Rolling-circle plasmids from Bacillus subtilis: complete nucleotide sequences and analyses of genes of pTA1015, pTA1040, pTA1050 and pTA1060, and comparisons with related plasmids from Gram-positive bacteria
Wilfried J.J Meijer,
Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, Kerklaan 30, 9751 NN Haren, The Netherlands
Most small plasmids of Gram-positive bacteria use the rolling-circle mechanism of replication and several of these have been studied in considerable detail at the DNA level and for the function of their genes. Although most of the common laboratory Bacillus subtilis 168 strains do not contain plasmids, several industrial strains and natural soil isolates do contain rolling-circle replicating (RCR) plasmids. So far, knowledge about these plasmids was mainly limited to: (i) a classification into seven groups, based on size and restriction patterns; and (ii) DNA sequences of the replication region of a limited number of them. To increase the knowledge, also with respect to other functions specified by these plasmids, we have determined the complete DNA sequence of four plasmids, representing different groups, and performed computer-assisted and experimental analyses on the possible function of their genes. The plasmids analyzed are pTA1015 (5.8 kbp), pTA1040 (7.8 kbp), pTA1050 (8.4 kbp), and pTA1060 (8.7 kbp). These plasmids have a structural organization similar to most other known RCR plasmids. They contain highly related replication functions, both for leading and lagging strand synthesis. pTA1015 and pTA1060 contain a mobilization gene enabling their conjugative transfer. Strikingly, in addition to the conserved replication modules, these plasmids contain unique module(s) with genes which are not present on known RCR plasmids of other Gram-positive bacteria. Examples are genes encoding a type I signal peptidase and genes encoding proteins belonging to the family of response regulator aspartate phosphatases. The latter are likely to be involved in the regulation of post-exponential phase processes. The presence of these modules on plasmids may reflect an adaptation to the special conditions to which the host cells were exposed.
Numerous plasmids of relatively small size have been identified in a wide range of Gram-positive bacteria. Most of these use the rolling-circle mechanism of replication which is characterized by the synthesis of single-stranded (ss) DNA intermediates. Several of these plasmids have been studied extensively (for reviews see [1–4]). Naturally occurring rolling-circle replicating (RCR) plasmids form the basis of many cloning and expression vectors for Gram-positive bacteria. Although the highly transformable Bacillus subtilis strain 168, which has become the paradigm for most studies, does not harbor endogenous plasmids, the existence of B. subtilis plasmids has been known for quite some time [5–11]. Since these plasmids do not confer easily selectable phenotypes to their hosts, most cloning vectors used today for B. subtilis are based on RCR plasmids from other Gram-positive bacteria, like staphylococci and streptococci [1–3, 12–15]. These plasmids can replicate and express their antibiotic resistance genes in B. subtilis. However, vectors based on such plasmids are frequently unstable [1, 3, 16–19].
So far, small cryptic plasmids from B. subtilis have attracted relatively little attention. Such plasmids have raised our interest for several reasons: first, for studies on plasmid replication and plasmid stability, and the construction of stable cloning vectors for B. subtilis. One of the factors underlying plasmid instability of non-native plasmid vectors is non-optimal plasmid–host interaction [18, 20–23]. Thus, vectors based on endogenous B. subtilis RCR plasmids are expected to be superior to vectors based on non-native plasmids. Second, interesting novel traits may be present on cryptic B. subtilis plasmids. Such studies on the gene functions on plasmids would nicely complement ongoing research on the function analysis of B. subtilis chromosomal genes. In the latter project, coordinated by S.D. Ehrlich (INRA, Jouy-en-Josas, France) and N. Ogasawara (Nara Institute of Science and Technology, Nara, Japan), about 25 groups in Europe and Japan cooperate to identify the function of as many chromosomal genes as possible. This project is a follow-up of the recently finished combined European–Japanese project aimed at the determination of the entire sequence of the B. subtilis chromosome , which was coordinated by F. Kunst (Institut Pasteur, Paris, France) and N. Ogasawara (Nara Institute of Science and Technology, Nara, Japan).
Of the plasmids studied here, several are mobilizable and, hence, can contribute not only to the total gene pool of the host but also to the horizontal gene pool of the bacterial population. Since B. subtilis plasmids are generally several kilobasepairs larger than most RCR plasmids from other Gram-positive bacteria, they can easily accommodate genes other than those required for replication and mobilization. Such genes may facilitate molecular adaptation of the host to certain conditions. These genes might even be of industrial relevance, since many of the endogenous plasmids of B. subtilis known today were isolated from industrial strains. It is known that several valuable traits for dairy industry are located on plasmids, in that case from lactic acid bacteria . Third, the comparison of several B. subtilis plasmids and the determination of their complete nucleotide sequence will give insight in their structural organization and evolutionary relationship.
In the present work we compare the complete DNA sequences of pTA1015 (5.8 kbp), pTA1040 (7.8 kbp), pTA1050 (8.4 kbp; also known under the name pPOD2000 ), and pTA1060 (8.6 kbp), and describe computer-based and experimental analyses of the function of the identified putative genes. By including information available from other B. subtilis RCR plasmids, we present an overview of our current knowledge on B. subtilis RCR plasmids.
2Classification of B. subtilis RCR plasmids
Uozumi et al. analyzed twenty B. subtilis strains that harbor relatively small plasmids. Based on size and restriction profiles these plasmids have been classified into six groups, represented by pTA1015, pTA1020, pTA1030, pTA1040, pTA1050, and pTA1060. The production of trace amounts of ssDNA replication intermediates by each of these plasmids indicates that they all use the rolling-circle mechanism of replication . Plasmids isolated in independent studies by Hara et al. (pUH series); Tanaka et al. [5, 6](pLS series); Darabi et al. (pBS2); Devine et al. (pBAA1), and by Poluektova et al. (p1410) can also be classified into one of these six groups. An RCR plasmid belonging to a seventh group, represented by pFTB14, was isolated from the Bacillus amyloliquefaciens strain S294 . B. subtilis and B. amyloliquefaciens are highly related. In fact, based on DNA homology studies, the B. subtilis strain IFO3022 harboring pTA1060 is now considered to be a B. amyloliquefaciens strain . Whereas most of these plasmids were identified in industrial isolates, Nezametdinova et al. identified small cryptic plasmids in 21 out of the 32 natural B. subtilis strains isolated from soil by Kozlovsky and Prozorov . Six of these, p1410 and others , belong to the pTA1020 group. An overview of the classification of the B. subtilis RCR plasmids is shown in Table 1. Plasmids of identical size and restriction profile are likely to be identical or at least highly similar to each other, and are placed in the same group. In addition, plasmids have been isolated from other Bacillus strains that cannot be placed in one of these groups [33, 34]. Since only limited information is available for the latter plasmids they are not discussed here.
