinose (white bars, none; light gray bars, 0.02%; black bars, 0.2%) Activity of -galactosidase was assayed as described in Materials and Methods.

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Testing mutants defective in acetyl phosphate metabolism for expression of an RpoS-dependent reporter. All strains are LT2A derivatives that contain the reporter fusion katE-lac [op]: TE6756 (wild-type) (white bars), TE6850 (clpX1::Tn10d-Cam) (dark gray bars), TE6851 (mviA22::Tn10d-Cam) (speckled bars), TE7109 (pta-209::Tn10) (black bars), and TE7110 (ack-408::Tn10) (light gray bars). Cultures were grown overnight to stationary phase in minimal MOPS medium containing 0.2% glucose, pyruvate, or acetate as indicated, and -galactosidase activity was assayed as described in Materials and Methods.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   Testing the effect of a ptsI mutant on expression of an RpoS-dependent reporter. All strains are LT2A derivatives that contain the reporter fusion katE-lac [op]: TE6756 (wild type) (white bars), TE6851 (mviA22::Tn10d-Cam) (dark gray bars), TE7382 (ptsI421::Tn10) (light gray bars), and TE7387 (ptsI421::Tn10 mviA22::Tn10d-Cam) (black bars). Cultures were grown overnight to stationary phase in minimal MOPS medium, with 0.2% galactose as the source of carbon and energy, and -galactosidase activity was assayed as described in Materials and Methods.

Induction of RpoS during growth in acetate. We first tested whether the apparent effect of growth in acetate on RpoS function was reporter specific. To do this, expression of katE-lac was compared with that of three other RpoS-dependent reporters. The proV-lac fusion is strongly dependent on RpoS function for its expression in both E. coli and S. typhimurium (37 and data not shown). The other lac fusions employed were formed by insertion of the transposon MudJ in otsA and ORF o186; these were isolated and characterized previously for their dependence on RpoS (18). Each of the four reporters was studied in a strain derived from LT2A that also carries an insertion in clpX to knock out the RpoS turnover pathway. The presence of the clpX mutation magnifies the signal from each reporter and ensures that the observed regulation is not due to effects on RpoS turnover. Each reporter strain showed five- to sixfold-higher expression of -galactosidase after overnight growth to stationary phase in minimal acetate medium compared with growth in minimal glucose (Fig. 6).

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Expression of four different RpoS reporter fusions in cells grown on glucose or acetate as the sole carbon and energy source. All strains are LT2A containing the clpX1::Tn10d-Cam insertion as follows: TE6850 (katE-lac [op]), TE7632 (proV-lac [op]), TE7633 (orfo186::MudJ), and TE7634 (otsA::MudJ). Cultures were grown overnight to stationary phase in minimal MOPS medium containing glucose (white bars) or acetate (black bars), and -galactosidase activity was assayed as described in Materials and Methods.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Western (immunoblot) analysis of RpoS abundance during growth on different carbon sources. The arrow indicates RpoS protein. Cultures of TE6357 (LT2A) were grown to exponential phase (OD600 = 0.4) in minimal MOPS medium with 0.2% of each of the following compounds as the carbon and energy source: (a) glucose, (b) glycerol, (c) pyruvate and (d) acetate. Lane (m), molecular mass markers (with sizes as indicated, in kilodaltons).

View larger version (10K): [in this window] [in a new window]   FIG. 8.   Pulse-labeling and immunoprecipitation of RpoS. Cultures were grown in minimal MOPS medium with 0.2% glucose (lanes a and b) or 0.2% acetate (lanes c and d) to an OD600 of 0.4, pulse-labeled with L-[35S]methionine and L-[35S]cysteine for 1 min, and then immunoprecipitated with a monoclonal anti-RpoS antibody. The samples analyzed in lanes b and d received unlabeled methionine and cysteine and were chased for 20 min. The strain used was TE6357 (S. typhimurium LT2A).

Growth rate regulation of RpoS that is independent of hfq function. At least three general mechanisms might explain the observed regulation. CRP-mediated transcription or repression might regulate some unknown regulator of RpoS, in a manner dependent on very high levels of cAMP, as would be achieved during growth in acetate. Alternatively, RpoS might respond to an acetate-specific regulator such as FadR or IclR. Finally, RpoS might be responsive to growth rate but not to the carbon source per se. The level of cAMP and the presence of a cya mutation were previously reported to affect RpoS expression in E. coli (32), although the reported effect was negative rather than positive as predicted by this model. Since crp and cya mutants are unable to grow with acetate as their sole carbon and energy source, we tested the model by examining a crp* strain, which is predicted to have inappropriately high expression of RpoS during growth in glucose. We compared expression of the katE-lac reporter fusion in strain TE6756 (LT2A, otherwise wild-type) with that in strain TE7631 (LT2A cya::Tn10 crp*) during exponential growth in both glucose and acetate. These two strains did not show significant differences in katE-lac expression in the same medium, and both showed equivalent acetate induction (data not shown). We conclude that CRP does not play an important role in the acetate regulation of RpoS in S. typhimurium.

View larger version (29K): [in this window] [in a new window]   FIG. 9.   Western (immunoblot) analysis of RpoS abundance during growth on glucose restricted by -methylglucoside. Samples analyzed here were obtained from the same cultures as those analyzed in the experiment whose results appear in Table 3. The single arrow at the right indicates native RpoS protein. Cultures of TE7579 (LT2A rpoS-lac [pr]) were grown to exponential phase (OD600 = 0.3) in minimal MOPS containing 2 mM glucose as the carbon and energy source and supplemented with no addition (a), or with 40 mM (b), 80 mM (c), or 120 mM (d) -methylglucoside. The same strain was also grown in minimal MOPS containing 0.2% acetate as the carbon and energy source (e). The positions of molecular mass markers run on the same gel are also shown (with sizes as indicated, in kilodaltons).

