Vol. 174, No. 11JOURNAL OF BACTERIOLOGY, June 1992, p. 3695-37060021-9193/92/113695-12$02.00/0Copyright © 1992, American Society for Microbiology

Activation of Bacillus subtilis Transcription Factor 0FB by a

Regulatory Pathway Responsive to Stationary-Phase SignalsSHARON A. BOYLAN, ANN RUTHERFORD, SUSAN M. THOMAS, AND CHESTER W. PRICE*

Department ofFood Science and Technology, University of California, Davis, California 95616

Received 28 January 1992/Accepted 25 March 1992

Alternative transcription factor cB of Bacillus subtilis controls a stationary-phase regulon induced undergrowth conditions that do not favor sporulation. Little is known about the metabolic signals and protein factorsregulating the activity of &-B. The operon containing the cM structural gene has the gene order orJV-orJW-sigB-rsbX, and operon expression is autoregulated positively by cB and negatively by the rsbX product (rsbX= regulator of sigma B). To establish the roles of the orJVand orjW products, orJVand orJW null and missensemutations were constructed and tested for their effects on expression of the crB-dependent genes ctc and csbA.These mutations were tested in two contexts: in the first, the sigB operon was under control of its wild-type,rB-dependent promoter, and in the second, the sigB operon promoter was replaced by the inducible Pspacpromoter. The principal findings are that (i) the orjY (now called rsbV) product is a positive regulator ofcMB-dependent gene expression; (ii) the orfW (now called rsbW) product is a negative regulator of suchexpression; (iii) cB is inactive during logarithmic growth unless the rsbW product is absent; (iv) the rsbX, rsbV,and rsbW products have a hierarchical order of action; and (v) both the rsbV and rsbW products appear toregulate cB activity posttranslationally. There are likely to be at least two routes by which information canenter the system to regulate cM: via the rsbX product, and via the rsbV and rsbW products.

As the enzyme of central importance in bacterial geneexpression, DNA-dependent RNA polymerase is the com-mon target of diverse regulatory mechanisms. Among themost potent of these is the association of core RNA poly-merase with alternative sigma (a-) factors, which conferdifferent promoter recognition specificities on the polymer-ase holoenzyme and reprogram the pattern of gene expres-sion in response to environmental, cell cycle, and morpho-logical signals (18). Because alternative a factors are suchpowerful regulatory effectors, the molecular mechanismscontrolling their synthesis or activity in response to suchsignals are of fundamental importance.

In Bacillus subtilis, a cascade of at least five differentalternative a- factors controls the sporulation process, whichis a complex program of morphological and structuralchange occurring under certain stationary-phase growthconditions (30). The regulation of both the synthesis andactivity of these cr factors is chiefly responsible for couplingdevelopmental gene expression to morphological events(46). But sporulation is only one of several patterns of geneexpression manifested by B. subtilis in stationary phase.Other examples include development of genetic compe-tence, production of antibiotics and extracellular enzymes,and the necessary adaptation to nutrient limitation (43). Theimportant question remains how the cell senses and inte-grates the information required to achieve the appropriatebalance of these myriad stationary-phase activities.Our approach to understanding how stationary-phase sig-

nals are conveyed to the transcription apparatus is to iden-tify the protein and metabolic effectors that regulate theactivity of the alternative transcription factor caB. a-B, for-merly called a37, was originally discovered biochemicallythrough its association with a unique RNA polymeraseactivity present in early-stationary-phase cells (17). Theexact physiological role of -B has not yet been established,

* Corresponding author.

and genetic analysis has shown that cr' is one of two knownstationary-phase o factors that is not essential for sporula-tion (4, 15, 23, 27). (The other is crD, which controls thechemotaxis-motility regulon [32].) Rather, cB is required formaximal expression of a class of genes called csb (forcontrolled by sigma B), which constitute a large regulonexpressed in early stationary phase under conditions notconducive to sporulation or to formation of tricarboxylicacid cycle enzymes (6, 8, 23, 24).We previously reported that the (rB structural gene (sigB)

lies third in a four-gene operon which is subject to complexregulation (27). The genetic organization of the sigB operonis shown in Fig. 1. CrB positively regulates expression of itsown operon from a uB-dependent promoter immediatelyupstream, whereas the product of the fourth gene (orX) is anegative regulator of operon expression. Our most strikingfinding was that the predicted products of the first two genesof the sigB operon (orJV and orJTV) have significant sequenceidentity with the products of the first two genes of the B.subtilis spoIL4 operon (spoIIAA and spoILAB), the thirdgene of which (spoILaC) encodes the sporulation-essentialaF. On the basis of this clear evolutionary relationship andfrom the known phenotypes of spoIIAA and spoIIAB mu-tants (38, 50), we suggested that crB and aF might beregulated by a common mechanism involving the products oftheir respective upstream open reading frames.Evidence confirming part of this suggestion was reported

by Schmidt et al. (42), who demonstrated that the spoIIAAand spoIIAB products are potent effectors of CnF-dependentgene expression. Here we present genetic evidence that theorfV product is a positive regulator and the orJW product isa negative regulator of uB-dependent gene expression andthat the orfX, orJV, and orJW products form a hierarchicalregulatory pathway. Related experiments conducted inde-pendently by Benson and Haldenwang (3) are consistentwith these data. We have further shown that both the orJVand orJW products act at the posttranslational level tocontrol (rB activity, that uB is activated by cellular or

3695

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3696 BOYLAN ET AL.

TABLE 1. B. subtilis strains

Strain Relevant genotypea Reference or constructionb

PB2 trpC2 Wild-type Marburg strain (37)PB79 lys::Tn917 trpC2 5PB114 sigB::pMD10 27PB155 csbA::Tn9I7lacZ trpC2 8PB197 SPI3ctc-lacZc trpC2 SPlctc-lacZ .PB2cPB198 amyE::pDH32-ctc trpC2 pDH32-ctc-*PB2PB199 rsbVAI SPlctc-lacZ This studyPB200 rsbV42 SPpctc-lacZ This studyPB201 rsbWl amyE::pDH32-ctc This studyPB202 sigBA2::cat dal-1 trpC2 AG232-*PB2PB203 rsbV42 This studyPB204 rsbV42 amyE::pDH32-ctc pDH32-ctc-->PB201PB205 rsbVA&l This studyPB206 rsbVAl amyE::pDH32-ctc pDH32-ctc--*PB203PB207 amyE::pDH32-ctc trpC2 pDG148-rsbW pDG148-rsbW--.PB198PB208 amyE::pDH32-ctc trpC2 pDG148 pDG148-*PB198PB209 socB trpC2 EU100- PB202PB210 socB amyE::pDH32-ctc trpC2 pDH32-ctc-*PB209PB212 Pspac (rsbVr rsbW+ sigB+ rsbX) amyE::pDH32-ctc trpC2 pAR7--PB198PB213 Pspac (rsbVW rsbWAl sigB+ rsbX+) amyE::pDH32-ctc trpC2 pAR7-+PB198PB214 Pspac (rsbVAl rsbW+ sigB+ rsbX+) amyE::pDH32-ctc trpC2 pAR6-*PB198PB216 P ac (rsbVr rsbW+ sigB+ socB) amyE::pDH32-ctc trpC2 pAR7-*PB210PB217 rstVAl rsbX::ery amyE::pDH32-ctc trpC2 pSK15-.PB206PB218 rsbX::ery amyE::pDH32-ctc trpC2 PB217- PB198PB219 Pspac (.[rsbV-rsbW] sigB+ rsbX+) amyE::pDH32-ctc trpC2 pAR8- PB198PB221 Pspac (rsbVr rsbW+ sigB::pMD10) pAR7- PB114PB222 Pspac (rsbVAI rsbW+ sigB::pMD10) pAR6--PB114PB226 amyE::pAWB trpC2 pAWB--PB2PB238 amyE::pAWB trpC2 pDG148-rsbW pDG148-rsbW-->PB226PB239 csbA::Tn9l71acZ trpC2 pDG148-rsbW pDG148-rsbW-- PB155PB240 spoVG+::pZL207 spoOJ93 pheAI trpC2 pDG148-rsbW pDG148-rsbW--*ZB223AG232 sigBA2::cat dal-1 pheAl trpc2 Alan GrossmanEU100 socB pheAI trpC2 23ZB223 spoVG'::pZL207 spoOJ93 pheAI trpC2 51

a p c refers to the replacement of the wild-type sigma B operon promoter by the inducible Pspac promoter from pAG58 (26).b The arrow indicates transformation from donor to recipient.SPfIc2 del2::Tn9l7(cat) ctc::lacZA798 (24).

environmental signals present at the end of logarithmicgrowth, and that inactivation of the orW product is aprincipal mechanism by which these signals are transmittedto er1. Because the orfV, orW, and o42[ products all controler activity, we have renamed these genes rsbV, rsbW, andrsbX (for regulator of sigma B).