Table 1. Classification of cryptic RCR plasmids from B. subtilis
Type of SSOa
Plasmids with identical sizes and restriction profiles
a1 and 2 represent the palT1- and palT2-type SSOs, described in Section 5.
bPlasmid size in kbp. pTA series according to Uozumi et al. ; pLS series according to Tanaka and coworkers [5, 6]; pUH series are according to Hara et al. .
Tanaka et al. noticed that several B. subtilis strains contained, in addition to a relatively small plasmid (the pLS series), a larger plasmid with an average size of approximately 55 or 77 kbp. The minimal replicon of one of these, pLS20, appeared to represent a novel type of theta replicon . As judged from their size, it is likely that also the other large plasmids use theta replication. Thus, the co-residence of two plasmids, one using the theta mode and the other the rolling-circle mode of replication, appears to be common in B. subtilis. To our knowledge, the co-residence of two RCR plasmids or two theta-replicating plasmids in cells of natural B. subtilis or B. amyloliquefaciens strains is rare. A possible exception has been observed by E. Poluektova who noticed the presence of two related replicons in one and the same B. subtilis strain isolated from soil (personal communication; A.A. Prozorov). This is in contrast to other Gram-positive bacteria, like lactococci, which can harbor several theta-type plasmids within one cell [36, 37].
For a number of B. subtilis RCR plasmids additional information is available. First, the replication regions of pBAA1 (pTA1020 group; ); pBS2 (pTA1060 group; ); pLS11 (pTA1060 group; ); pUH1 (pTA1015 group; ); and pFTB14 have been sequenced. Second, the single-strand origins (SSOs) (see Section 5) of pTA1060 ; pTA1015 ; pLS11 (pTA1060 group; ); pBAA1 (pTA1020 group; [28, 42]), and pTA1040 have been analyzed. Third, a 1.73-kbp region of plasmid pUH1 (pTA1015 group), claimed to contain the γ-glutamyltranspeptidase (γ-gtp) gene, has been sequenced .
3Sequences and structural organization of pTA1015, pTA1040, pTA1050, and pTA1060
RCR plasmids consist of several interchangeable modules which frequently show considerable homology at the DNA and/or deduced protein level [1, 3, 44]. An essential module comprises the replication initiation gene (rep) and its cognate target site, the double-strand origin (DSO). Another module contains the SSO which functions as the major initiation site for lagging-strand synthesis. Several RCR plasmids contain, in addition, a module with a gene, denoted mob or pre, which is involved in conjugative mobilization and site-specific recombination, respectively.
For the analysis of the structural organization of the B. subtilis plasmids pTA1015, pTA1040, pTA1050 and pTA1060, we determined their complete nucleotide sequence. The sizes of pTA1015, pTA1040, pTA1050, and pTA1060 are 5807 bp, 7837 bp, 8397 bp, and 8737 bp, respectively. The G+C contents are 40.7% (pTA1015), 36.9% (pTA1040), 38.1% (pTA1060), and 40.0% (pTA1050). The sequences were deposited in the EMBL/GenBank/DDBJ database under the accession numbers U32379 (pTA1015), U32378 (pTA1040), U32380 (pTA1060), and U55043 (pTA1050). For pTA1015, pTA1040 and pTA1060 the putative nick sites for replication initiation were given coordinate number 3; in the database entry of pTA1050 (U55043) this position corresponds to coordinate 2776 and is located in the opposite strand relative to the other plasmids.
The DNA sequences and deduced amino acid sequences of identified open reading frames (ORFs) of the four pTA plasmids were compared to known sequences in the databases. This revealed the following putative genes and non-coding DNA sequences, which are also commonly found in other RCR plasmids: (i) a rep gene and cognate DSO located upstream of the gene (pTA1015, pTA1040, pTA1060, and pTA1050); (ii) an SSO (pTA1015, pTA1040, pTA1060, and pTA1050); and (iii) a mob gene (pTA1015 and pTA1060). In addition, other putative genes were identified, some of which showed significant homology to previously identified genes. These plasmid-located genes were given names according to the nomenclature of their homologues. The directions of transcription and the localization of the genes, as well as the positions of the DSOs and the SSOs (indicated as palT1 or palT2), are shown in Fig. 1. Characteristics of the identified genes are shown in Table 2. The structural organization of the four pTA plasmids, i.e. Rep-encoding gene with an upstream located DSO, a separated SSO, and the majority of the genes being transcribed in the direction coinciding with the direction of replication, is similar to most other RCR plasmids. In the following sections comparisons and experimental analyses of the different genes and modules are discussed.
Table 2. Characteristics of the ORFs on pTA plasmids
Mol. mass (kDa)b
Putative start codon and RBSc
aCoordinate numbers for pTA1015, pTA1040, pTA1060 and pTA1050 correspond to those in the EMBL/GenBank Data Library; accession numbers U32379, U32378, U32380, and U55043, respectively. The origin nick sites were taken as position 3 for all four plasmids. In the case of pTA1050, these coordinates differ from those under accession number U55043, since in the database submission a different starting point, on the opposite strand, was used. For reasons of comparison we have used here a coordinate numbering for pTA1050 which corresponds to that of the other plasmids.
bMolecular mass was calculated based on the deduced amino acid sequence.
cPutative initiation codons (capital letters) and the 22-bp upstream sequences are shown; nucleotides complementary to the 3′-end of B. subtilis 16S rRNA (UCUUUCCUCCACUAG) region are underlined.
dThe spacing is calculated as the distance from the first nucleotide to the right of the AGGA sequence (or the equivalent) to the nucleotide at the 5′-side of the initiation codon.
e‘C’ indicates that the ORF is located on the opposite strand.
9 or 15
4.1Rep genes, Rep proteins and DSOs
4.1.1Characterization of rep genes and Rep proteins
All RCR plasmids studied so far contain a rep gene encoding a Rep protein that is essential for the initiation of replication. Through their DNA-binding and nicking/closing activity the Rep proteins introduce a strand- and site-specific nick in the plasmid at the DSO. The 3′-OH end of the nick site is subsequently used for the initiation of leading strand synthesis.
Based on sequence similarity of their DSOs and homology of the replication proteins, RCR plasmids from Gram-positive bacteria can be divided into five classes, represented by pT181, pC194, pE194, pSN2, and pTX14-3 [1, 45]. Based on these criteria the four pTA plasmids belong to the pC194 class of plasmids. Also the other B. subtilis/B. amyloliquefaciens RCR plasmids of which the replication regions have been sequenced so far belong to the pC194 class . An alignment of the deduced Rep protein sequences of the four pTA plasmids described here with those of published sequences of other Bacillus plasmids is shown in Fig. 2A. This alignment shows that the four pTA plasmids are highly homologous to each other and to the known Rep proteins of the other B. subtilis/B. amyloliquefaciens RCR plasmids.
