Dual control of subtilin biosynthesis and immunity in Bacillus subtilis



The production of the peptide antibiotic (lantibiotic) subtilin in Bacillus subtilis ATCC 6633 is highly regulated. Transcriptional organization and regulation of the subtilin gene cluster encompassing 11 genes was characterized. Two polycistronic mRNAs encoding transcript spaBTC (6.8 kb) and encoding transcript spaIFEG (3.5 kb) as well as the monocistronic spaS (0.3 kb) mRNA were shown by Northern hybridization. Primer extension experiments and β-galactosidase fusions confirmed three independent promoter sites preceding genes spaB, spaS and spaI. β-Galactosidase expression of spaB, spaS and spaI promoter lacZ fusions initiated in mid-exponential growth. Maximal activities were reached at the transition to stationary growth and were collinear with subtilin production. The lacZ activity was dependent on co-expression with the two-component regulatory system spaRK. The presence of subtilin was needed for efficient expression of all three promoter lacZ fusions. This suggests a transcriptional autoregulation according to a quorum-sensing mechanism with subtilin as autoinducer and signal transduction via SpaRK. Additionally, spaR expression was found to be under positive control of the alternative sigma factor H. Deletion of sigma H strongly decreased subtilin production. Full subtilin production could be restored after in-trans complementation of spaR. Deletion of the major B. subtilis transition state regulator AbrB strongly increased subtilin production. These results show that the spaRK two-component regulatory system, and hence subtilin biosynthesis and immunity, is under dual control of two independent regulatory systems: autoinduction via subtilin and transcriptional regulation via sigma factor H.


Biosynthesis of antimicrobial peptides is a common phenomenon among many Gram-positive bacteria (Hancock, 1997; Hancock and Lehrer, 1998). Subtilin produced by Bacillus subtilis ATCC 6633 (Jansen and Hirschmann, 1944; Gross et al., 1973), is ribosomally synthesized as a precursor peptide and undergoes extensive post-translational modifications (for reviews, see Entian and de Vos, 1996; Sahl and Bierbaum, 1998; Jack and Jung, 2000). Matured subtilin contains a series of unusual amino acids such as dehydrated serine/threonine residues and (β-methyl-) lanthionine thioether bridges, characteristic structural elements for lantibiotics. Two classes of lantibiotics are distinguished. The class of linear lantibiotics includes subtilin from B. subtilis, nisin from Lactococcus lactis (Buchman et al., 1988; Kaletta and Entian, 1989; Dodd et al., 1990), epidermin (Allgaier et al., 1985; Allgaier et al., 1986; Schnell et al., 1988) and Pep5 (Sahl and Brandis, 1981) produced by Staphylococcus epidermidis, and gallidermin from Staphylococcus gallinarum (Kellner et al., 1988; Schnell et al., 1989). Streptomyces species produce a series of globular type lantibiotics like cinnamycin (Kessler et al., 1992), duramycin (Fredenhagen et al., 1990) and ancovenin (Wakamiya et al., 1985). Mersacidin produced by B. subtilis sp. (Chatterjee et al., 1992; Altena et al., 2000) belongs to an intermediary structural type. Cytolysin from Enterococcus faecalis (Gilmore et al., 1994; Booth et al., 1996) and lactocin 34 from L. lactis subsp. lactis DPC3147 (McAuliffe et al., 1998; Ryan et al., 1999) represent two-component lantibiotics, which need two peptides for the antimicrobial activity. Recently, a gene cluster encoding structural genes for two different lantibiotics (ericin A and S) modified by a common modification machinery was identified in B. subtilis A 1/3. Ericin S is highly homologous to subtilin, whereas ericin A exhibits a different structure (Stein et al., 2002). Lantibiotics act against several Gram-positive bacteria by the formation of pores into the cytoplasma membrane (for a review, see Breukink and de Kruijff, 1999). For nisin, pore formation was biophysically studied with black lipid bilayers and membrane vesicles (Driessen et al., 1995; Breukink et al., 1998; van Kraaij et al., 1998). Additionally, specific binding to the peptidoglycan precursor lipid II was shown for nisin (Breukink and de Kruijff, 1999; Breukink et al., 1999), which inhibited cell wall biosynthesis (Wiedemann et al., 2001).

In the subtilin-producing strain B. subtilis ATCC 6633, a gene cluster was identified (for summary, see Entian and de Vos, 1996), which contains the structural gene spaS (Banerjee and Hansen, 1988), genes required for biosynthesis and transport of subtilin spaBTC (Klein et al., 1992; Gutowski-Eckel et al., 1994) and genes encoding a two-component regulatory system spaRK (Klein et al., 1993), as well as genes involved in immunity against the produced lantibiotic spaIFEG (Klein and Entian, 1994). A multimeric protein complex consisting of proteins SpaB, C and T was shown, which physically interacts with the subtilin prepeptide and which catalyses presubtilin modification and transport through the cytoplasmic membrane (Kiesau et al., 1997). Despite the mapping of mRNA initiation of spaS and spaB (Banerjee and Hansen, 1988; Chung and Hansen, 1992), little is known about the transcriptional regulation of subtilin biosynthesis and immunity. For nisin biosynthesis, nisin was identified as an autoinducing molecule (Kuipers et al., 1995) acting on the transcriptional units nisABTCIP, nisRK and nisFEG (de Ruyter et al., 1996).