MATERIALS AND METHODS

Bacterial strains and genetic methods. We used Escherichiacoli TG-1 (Amersham) as the host for pDG148 (44) and itsderivatives and E. coli DH5a (Bethesda Research Labora-tories) as the host for all other plasmid constructions. The B.subtilis strains used are listed in Table 1. For strain construc-tions, B. subtilis PB2 and its derivatives were the recipientsfor natural transformation with linear and plasmid DNA (13).Transformation selections for drug-resistant B. subtilisstrains were done on tryptose blood agar plates (DifcoLaboratories) containing either chloramphenicol (5 jig/ml),kanamycin (5 jig/ml), or erythromycin (0.5 p,g/ml) pluslincomycin (12.5 ,ug/ml). The minimal medium used for B.subtilis was that of Anagnostopoulos and Spizizen (1), andLuria broth (LB) and M9 minimal glucose medium werefrom Davis et al. (12).DNA methods. All standard recombinant DNA methods

have been described previously (7), and polymerase chainreactions (PCRs) were done by standard protocols (25).

DNA sequencing reactions were done by the dideoxynucle-otide chain termination method (41) with appropriate restric-tion fragments cloned into M13mpl9 and pUC19. We usedSequenase enzyme (U.S. Biochemical) and [a-35S]dATP(Amersham) to label sequencing reaction mixes. Primingwas done with custom oligonucleotide primers (from OperonTechnologies, Alameda, Calif.) or standard forward andreverse primers (from U.S. Biochemicals).a"-dependent promoter fusions. Five different e3-depen-

dent lacZ fusions were used to monitor e13 synthesis oractivity in wild-type and mutant strains. We primarily used asingle-copy transcriptional fusion of the ctc promoter in-serted at the amyE locus of strain PB198. This strain wasconstructed by moving a 149-bp EcoRI-HindIll fragmentcontaining the ctc promoter from pLC4 (39) into the EcoRIsite of pDH32, immediately upstream from the promoterless,hybrid spoVG-lacZ reporter gene of the vector (20). pDH32-ctc was cut with ScaT and transformed into a PB2 recipient,in which the linear transcriptional fusion was incorporatedinto the amyE chromosomal locus via a double crossoverevent, with regions ofamyE homology carried by the pDH32vector. pDH32 was also the basis of the second transcrip-tional fusion, pAWB, which was constructed by Arlene Wiseby using the 830-bp EcoRV-AflIl fragment carrying the sigBoperon promoter (27) and borne at the amyE locus of strainPB226. The third transcriptional fusion was the recentlydescribed csbA::Tn9171acZ in strain PB155 (8). The fourth

J. BACTERIOL.

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

REGULATION OF BACILLUS SUBTILIS aB 3697

si2B3

FIG. 1. sigB operon and strategy to replace the sigB operonpromoter. Physical map of the sigB operon (27) showing theo&B-dependent operon promoter (P.), the rsbV, rsbW, sigB, and rsbXcoding regions (arrows), and the operon terminator (stem-loop). TheorfU coding region lies immediately upstream from the PB promoter(49). The restriction sites shown in the 2.3-kb EcoRI fragmentcontaining PB were used to construct in-frame deletions of rsbV andrsbW and to replace PB with the inducible Pspac promoter cassette,shown at the bottom (not to scale). For the Pspac construction,regions from the B. subtilis chromosome up- and downstream fromPB were subcloned into pAR4, flanking the E. coli lactose repressorgene (lacl), the pE194 erythromycin resistance gene (ery), thepBR322 ampicillin resistance gene (bla), and the SPOl phagepromoter controlled by the E. coli lac operator (Pspac). We trans-formed B. subtilis with the linearized construction, replacing PBwith Pspac by the double crossover event shown. Similar Pspacconstructions were used to move rsbV and rsbW deletions into thechromosome by linked transformation with the ery marker of thePspac cassette (see text).

A.

fusion was the ctc-lacZ translational fusion constructed byIgo and Losick (24) and carried on the resident SPP pro-phage of strain PB197. The fifth was a translational fusionbetween codon 45 of sigB and codon 8 of lacZ, integrated insingle copy into the sigB locus and carried in strain PB114(27).

Construction of missense and deletion mutations. We mademissense mutations in rsbV and rsbW by oligonucleotidemutagenesis in vitro, using the dut-ung E. coli strain (29) asdescribed in the Muta-Gene kit (Bio-Rad). For the followingconstructions, all nucleotide (nt) numbers refer to Fig. 2 ofKalman et al. (27). The 1,396-bp PstI-EcoRI fragment con-taining rsbV, rsbW, and part of sigB (Fig. 1) was cloned intoM13mpl9. Mutagenic primers included a 21-mer comple-mentary to nt 467 to 487, containing a transition at nt 477 (Gto A), to generate the rsbV42 mutation (Gly-95 to Asp in Fig.2A) and a 22-mer complementary to nt 555 to 576, containinga transition at nt 566 (C to T), to generate the rsbWlmutation (Ala-14 to Val). We used the same method togenerate the 258-bp in-frame deletion in rsbV (rsbV.iJ) via a50-mer complementary to nt 211 to 235 and 494 to 518. Themutational alterations were confirmed by DNA sequenceanalysis, and the mutant PstI-EcoRI fragments were clonedinto pUC19.

Restriction digestion was used to make an in-frame dele-tion within rsbW (Fig. 2B) as well as a second in-framedeletion removing most of rsbV and rsbW. pSTR1 carriedthe rsbV, rsbW, and part of the sigB coding regions on the1,391-bp PstI-EcoRI fragment cloned into the PstI andEcoRI polylinker sites of pUC19. pSTR1 was digested withNaeI and BanII (Fig. 1), treated with T4 DNA polymerase to

[-r rsbVA 1

RsbV MNINVDVKQNENDIQVNIAGEIDVYSAPVLREKLVPLAEQGADLRICLK--DVSYMDSTG. .. . .. .. .: . .. .: :.

SpoIIAA MSLGIDMNVKESVLCIRLTGELDHHTAETLKQKVTQSLEKDDIRHIVLNLEDLSFMDSSG

D (rsbV42)

Va-iLGVFVGTFKMVKKQGGSLKLENLSERLIRLFDITGLKDIIDISAKSEGGVQ.:.. .: . . .:. . ....

LGVILGRYKQIKQIGGEMVVCAISPAVKRLFDMSGLFKIIRFEQSEQQALLTLGVAS

D (spolIA42)

B._rsbWA 1

v (rsbWl)

RabW MKNNADYIEMKVPAQPEYVGIIRLTLSGVASRMGYTYDEIEDLKIAVSEACTNAVQHAYK... . .: .. :.:. ... .. : .. .. :. .:::: : :. :.: .

SpoIlAs MKNEMHLEFSALSQNESFARVTVASFIAQLDPTMDELTEIKTVVSEAVTNAIIHGYE

v (spollABl) 'XEDKNGEVSIRFGVFEDRLEVIVADEGDSF-DFDQKQQDLGPYTPSHTVDQLSEGGLGLYL 119.. ...ET..T...I.EL..D. -..-.- ..

ENCEGKVYISVTLEDHVVTMTIRDEGLGITDLEEARQPL--FTTKPELER---SGMGFTI 112

METLMDEVRVQ--NHSGVTVAMTKYLNGERVDHDTTIKNYETN

T.AC.MENFMDDVS IDSSPEMGTTIRLTKHLSKSKALCN

160

146

FIG. 2. Mutations introduced into the rsbV and rsbW coding regions. The alignments of the rsbV and spoIIAA products and of the rsbWand spoIL4B products are from Kalman et al. (27). Sequences are given in the single-letter code, with identical residues indicated by two dotsand conserved substitutions by a single dot. (A) Gly-95 to Asp (spoIIAA42) alteration of SpoIIAA (50) and the corresponding Gly-93 to Asp(rsbV42) change we made in RsbV (see Materials and Methods). Also shown are the residues removed by the rsbVAJ deletion, enclosed byarrows. (B) Ala-11 to Val (spoIIABI) alteration of SpollAB (38) and the equivalent Ala-14 to Val (rsbWl) mutation we made in RsbW. Theresidues removed by the rsbWAl deletion are enclosed by arrows.

58

60

109

117

60

57

VOL. 174, 1992

pB

u -w P-.1

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3698 BOYLAN ET AL.

make blunt ends, and then ligated to make pSTR2. DNAsequence analysis indicated that the rsbWAJ mutation re-moved 306 bp (nt 564 to 869) from within rsbW, deleting theregion coding for residues most similar to SpoIIAB butleaving intact the region of dyad symmetry at the 3' end ofrsbW (27). The A(rsbV-rsbW) mutation was similarly madeby removal of the 539-bpAflIl-BanII fragment (Fig. 1; nt 331to 869) to make pSTR3. This deletion removed 66 codonsfrom the 3' end of rsbV and 115 codons from the 5' end ofrsbW.