Despite their strong overall homologies, the published amino acid sequences of the Rep proteins specified by pTA1060, pLS11, and pBS2, all belonging to the pTA1060 group, seem to deviate in their termini (Fig. 2A). These differences can, at least in part, be attributed to sequence errors. The 3′-end of rep (pBS2), compared to the corresponding region of rep (pTA1060), contains three differences (one additional C at position 1943, an A/T substitution at position 1985, and a deletion of one T at position 2011 ), which are likely to result from sequence errors in pBS2. Even more variation seems to be present in the N-termini of these proteins. The differences between pLS11, pBS2 and pTA1060 in this region can be traced back to a few basepairs. Repeated sequencing gives us confidence that the pTA1060 sequence is correct. Introduction of the corresponding corrections in pBS2 and pLS11 results in N-termini of their Rep proteins identical to that of Rep (pTA1060). The published sequence of the Rep protein of pBAA1 (pTA1020 group) is at its N-terminus more than 30 amino acids shorter than the Rep proteins of pTA1060, pTA1015, pTA1040, and pFTB14 . However, the reading frame of rep (pBAA1) is open upstream of its published ATG start codon. In fact, in this region the pBAA1 sequence differs at only one position from that of pTA1060. It is, therefore, likely that also the Rep protein specified by pBAA1 is identical to the other Rep proteins discussed here. Similarly, the published sequence of the Rep protein of pUH1, belonging to the pTA1015 group, deviates at the N-terminus from that of pTA1015. Also in this case the differences can be traced back to a few basepairs. Since the DNA sequences of pTA1015 and pUH1 are 99.3% identical over 5807 bp, we consider it likely that these differences also result from sequence errors and that the Rep proteins of these plasmids are identical.
A consequence of the corrections following from these considerations is that the actual size of the Rep proteins encoded by pLS11, pBS2, pBAA1, and pUH1 is larger than their published sequence (all contain from 339 to 341 amino acids). Additional support for the corrections is the observation that in all cases a potential ribosomal binding site (RBS) is located upstream of the proposed new start codons. Another implication is that only the rep genes of pTA1015, pUH1 and pFTB14 have an ATG start codon; the others have TTG as start codon.
The fact that the various Rep proteins encoded by plasmids of different groups are highly homologous to each other suggests that these plasmids are derived from a common ancestor. The dendrogram (Fig. 2B) shows that the Rep proteins of pBAA1 and pTA1060 are the most closely related. These plasmids also contain an almost identical gene downstream of their rep gene (see Section 10). Among the Rep proteins compared in Fig. 2B, those of pTA1040 and pTA1050 deviate most from the others and these can be grouped separately. In agreement with this classification into a subgroup is the observation that pTA1040 and pTA1050 contain a palT2-type SSO (see Section 5), whereas the other plasmids (the SSO of pFTB14 is not known) contain a palT1-type SSO .
Ilyina and Koonin compared a considerable number of replication initiation proteins required for rolling-circle replication from bacteria, eukarya and archaea. This revealed three conserved amino acid sequence motifs which are also conserved in the deduced sequences of the Rep proteins of the four pTA plasmids and the other plasmids shown in Fig. 2A. The middle motif (II) consists of the sequence His-Hy-His-Hy-Hy-Hy (His and Hy representing a histidine and a bulky hydrophobic residue). Based on analogies with metalloenzymes, the authors hypothesized that the two conserved histidine residues may be involved in the binding of metal ions required for Rep activity. The tyrosine residue present in the C-terminal motif (III) corresponds to a tyrosine residue in the Rep proteins of the RCR plasmids pT181 and pC221. This Tyr residue has been shown to become covalently linked to the nicked DNA at the DSO . Gros et al. showed that at least three amino acid residues of the pC194 replication protein have a catalytic role. One of these is the conserved tyrosine residue in motif III; the other two are glutamates. The corresponding amino acids (indicated in Fig. 2A) are, except for one glutamate in pLS11, conserved in all plasmids included in the present comparison.
Murai et al. delineated the promoter region of the pFTB14 rep gene to about 95 bp. Within this region, sequences were identified resembling the −35 and −10 consensus sequences of B. subtilisσA promoters. Highly homologous sequences are present upstream of the rep genes of the four pTA plasmids and the other listed plasmids (Fig. 2C), indicating that the expression of these rep genes is driven by highly homologous promoters.
Characteristically, DSOs are present in a plasmid region that has the potential to form secondary structures. The sequence 5′-TCTTGATA-3′ is found at the original nick site of most members of the pC194 group of plasmids [1, 46]. As shown in Fig. 3, this consensus nick-site sequence is also conserved in the four pTA plasmids and present within a region of dyad symmetry located upstream (about 160 to 170 bp) from the start codon of its cognate rep gene.
4.1.4Minimal replication regions
The region containing the rep gene and its cognate DSO is sufficient to drive replication of plasmids pUH1 , pFTB14 , pLS11 , and pBAA1 . Similarly, the replication functions of pTA1060, present on a 2.2-kbp HindIII fragment , which contains the rep gene and its DSO (Fig. 1), are sufficient for replication. Likewise, the 2.4-kbp PstI region of pTA1015, containing its rep gene and DSO (Fig. 1), is sufficient to drive autonomous replication in B. subtilis. As described in more detail elsewhere (Bron et al., J. Biotechnol., in preparation) the replication region of pTA1060 was used to construct the versatile B. subtilis/E. coli shuttle vector pHB201, which can be obtained from the Bacillus Genetic Stock Center (Dr. R. Zeigler, Ohio State University, Department of Biochemistry, 484 West 12th Ave, Columbus, OH; E-mail: email@example.com).
4.2Plasmids within one group are highly homologous and may be identical
As described in Section 2, most B. subtilis plasmids can be classified into 7 groups based on their size and restriction profile. This suggests that plasmids within one group may be highly homologous to each other. Comparison of known DNA sequences of corresponding regions of various plasmids belonging to the same group supports this idea. This is clearly illustrated with pTA1015 and pUH1, the entire sequences of which are known (this work; ). The level of sequence identity of these plasmids is 99.3% over 5807 bp. The few discrepancies between the two plasmid sequences are predominantly located in regions that have the potential to form secondary structures and are likely to be due to sequence errors. Very high levels of DNA homology are also observed when the available DNA sequences of pTA1060, pBS2 and pLS11 (i.e. the rep gene region), all belonging to the pTA1060 group, are compared: pTA1060 and pBS2 are 97.5% identical over 2279 bp; pTA1060 and pLS11 96.9% over 1606 bp; and pBS2 and pLS11 93% over 1606 bp. Again, the differences between these plasmids are mainly located in regions that have the potential to form secondary structures. In addition, DNA sequences upstream of the rep gene of pBS2 and pTA1060 (this work) are almost identical.