Here, we report on the regulation of subtilin biosynthesis and immunity, which is closely related to nisin with respect to autoinduction, but is more complex with respect to the regulation of the spaRK two-component regulatory system whose expression is pleiotropically controlled by the major transition state regulator AbrB and sigma factor H.


Transcription units of the spa gene cluster

The transcriptional organization of the subtilin gene cluster and the expression of genes spaB, T, C, S, I, F, E, G, R and K (Fig. 1) was studied in B. subtilis ATCC 6633 at various growth phases. Total RNA was isolated at different time-points from the wild type and as a control from deletion mutants of spaS, spaB and spaI.

Figure 1.

Genomic organization of the subtilin gene cluster.

A. Genome comparison between subtilin non-producer Bacillus subtilis 168 (3456.55 kb and 3459.51 kb; Kunst et al., 1997) and subtilin producer B. subtilis ATCC 6633 indicating the positions of spa gene cluster integration. In B. subtilis ATCC 6633, most of yvaQ from B. subtilis 168 is replaced by the subtilin gene cluster.

B. Putative promoter regions of the subtilin gene cluster were fused to lacZ and integrated into the subtilin producer B. subtilis ATCC 6633 and the B. subtilis 168 derivative B. subtilis MO1099, respectively, as indicated in Experimental procedures.

With spaB as a probe for Northern hybridization, an intensive band of 6.8 kb was obtained with RNA isolated in the mid-logarithmic growth phase (4–5 h after inoculation; Fig. 2B, lanes 5–6). This signal disappeared after 5 h and was also no longer present in RNA samples taken later then 6 h (Fig. 2B, lanes 7–8). No such signal was found with RNA isolated from the ΔspaB mutant (Fig. 2B, lanes 1–4). The large transcript is in accordance with a polycistronic mRNA compassing the three genes spaB, spaT and spaC (theoretically 6.2 kb).

Figure 2.

Northern hybridization analyses of spa gene transcription.

A. Growth curves of the wild-type B. subtilis ATCC 6633 (•) and gene deletion mutants ΔspaB (○), ΔspaS (▴), and ΔspaI (▵).

B. spaB hybridization. RNAs were isolated from a ΔspaB mutant (lanes 1–4) and wild type (lanes 5–8). Samples were taken after 4 h (lanes 1 and 5), 5 h (lanes 2 and 6), 6 h (lanes 3 and 7), and 7 h (lanes 4 and 8).

C. spaS hybridization. RNAs were isolated from a ΔspaS mutant (lanes 1 and 2) and wild type (lanes 3 and 4). Samples were taken after 4 h (lanes 1 and 3) and 5 h (lanes 2 and 4).

D. spaI (lanes 1 and 2) and spaE (lanes 3 and 4) hybridization. RNAs were isolated from a ΔspaI mutant (lanes 1 and 3) and wild type (lanes 2 and 4) after 5 h of growth.

Northern hybridization using a spaS-specific probe revealed a band of 0.35 kb (Fig. 2C). Here, the most intensive band was detected after 5 h of growth. The transcript was not observed using RNA of a ΔspaS strain (Klein et al., 1992). Therefore, the structural gene of subtilin is located on a small transcript, which confirms previous findings (Banerjee and Hansen, 1988).

To identify the transcript of the immunity genes, probes were directed against spaI and spaE (Fig. 2D). Using wild-type RNA, a transcript of 3.5 kb was obtained in both cases, which is in accordance with a polycistronic RNA transcript comprising all four immunity genes spaI, spaF, spaE and spaG (theoretically 3 kb). As a control, RNA isolated from a ΔspaI deletion mutant was used and revealed no respective signal with the spaI probe. The spaE probe also provided no signal, because of a transcription terminator behind the chloramphenicol cassette, which was used for spaI deletion (Klein and Entian, 1994). Unspecific hybridization of spaI/spaE probes with 23S and 16S B. subtilis rRNA gave signals at 1.5 and 2.9 kb (theoretically 2.96 and 1.54 kb respectively).

To determine the 5′-start sites of the spa gene transcripts, total RNA of B. subtilis ATCC 6633 was isolated after 5 h of growth, where a maximal Northern signal was observed. For spaS primer extension, an oligonucleotide complementary to the coding strand of spaS was used (see Table 2). Assuming that the most intensive band is the primer extension product, transcription of the spaS gene initiates at the T residue 74 bases upstream of the ATG start codon (Fig. 3A). For spaI, the most intensive signal corresponds to a primer extension product starting at an A residue 144 bases upstream the spaI start codon (Fig. 3B). Although no clear consensus sequences were found, comparison of the promoters of spaS and spaI with that of spaB (Chung and Hansen, 1992) revealed some similarities within the –10 and –35 region (Fig. 3C).

Table 2. Primers used in this work.
UsageUpstream primeraDownstream primera
  • a.