Replacement of chromosomal rsbV and rsbW alleles bycongression. We replaced the wild-type chromosomal copiesof rsbV or rsbW with the appropriate rsbVAI, rsbV42, andrsbWl alleles by congression (cotransformation) with theunlinked trpC+ marker. In all constructions, the presence ofeach mutant allele in the chromosome was confirmed byDNA sequence analysis of PCR products amplified from thesigB chromosomal locus. The trpC2 recipients PB197(SPPctc-lacZ) and PB198 (pDH32ctc-lacZ) were trans-formed with a mixture of 0.1 p.g of linearized pTrp-H3plasmid DNA (2) per ml and 200 ,ug of linearized pUC19plasmid containing the rsbV or rsbW mutation per ml. Trp+transformants were selected on minimal glucose plates;these plates also contained X-Gal (5-bromo-4-chloro-3-in-dolyl-r3-D-galactopyranoside; 80 ,ug/ml) to allow screeningfor colonies which differed in color from the parent. ThersbVAl mutation introduced into PB197 resulted in coloniesthat were white compared with the medium blue of theparent, whereas the rsbV42 allele rendered the colonies lightblue. These strains were called PB199 and PB200, respec-tively. The rsbWl transformants of PB198 had darker bluecolonies than the parent; this strain was called PB201. Thefrequency of congression of these mutant colony phenotypesranged from 2 to 5%.We placed the rsbV42 and rsbVIXJ alleles in backgrounds

containing the pDH32ctc-lacZ transcriptional fusion as fol-lows. For the rsbV42 allele, PB200 was cured of its residentSPp3ctc-lacZ fusion by overnight growth in LB at 50°C; lossof the prophage was scored by chloramphenicol sensitivity.The cured strain, PB203, was transformed with linearizedplasmid pDH32ctc-lacZ to make PB204. For the rsbVAJallele, we took advantage of the linkage of the sigB locus tothe dal-] marker (4, 15). Strain PB202 (dal-] sigBzA2::cat)was transformed with chromosomal DNA from PB199(rsbVzv1), selecting for Dal'. Scoring for chloramphenicolsensitivity allowed identification of transformants in whichthe sigBA2::cat region of PB202 had been replaced by thesigB region of PB199. The presence of the rsbVAM mutationin the chromosome was confirmed by DNA sequence anal-ysis of PCR-amplified chromosomal products in six of sixchloramphenicol-sensitive colonies tested. One of thesestrains, PB205, was transformed with linearized pDH32ctc-lacZ to yield PB206.

PB217, carrying the double mutation rsbVAI rsbX::ery,was constructed by transforming PB206 (rsbVAJ pDH32ctc-lacZ) with linearized pSK15, which contains the insertion-deletion mutation rsbX::ery (27). Transformants were se-lected for erythromycin-lincomycin resistance and screenedby PCR to confirm the presence of both the rsbVAI andrsbX::ery mutations. The single rsbX::ery mutation was alsomoved into the PB198 (pDH32ctc-lacZ) background bytransforming PB198 with chromosomal DNA from PB217.Replacement of chromosomal rsbV and rsbW alleles by

linked transformation with Pspac. As shown in Fig. 1, weplaced the chromosomal copy of the sigB operon undercontrol of the inducible Pspac promoter by using pAR7, a

derivative of plasmid pAG58 (26). pAR7 and similar con-structions also allowed us to use linked transformationrather than congression to replace the chromosomal copiesof rsbV and rsbW with their mutant alleles. To begin theconstruction of pAR7, we first made pAR4, in which the1,443-bp TaqI fragment encoding the macrolide resistancegene from pE194 (21) was substituted for the 1,587-bpStuI-AvaI fragment containing the chloramphenicol resis-tance gene in pAG58; restriction sites generating protrudingends were blunted before ligation.The uB-dependent promoter of the sigB operon has an

SpeI site at the exact position of transcription initiation (27).We took advantage of this site to flank the Pspac promoter ofpAR4 with sigB operon fragments from up- and downstreamof the transcription initiation region. The upstream fragmentcame from pSK21, a pUC19 derivative in which the 2.3-kbEcoRI fragment containing the 5' half of the sigB operon (27)was subcloned into the EcoRI site so that the polylinkerregion lay upstream from the operon. A 1,129-bp SmaI-SpeIfragment was removed from pSK21, the SpeI end was filledin, and the fragment was ligated into the flush SphI site ofpAR4 to create pAR5. This SmaI-SpeI fragment containedpart of the pSK21 polylinker region, the 3' end of anupstream open reading frame, orfU (49), and the sigB operonpromoter. A 920-bp SpeI-EcoRI fragment from pSTR2-beginning at the +1 transcription initiation site of the sigBoperon, containing the rsbWAI deletion, and extending wellinto the sigB coding region (Fig. 1)-was blunt-end ligatedinto the filled Sall site of pAR5 to make pAR7.We linearized pAR7 at the KpnI site introduced via the

pSK21 polylinker and transformed strain PB198 (containingthe pDH32ctc-lacZ fusion). As shown in Fig. 1, selection forerythromycin-lincomycin resistance forced the linearizedPspac construction into the chromosome via a double cross-over event. Because the SpeI-EcoRI fragment lacks the sigBoperon promoter, this construction placed the sigB operonunder the control of Pspacs We used PCR to choose twotransformants for further study. In strain PB212, the secondcrossover occurred between Pspac and rsbWAl, regeneratingan operon wild type for all four gene products but underPspac control. In strain PB213, the second crossover oc-curred promoter-distal to rsbW.U, transferring the deletionto the chromosome. A similar strategy was used for the twoother mutant replacement transformations, substituting theSpeI-EcoRI region from the following plasmids for thecorresponding region of pAR7: pAR6 (rsbV/l), to makePB214; and pAR8 [A(rsbV-rsbW)], to make PB219. pAR6and pAR7 were also transformed into strain PB114, placingthe sigB-lacZ translational fusion at the sigB locus underPspac control. In strain PB221, the second crossover oc-curred between Pspac and rsbWAl, to regenerate wild-typersbV and rsbW preceding the sigB-lacZ translational fusion,and in PB222, the promoter-distal second crossover trans-ferred rsbV.I into the chromosome of the fusion, replacingthe wild-type rsbV. We confirmed the expected structure ofeach construction and mutation by PCR amplification of thesigB chromosomal locus.

Overexpression of the rsbW product in B. subtilis. We madepDG148-rsbW to place rsbW under control of the induciblePspac promoter of the multicopy expression vector pDG148(44). The 676-bp SpeI-HincII fragment (nt 166 to 1099)containing the rsbVAl mutation was subcloned from pUC19into the XbaI and SmaI sites of pMTL23 (10). The clonedfragment was cut at the flanking SalI-SphI sites and moveddownstream of the spoIlIG gene between the Sall and SphIsites of pDG298 (47) to obtain pDG67. We then deleted the

J. BACTERIOL.

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

REGULATION OF BACILLUS SUBTILIS cB 3699

spoIIIG coding region from pDG67 by cutting with HindlIl.The resulting plasmid, pDG148-rsbW, retained the rsbVribosome-binding site together with a large, in-frame dele-tion of rsbV, in order to support rsbW expression if theframes are translationally coupled, as suggested by the DNAsequence (27).Enzyme assays. B. subtilis strains were grown to the late

logarithmic stage and diluted 1:25 into fresh medium. Sam-ples taken in both logarithmic-growth and stationary phaseswere assayed for P-galactosidase assay by the method ofMiller (34), using sodium dodecyl sulfate and chloroform topermeabilize the cells. Activity was expressed in Millerunits, defined as 103 AA420 per minute per milliliter per unitof optical density at 600 nm.

RESULTS

Experimental design. Our genetic analysis of oC regulationwas complicated by three features of sigB operon organiza-tion. First, because cB is required for transcription of its ownstructural gene, it is ordinarily difficult to determine whetherregulatory mutations affecting aB-dependent gene expres-sion act at the level of (uB synthesis or o- activity (27). Wetherefore carried out our analysis in two different geneticbackgrounds, one containing the wild-type, cB-dependentsigB operon promoter and another in which the autocatalyticregulatory circuit was broken by replacing the wild-typepromoter with the inducible Pspac promoter (19, 26).

Second, an operon just upstream from the sigB operonmight influence a&-dependent gene expression, either byreadthrough transcription or by the potential trans-actingfactors it might encode (49). It was therefore important tocompare the effect of selected mutations in the wild-typecontext with the effect of the same mutations in the Pspacconstruction, in which readthrough transcription was pre-vented. The Pspac promoter construction was also designedto maintain the integrity of the coding regions in both thesigB and upstream operons.