These observations show that, at least in the cases discussed here, plasmids within one group are highly homologous and possibly even identical.
Taken together, the data described in this section on replication modules show that the minimal replicons of all B. subtilis/B. amyloliquefaciens plasmids analyzed so far are highly conserved, indicating that they are derived from a common ancestor.
5.1PalT1- and palT2-type SSOs
Efficient conversion of ssDNA into duplex plasmid DNA of RCR plasmids is initiated from specific, non-coding plasmid regions, the SSOs, which have a high potential to form secondary structures. Previously, we have reported the cloning and sequencing of the SSOs of pTA1015, pTA1040, and pTA1060 using a specially designed vector, pWM100 . The results showed that whereas the SSOs of pTA1015 and pTA1060, designated palT1, are almost identical to each other, the SSO of pTA1040, designated palT2, is less homologous (77%; see Fig. 4). Despite the differences between palT1 and palT2 both are highly efficient ssDNA conversion signals in B. subtilis. DNA sequences of SSO-containing regions of the following plasmids have been published: pBAA1 (pTA1020 group; [28, 42]); pLS11 (pTA1060 group; ), and pUH1 (pTA1015 group; ). The SSO of pBAA1 is almost identical to those of pTA1015 and pTA1060 and, therefore, belongs to the palT1-type. As expected, pLS11 (pTA1060 group) and pUH1 (pTA1015 group) also contain an SSO of the palT1-type . To study the nature of the SSOs of pTA1030 and pTA1050, the six representative pTA plasmids were compared in Southern hybridization studies using either palT1 of pTA1015 or palT2 of pTA1040 as a probe under stringent conditions. The results showed that pTA1030 and pTA1050, like pTA1040, contain a palT2-type SSO . This conclusion has now been confirmed by sequencing pTA1050 which showed that pTA1050 contains a palT2-type SSO (Fig. 4). This indicates that all plasmids in the groups represented by pTA1015, pTA1020, and pTA1060 contain the palT1-type SSO, whereas plasmids represented by pTA1040, pTA1030, and pTA1050 contain a palT2-type SSO.
5.2Delineation of palT SSOs
The alignment of the SSO-containing regions of pTA1015, pBAA1 (pTA1020 group), pTA1060, pTA1040, and pTA1050 reveals a highly conserved region of about 200 bp (Fig. 4, coordinates 74 to about 270). This suggests that this region constitutes the functional SSO. This idea is supported by the fact that this region encompasses the experimentally determined minimal sequence required for full SSO activity determined for pTA1060 . The experimentally determined SSO sequences of pLS11 (pTA1060 group; ), and pBAA1 (pTA1020 group; ) are about 30 bp shorter at the 3′-end. Since the contribution of this additional 30-bp conserved region to SSO activity is also relatively low in pTA1060 , it may have been below the level of detection in the other plasmids studied.
Seery et al. identified three motifs (marked as motif 1, 2, and 3 in Fig. 4) which are highly conserved between palT1 of pBAA1 and the putative SSO of pGI2, a plasmid isolated from Bacillus thuringiensis. Madsen et al. noticed that the same motifs were also highly conserved in the SSO of plasmid pTX14-3, another endogenous plasmid from B. thuringiensis. Within motifs 1 and 2, only two basepairs differ between palT1 of pBAA1 and the corresponding sequences of pGI2 and pTX14-3. Interestingly, these bases are conserved between palT2 of pTA1040/pTA1050 and pGI2 and pTX14-3 (bp 94 and 109 in Fig. 4).
Whereas in pTA1015, pTA1020, and pUH1 (pTA1015 group), the SSO and DSO sequences are located in close proximity, these primary and secondary replication origins are separated in pTA1040 and pTA1060 by several kilobasepairs (Fig. 1). We showed that the conversion of ssDNA occurred with high efficiency in pTA1015, pTA1040 and pTA1060 . This corroborates the general view that the position of SSOs relative to their DSOs usually does not affect their functionality.
6Mobilization modules (pTA1015 and pTA1060)
6.1Absence of γ-gtp genes (pTA1015 and pTA1060)
γ-Glutamyl transpeptidase (γ-GTP) catalyzes the hydrolysis of glutathione to glutamic acid and the transfer of the γ-glutamyl group of glutathione to an amino acid or peptide [56, 57]. Hara et al. [9, 33]reported that some B. subtilis strains produce active γ-GTP and claim to have identified a γ-gtp gene on plasmid pUH1 (pTA1015 group ). It is doubtful, however, whether the latter claim is correct. First, the proposed γ-gtp gene on pUH1 has neither significant homology with the B. subtilis SJ138 chromosomal γ-gtp gene , nor with that of E. coli. Second, pTA1015 and pTA1060 encode a putative protein which is almost identical to that of the postulated γ-GTP protein of pUH1. This protein, indicated as Mob in Fig. 1, is required for the conjugative mobilization of pTA1015 and pTA1060 (see Section 6.2).
To prove that the mob genes of pTA15 and pTA1060 do not specify γ-GTP, the pTA1015 derivatives pTAB11A, pTAB11B, pTAB31 and pTAB13 were constructed, which contain either an intact (the first three constructs) or a disrupted mob gene (the latter construct; Fig. 5). In these constructs, genes flanking the mob gene were also inactivated by disruption. These derivatives were introduced into B. subtilis strain BD630 and γ-GTP assays of growing cells were carried out. Various media were tested: Luria broth, minimal medium, sporulation medium and DGYP (described to be optimal for γ-GTP assays ). As a positive control, commercially available γ-GTP was used. Whereas γ-GTP activity was clearly detectable with the commercial enzyme, no γ-GTP activity was detectable in any of the B. subtilis samples tested (cell fractions and culture supernatants were assayed as a function of growth time). Koehler and Thorne have reported that pLS19, another plasmid belonging to the pTA1015 group, was also unable to endow its host with γ-GTP activity. Together, these observations make it highly unlikely that the region of pUH1 analyzed by Hara et al. encodes a γ-GTP enzyme.
6.2Functional mob genes (pTA1015 and pTA1060)
The N-terminal parts of the deduced protein sequences of mob genes in pTA1015 and pTA1060 reveal significant homology with corresponding regions deduced from mob genes present on several RCR plasmids (Fig. 6A). In several cases these genes have been shown to be required for conjugative mobilization [61–64]. To study the function of the pTA1015-located mob, the constructs pTAB11A, pTAB11B, pTAB13 and pTAB31, described in the foregoing section, were assayed for their ability to be mobilized by the conjugative plasmid pLS20. The latter enables co-resident RCR plasmids, like pUB110, pTB913, pBC16, and pLS19 (pTA1015 group), to be transferred to other cells during matings [60, 61]. pTAB11A, pTAB11B, pTAB13, pTAB31, and pUB110 (positive control) were introduced into B. subtilis strain 3335 UM4, which contains pLS20, and the resulting strains were used as donors in matings with the chloramphenicol-resistant B. subtilis strain 1012Cm. The results of the matings (Table 3) demonstrated that pTA1015 derivatives can be mobilized in a process requiring the intact mob gene. A derivative of pTA1060, pBB2, in which the homologous ORF is intact, was also shown to be mobilizable (results not shown). Together, these results show that these ORFs of pTA1015 and pTA1060 are required for mobilization. The corresponding genes were designated mob15 and mob60, respectively. These are the first mobilization genes from endogenous B. subtilis plasmids which have been sequenced and functionally analyzed.