    Primer sequences are given in the direction 5′-3′; changed nucleotides are given in bold face.

spaS-(3)-promoter regionGGTATCGGTCGAATTCTGCTTGC (EcoRI) 
Figure 3.

Mapping the transcriptional start sites of spaS and spaI by primer extension analysis. RNA was prepared after growth in TY medium for 5 h.

A and B. The first four lanes show autoradiography of dideoxynucleotide sequencing reactions with primers complementary to the 5′-region of spaS (A) and spaI (B) genes (see Table 2). The respective DNA sequences are shown in the outside lanes and the transcriptional start sites are marked by arrows.

C. Comparison of DNA regions upstream from the transcriptional start sites (+1) of spaB (Chung and Hansen, 1992), spaS and spaI. The most probable –10 and –35 promoter sequences are underlined. Supposed sigma A binding sites are indicated in bold face and compared with the sigma A consensus motif (bottom line).

Expression of spa genes

Growth phase dependent expression of lacZ under control of different spa promoters (Fig. 1, Table 1) was investigated after chromosomal integration into the amyE locus of subtilin-producing strain B. subtilis ATCC 6633. As control a promoterless lacZ gene was integrated at the amyE locus (strain 33-Z) without any effects (Fig. 4A).

Table 1. Strains and plasmids used in this study.
StrainsDescription or relevant genotypeSource or reference
Bacillus subtilis
ATCC 6633 and derivatives:
ATCC 6633Subtilin producerBacillus Genetic
  Stock Center
ATCC 6633 ΔspaS spaS::cm (Cmr) Klein et al. (1992)
ATCC 6633 ΔspaB spaB::cm (Cmr) Klein et al. (1992)
ATCC 6633 ΔabrB abrB::neoThis work
ATCC 6633 ΔsigHsigH::cmThis work
ATCC 6633 ΔsigHPsposigHThis work
33-Z amyE::lacZ (Cmr)This work
33-S2amyE::PspaS2-lacZ (Cmr)This work
33-B2amyE::PspaB2-lacZ (Cmr)This work
33-I amyE::PspaI-lacZ (Cmr)This work
MO1099 and derivatives:
MO1099 amyE::erm trpC2 pheA1; JH642 derivative (Mlsr) Gonzy-Treboul et al. (1992)
99-SB7with pSB7 (Mlsr, Kanr, Phleor)This work
99-SB10with pSB10 (Mlsr, Kanr, Phleor)This work
99-Z amyE::lacZ (Cmr)This work
99-S2amyE::PspaS2-lacZ (Cmr)This work
99-S3amyE::PspaS3-lacZ (Cmr)This work
99-B2amyE::PspaB2-lacZ (Cmr)This work
99-I amyE::PspaI-lacZ (Cmr)This work
99-S2-RamyE::PspaS2-lacZ (Cmr) with pSB7 (Kanr, Phleor)This work
99-B2-RamyE::PspaB2-lacZ (Cmr) with pSB7 (Kanr, Phleor)This work
99-I-R amyE::PspaI-lacZ (Cmr) with pSB7 (Kanr, Phleor)This work
99-S2-RKamyE::PspaS2-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
99-B2-RKamyE::PspaB2-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
99-I-RK amyE::PspaI-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
JH651 spoOH81 J. Hoch
   Marahiel et al. (1987)
JH651-S2-RKspoOH81 amyE::PspaS2-lacZ (Neor, Cmr) with pSB10 (Kanr, Phleor)This work
JH651-R-RKspoOH81 amyE::PspaR-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
TT71 abrB::neoT. Tanaka
TT71-S2-RK abrB::neo amyE::PspaS2-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
TT71-R-RKabrB::neo amyE::PspaR-lacZ (Cmr) with pSB10 (Kanr, Phleor)This work
Escherichia coli
RR1Fhsd520 supE44 ara-14 proA2 lacY1 galK2 rpsL20 xyl-5 mtl-1 
BL21 DE3 pLysS host E. coli B F- dcm ompT supE hsdS (rB- mB-) galλ (DE3) Studier et al. (1990)
pDG148 E. coli and B. subtilis shuttle vector, IPTG-inducible Pspac
promoter (Apr, Kanr, Phleor)
Stragier et al. (1988)
pDG268pBR322 derivative with promoterless lacZ; for marker exchange
recombination into B. subtilis amyE (Apr Cmr)
Antoniewski et al. (1990)
pPKiS3 spaS (3) promoter region (318 bp), EcoRI/BamHI in pDG268 (Cmr)This work
pSB5 spaS (2) promoter region (202 bp), EcoRI/BamHI in pDG268 (Cmr)This work
pSB6 spaB (2) promoter region (251 bp), EcoRI/BamHI in pDG268 (Cmr)This work
pHZ52 spaI promoter region (486 bp) EcoRI/HindIII in pDG268 (Cmr)This work
pSB7 spaR HindIII/SphI in pDG148 (Apr, Kanr, Phleor)This work
pSB10 spaRK HindIII/XbaI in pDG148 (Apr, Kanr, Phleor)This work
pCE26 B. subtilis/E. coli shuttle vector (Apr, Cmr) Klein et al. (1992)
pCE82pBSR derivative with sigH disrupted by a cm cassette (Apr, Cmr)This work
pCE20 B. subtilis/E. coli shuttle vector (Apr, Ampr) Klein et al. (1992)
pCE80 sigH BamHI/XbaI in pCE20 (Apr, Kanr)This work
Figure 4.