Third, expression of the coding regions in the sigB oper-on-the rsbV and rsbW regions in particular-appears to betranslationally coupled (27). We therefore studied the effectsof both missense and null mutations in rsbV and rsbW, andthe null mutations we constructed were large, in-framedeletions which should have little effect on translationalexpression of downstream genes.We generally used the well-characterized crB-dependent

ctc promoter (36) to assay the effect of the various mutationson oB activity in vivo. Promoter activity was measured byusing a transcriptional fusion of the ctc promoter and a lacZreporter gene in the single-copy vector pDH32, integrated atthe amyE locus. Although the function of ctc remainsunknown (22), the interactions of cB-containing holoenzymewith the ctc promoter are well defined, and some of themetabolic signals important for ctc expression are known.ctc is highly expressed early in stationary phase when cellsare grown in rich medium containing high levels of glucoseand glutamine, conditions which repress both the synthesisof tricarboxylic acid cycle enzymes and the sporulationprocess (24). Expression of a ctc-lacZ fusion is abolished ina sigB null mutant (23), and DNA methylation protectionexperiments have defined (rB holoenzyme contacts withinthe ctc promoter (35). Mutational analysis has found thatmany of the bases identified by these contacts are importantfor transcription initiation in vitro and in vivo (39, 48). Insome experiments, we also used transcription fusions of thecsbA and sigB operon promoters (8, 27) to establish that the

tested mutations globally affected cB-dependent geneexpression and were not restricted to ctc. A sigma-H-dependent spoVG-lacZ fusion (9, 14, 51) was used to deter-mine that the effects observed were specific to aB-dependentgenes.

Effect of rsbV and rsbW mutations on CFB-dependent geneexpression. The rsbV and rsbW products have significantsequence identity with their counterparts in the spoIL4operon, SpoIIAA and SpoIIAB, and residues important forSpoIIAA and SpoTIAB function are conserved in the RsbVand RsbW proteins (27). The spoIL4A42 mutation is analteration of Gly-95 to Asp, which decreases sporulationfrequency by 103 (16, 50). We made the correspondingGly-93 to Asp change in rsbV(rsbV42) to determine whetherthis residue was also important for rsbV function (Fig. 2A).Similarly, the spoIL4B1 mutation is a change of Ala-11 toVal, which increases expression of forespore-specific geneswhen cells are grown in medium that does not supportsporulation (38). We therefore made the correspondingAla-14 to Val change in rsbW (rsbWl; Fig. 2B).

In order to test the possible regulatory role of the rsbV andrsbW products, we replaced their wild-type chromosomalalleles with the rsbV42 and rsbWl missense alleles bycongression, as described in Materials and Methods. Wealso introduced the rsbVAI mutation, an in-frame deletionwhich removed 86 of the 109 codons from the rsbV codingregion (Fig. 2A). In all three mutant strains, the presence ofeach mutation in the chromosome was confirmed by se-quencing the amplified, subcloned PCR product of the re-gion. The operon remained otherwise wild type and alsoremained under control of the wild-type sigB promoter. Incontrast to our success with the rsbV42, rsbV/1, and rsbWlmutant alleles, we were unable to substitute the rsbWAlin-frame deletion for the rsbWwild-type allele. Rather, whenusing the ctc-lacZ fusion strain PB198 as the transformationrecipient for the rsbWAl allele, we noted the appearance, atcongression frequency, of pinpoint colonies of lysing cells.These lysing colonies were intensely blue on X-Gal platescompared with the medium blue of the PB198 parent. Weinfer from these results that the rsbWAl mutation is lethal inthe context of the wild-type sigB operon promoter. We alsoinfer that the rsbWAJ mutation causes increased expressionof the aB-dependent ctc fusion. These inferences were laterconfirmed in experiments which introduced the rsbWAIallele by linked transformation with the Pspac promoter,reported below.We measured the effects of the rsbV42, rsbWl, and

rsbVAl mutations on cB activity in the wild-type promotercontext by monitoring expression of the ctc-lacZ transcrip-tional fusion in two different media. In unsupplemented LB,ctc transcription is relatively low, and in LB supplementedwith glucose and glutamine, ctc transcription is normallyinduced fivefold (24). As shown in Fig. 3, results with thersbVAI allele indicate that the rsbV product is a positivefactor required for ctc expression in both media. In separateexperiments, we also found that rsbVAl abolished CrB_dependent expression of the csbA and sigB transcriptionalfusions (not shown). Thus, the rsbV product is generallyrequired for CrB-dependent gene expression.The effect of the rsbV42 missense allele was both less

severe than that of rsbV.IJ and more dependent on thegrowth medium. ctc expression in the rsbV42 mutant wassignificantly lower than in the wild type in LB medium (Fig.3A), whereas in LB containing glucose and glutamine, thedifference between mutant and wild-type activity was lesspronounced (Fig. 3B). These results indicate that the con-

VOL. 174, 1992

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3700 BOYLAN ET AL.

0)

-1 0 1 2 3 4 5 6

Hours in Stationary Phase

b-1 0O 1 2 3 4 5 6

Hours in Staffonary Phase

FIG. 3. Effect of rsbV and rsbW mutations on ctc-lacZ expression. eI3 activity was measured by monitoring 1-galactosidase productionfrom the single-copy pDH32-ctc-lacZ transcriptional fusion. Cultures were grown in LB (A) and in LB supplemented with 5% glucose and0.2% glutamine (B). Samples were removed at various times during the logarithmic and stationary phases to assay for 3-galactosidase activity;the activity of the PB2 negative control was subtracted for each time point. The strains used were PB198 (rsbV' rsbW ) (0); PB201 (rsbWl)(S); PB204 (rsbV42) (0); and PB206 (rsbVA1) (A).

served Gly-93 residue is important for RsbV function, espe-cially under noninducing growth conditions.The effect of the rsbWl missense allele was also depen-

dent on the growth medium. In LB medium, ctc expressionin the rsbWl mutant was elevated three- to fivefold com-pared with that in the wild type (Fig. 3A). But the effect ofrsbWl was less obvious when cells were grown in LBcontaining glucose and glutamine (Fig. 3B). Consistent withour results with the rsbWl allele, Rather et al. (38) found thatthe equivalent spoILUBI mutation affected forespore-specificgene expression only when cells were grown under nonin-ducing conditions for their system. We conclude that theconserved Ala-14 residue is important for RsbW functionprimarily under noninducing growth conditions.

Increased levels of rsbW inhibit e3-dependent gene expres-sion. The effect of the rsbWl missense allele on ctc expres-sion suggests that the rsbWproduct is a negative regulator ofe activity. However, it is possible that rsbWl causes analteration rather than a loss of RsbW function. IfRsbW werein fact a negative effector, we would predict that overpro-duction of the wild-type RsbW protein would inhibit expres-sion of &r-dependent genes in trans.

This prediction was verified in the experiment shown inFig. 4, for which we constructed a multicopy plasmid thatplaced rsbW expression under control of the inducible Pspacpromoter. In this construction, we maintained the transla-tional coupling of rsbW to the upstream rsbV, which wasrendered null by the rsbVAl in-frame deletion (see Materialsand Methods). We introduced this plasmid into strain PB198,which bore the ctc-lacZ transcriptional fusion, creatingstrain PB207. As shown in Fig. 4, when PB207 cells weregrown without the IPTG (isopropyl-13-D-thiogalactopyrano-side) inducer of the Pspac promoter, ctc was expressed asusual in the early stationary phase of growth. However,induction of rsbW completely abolished detectable ctcexpression, both in LB medium supplemented with glucoseand glutamine (Fig. 4) and in unsupplemented LB (notshown). rsbW induction had an equivalent negative effect onthe cr'-dependent csbA and sigB transcriptional fusions buthad no influence on expression of a sigma H-dependentspoVG-lacZ fusion (not shown). Thus, an rsbW missensemutation and an rsbW overexpression system fulfill thepredicted properties for a negative regulatory element whichspecifically affects e-dependent gene expression. Because

is responsible for transcription of its own structural gene,

the RsbW product could exert its negative regulatory effecton either the synthesis or the activity of er.

e" is activated by stationary-phase signals and factors. Wereplaced the wild-type sigB operon promoter with the induc-ible Pspac promoter for two reasons. First, this constructionallowed us to break the autocatalytic transcriptional loopand analyze the level at which the rsbV and rsbW productsregulated aE-dependent genes. Second, this constructionallowed the introduction of potentially lethal mutations, suchas rsbWAl, into the operon. As shown in Fig. 1, weessentially substituted the E. coli lac repressor-operator

300

200-

C

100-

-1 0 1 2 3 4 5

Hours in Stationary Phase

FIG. 4. Effect of increased RsbW levels on ctc expression.Strain PB207 contains the multicopy plasmid pDG148-rsbW, whichplaces rsbW expression from the plasmid under control of theIPTG-inducible Pspac promoter. PB207 was grown until mid-logphase in LB medium supplemented with 5% glucose and 0.2%glutamine. The culture was then diluted 1:25 into fresh medium andsplit in half; one half received no IPTG (0), and the other receivedIPTG at 1 mM (final concentration) (U). 13-Galactosidase activityfrom the resident pDH32-ctc-lacZ transcriptional fusion was mea-sured at various times after the addition of IPTG. As a negativecontrol, we followed the same growth and assay procedure withstrain PB208, which contained the pDH32-ctc-lacZ transcriptionalfusion and the parental plasmid pDG148, which carries no B. subtilisgenes. 1-Galactosidase activities measured in PB208 grown with orwithout IPTG were similar to those measured in PB207 grownwithout IPTG (data not shown).