Table 3. Mobilization frequencies
aFrequencies are given as the number of transconjugants per donor cell and represent the mean of five independent experiments. (−) denotes interruption of the ORF by the KmR gene (ORF2c, ORF1), or deletion of an internal fragment from the ORF (mob15).
bInstead of mob15, pUB110 carries another mob gene enabling conjugative transfer.
The results presented in Table 3 show that the frequency of mobilization was significantly higher (almost ten-fold) with pTAB11A (interrupted ORF2C) than with pTAB31 (intact ORF2C). The possibility that readthrough transcription activity from the promoter of ORF2C in pTAB31 interferes with the expression of the convergently transcribed mob gene is unlikely because pTAB11A and pTAB11B, which differ only by the orientation of the cloned KmR gene in ORF2C, showed similar mobilization frequencies. Since neither clear differences in the copy numbers of these plasmids, nor obvious differences in growth rate or number of colony-forming units per OD600 were observed between the various plasmid-containing B. subtilis cultures, it is unlikely that these factors underlie the observed differences in mobilization frequencies. An alternative explanation, which implies a direct effect of the gene product of ORF2C on mobilization frequencies, is described in Section 7.
6.2.1Relatedness of Mob proteins
To obtain insight in the relatedness of the Mob proteins encoded by pTA1015, pTA1060, and other RCR plasmids, their deduced amino acid sequences were compared. The Mob proteins of pTA1015 and pTA1060 are highly homologous to each other (95% identity), and somewhat more distantly related to those of the other groups (see also legend to Fig. 6B). Although the homology between all sequences of the 14 Mob proteins shown here is rather low (19% identity), pairwise identities are considerably higher (legend to Fig. 6B). Also, much higher levels of identity are observed within the first 200 amino acids of these proteins (46% identity), suggesting that this may constitute an important functional region of the proteins. A schematic overview of the relatedness between the various Mob proteins is presented in the dendrogram of Fig. 6B.
6.2.2Conserved modules in Mob proteins
Koonin and Ilyina [47, 65]compared the deduced amino acid sequences of a considerable number of proteins involved in plasmid transfer. This comparison included, in addition to the mobilization proteins encoded by several RCR plasmids of Gram-positive bacteria, also proteins involved in conjugation that are encoded by plasmids from Gram-negative bacteria (including virulence proteins specified by the Agrobacterium tumefaciens pTi plasmid). Interestingly, two of the three conserved amino acid sequence motifs identified by the same authors [47, 66]in the replication initiation proteins of RCR plasmids (motifs II and III in Fig. 2A) were also present in proteins involved in plasmid transfer, albeit in inverse orientation (i.e. the motif containing the putative DNA-binding tyrosine residue of motif III is located N-terminally from motif II). These two motifs are also conserved in Mob15 and Mob60 (Fig. 6A). The observation that replication initiation proteins of RCR plasmids and proteins involved in plasmid transfer share amino acid motifs suggests a functional similarity. Most likely, the basis for this similarity is that in both cases the mechanism of DNA transfer involves a rolling-circle-like mechanism. Recently, an alignment of the regions encompassing motifs II and III present in various Mob proteins was also published by Guzmán and Espinosa .
pE194 [60, 68]and pT181 specify proteins with N-termini showing strong homology with the N-termini of Mob proteins (Fig. 6A). The corresponding genes are involved in site-specific, recA-independent, plasmid cointegrate formation and were named pre (for: plasmid recombination enzyme). Pre proteins act at specific recombination sites (called RSA), which overlap with the −10 promoter sequence of the pre genes . Sequences similar to RSA were detected upstream of all mob/pre genes shown in Fig. 7. Selinger et al. showed that a plasmid containing the RSA site of pUB110 can be mobilized by the pUB110 Mob protein provided in trans. This indicates that, at least in this case, the RSA-like sequence is the target site for the Mob protein. These sites have been called oriT. Recently, Guzmán and Espinosa have demonstrated unambiguously that the Mob protein of the streptococcal plasmid pMV158, MobM, attacks the 5′-GpT-3-dinucleotide within its RSA-like sequence 5′-TAGTGTG/TTA-3′. A list of oriT/RSA sites present on various RCR plasmids has also been provided in that paper .
Analysis of pTA1015 and pTA1060 revealed sequences that are highly homologous to known oriT/RSA sites at a position 57 bp upstream of the potential start codons of the mob genes. These sequences, which are identical in these plasmids, are likely to constitute the target sites for the pTA1015- and pTA1060-encoded Mob proteins. Like other oriT/RSA sites [63, 69, 70], the putative oriT sites of pTA1015 and pTA1060 contain an inverted repeat (Fig. 7). The stem of the possible stem-loop structure of RSA/oriT sites varies from 7 to 10 nucleotides and the loop usually consists of 6 nucleotides. Based on the complementary nucleotide pairs located at the third position relative to the loop sequence, the RSA/oriT sites can be classified in the three groups indicated in Fig. 7. Although direct evidence is lacking, it is conceivable that these differences are important for the correct recognition by the corresponding Mob proteins.
Since sequences highly homologous to oriT/RSA sites are also present on plasmids pT48 , pNE131 , and pFX2 , which do not contain mob or pre genes, we screened pTA1040 and pTA1050 (which, likewise, do not contain a mob gene) for the presence of sequences homologous to oriT/RSA. Such sequences were not identified on these plasmids.
The high level of homology in the Mob-encoding regions and oriT of pTA1015 and pTA1060 extends for approximately 150 bp further upstream of the mob genes. These homologous sequences have the potential to form secondary structures, but a possible function of these regions is not known.