Transcriptional regulation of different spa-promoter–lacZ fusions in subtilin producing B. subtilis strain ATCC 6633 (A) and the non-producing strain MO1099 (B-D). Filled symbols correspond to growth curves following inoculation of 200 ml sporulation medium (DSM, with or without IPTG) with stationary B. subtilis cultures. Open symbols correspond to β-galactosidase activities.

A. Analysis of genomically integrated spa promoter lacZ fusions in subtilin-producing strain B. subtilis ATCC 6633: spaS (strain 33-S2, triangles); spaB (strain 33-B2, boxes); and spaI (strain 33-I, circles).

B–D. Analysis of integrated spaB promoter lacZ fusions (B), spaS promoter lacZ fusions (C) and spaI promoter lacZ fusions (D) coexpressed with plasmid-encoded transcriptional regulator SpaR under Pspac control. The respective transformants of subtilin non-producing strain MO1099 were analysed without (□) and after (▵) IPTG induction.

The lacZ expression under control of spaB-, spaS- or spaI promoters was strongly dependent on the growth phase and coincided with the intensity of the Northern hybridization signals (see Fig. 2). There was no lacZ activity in the early growth phase; expression initiated in the mid-exponential growth and reached maximal β-galactosidase activities at the transition to the stationary growth phase. Interestingly, for spaS and spaB, maximal activities were observed after 6 h, whereas in the case of spaI maximal induction was reached 1–2 h later (Fig. 4A).

To study the influence of spaR on spa gene transcription, lacZ expressions under control of the spa promoters were investigated in B. subtilis MO1099, which originally contains no genes from the subtilin gene cluster. No β-galactosidase activity was observed without IPTG-induced spaR expression. After SpaR induction (with IPTG via Pspac) the lacZ activity strongly increased (Fig. 4B and C). The decrease of activity in the stationary phase is most likely as a result of sporulation of the cells in DSM medium. The expression of spaS and spaI promoter lacZ fusions in B. subtilis MO1099 (Fig. 4C) led to significant higher β-galactosidase activities as compared with expression of the same constructs in B. subtilis ATCC 6633 (Fig. 4A). This is as a result of a higher copy number of plasmid-encoded SpaR in the MO1099 derivatives. Furthermore, the influence of spaK coexpression was investigated as exemplified for the spaI-lacZ construct (strain 99-I-RK) in Fig. 4D. Coexpression of spaR with spaK under the control of the inducible Pspac promoter did not further increase the lacZ activity. This shows that under conditions of strongly induced SpaR, there is no necessity for the SpaK histidine kinase; a phenomenon which has been also described for two-component regulator proteins OmpR in Escherichia coli (Forst et al., 1990) and ComA for B. subtilis (Weinrauch et al., 1990).

Autoinduction of subtilin biosynthesis and immunity

As shown here, subtilin biosynthesis and subtilin immunity require the two-component signal transduction system spaRK. To investigate if subtilin provides the signal for the SpaRK activation, we analysed spa-promoter lacZ fusions in the non-producing strain B. subtilis MO1099, under semi-induced conditions of the Pspac spaRK genes. As shown in Fig. 5A, for the spaS promoter construct (strain 99-S2-RK), no activity was obtained without spaRK induction (no IPTG). At partially induced conditions (0.1 mM IPTG), extracellular addition of subtilin (100 μg l−1) induced β-galactosidase activity twofold. No further increase of activity was obtained with higher subtilin concentrations, which reflects saturation of spaS gene transcription. Similar results were obtained with spaB- and spaI-promoter lacZ fusions. However, spaI and spaB promoter lacZ fusions were further improved even after stronger induction of Pspac spaRK (0.1 and 0.5 mM respectively). Promoters spaB and spaI were obviously more sensitive to subtilin autoinduction than the spaS structural gene.

Figure 5.

Subtilin response of genomically integrated spaS promoter lacZ fusions (A), spaB promoter lacZ fusions (B) and spaI promoter lacZ fusions (C) coexpressed with plasmid encoded spaRK under Pspac control. β-Galactosidase activities were followed in the respective transformants of subtilin non-producing strain B. subtilis MO1099 with various subtilin concentrations (0, 0.1 and 0.3 mg subtilin l−1) under different conditions of spaRK induction (0, 0.1 and 0.5 mM IPTG). Cells were grown in DSM medium (see Fig. 4) and subtilin was added in the mid-exponential growth phase (OD600 = 1.0).

These data indicate that subtilin acts as a pheromone for the expression of its own structural gene (autoinduction) and for expression of biosynthetic (spaBTC) and subtilin immunity (spaIFEG) genes.