200

C

0.S

.

J. BACTERIOL.

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

REGULATION OF BACILLUS SUBTILIS &B 3701

50 ~~~~~c

0--1 0 1 2 3 4

Hours in Stationary Phase

FIG. 5. (B activity with sigB operon expression under Pspaccontrol. Plasmid pAR7 was integrated into the sigB operon locus bya double crossover event, replacing the &3-dependent sigB operonpromoter with the IPTG-inducible Pspac promoter (see text). Theresultant strain, PB212 (Pspac rsbV+ rsbW+ amyE::pDH32-ctc) wasgrown in triplicate in LB supplemented with 5% glucose and 0.2%glutamine. sigB operon transcription was induced by adding IPTG (1mM, final concentration) to the cultures at -0.7 h (arrow A), -0.2h (arrow B), or 1.25 h (arrow C); parallel control cultures weretreated identically but were not induced with IPTG. The ,-galac-tosidase activity shown for each culture is the difference betweenthe activity in the induced culture and that in the uninduced control:A (0), B (0), and C (U).

system for the sigB operon promoter, placing the Pspacpromoter under lacI control. In this construction, the re-gions immediately up- and downstream from the sigB operonpromoter flanked the lac control cassette of a pAG58 deriv-ative (see Materials and Methods). When this constructionwas linearized and integrated into the B. subtilis sigB regionby a double crossover event, the integrity of the sigBpromoter was destroyed and the sigB operon was placedunder Pspac control. The orfU reading frame encoded by theregion upstream of the sigB operon (49) remained intact, asdid the downstream rsbV, rsbW, sigB, and rsbX frames. Ifthe downstream region of the linearized construction bore amutant rsbV or rsbW allele, this allele was introduced intothe sigB operon together with Pspac when the second cross-over occurred downstream from the mutation. PCR experi-ments verified the presence of either the wild-type alleles orthe rsbVA1, rsbWMl, or l(rsbV-rsbW) mutation in the sigBoperon under Pspac control.With the wild-type sigB operon under Pspac control, the

timing of induction was critical for oB activity (Fig. 5). Whenthe operon was induced by adding IPTG to logarithmicallygrowing cells, ctc expression was entirely lacking through-out the growth cycle, even though this experiment wasconducted under growth conditions that maximize ctcexpression. In contrast, when the operon was induced at theend of logarithmic growth, ctc was expressed during earlystationary phase in a manner similar to that observed whenthe sigB operon was under control of its wild-type promoter(see Fig. 3). Induction of the sigB operon 1 h into stationaryphase led to reduced expression of ctc, probably due todecreased permeability of the cells to IPTG (not shown).We conclude from these results that orB activity requires

signals or factors present in early-stationary-phase cells andthat induction of the operon at an inappropriate growth stagecannot overcome the absence of these signals. Indeed, whenthe operon was induced during logarithmic growth, the lackof any o-B activity, even in stationary phase, was particularly

30

a,_)

2

20

10

10

10

10

0 aa --

0 1 2 3Hours After Addition of IPTG

FIG. 6. Effect of rsbV, rsbW, and rsbX mutations on crB activityduring logarithmic growth. Cultures were grown to the late logarith-mic growth phase in unsupplemented LB. Cultures were then splitand diluted 1:25 into fresh medium, and IPTG (1 mM, final concen-tration) was added to one half. Samples were removed from eachculture at the indicated time after IPTG addition and assayed forf8-galactosidase activity. For each point shown, the activity of theuninduced culture was subtracted from the activity of the corre-sponding induced culture. Symbols: El, PB212 (Pspac rsbV+ rsbW+amyE::pDH32-ctc); A, PB213 (Pspac rsbV+ rsbWAI amyE::pDH32-ctc); A, PB214 (Pspac rsbVAM rsbW+ amyE::pDH32-ctc); *, PB216(Pspac rsbV+ rsbW+ socB amyE::pDH32-ctc). socB is a frameshiftmutation early in the rsbX reading frame (23).

striking. One explanation of this result is that at least one ofthe sigB operon products inhibits uB activity during logarith-mic growth and that the stationary-phase signals whichnormally activate o(B cannot overcome early expression ofthis product.To determine whether any of the sigB operon products

was responsible for inhibiting UB activity during logarithmicgrowth, we measured ctc expression in various sigB operonmutants in which the operon was induced under Pspaccontrol. This experiment was conducted under growth con-ditions in which ctc is not optimally induced in order toprovide a stringent test of negative regulator function. Asshown in Fig. 6, strains carrying null rsbV and rsbX muta-tions behaved essentially the same as the wild type, and themutations did not relieve the block to uB activity in logarith-mically growing cells. However, the rsbWIsJ mutation diddramatically reverse the inhibition of activity for at least 2 hafter operon induction, at which point the cells stoppedgrowing. We conclude that one role of the rsbW product is toprevent aB activity at inappropriate times and that the loss ofrsbW is deleterious in logarithmically growing cells.We also showed that the rsbW product has a comparable

negative role in stationary-phase cells. In contrast to theinduction of the rsbWz\l mutant in logarithmically growingcells, induction of the Pspac construction at the end oflogarithmic growth had no obvious deleterious effect on cellgrowth or viability. The experiment in Fig. 7 showed that inLB medium, the rsbWAl mutation strongly increased 0Bactivity measured at the ctc promoter. The rsbWAI mutationhad a similar but less potent effect on ctc expression whencells were grown in LB containing glucose and glutamine. Incontrast, the phenotype of the rsbVAI mutation under Pspaccontrol established that the rsbV product is an essentialpositive effector of aB activity in both media. Notably, theresults shown in Fig. 7 were consistent with those in Fig. 3.Other than a change in the operon promoter, the principaldifference between these experiments was that the rsbWlmissense mutation was tested in one (Fig. 3) and the rsbWAJ

VOL. 174, 1992

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3702 BOYLAN ET AL.

.s

200

c

a!i

-1 0 1 2 3 4 5 6 0 1 2 3 4 5

Hours in Stationary Phase Hours in Stabonary Phase

FIG. 7. e activity in stationary phase with the sigB operon under P.pac control. IPTG (1 mM, final concentration) was added to one oftwo parallel cultures at the end of logarithmic growth (TO). Samples were removed from each culture at the indicated times and assayed for3-galactosidase activity of the resident pDH32-ctc-lacZ transcriptional fusion. For each strain, 3-galactosidase activities from the uninduced

culture were subtracted from the activities of the corresponding induced culture. Symbols: El, PB212 (Pspac rsbV' rsbW' amyE::pDH32-ctc);A, PB213 (Ps rsbV+ rsbWAI amyE::pDH32-ctc); A, PB214 (P8p5c rsbVAI rsbW' amyE::pDH32-ctc). (A) Cultures grown in unsupple-mented LB; (S) Cultures grown in LB supplemented with 5% glucose and 0.2% glutamine.

mutation was tested in the other (Fig. 7). Together, theresults of the experiments shown in Fig. 3 through 7 are inaccord with the view that the rsbW product inhibits theactivity of &rB in both logarithmic- and stationary-phase cells.Because different promoters were used to express the sigBoperon in the experiments shown in Fig. 3 and 7, we can ruleout the possibility that RsbV and RsbW act primarily bycontrolling the rate of transcriptional initiation from the sigBoperon promoter.

Relationship of rsbV, rsbW, and rsbX null mutations. Wedid tests of genetic epistasis to determine the relationshipsamong the two regulatory gene products described here,RsbV and RsbW, and the rsbX gene product describedearlier (15, 23, 27). Epistasis is the masking of the phenotypeof one allele by the phenotype of another, and such testsallow inferences to be made about the order of action ofregulatory elements from phenotypic comparison of a doublemutation with the single mutations used in its construction.As shown in Fig. 8A, we used strains bearing the wild-type

sigB promoter to test the epistasis between rsbVAl andrsbX::ery, null regulatory mutations which had clear andopposite effects on the expression of e'-dependent geneexpression. This experiment (and the ones shown in Fig. 8Band Fig. 9) were carried out in unsupplemented LB medium,

600

D 400

200-

-1 0 1 2 3 4

a growth condition in which ctc ordinarily is not highlyexpressed. This condition gave a more pronounced differ-ence in the effects of the single mutations, and thus thephenotype of a double mutation would be more readilyapparent. eB activity in the rsbVAI rsbX::ery double mutantwas identical to that in the rsbVAl single mutant, indicatingthat rsbVAI was epistatic to rsbX::ery (Fig. 8A). The rela-tionship of rsbX::ery and rsbVAl determined from the activ-ity assays shown in Fig. 8A was the same as that deducedfrom the mutant growth phenotypes on LB plates. ThersbX::ery mutation causes a small-colony phenotype (27),and colony size was restored to wild type in the rsbVA1rsbX::ery double mutant. These epistasis results imply thatthe rsbX product acts before the rsbVproduct in an orderedpathway regulating eB activity or that loss of the rsbVproduct negates any control exerted by the rsbX product.As shown in Fig. 8B, we used the Ps ac constructions to

test the epistasis between the rsbVAJ and rsbWAlI nullalleles, which also had opposite regulatory effects. The eactivity of the A(rsbV-rsbW) double mutant was very similarto that of the rsbWAl single mutant, indicating that rsbWAIwas epistatic to rsbVAL. The relationship of rsbV4l andrsbWA&l determined from the activity assays agreed with thatinferred from the mutant growth phenotypes on LB plates.