7ORF2C modules (pTA1015, pTA1040 and pTA1060)
Except pTA1050, the pTA plasmids contain a highly homologous ORF2C with an orientation of transcription opposite to that of most other ORFs/genes on these plasmids (Fig. 1). In pTA1015 and pTA1060, the stop codons of the convergently oriented mob gene and ORF2C are separated by only 10 and 21 basepairs, respectively. These short intergenic regions are part of sequences that have the potential to form secondary structures which could function as transcriptional terminators. An alignment of the deduced ORF2C protein sequences is shown in Fig. 8. A potential DNA-binding helix-turn-helix motif is present in the N-terminal part of these proteins. Such motifs are often present in bacterial transcription regulator proteins. Since in at least two of these plasmids (pTA1015 and pTA1060) ORF2C is located adjacent to the mob genes, and interruption of ORF2C of pTA1015 resulted in increased mobilization frequencies (Table 3), we considered the possibility that the ORF2C products might be negative regulators of mob gene expression. Support for this idea was not obtained, however, from an analysis on the effects of intact and interrupted ORF2C on β-galactosidase expression driven by the promoter of the mob gene of pTA1015 (results not shown). Interruption of ORF2C had no clear effect on plasmid copy number or plasmid maintenance (results not shown). Therefore, the biological function of the ORF2C remains obscure. In pTA1040 ORF2C is convergently transcribed with the ORF1/sipP module (see Section 8). We have no indications that the expression of this module is affected by ORF2C.
8ORF1/Sip-P modules (pTA1015 and pTA1040)
8.1Orf1 encodes a putative export protein
Plasmids pTA1015 and pTA1040 contain a homologous region encompassing ORF1 and sipP. An alignment of the deduced ORF1 products, which show 56% identity, is presented in Fig. 9A. Since they contain a putative signal peptide with a positively charged n-region, a hydrophobic h-region, and a c-region containing several potential type I signal peptidase (SPase I) cleavage sites, it is likely that these products are export proteins. These proteins show no similarity to previously identified proteins and they have a remarkably high calculated isoelectric point (10.4 and 10.7, respectively). Despite intensive searches, we have not been able to detect ORF1-encoded proteins in the growth medium, cytoplasmic membranes, or unfractionated cellular extracts of B. subtilis cells harboring pTA1015 or pTA1040. Consequently, the biological function of ORF1-encoded proteins remains obscure.
8.2Analysis of SipP
Interestingly, downstream of ORF1, pTA1015 and pTA1040 another homologous ORF is present , which is highly homologous to the chromosomally-located sipS gene of B. subtilis [denoted: sipS (Bsu)], which encodes a type I signal peptidase (SPase, ). Type I SPases are membrane-bound proteases which specifically remove signal peptides of export proteins. For reviews on protein secretion in B. subtilis we refer to Simonen and Palva ; van Dijl et al. ; and Nagarajan . SPases have been reviewed by von Heijne [79, 80].
An alignment between SipS (Bsu) and the homologous products of pTA1015 and pTA1040 is shown in Fig. 9B. It should be realized that, in addition to SipS and the two plasmid-specified SPases, at least four other type I SPases, SipT, SipV, SipW and SipU , are encoded by chromosomal B. subtilis genes. All these SPases are related but slightly different from each other .
Functional activity of the pTA1015- and pTA1040-encoded SPase homologues was demonstrated in vivo, in both B. subtilis and E. coli, using a hybrid β-lactamase precursor that is processed by SipS(Bsu) but not by its homologue, the leader peptidase of E. coli[74, 75, 83]. The plasmid-located genes were named sipP (pTA1015) and sipP (pTA1040); sipP stands for signal peptidase of plasmid.
Several features are shared by sipS and the plasmid-encoded sipP genes. First, all three sip genes possess an atypical start codon (TTG for sipS and sipP [pTA1015], and GTG for sipP [pTA1040]). Second, no sequences corresponding to the major classes of B. subtilis promoters could be detected upstream of their RBS. And third, inverted repeats are present downstream of all three sip genes that are potential rho-independent transcriptional terminators. The alignment presented in Fig. 9B shows that also on the amino acid level SipS and the plasmid-encoded SipP proteins have several features in common. All three SPases possess only one putative membrane spanning domain (indicated as ‘H2’ in Fig. 9B). In this respect, the Sip proteins are distinct from the leader peptidases of the Gram-negative bacteria Pseudomonas fluorescens, Salmonella typhimurium, and E. coli[84–86], which possess two membrane spanning domains and are considerably larger (284, 324 and 323 amino acids, respectively) than SipS, SipP (pTA1015) and SipP (pTA1040).
Patterns of conserved amino acids are present in all known type I SPases of prokaryotic and eukaryotic origin [74, 75, 77, 87, 88]. These patterns are also conserved in the plasmid-encoded SPases (Fig. 9B). Finally, in SipS only three out of 30 amino acid residues which are conserved within all known type I SPases are important for functionality . These amino acid residues are also conserved in the two SipP enzymes (Fig. 9B).
8.3Possible function of the ORF1/sipP modules
In pTA1015 and pTA1040 the homologous region comprising ORF1 and sipP extends from about 100 bp upstream of the start codon of ORF1 to 10 bp downstream of the stop codon of sipP (overall identity: 68%). The intergenic region between these genes is short (102 bp for pTA1015 and 124 bp for pTA1040) and contains an inverted repeat. Moreover, single transcription products have been identified from which both the ORF1 product and SipP can be translated, indicating that these genes are organized in an operon (H. Tjalsma, S. Bron, J.M. van Dijl, unpublished results). Together, these findings suggest that ORF1 and sipP form one structural module on pTA1015 and pTA1040 and that a functional relationship between the products of these genes may exist. Southern hybridization showed that pTA1020 is also likely to possess a sip gene but, based on Southern hybridization and PCR analyses, none of the plasmid-located sip genes are present on the B. subtilis chromosome (results not shown).
The fact that the ORF1/sipP module is present on several of the plasmids tested, suggests that this module may occur rather frequently on RCR plasmids of B. subtilis and may provide the cell with a specific advantageous trait. We can only speculate about this possible function. The pTA series of plasmids are all isolated from B. subtilis strains that are used for the production of natto, a traditional japanese food product based on fermented soybeans. This process requires the production of large amounts of extracellular polymers, such as polyglutamate and levan. In particular plasmids of the pTA1015 group have been implicated in this process . Since the ORF1-encoded proteins are likely to be exported from the cytoplasm, it is conceivable that these are directly or indirectly involved in the production of the extracellular polymers. A possible function of the plasmid-encoded SPases might be the processing of the ORF1-specified product, which may be a preferred substrate. In fact, substrate preference by SPases of B. subtilis does occur, as was recently demonstrated for the chromosomally-encoded SipS of this bacterium [81, 90].
9Rap modules (pTA1040, pTA1050, and pTA1060)
9.1Rap family of proteins
pTA1040, pTA1050, and pTA1060 contain an ORF showing significant homology to the chromosomally-located rapA and rapB genes of B. subtilis[91–93]. In the deduced amino acid sequences, the levels of identity amount to about 40%. In addition, we have identified a gene homologous to rapA/rapB on yet another plasmid from B. subtilis, pLS20, which replicates according to the theta mechanism . We denote these genes from pTA1040, pTA1050, and pTA1060 as rap40, rap50 and rap60.