To prove that subtilin autoinduction also occurs in the producer strain B. subtilis ATCC 6633, we determined the amount of SpaC protein in the wild type and in a ΔspaS mutant after induction with sublethal amounts of subtilin (0.3 mg l−1). Lysates of stationary-grown cells were analysed using SDS–PAGE and immunoblotting. Wild-type B. subtilis ATCC 6633 revealed an intensive band for SpaC, whose intensity is similar in both cases, with and without subtilin induction (Fig. 6, lanes 1–2). For the ΔspaS strain, a weak SpaC signal was observed without induction of subtilin (Fig. 6, lane 4), whose intensity increased fourfold in the presence of subtilin (Fig. 6, lane 3). However, the slight SpaC band in the ΔspaS cells indicated that, even without subtilin in the extracellular medium, spa genes were expressed at a low level. As control, the spaB mutant showed no signal for SpaC because a transcription terminator behind the chloramphenicol cassette disrupting the spaB gene (Klein et al., 1992) blocks expression of the following genes spaT and spaC.

Figure 6.

Subtilin-induced expression of SpaC protein. Lysates of stationary cells from B. subtilis 6633 (wild type, lanes 1 and 2), and the respective mutants ΔspaS (lanes 3 and 4) and ΔspaB (lanes 5 and 6) were separated by SDS–PAGE. A SpaC-directed immunoserum was used for Western blot analyses without subtilin (– lanes) and after addition of 0.5 mg l−1 subtilin (+ lanes) in the mid-exponential growth phase (OD600 = 1.0).

Expression of spaR is regulated by SigH and AbrB

The production of many antibiotics in B. subtilis is under regulatory control of abrB, which encodes the transition state regulator of late-growth gene transcription (Weir et al., 1991; Strauch, 1995). During vegetative growth, AbrB acts as a repressor of sigH (Strauch et al., 1990), encoding sigma factor H, a regulator of late-growth activities (Carter and Moran, 1986; Dubnau et al., 1988).

To analyse the effect of abrB and sigH mutations on subtilin production, a ΔabrB deletion mutant was constructed after transforming competent B. subtilis ATCC 6633 cells with chromosomal DNA of strain TT71 (abrB::neo) and selecting on neomycin resistance and subtilin production. For sigH disruption, the gene was replaced after insertion of a chloramphenicol cassette (see Experimental procedures and Table 1).

As indicated in Fig. 7, subtilin production strongly increased in an ΔabrB deletion mutant, whereas subtilin production reached only 5% of the wild type in a ΔsigH mutant. This clearly showed that subtilin biosynthesis is negatively regulated via AbrB and needs sigma H for its expression.

Figure 7.

Effects of AbrB and sigma factor H on subtilin production.

A. Growth inhibition assays with wild type (WT) and mutants of B. subtilis ATCC 6633. Subtilin production is indicated by growth inhibition haloes with indicator organism Micrococcus luteus. The halo diameter increased in ΔabrB mutants and is beyond detection in ΔsigH mutants. The subtilin-deficient ΔsigH phenotype is complemented after constitutive spaR expressed spaR (ΔsigH + spaR).

B. Quantitation of subtilin production in B. subtilis ATCC 6633 and the respective ΔabrB and ΔsigH mutants. Culture supernatants of the strains were separated on an analytical RP-HPLC column as indicated in Experimental procedures. Subtilin eluted within the peak at 21.3–22.5 min.

C. Western blot analysis with SpaR antibodies. Lysates of stationary cells from B. subtilis 6633 (lane 1) and the respective ΔsigH mutant (lane 2) were separated by SDS–PAGE and analysed with a SpaR-directed immunoserum. The position of SpaR (calculated: 25.6 kDa) is marked by an arrow.

Whereas the promoters of spaS, spaBTC and spaIFEG contain consensus motifs for sigma A-dependent transcription (Moran et al., 1982; Jarmer et al., 2001), a sigma H consensus motifs was found in the spaRK promoter. To investigate the possible function of sigma H on the expression of spaR, the respective spaR promoter was replaced by the sigma H independent constitutive spo promoter. After transformation with the constitutive spaR gene, subtilin production was completely restored in a sigH mutant of B. subtilis ATCC 6633 (Fig. 7A). Western blot analysis provided a clear SpaR signal in the wild type but not in the sigH mutant (Fig. 7C), which provided evidence that spaRK transcription is sigma H-dependent.


Production of the lantibiotic subtilin in B. subtilis ATCC 6633 starts in the middle of the logarithmic growth phase and reaches its maximum in the stationary phase (Gutowski-Eckel et al., 1994). Using Northern analysis, two polycistronic mRNAs encoding genes spaB, spaT and spaC (6.8 kb), and spaI, spaF, spaE and spaG (3.5 kb), were identified. The subtilin structural gene, spaS, was transcribed as a single transcript of 0.35 kb. Primer extension analyses identified the –10 and –35 regions of spaS and the spaBTC operon which revealed perfect consensus sequences for B. subtilis ATCC 6633 sigma factor A (Moran et al., 1982; Jarmer et al., 2001). The transcription start site of spaS was identified 74 bp upstream of the translational start codon ATG which is different to the start site previously described 223 bp upstream of the start codon (Banerjee and Hansen, 1988). To investigate this discrepancy, two spaS promoter lacZ- fusions were constructed. Expression of β-galactosidase from both constructs, which either contained (33-S3) or lacked (33-S2) the previously described promoter region, was completely functional. This provides strong evidence for the –74 position to be sufficient for regulated spaS transcription.