0)

Hours in Staionary Phase Hours in Stationary Phase

FIG. 8. Epistatic relationships among regulatory mutations in the sigB operon. (A) P-Galactosidase activities from strains PB198 (rsbVtrsbX+ amyE::pDH32-ctc; O), PB206 (rsbVAlI rsbX' amyE::pDH32-ctc; A), PB217 (rsbVA1 rsbX::ery amyE::pDH32-ctc; 0), and PB218(rsbVl rsbX::ery amyE::pDH32-ctc; A). Strains PB198 and PB206 were grown in LB, and strains PB217 and PB218 were grown in LBcontaining erythromycin (0.5 p,g/ml) and lincomycin (12.5 p,g/ml) to retain the rsbX::ery marker. (B) P-Galactosidase activities from strainsPB213 (Pspac rsbV' rsbWAI amyE::pDH32-ctc; A) and PB214 (Pspac rsbVAI rsbW' amyE::pDH32-ctc; A) are from Fig. 7A. Strain PB219(P5p,c8 A[rsbV-rsbW] amyE::pDH32-ctc; *) was grown in LB and assayed as described in the legend to Fig. 7.

A

n -_v

J. BACTERIOL.

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

REGULATION OF BACILLUS SUBTILIS o&B 3703

Under conditions in which was Pspac was uninduced, thersbVAl, rsbWA] , and A(rsbV-rsbW) mutants had no obviousgrowth defect. However, when Pspac was induced, both thersbWAl mutant and the A(rsbV-rsbW) double mutant lysed(not shown), while the rsbVAl mutant grew normally. Theseepistasis results suggest that the rsbV product acts in aregulatory pathway before the rsbW product or, alterna-tively, that loss of the rsbW product overrides any controlimposed through rsbV.To further characterize the manner in which RsbW and

RsbX inhibited (rB activity, we found that the increase inactivity caused by the simultaneous loss of rsbW and rsbXfunction was additive. If RsbX functions as a negativeregulator in a pathway containing RsbW as the final negativeeffector of a& activity, then loss of rsbX function in an rsbWmutant should cause no additional increase in crB activity.However, if RsbX regulates orB activity by a mechanismindependent of RsbW, then the increase in e activitycaused by loss of both RsbX and RsbW should be additive.We therefore attempted to construct a strain carrying thersbX: :ery null mutation together with the rsbWl missensemutation by transforming strain PB201 (rsbWl amyE::pDH32-ctc) with plasmid pSK15, bearing the rsbX: :eryinsertion-deletion mutation. On plates selective for thersbX: :ery marker, we found only intensely blue, pinpointcolonies that proved inviable on restreaking. These resultssuggest that the increases in cu' activity (and decreases incell viability) caused by the loss of RsbX and RsbW functionare additive. However, this result-implying independentaction-must be interpreted with caution, because thersbWl missense allele was not a total-loss-of-function muta-tion. Our preliminary results with an rsbWAl rsbX::erydouble null mutation under Pspac control confirm the additiveeffects of the two alleles, suggesting that the rsbW and rsbXproducts act independently (6).rsbV and rsbW products regulate crB posttranslationally.

The rsbV and rsbW products could plausibly regulate orBdependent gene expression by modulating either aB synthe-sis or e' activity. Our first indication of the level at whichRsbV and RsbW act comes from the Pspac studies. The rsbVand rsbW mutations had a similar effect on e3-dependentgene expression whether the sigB operon was under controlof the wild-type sigB promoter or Pspac (Fig. 3 and 7). Thus,RsbV and RsbW cannot act primarily through the regionreplaced in the Pspac construction, that is, through the sigBpromoter sequences upstream from the Spel site, the site oftranscription initiation with the wild-type promoter (27).We next tested whether the rsbV product regulated trans-

lation of the sigB message. We indirectly measured the effectof the rsbVAl allele on cB protein levels by using a sigB-lacZtranslation fusion. This fusion was constructed by using thepJF751 vector, which upon integration into the B. subtilischromosome both created the sigB-lacZ fusion and simulta-neously abolished sigB and rsbX function (27). Thus, thePB114 strain carrying this fusion is a sigB rsbX double nullmutant. We used the Ps constructions both to place thesigB-lacZ fusion under the control of a functional promoterand to move the rsbVAI mutation into the fusion back-ground. Because the epistasis results indicated that thenature of the rsbX allele had no effect on the phenotype of anrsbV null mutation (Fig. 8A), for the resolution of thisexperiment we could disregard the absence of a functionalrsbX product in the fusion strain.As shown in Fig. 9, there was no significant difference in

expression of the sigB-lacZ fusion in the rsbVAI mutantcompared with the strain that carried the wild-type rsbV

100-

00 1 2 3 4 5

Hours in Stationary Phase

FIG. 9. Effect of rsbVAI mutation on expression of a sigB-lacZtranslational fusion under Pspac control. 1-Galactosidase activities ofa sigB-lacZ translational fusion (pMD10) integrated at the sigB locus(27), with the aB-dependent promoter of the sigB operon replacedwith the inducible Pspac promoter. For each strain, parallel cultureswere grown in LB, and IPTG (1 mM, final concentration) was addedto one of the cultures at the end of logarithmic growth. ,-Galactosi-dase activities from the uninduced control were subtracted from theactivities of the corresponding induced culture. Symbols: El, PB221(PSpac rsbV+ rsbW+ sigB::pMD10); A, PB222 (Pspac rsbVA1 rsbW+sigB::pMD10).

allele. These results strongly suggest that the rsbV productacts on aB posttranslationally and thus works at the level of&` activity. Based on the epistatic relationship between rsbVand rsbW (Fig. 8B), we can also conclude that the rsbWproduct acts posttranslationally.

DISCUSSION

The genetic results reported here establish that all fourgenes in the B. subtilis sigB operon have important regula-tory roles in exponential- and stationary-phase cells. Ourexperiments focused primarily on the roles of the first twogenes of the operon, rsbV and rsbW, and their relationship tothe last two genes, sigB and rsbX, which were previouslyshown to autoregulate operon expression (27). With regardto rsbV and rsbW function, our principal results are that (i)overexpression of the rsbW product blocks CrB-dependentgene expression, whereas an rsbW null mutation increasesexpression; (ii) an rsbVnull mutation prevents cuB-dependentgene expression; and (iii) an rsbWnull mutation reverses theblock in oB-dependent gene expression caused by the loss ofrsbV. From these genetic results, we conclude that RsbW isan inhibitor of oB-dependent gene expression and that RsbVis a positive effector which is required for or' activity only ifRsbW is present. In addition, these results suggest that thefunction of RsbV is to counter the action of RsbW.With regard to the regulatory relationships among the

other products of the sigB operon, we first confirmed thefinding of Igo et al. (23) that an rsbX null mutation increasedthe expression of &3-dependent genes and then determinedthat an rsbV null mutation reversed the effect of the rsbXmutation. Thus RsbX, RsbV, and RsbW form a sequential orhierarchical regulatory system which controls aB-dependentgene expression. Although our preliminary results suggestthat RsbX and RsbW act separately, additional geneticanalysis is needed to firmly establish whether the rsbX,rsbV, and rsbW products define independent regulatory

VOL. 174, 1992

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3704 BOYLAN ET AL.

pathways or a single, dependent pathway (see reference 40for an analysis which distinguishes dependent and indepen-dent pathways). In the absence of such analysis, we canconstruct two different types of models which are consistentwith the available data. In one model, the three productsform a dependent negative regulatory cascade, with RsbWinhibiting &rB activity, RsbV inhibiting the action of RsbW,and RsbX inhibiting RsbV:

RsbX I RsbV RsbW - sigma B

Alternatively, in the other model, the three products actsequentially yet independently to influence the activation ofe, with the absence of the downstream product overridingthe need for the upstream product. In this model, theupstream product counteracts but does not need to exactlyreverse the action of the downstream product:

RsbX RsbV RsbW

sigma B (inactive)I

* sigma B (active)