The chromosomal rapA gene was initially called gsiA (glucose starvation induced protein A; [94, 95]). Expression of this gene delays the onset of sporulation , and its gene product is a protein-aspartate phosphatase acting on phosphorylated SpoOF (SpoOF∼P), a response regulator component of the phosphorelay system [92, 93]. Hence, the gene was renamed rapA (response regulator aspartate phosphatase protein A). Dephosphorylation of SpoOF∼P prevents the transfer of phosphate to SpoOA and lowers the level of SpoOA∼P, which is a key factor in the induction of sporulation . Like RapA, RapB is a protein-aspartate phosphatase active on SpoOF∼P and it also delays the onset of sporulation .
Within the framework of the B. subtilis genome sequencing project , nine additional members of this gene family were recently identified [91, 92, 97, 98], including rapC, rapD, rapE, rapF and rapG. An alignment of the deduced protein sequences of the eleven currently known Rap proteins is shown in Fig. 10A, and a dendrogram showing their relatedness in Fig. 10B. The RapA and RapB proteins are clearly highly related to each other (about 50% identity), as are the three plasmid-encoded proteins Rap40, Rap50, and Rap60 (pairwise identities 50% to 60%). The group of plasmid-specified Rap proteins showed the highest similarities with RapA and RapB from the chromosome (about 40% identity). The other chromosomally specified Rap proteins show pairwise levels of identity ranging from 20% (RapD–RapE; RapD–RapG) to about 40% (examples: RapB–RapC; RapB–RapE; RapC–RapE).
9.2Expression and function of Rap genes
Although, so far, phosphatase activity has only been demonstrated for RapA and RapB, it is tempting to speculate that all Rap proteins belong to a large family of phosphatases which may be expressed under different physiological conditions and may act on different targets. In addition to the demonstrated effects of rapA and rapB on the initiation of sporulation, rapC has been shown to act as a negative regulator of srfA expression , which is required for the initiation of competence development in B. subtilis. Conceivably, other members of the Rap family also have roles in the signal transduction pathways leading to, for instance, chemotaxis, competence, synthesis of secretory proteins, synthesis of antibiotic peptides, and sporulation. The plasmid location, resulting in increased copy numbers, of some of these genes may provide the cell with an additional tool for controlling these global gene expression systems, and extend its adaptative capacity to react to changing conditions.
Evidence that different rap genes, including the plasmid-located rap40, are expressed under different physiological conditions is already available. Whereas rapA is controlled by the ComP–ComA two-component signal transduction system , rapB seems to be regulated by the AbrB transition state regulator [92, 100]. Therefore, the synthesis of these proteins is likely to be induced under different physiological conditions. Analysis in our group showed that the regulation of rap40 expression differs from that of the chromosomal rapA gene: whereas rapA is induced under starvation conditions, in particular glucose deprivation [94, 95], similar conditions did not induce the rap40 gene. Moreover, maximal expression of this plasmid-located gene occurred at the transition between the exponential and stationary phase, approximately 2 h before maximal expression of rapA.
In response to nutrient exhaustion, several developmental programs such as development of genetic competence, motility and chemotaxis, degradative enzyme synthesis, antibiotic production and sporulation can be initiated by the B. subtilis cell. The various interconnected programs are generally regulated by so-called two-component regulatory systems consisting of a histidine protein kinase and a cognate response regulator. Histidine kinases can undergo autophosphorylation by certain environmental or intracellular stimuli. Subsequently, the phosphate group is transferred to the cognate response regulator which, in many cases, modulates the activity of that protein to a transcriptional activator or repressor (for review see ). For the development of genetic competence the ComP and ComA proteins are important as histidine kinase and response regulator, respectively .
Like the expression of rapA, expression of rapC also depends on ComA (mentioned in reference ), and a putative ComA∼P binding site (ComA-box) was identified upstream of the rapA gene . Analysis of the DNA sequences upstream of other rap-like genes revealed sequences homologous to ComA∼P binding sites for rap40, rap50, rap60, rapE and rapF.
An alignment of putative ComA boxes is shown in Fig. 11. The presence of ComA boxes upstream of the other mentioned rap genes suggests that they are also under control of the ComA–ComP system. In addition, we noticed that the consensus SpoOA∼P recognition sequence, 5′-TGNCGAA-3′, is present just upstream of the putative RBS of rap40 and thus this gene may be under the control of SpoOA∼P. This so-called ‘OA box’ is not present upstream of rap60 or rap50.
9.4Translationally coupled small genes
A small ORF, called gsiAB by Mueller et al. , which could encode a peptide of 44 amino acids, is located at the 3′-end of the rapA gene. Expression of this ORF is translationally coupled to that of rapA. Small translationally-coupled putative genes are also present at the 3′-ends of the members of the rap family shown in Fig. 12, including those on pTA1040, pTA1050, and pTA1060. The deduced protein sequences of these genes contain a potential signal peptide with appropriate SPase I cleavage sites (the small peptide associated with rapB could be an exception, since no clear SPase I cleavage site was found in this case).
The presence of potential signal peptides in most of the small gene products indicates that these are secreted into the medium. Support for this hypothesis has recently been obtained for the products of the small genes at the 3′-ends of rapAand rapC. In the case of rapA it was shown that RapA phosphatase activity is negatively controlled by the product of the small gene at its 3′-end, and that the control requires the uptake from the medium via the oligopeptide transport system of at least part of the cognate peptide (minimally the last six amino acid residues from the C-terminus of the peptide are required [91, 93]). The rapA-associated small gene was, therefore, renamed phrA (for phosphatase regulator). As with RapA, the activity of RapC (negative regulator of srfA) is antagonized by expression of the small gene located at its 3′-end and, by analogy, this gene was named phrC. The phrC gene appeared to specify a precursor of the extracellular CSF (competence and sporulation stimulating factor), which serves as a cell density signal for both competence development and sporulation . CSF turned out to be a pentapeptide, corresponding to the carboxy terminus of PhrC which, like PhrA, requires the SpoOK oligopeptide permease for its uptake by the cell .
Although no clear homology exists between the deduced amino acid sequences of the various phr gene products, an ‘AxRxxT’ motif was noticed by Perego and Hoch in the C-terminal ends of PhrA, PhrC, and PhrE. As indicated in Fig. 12 this motif is also present in Phr40, Phr50 and Phr60 and can be more accurately described as an ‘ASRxAT’ motif in the latter cases. In addition, a different C-terminal motif, ‘QxxVAQxx[R/K]GMx’, may be present in PhrF and PhrLS20. Since the conserved motifs in PhrA and PhrC are of functional significance [91, 99], it can be conceived that Phr peptides sharing a certain motif can down-regulate the activity of one group of Rap proteins, and peptides sharing another motif the activity of other Rap protein(s).