The polycistronic transcriptional units identified here coincide with the proteins within the multimeric protein complexes for lantibiotic biosynthesis (SpaC, SpaT, SpaB) and immunity (SpaF, SpaE, SpaG), which is useful for the stoichiometric synthesis of the subunits. The structural gene spaS is much more strongly transcribed than the gene cluster encoding the lanthionine–synthetase complex, which ensures that the amount of substrate (presubtilin) strongly exceeds the amount of enzyme (lanthionine–synthetase complex). The expression of immunity genes was delayed as compared with the biosynthetic genes. This ensures that subtilin immunity is established when the amount of subtilin reaches a critical concentration for the producing strain.

To study spa gene expression independent of subtilin production, promoter lacZ fusions were expressed in B. subtilis MO1099, a strain lacking all subtilin genes. Here, gene expression was studied in the presence or absence of spaR, the regulator protein of the two-component regu-latory system spaRK. From the data presented, we conclude that spa gene expression depends on coexpression with spaR. Under conditions in which SpaR expression was semi-induced, we could show that subtilin further induced spaS and spaB transcription. A concentration of less than 100 μg l−1 subtilin was sufficient to induce spaS, spaBTC and SpaIFEG expression. The biosynthetic and immunity genes were significantly more sensitive to subtilin induction as compared with spaS. This is possibly as a result of a much lower expression rate of spaBTC and spaIFEG, which makes these operons more susceptible to further stimulation.

This shows that subtilin acts as an autoinducer for its own biosynthesis. This is similar as previously described for nisin biosynthesis and immunity, in which nisin was identified as the signal inducing its own biosynthesis and immunity (Kuipers et al., 1995; de Ruyter et al., 1996). Although the genetic components of subtilin and nisin biosynthesis and immunity are very specific for the respective lantibiotic, this result shows that the genetic organization and the regulation of the two lantibiotics is closely related.

Most probably the histidine kinase SpaK represents the subtilin sensor, which transfers the signal to the regulatory (DNA binding) SpaR protein. SpaK contains two transmembrane domains and characteristic sequence motifs, which strongly suggest a membrane localization (Klein et al., 1993). After sensing extracellular subtilin, SpaK is probably autophosphorylated at a conserved histidine residue. Subsequently, the response regulator SpaR is activated by transfer of the phosphoryl group to a conserved asparagine residue, by which SpaR is able to bind to promoter elements and induce gene transcription. A repeated pentanucleotide motif WTGAT (W = A/T) spaced by six basepairs was recently identified as the DNA sequence recognized by SpaR, which is present in front of each transcriptional unit (T. Stein, S. Heinzmann, P. Kiesau and K.-D. Entian, manuscript in preparation).

Interestingly, we found subtilin production under positive control of B. subtilis sigma factor SigH. This alter-native sigma factor represents a typical regulator of late-growth activities (Carter and Moran, 1986; Dubnau et al., 1988) affecting a broad range of genes involved in sporulation, competence development, motility and production of degradative enzymes. Additionally, sigma factor H is required for transcription of stress-inducible genes (Solomon et al., 1995). Deletion of sigH in B. subtilis ATCC 6633 diminished subtilin production by 95%. Transcription of sigH is under negative control of the general transition state regulator AbrB (Weir et al., 1991; Strauch, 1995), whose synthesis is repressed by the key regulator of late-growth processes, Spo0A (Strauch et al., 1990). Consequently, a abrB deletion mutant revealed a fourfold increased subtilin production and production started already in the logarithmic growth phase.

As the target for AbrB/sigma H-dependent regulation, spaRK was identified. This was indirectly shown by restored subtilin production in a ΔsigH mutant after constitutive expression of SpaR and directly proven by the absence of the SpaR protein in a ΔsigH mutant. However, an indirect effect of the global B. subtilis regulators AbrB/sigma H on subtilin production (via SpaR) cannot be excluded.

Subtilin autoinduction shows characteristic properties of quorum-sensing systems, which have been described in several bacteria for important physiological processes, for example, competence (Tortosa et al., 2001), sporulation, motility, biofilm formation, bioluminescence and virulence (Kleerebezem and Quadri, 2001). In Bacillus subtilis, quorum sensing by autoinducing peptides was described for phosphatase regulation (phr) (Lazazzera, 2001; Perego and Brannigan, 2001) controlled by sigma factor H (McQuade et al., 2001) and also for ComX, a natural competence regulating pheromone (Solomon et al., 1995; Tortosa et al., 2001). For nisin of L. lactis, autoinduction of the two-component system nisR and nisK was also discussed in the light of a quorum-sensing mechanism (Kuipers et al., 1995; de Ruyter et al., 1996), although the strong definition of quorum sensing allows bacteria to sense their population density and to respond through appropriate changes in gene expression (Kleerebezem et al., 1997). The AbrB-dependent regulation for subtilin shows the interaction of growth phase-dependent regulation and autoinduction. Quorum sensing with subtilin would also enable producing cells to communicate with neighbouring cells, for example to signal/ sense non-ideal conditions, and may provide an example for a B. subtilis peptide antibiotic used as a pheromone for quorum sensing.