Neither of these formal genetic models implies the molecularmechanism by which RsbX, RsbV, and RsbW act to controlrB-dependent gene expression, and there is no biochemicalevidence with respect to possible direct interactions of thefour proteins in the system. We are therefore mindful of thepossibility that RsbX, RsbV, and RsbW act indirectly oneand that the eB system may include additional regulatorycomponents.Our results further show that RsbV and RsbW act post-

translationally, that RsbW is responsible for blocking eB-dependent gene expression during logarithmic growth, andthat loss of this RsbW control is deleterious to the cell. Thedata indicating that RsbV and RsbW affect eB activity ratherthan synthesis comes from experiments in which a transla-tional fusion was used to indirectly monitor er proteinlevels. Therefore, we cannot rigorously rule out some formsof control, such as retroregulation (33), that might influenceo- synthesis through sequences replaced by the lacZ fusion.However, for this discussion we assume the more likelyexplanation that RsbW inhibits eB activity. We can thenimagine two broad classes for the mechanisms of RsbWaction. In the first, the ultimate RsbW target is es or

e'-containing holoenzyme, and RsbW would modulate tar-get activity either by direct protein-protein interaction, bycatalytic addition or removal of a posttranslational modifi-cation, or by controlling the stability of the target protein. Inthe second class, the RsbW targets are e3-dependent pro-moters, at which RsbW would function directly as a repres-sor protein or indirectly by modulating such a protein.Arguing against the promoter-target mechanism are the lackof any obvious DNA-binding motifs in RsbW (27) and thelack of any common sequences which might define cis-actingRsbW binding sites in promoters known to be regulated byer (8, 22, 27). If the final RsbW target is either &B or crholoenzyme, this system-together with the analogous &r

system (42)-would define a new mechanism for controllingsigma factor activity.A significant finding of our current work is that &e and <r'

are regulated by similar molecular mechanisms. This con-clusion is supported by three lines of evidence. First, the

sigB and spoIlA (or) operons have a similar genetic organi-zation, with significant sequence identity between RsbV andSpoIIAA and between RsbW and SpoIIAB, the counterpartproducts of the two operons (27). Second, residues known tobe important for SpoIIAA and SpoIIAB function are bothconserved and important for RsbV and RsbW function (Fig.2 and 3). Third, analysis of the eF system by Schmidt et al.(42) has shown that spoIL4A and spoIL4B mutations haveregulatory consequences and epistatic relationships compa-rable to those of rsbV and rsbWmutations and that SpoIIAAand SpoIIAB likely affect o- activity at the posttranslationallevel.Although both the &B and ao regulons are expressed in the

stationary growth phase, at least some of the genes in eachregulon have different physiological roles and most likelyrespond to different stationary-phase signals. erF activity isessential for successful completion of the sporulation pro-cess, and e'F-directed transcription is thought to be crucial inestablishing gene expression specifically in the foresporecompartment (31). In contrast, e3 is not required for sporu-lation (4, 15, 23, 27), and the genes it controls appear to bemost highly expressed under stationary-phase growth con-ditions, in which sporulation is not favored (6, 8, 24).A clear regulatory difference between the two operons is

the presence ofrsbX as a fourth gene in the sigB operon. Theadditional layer of negative control provided by RsbX maybe required by the positive autoregulation that eB exertsover its own structural gene, a feature it shares with u'l (28).oM1c controls essential sporulation genes expressed in themother cell compartment, and the autocatalytic nature ofsigK expression is thought to allow rapid amplification of orKactivity, facilitating competition with a factors from earlierstages of development. Like a0B, the autocatalytic r1K activ-ity is subject to multiple levels of control, and disabling a keyoK control leads to the production of defective spores byinterfering with the timing of the sporulation process (11, 28,45). In the case of &rB, disabling the rsbX control leads toslow growth and a small-colony phenotype (15, 23, 27), whiledisabling rsbW control is apparently lethal to cells in loga-rithmic growth.The additional complexity of the sigB operon also suggests

at least two possible routes of information entry into thesystem: through RsbX and through the RsbV-RsbW pair.From our preliminary genetic results, we presume that RsbXand RsbW constitute independent regulatory pathways.And, although the absolute dependence of RsbV on RsbWremains to be established by genetic or biochemical criteria,our epistasis experiments and the apparent translationalcoupling of rsbV and rsbW expression suggest that RsbVacts with RsbW. Two lines of experiments indicate that theinferred RsbV-RsbW and RsbX pathways each respond tostationary-phase signals. First, with the wild-type sigB pro-moter, loss of RsbX leads to greatly increased expression ofsigma B-dependent genes in exponential growth (27). Thus,in the wild-type context, RsbX is important for preventingthe appearance of eB activity until the stationary phase.Second, experiments with the P promoter (Fig. 6)showed that RsbW was responsible for preventing eB activ-ity in exponentially growing cells. The inhibitory effects ofRsbW and RsbX could be either potentiated by a metabolicsignal or protein factor present in logarithmic growth orreversed by a signal or factor in stationary phase. RsbW andRsbX continue to affect &B activity in stationary phase,because loss of either factor increases stationary-phaseexpression of or-dependent genes (Fig. 7 and 8). Thus,

J. BACTERIOL.

v I. I

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

REGULATION OF BACILLUS SUBTILIS aB 3705

RsbW and RsbX are important elements in regulating (rBactivity in both logarithmic- and stationary-phase cells.To what signals and factors might the RsbV-RsbW and

RsbX pathways respond? Although most o'-dependentgenes are highly expressed in LB containing glucose andglutamine, the available data do not support the view thatthis response is mediated solely through RsbV, RsbW, orRsbX. A null rsbX mutant retains the glucose and glutamineregulation of ctc expression characteristic of the wild-typestrain, and the null rsbWAl mutant in the Pspac backgroundalso retains the full glucose-glutamine response when partlyinduced (6). Because the rsbV null mutant lacks detectablecrB activity, it is not useful in establishing the signals towhich RsbV might respond. However, we did observeglucose-glutamine induction of ctc expression in the mutantcarrying the rsbV42 missense allele (Fig. 3). If the rsbV42allele is a loss-of-function mutation, as suggested by ourdata, this result argues that RsbV alone cannot mediate theglucose-glutamine response. Thus, additional regulatory el-ements remain to be discovered in the (B system.Our results indicate that RsbW is a key element in a'

regulation. Further analysis will determine whether signalsthat activate orB actually pass through RsbW or whetherthese signals act first on RsbV, which in turn transmits thesignal to RsbW. Determination of the molecular mechanismby which RsbW regulates &- activity may well define a newmeans for controlling transcription in response to environ-mental or cell cycle signals.

ACKNOWLEDGMENTS

We thank Richard Losick, William Haldenwang, and CharlesMoran, Jr., for sharing their unpublished data; Patrick Stragier forhis help in constructing pDG148-rsbW; and Arlene Wise for con-structing the sigB transcriptional fusion in PB226. We particularlythank Alan Grossman and Patrick Stragier for their helpful com-ments on the manuscript.

This research was supported by Public Health Service grantGM42077 from the National Institute of General Medical Sciences.Ann Rutherford was a predoctoral trainee supported in part byPublic Health Service training grant GM07377 from the NationalInstitute of General Medical Sciences, and some costs of herresearch were met by a Jastro-Shields Graduate Research Award.

REFERENCES1. Anagnostopoulos, C., and J. Spizizen. 1961. Requirements for

transformation in Bacillus subtilis. J. Bacteriol. 81:741-746.2. Band, L., H. Shimotsu, and D. J. Henner. 1984. Nucleotide

sequence of the Bacillus subtilis trpE and trpD genes. Gene27:55-65.

3. Benson, A. K., and W. G. Haldenwang. 1992. Characterizationof a regulatory network that controls oa expression in Bacillussubtilis. J. Bacteriol. 174:749-757.

4. Binnie, C., M. Lampe, and R. Losick. 1986. Gene encoding thesigma-37 species of RNA polymerase sigma factor from Bacillussubtilis. Proc. Natl. Acad. Sci. USA 83:5943-5947.

5. Boylan, S. A., K. T. Chun, B. A. Edson, and C. W. Price. 1988.Early-blocked sporulation mutations alter expression of en-zymes under carbon control in Bacillus subtilis. Mol. Gen.Genet. 212:271-280.

6. Boylan, S. A., A. Rutherford, and C. W. Price. Unpublisheddata.

7. Boylan, S. A., J.-W. Suh, S. M. Thomas, and C. W. Price. 1989.Gene encoding the alpha core subunit of Bacillus subtilis RNApolymerase is cotranscribed with the genes for initiation factor1 and ribosomal proteins B, S13, SlL, and L17. J. Bacteriol.171:2553-2562.

8. Boylan, S. A., M. D. Thomas, and C. W. Price. 1991. Geneticmethod to identify Bacillus subtilis regulons controlled bynonessential elements: isolation of a gene dependent on alter-

nate transcription factor CrB. J. Bacteriol. 173:7856-7866.9. Carter, H. L., III, and C. P. Moran, Jr. 1986. New RNA

polymerase a factor under spoO control in Bacillus subtilis.Proc. Natl. Acad. Sci. USA 83:9438-9442.