Intriguing questions are why rap/phr modules are located on several B. subtilis plasmids, and why the Phr peptides are secreted. Since the rap/phr modules appear to play important roles in global gene regulation in B. subtilis, in particular of regulons that are expressed in the post-exponential growth phase, the increase in copy number of these genes once they are plasmid-located may well extend the cells' ability to adapt to certain conditions. Concerning the secretion of the PhrA peptide, Perego and Hoch proposed that these could serve as ‘quorum sensors’. They envisage that if only a small portion of the total cell population produces the peptide, the internal peptide concentration in producing cells will be too low to down-regulate the Rap phosphatase and sporulation will not be initiated. However, if a substantial fraction of the cells produces the peptide, it will be secreted and re-internalized by other cells which can then prevent the synthesis or activity of the Rap phosphatase, resulting in the initiation of sporulation. The peptide could thus be considered to form a check point for sporulation .
10Unique modules (pTA1040, pTA1050 and pTA1060)
Insertional mutagenesis on pPOD2000 (=pTA1050) implicated that the parABC region is involved in segregational stability . The predicted products of all three ORFs have signal peptides. ParA has significant similarity to the serine protease subtilisin, while ParB has a probable transmembrane domain. ParC appears to regulate the effect of these polypeptides at a level different from transcription or translation. It may thus provide some sort of immunity function . It is possible that the genes act as a post-segregational killing system acting against plasmid-free bacteria.
Downstream of rep(pTA1040) ORF3C40 is located which could encode a protein of 201 amino acids. The rep gene and ORF3C40 are convergently transcribed and the reading frames are separated by 39 bp in a region showing dyad symmetry. No significant homology of the deduced protein product was found with other proteins, but it contains a putative signal peptide, indicating that it is secreted into the medium. The deduced amino acid sequence, together with the potential SPase cleavage sites, is shown in Fig. 13A.
As in pTA1040, in pTA1060 an ORF (ORF7C60) is located downstream of and convergently transcribed with the rep gene. This ORF could encode a protein of 180 amino acids. Although in the two plasmids these ORFs and rep have a very similar structural organization, including a short intergenic region, the deduced protein sequences of ORF3C40 and ORF7C60 show little homology. The only similarity is that the ORF7C60 product also contains a putative signal peptide. Unlike the putative ORF3C40 product, that of ORF7C60 has a potential transmembrane spanning segment (amino acids 95 to 115), the topology of which is likely to be such that its N-terminal part is located at the outer side of the membrane and the C-terminal part in the cytoplasm (according to the positive-inside-out-rule of Claros and von Heijne ). No significant homology was found with other protein sequences. However, an ORF is present downstream of the rep gene of pBAA1, the deduced amino acid sequence of which is almost identical to the C-terminal part of the ORF7C60 product (further sequence information is not available for pBAA1). As in pTA1060, in pBAA1 the directions of transcription of this ORF and rep are convergent and in both plasmids a short intergenic region containing dyad symmetry separates the genes. An alignment of the deduced amino acid sequences of the two genes is shown in Fig. 13B. So far, the biological function of ORF7C60 is unknown.
10.4ORF4.60 and ORF5.60 (pTA1060)
pTA1060 contains unique ORFs between the SSO and ORF2C, designated ORF4.60 and ORF5.60. ORF4.60 could encode a protein of 305 amino acids and ORF5.60 a protein of 132 amino acids (Fig. 13C). The reading frames of ORF4.60 and ORF5.60 are separated by only 20 bp on which two possible RBSs for ORF5.60 are located. This organization suggests that the two ORFs form an operon. A potential σA promoter is located 29 bp upstream of the start codon of ORF4.60 (TTGAGT–17 nt–TATGAT). Neither the deduced protein sequence of ORF4.60, nor that of ORF5.60 shows significant homology with sequences available in the databases. Until now, the biological function of these putative genes is unknown.
Until recently, available data for small endogenous B. subtilis plasmids was limited to size, restriction maps and DNA sequences of specific regions. Now, the DNA sequences of the regions encompassing the DSO and cognate rep gene of at least one member of the following groups of B. subtilis are known: pTA1015, pTA1020, pTA1040, pTA1050, pTA1060 and pFTB14. All plasmids in these groups belong to the pC194 family of RCR plasmids.
The six representative pTA plasmids contain either a palT1- (pTA1015, pTA1020, and pTA1060) or a palT2-type SSO (pTA1030, pTA1040 and pTA1050). The similarities in the DNA sequences involved in replication indicate that all the plasmids described here are highly related to each other, and are likely to be derived from a common ancestor. Like several other RCR plasmids, pTA1015 and pTA1060 contain a mob gene and its putative target sequence (oriT), enabling the conjugative transfer of the plasmids.
In addition to the replication and mobilization modules, other interesting modules were identified on pTA1015, pTA1040, pTA1050, and pTA1060. These may be unique for RCR plasmids from B. subtilis since analogous modules have not been identified on RCR plasmids from other Gram-positive bacteria. At least three of these modules contain genes which are homologous to chromosomally-located genes. One of these, sip, present on pTA1015 and pTA1040, encodes a functional type I SPase. A second gene, rap, present on pTA1040, pTA1050, and pTA1060, is likely to encode a phosphatase involved in the regulation of stationary phase processes, and a third gene, parA, present on pTA1050, has significant similarity to the serine protease subtilisin and seems to be involved in the stability of the plasmid. Although the actual function of these modules and the reason why they are located on plasmids is, at present, not fully clear, we consider it likely that these modules extend the possibilities of the host to adapt to specific conditions. Since the B. subtilis strains harboring these plasmids were industrial isolates, selection for specific (industrial) traits may have been the driving force for the establishment of these genes on plasmids.
In conclusion, the systematic sequence analyses of four cryptic RCR plasmids of B. subtilis presented here, has proven to be a valuable approach for increasing our knowledge of RCR plasmids in general and of B. subtilis RCR plasmids in particular. These studies provided detailed insight in the structural organization and relatedness of these plasmids and allowed a classification of many B. subtilis RCR plasmids.
We thank Henk Mulder for preparing the figures. We are indebted to Leendert Hamoen, Bert-Jan Haijema, Rob Meima, Jan Maarten van Dijl, Albert Bolhuis, Harold Tjalsma, and Steven de Jong for helpful discussions. Philippe Glaser, Marta Perego, and James Hoch provided us with information on the rap/phr genes prior to publication. W.J.J.M. was supported by STW (Stichting Technische Wetenschappen, The Netherlands) and Gist-Brocades BV (Delft, The Netherlands). S.B. was supported by CEU Grants BRIDGE Program BIOT-CT910268 and BIO2 CT93 0254. P.B.T. was supported by the UK Biotechnology and Biological Sciences Research Council Grant GR/J62838.