In summary, the expression of subtilin biosynthesis and immunity genes is regulated by two different mechanisms (Fig. 8). During exponential growth phase, the transition state regulator protein AbrB acts as a repressor of subtilin biosynthesis and only a basal level of biosynthetic proteins SpaBTC are synthesized (see SpaC content in ΔspaS strain). Consequently, low amounts of subtilin appeared in the culture, which were not sufficient to induce the regulatory cascade. At the end of exponential growth, AbrB synthesis is repressed (Perego et al., 1988; Strauch et al., 1989; Strauch et al., 1990), followed by derepression of the alternative sigma factor H (Weir et al., 1991; Strauch, 1995). Sigma factor induces spaRK transcription and the SpaK histidine kinase is expressed. From a critical concentration, subtilin acts via SpaRK as inducer of spaS and spaBTC transcription, resulting in drastically increased subtilin production. Additionally, the immunity system spaIFEG is induced to protect the producer from the lethal activity of its own lantibiotic. The dual control tunes subtilin production and the physiological state of the cell.

Figure 8.

Dual control of subtilin biosynthesis and immunity. During exponential growth, the expression of subtilin (spa) genes is repressed. The transition state regulator protein AbrB acts as a repressor of subtilin biosynthesis and immunity via the alternative sigma factor H. At a critical level of the extracellular subtilin concentration, the lantibiotic acts as a pheromone and induces the expression of spaBTC, spaS and spaIFEG via the two-component regulation system SpaRK. The dual control allows quorum sensing to co-ordinate subtilin production and the physiological state of the cell.

Experimental procedures

Bacterial strains, growth conditions and construction of plasmids

Bacillus subtilis cells were grown in Difco sporulation medium (DSM, Difco; Harwood and Cutting, 1990) and TY medium (0.8% tryptone, 0.5% yeast extract, 0.5% NaCl). Escherichia coli strains were grown on Luria–Bertani (LB) medium (Gibco). The concentrations of antibiotics used were: ampicillin (40 μg ml−1) and chloramphenicol (20 μg ml−1) for E. coli, and chloramphenicol (5 μg ml−1), spectinomycin (25 μg ml−1), neomycin (7 μg ml−1), phleomycin (5 μg ml−1) and kanamycin (10 μg ml−1) for B. subtilis. The B. subtilis strains used in this study are listed in Table 1; a schematic representation of the different strains is given in Fig. 1. EcoRI–BamHI (BamHI–HindIII in the case of spaI) fragments of putative spa gene promoters were amplified by polymerase chain reaction (PCR) (primer sequences are detailed in Table 2) and cloned into pDG268 (Antoniewski et al., 1990), which contains a promoterless lacZ gene, an upstream ribosomal binding-site for B. subtilis and a chloramphenicol resistance cassette for chromosomal integration into B. subtilis amyE locus (Table 2). Promoter regions consisted of promoter regions of spaB comprising 251 bp, the promoter region of spaS with two different lengths comprising 110 and 202 bp, the spaI promoter (486 bp), and the putative spaR promoter (497 bp). Linearized plasmids were used to transform competent cells of B. subtilis ATCC 6633 and MO1099; resulting strains are summarized in Table 1. Correct integration of the spa promoter lacZ fusions were analysed by PCR. Recombinant plasmids were amplified in E. coli RR1 cells. For expression of spaR and spaRK, genes were PCR-amplified and cloned into pDG148 (Stragier et al., 1988) under control of an IPTG-inducible promoter (Pspac).

Molecular biology techniques

Established protocols were followed for molecular biology techniques (Sambrook et al., 1989). DNA was cleaved according to the conditions recommended by the commercial supplier of the restriction enzymes (Boehringer GmbH). Restriction endonuclease-digested DNA was eluted from 0.7% agarose gels by the freeze-squeeze method.

Plasmid isolation and PCR

The alkaline extraction procedure (Birnboim and Doly, 1979) was followed for E. coli plasmid isolation. PCR was carried out following standard procedures (Sambrook et al., 1989) in a Hybaid combi-thermal-reactor R2, Taq DNA polymerase (Boehringer Mannheim) was used.

RNA isolation

Total RNA of B. subtilis ATCC 6633 was isolated by the acid/phenol method (Volker et al., 1994). For primer extension analysis, RNA was isolated from cells at the end of exponential growth. To exclude contamination by DNA, each sample was treated by DNase I (2 U per μg RNA, Pharmacia, Biotech) in assay buffer (40 mM Tris-HCl pH 7.5, 60 mM MgCl2) for 10 min at 37°C. After additional phenol/ chloroform extraction and ethanol precipitation, the RNA was dissolved in 20 μl of water. RNA concentrations were measured photometrically at 260 nm or estimated by comparison with 16 S and 23 S RNA band intensities.