10. Chambers, S. P., S. E. Prior, D. A. Barstow, and N. P. Minton.1988. The pMTLnic- cloning vectors. I. Improved pUC poly-linker regions to facilitate the use of sonicated DNA for nucle-otide sequencing. Gene 68:139-149.

11. Cutting, S., V. Oke, A. Driks, R. Losick, S. Lu, and L. Kroos.1990. A forespore checkpoint for mother cell gene expressionduring development in B. subtilis. Cell 62:239-250.

12. Davis, R. W., D. Botstein, and J. R. Roth. 1980. Advancedbacterial genetics: a manual for genetic engineering. Cold SpringHarbor Laboratory, Cold Spring Harbor, N.Y.

13. Dubnau, D., and R. Davidoff-Abelson. 1971. Fate of transform-ing DNA following uptake by competent Bacillus subtilis. J.Mol. Biol. 56:209-221.

14. Dubnau, E., J. Weir, G. Nair, L. Carter III, C. Moran, Jr., andI. Smith. 1988. Bacillus subtilis sporulation gene spoOH codesfor a30 (C.H). J. Bacteriol. 170:1054-1062.

15. Duncan, M. L., S. S. Kalman, S. M. Thomas, and C. W. Price.1987. Gene encoding the 37,000-dalton minor sigma factor ofBacillus subtilis RNA polymerase: isolation, nucleotide se-quence, chromosomal locus, and cryptic function. J. Bacteriol.169:771-778.

16. Errington, J., and J. Mandelstam. 1983. Variety of sporulationphenotypes resulting from mutations in a single regulatorylocus, spoIIA, in Bacillus subtilis. J. Gen. Microbiol. 129:2091-2101.

17. Haldenwang, W. G., and R. Losick. 1980. A novel RNA poly-merase sigma factor from Bacillus subtilis. Proc. Natl. Acad.Sci. USA 77:7000-7005.

18. Helmann, J. D., and M. J. Chamberlin. 1988. Structure andfunction of bacterial sigma factors. Annu. Rev. Biochem. 57:839-872.

19. Henner, D. J. 1990. Inducible expression of regulatory genes inBacillus subtilis. Methods Enzymol. 185:223-338.

20. Henner, D. J. Personal communication.21. Horinouchi, S., and B. Weisblum. 1982. Nucleotide sequence

and functional map of pE194, a plasmid that specifies inducibleresistance to macrolide, lincosamide, and streptogramin type Bantibiotics. J. Bacteriol. 150:804-814.

22. Igo, M., M. Lampe, and R. Losick. 1988. Structure and regula-tion of a Bacillus subtilis gene that is transcribed by the EaBform of RNA polymerase holoenzyme, p. 151-156. In A. T.Ganesan and J. A. Hoch (ed.), Genetics and biotechnology ofthe bacilli, vol. 2. Academic Press, Inc., New York.

23. Igo, M., M. Lampe, C. Ray, W. Shafer, C. P. Moran, Jr., and R.Losick. 1987. Genetic studies of a secondary RNA polymerasesigma factor in Bacillus subtilis. J. Bacteriol. 169:3464-3469.

24. Igo, M., and R. Losick. 1986. Regulation of a promoter that isutilized by minor forms of RNA polymerase holoenzyme inBacillus subtilis. J. Mol. Biol. 191:615-624.

25. Innes, M. A., D. H. Gelfand, J. J. Sninsky, and T. J. White.1990. PCR protocols: a guide to methods and applications.Academic Press, Inc., New York.

26. Jaacks, K. J., J. Healy, R. Losick, and A. D. Grossman. 1989.Identification and characterization of genes controlled by thesporulation regulatory gene spoOH in Bacillus subtilis. J. Bac-teriol. 171:4121-4129.

27. Kalman, S., M. L. Duncan, S. M. Thomas, and C. W. Price.1990. Similar organization of the sigB and spoIIA operonsencoding alternate sigma factors of Bacillus subtilis RNA poly-merase. J. Bacteriol. 172:5575-5585.

28. Kroos, L., B. Kunkel, and R. Losick. 1989. Switch protein altersspecificity of RNA polymerase containing a compartment-spe-cific sigma factor. Science 243:526-529.

29. Kunkel, T. A., J. D. Roberts, and R. A. Zakour. 1987. Rapid andefficient site-specific mutagenesis without phenotypic selection.Methods Enzymol. 154:367-382.

30. Losick, R., P. Youngman, and P. J. Piggot. 1986. Genetics ofendospore formation in Bacillus subtilis. Annu. Rev. Genet.20:625-670.

VOL. 174, 1992

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3706 BOYLAN ET AL.

31. Margolis, P., A. Driks, and R. Losick 1991. Establishment ofcell type by compartmentalized activation of a transcriptionfactor. Science 254:562-565.

32. Marquez, L. M., J. D. Helmann, E. Ferrari, H. M. Parker,G. W. Ordal, and M. J. Chamberlin. 1990. Studies of sigma-D-dependent functions in Bacillus subtilis. J. Bacteriol. 172:3435-3443.

33. Mattheakis, L., L. Vu, F. Sor, and M. Nomura. 1989. Retroreg-ulation of the synthesis of ribosomal proteins L14 and L24 byfeedback repressor S8 in Escherichia coli. Proc. Natl. Acad.Sci. USA 86:448452.

34. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

35. Moran, C. P., Jr., W. C. Johnson, and R. Losiclk 1982. Closecontacts between '7-RNA polymerase and a Bacillus subtilischromosome promoter. J. Mol. Biol. 162:709-713.

36. Moran, C. P., Jr., N. Lang, and R. Losiclk 1981. Nucleotidesequence of a Bacillus subtilis promoter recognized by Bacillussubtilis RNA polymerase containing O3. Nucleic Acids Res.9:5979-5990.

37. Price, C. W., and R. H. Doi. 1985. Genetic mapping of rpoDimplicates the major sigma factor of Bacillus subtilis RNApolymerase in sporulation initiation. Mol. Gen. Genet. 210:88-95.

38. Rather, P. N., R. Coppolecchia, H. DeGrazia, and C. P. Moran,Jr. 1990. Negative regulator of cQ-controlled gene expression instationary-phase Bacillus subtilis. J. Bacteriol. 172:709-715.

39. Ray, C., R. E. Hay, H. L. Carter III, and C. P. Moran, Jr. 1985.Mutations that affect utilization of a promoter in stationary-phase Bacillus subtilis. J. Bacteriol. 163:610-614.

40. Russell, P., and P. Nurse. 1987. The mitotic inducer niml+functions in a regulatory network of protein kinase homologuescontrolling the initiation of mitosis. Cell 49:569-576.

41. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.

USA 74:5463-5467.42. Schmidt, R., P. Margolis, L, Duncan, R. Coppolecchia, C. P.

Moran, Jr., and R. LosicLk 1990. Control of developmentaltranscription factor erF by sporulation regulatory proteins Spol-IAA and SpoIIAB in Bacillus subtilis. Proc. NatI. Acad. Sci.USA 87:9221-9225.

43. Sonenshein, A. L. 1989. Metabolic regulation of sporulation andother stationary-phase phenomenon, p. 109-130. In I. Smith,R. A. Slepecky, and P. Setlow (ed.), Regulation of prokaryoticdevelopment. American Society for Microbiology, Washington,D.C.

44. Stragier, P., C. Bonamy, and C. Karmazyn-Campelli. 1988.Processing of a sporulation sigma factor in Bacillus subtilis: howmorphological structure could control gene expression. Cell52:697-704.

45. Stragier, P., B. Kunkel, L. Kroos, and R. Losick. 1989. Chro-mosomal rearrangement generating a composite gene for adevelopmental transcription factor. Science 243:507-512.

46. Stragier, P., and R. Losick. 1990. Cascades of sigma factorsrevisited. Mol. Microbiol. 4:1801-1806.

47. Sun, D., P. Stragier, and P. Setlow. 1989. Identification of a newsigma factor involved in compartmentalized gene expressionduring sporulation of Bacillus subtilis. Genes Dev. 3:141-149.

48. Tatti, K. M., and C. P. Moran, Jr. 1984. Promoter recognitionby sigma-37 RNA polymerase from Bacillus subtilis. J. Mol.Biol. 175:285-297.

49. Wise, A. A., and C. W. Price. Unpublished data.50. Yudkin, M. D., K. A. Jarvis, S. E. Raven, and P. Fort. 1985.

Effects of transition mutations in the regulatory locus spoIL4 onthe incidence of sporulation in Bacillus subtilis. J. Gen. Micro-biol. 131:959-962.

51. Zuber, P., and R. Losick 1983. Use of a lacZ fusion to study therole of the spoO genes of Bacillus subtilis in developmentalregulation. Cell 35:275-283.

J. BACTERIOL.

on October 22, 2020 by guest

http://jb.asm.org/

Dow

nloaded from