View larger version (14K): [in this window] [in a new window]   FIG. 10.   Expression of rpoS-lac [pr] in cells grown in minimal MOPS medium containing 0.2% glucose (white bars) or acetate (black bars). Strains were as follows: TE6253 (LT2 rpoS-lac [pr], TE6266 (LT2 rpoS-lac [pr] hfq::Mud-Cam), TE7579 (LT2A rpoS-lac [pr]), TE7585 (LT2A rpoS-lac [pr] hfq::Mud-Cam). Cultures were grown to an OD600 of 0.4, and -galactosidase activity was assayed as described in Materials and Methods.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The initial objective of this study was to isolate mutants defective in the genes required for RpoS turnover in S. typhimurium, with the goal of understanding how the activity of this pathway is regulated in response to carbon starvation and osmotic challenge. The structure of the response regulator RssB/MviA and its ability to be phosphorylated in vitro by acetyl phosphate suggest that, as for other response regulators (26), MviA-P is the active form of the protein in vivo. Since virtually all response regulators are members of a two-component team, we would expect the genetic analysis to identify mutants defective in the cognate histidine kinase that acts on MviA. Such mutants should have a phenotype similar to that of an mviA mutant. However, our studies with S. typhimurium did not reveal any mutants lacking an MviA kinase, and previous studies with E. coli did not reveal any candidates. The mutant hunt was substantial, although clearly not exhaustive, since mutations in one gene shown to function in the pathway, hns, were not recovered. Of course, kinase mutants may exist but have escaped recovery or detection.

Although our results do not compel it, we suggest that other models of MviA action should be considered to account for the absence of mutations affecting a histidine kinase. Failure to obtain these might have several causes. One possibility is that MviA is active without being phosphorylated in vivo. To test this model, we constructed a mutant mviA gene encoding MviA D58N, which replaces aspartic acid with asparagine at the predicted phosphorylation site. This mutant protein is substantially defective in vivo in a complementation test. In another model MviA is directly phosphorylated by a small molecule such as acetyl phosphate or by a promiscuous phosphate donor such as a component of the PTS, circumventing the need for a dedicated MviA kinase. A role for acetyl phosphate has been proposed previously in E. coli based on a two- to threefold increase in RpoS half-life in strains defective in acetyl phosphate synthesis (8). We tested the acetyl phosphate model in S. typhimurium, using well-defined pta and ack mutants, but obtained no evidence to support it. S. typhimurium and E. coli may differ in this respect. However, it should be noted that neither the genes covered by the deletion nor the position of the linked Tn10 employed in the E. coli study are known (8), and no complementation test of the deletion mutant was performed.

Another model invokes a single kinase with multiple substrates. In this case, mutants defective in the kinase might not survive because its second substrate has an essential function, or the kinase mutants might not pass our screen if kinase activity also negatively controls the synthesis of RpoS. No response regulator controlling RpoS synthesis is envisioned currently, and this general class of models seems unlikely.

Third, there might be multiple kinases whose functions are redundant. If so, then a single mutant affecting only one kinase would have a wild-type or nearly-wild-type phenotype. A problem with the two-kinase model is that induction of RpoS protein stabilization should require the simultaneous inhibition of both kinases, so what advantage such an arrangement would have for the cell is not clear. The simplest solution, and one with a precedent in the enteric bacteria, is to imagine a single dedicated kinase, whose associated phosphatase activity normally prevents inappropriate activation of MviA by another kinase. More complex models can be imagined, with two kinases, for example (whether dedicated to MviA or not), and regulation exerted by control of MviA abundance. There might be a protein analogous to CheZ that promotes dephosphorylation of MviA or a protein that interferes with the formation of the postulated MviA-RpoS complex by binding to MviA. MviA phosphorylation might be required for its activity even though regulation is exerted through another type of modification (e.g., by acetylation [48]). Such models would be novel for two-component systems of enteric bacteria but familiar to those working with other bacteria such as Bacillus subtilis.

In the course of these studies, we have confirmed and extended the findings of a previous study which indicated that RpoS is growth rate regulated (45). Here, we observed a fivefold elevation in RpoS synthesis during growth with acetate as the sole carbon and energy source compared to the basal level obtained with several other carbon sources, including glucose, glycerol, and pyruvate. Actually, it has been known for some time that catalase HPII, the product of katE, is increased in cells growing on trichloroacetic acid (TCA) cycle intermediates and acetate (36). Previous work with E. coli has implicated acetate but also other nonmetabolized weak acids such as benzoate in the induction of RpoS synthesis (43, 52). The addition of acetate had little effect on RpoS in glucose medium and none in glycerol medium, in contrast to experiments with E. coli. This result suggested that carbon and energy restriction, rather than some other effect of acetate, is responsible for RpoS induction.

The finding of high RpoS during growth on acetate led to an experiment showing that RpoS levels are also sharply elevated (up to fivefold) during exponential growth in cells whose growth rate is limited by slow transport of glucose. This suggests that acetate regulation is actually a specific manifestation of a generalized growth rate-dependent regulation. Based on analysis of lac fusions, the regulation acts not on RpoS turnover but on its synthesis and occurs mainly at the posttranscriptional level. The two most notable features of growth rate-dependent regulation of RpoS are that (i) it occurs in exponentially (albeit slowly)-growing cells rather than in shocked or stationary-phase cells and (ii) it is independent of the Hfq protein required for most previously described posttranscriptional regulation of RpoS.