Northern hybridization

A total of 30 μg of RNA was fractionated on a 1.5% agarose gel containing formaldehyde (final concentration 1.2%) and stained with ethidium bromide. After washing with water (15 min), and five times with 10× standard saline citrate (SSC: 1.5 M NaCl, 150 mM sodium citrate, pH 7) RNA was transferred to Biodyne B transfer membrane (Hybond-N). RNA was cross-linked to the dried membrane by UV light (312 nm, 7 min) after briefly rinsing with 2× SSC buffer. The spaB probe (1066 bp) was obtained after EcoRI clearage of plas-mid pCE42 (Klein et al., 1992). For spaS, spaI, and spaE probes, PCR-amplified DNA was used (550, 300 and 300 bp respectively; for primers see Table 2). DNA probes were labelled with [γ-32P]-dCTP using the Ready-to-go-kit (Pharmacia Biotech). For hybridization, membranes were incubated for 3 h in a solution containing 5× SSC, 1× blocking agent (Boehringer), 0.02% (w/v) sodium dodecyl sulphate (SDS) and 0.1% (w/v) N-laurylsarcosine at 42°C. Hybridization was performed with a denatured DNA probe in the same buffer overnight. Hybridized blots were washed in 5× SSC buffer containing 0.01% (w/v) SDS. Radioactivity was detected using autoradiography.

Primer extension analysis

Primer extension experiments were performed according to (Sambrook et al., 1989). A total of 5 μg of B. subtilis RNA was incubated with 10 pmol oligonucleotide primer (Table 2) and 20 U RNase inhibitor (Boehringer Mannheim) for 5 min at 96°C. After annealing (cool down to 42°C in 60 min), 5 mM DTT, 2 μl of 10-fold reaction buffer (500 mM Tris-HCl, pH 8.3, 60 mM MgCl2, 400 mM KCl, and 500 μg ml−1 BSA) were added. To 20 μl of this reaction mixture, 20 U RNase inhibitor, 10 μCi [α-32P]-dCTP, 7.5 μM dNTP-mix without dCTP and 14.5 U AMVRT (reverse transcriptase from Avian Myeloblastosis virus; AGS GmbH) were added. After 5 min of pre-incubation at 42°C, the dNTP mix was added to a final concentration of 23 mM, and the incubation was extended to 25 min at 42°C. By adding 5 μl of stop-mix (95% formamid, 20 mM EDTA, 0.05% bromophenol blue, 0.05 xylen cyanol FF) and a denaturating step (5 min, 96°C), the reaction was stopped. Primer extension reaction and sequencing reactions with the corresponding primer were run on an 8% polyacrylamide gel, radioactivity was detected by autoradiography by exposure of dried gels to Kodak X-Omat-RP films at –80°C.

β-Galactosidase assays

Histochemical screening for β-galactosidase activity by selecting for blue colonies was performed by including isopropyl-β-D-thiogalactoside (IPTG) at a final concentration of 50 μg ml−1 in DSM plates. For quantitative β-galactosidase activities, standard methods were used (Zuber and Losick, 1983; Miller, 1992). A total of 200 ml of TY/DSM medium was inoculated (1:100) with an overnight culture of B. subtilis cells carrying the reporter gene. Then,1 ml samples were taken at different times of the growth curve, sedimented and washed with 1 ml of 50 mM Tris-HCl (pH = 8.0). Cells were solved in 1 ml Z-buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, 50 mM β-mercaptoethanol, pH 7.0). After appropriate dilution in 1 ml of the same buffer, the cells were permeabilized by adding 20 μl of toluene. After removal of toluene (37°C for 45 min), the β-galactosidase reaction was started with 200 μl of o-nitrophenyl-β-D-galactoside (ONPG, 4 mg ml−1 in Z-buffer) at 28°C and stopped with 500 μl of 1 M Na2CO3.

SDS–polyacrylamide gel electrophoresis (SDS–PAGE) and Western blot analysis

SDS–PAGE and Western blot analyses were performed as described previously (Kiesau et al., 1997). Selectivity of the SpaC-directed immunoserum was significantly improved after immunoaffinity purification (Harlow and Lane, 1999) using purified SpaC–GST fusion protein heterologously expressed in E. coli. For SpaR-directed antisera, SpaR was isolated from E. coli cells expressing spaR under control of the T7 promoter. Escherichia coli cell extract (incubation in 100 mM NaPO4, pH 7.0, 0.5%β-mercaptoethanol, 4% SDS, 8 M urea and 0.02% bromophenol blue for 10 min at 95°C) was separated by preparative SDS–PAGE; SpaR was isolated by electroelution and used for rabbit immunization (Eurogentec).

Analysis of subtilin production by reversed phase (RP)-HPLC

Stationary B. subtilis cultures were centrifuged at 10 000 g for 10 min. RP-HPLC analyses of the supernatant were performed by means of a Beckman Gold HPLC System, a semi-preparative reversed phase Lichrospher column (particle size: 10 μm, width and length: 4 × 200 mm, Merck) and an analytical ODS Hypersil column (particle size: 5 μm, width and length: 4 × 300 mm, Maisch). Samples were applied using solvent A (aqueous solution of 20% acetonitrile and 0.1% trifluoroacetic v/v). Elution was performed using linear gradients to solvent B (0.1% trifluoroacetic acid in acetonitrile v/v). Eluting peptides were detected measuring the absorption at 214 nm.