Staphylococcus aureus multiresistance plasmid pSK41: analysis of the replication region, initiator protein binding and antisense RNA regulation

Authors


Summary

The vast majority of large staphylococcal plasmids characterized to date appear to possess an evolutionarily common replication system, which has clearly had a major impact on the evolution of antimicrobial resistant staphylococci worldwide. Related systems have also been found in plasmids from other Gram-positive genera, including enterococci, streptococci and bacilli. The 46.4 kb plasmid pSK41 is the prototype of a family of conjugative staphylococcal multiresistance plasmids. The replication region of pSK41 encodes a protein product, Rep, which was shown to be essential for replication; mutations that truncated Rep could be complemented in trans. Rep was found to bind in vitro to four tandem repeat sequences located centrally within the rep coding region. An A + T-rich inverted repeat sequence upstream of rep was required for efficient replication, whereas no sequences downstream of rep were necessary. An antisense countertranscript, RNAI, encoded upstream of rep was identified and transcriptional start points for both RNAI and the rep-mRNA were defined.

Introduction

Plasmids replicate independently from the chromosome and have a small segment of DNA (1–3 kb), the replication region, that contains the genetic information required for replication and its control (Helinski et al., 1996; del Solar et al., 1998). In general, plasmid replication regions contain at least one gene that gives rise to a trans-acting replication initiation protein (Rep), an origin of replication (oriV ) that contains binding sites for Rep, and a negative control element that may be in the form of iterons, an antisense RNA molecule or an antisense RNA molecule and a protein (Helinski et al., 1996; del Solar et al., 1998; del Solar and Espinosa, 2000). In plasmids with iterons, the Rep protein interacts with iterons in the oriV to initiate replication. The iterons also have a role in replication control; when the number of plasmid molecules is greater than the characteristic copy number, then plasmids preferentially pair at the iterons through Rep–Rep protein interactions (‘handcuffing’), which is thought to sterically hinder replication initiation (Pal and Chattoraj, 1988; McEachern et al., 1989; Kittel and Helinski, 1992).

Antisense RNAs regulate plasmid replication by forming an RNA-RNA duplex with the target RNA, which may be a preprimer essential for replication as in plasmid ColE1 (Tomizawa and Itoh, 1981), or the 5′-leader region of the essential rep mRNA. Binding of the antisense RNAs to the leader region of the rep mRNAs can result in either transcriptional attenuation as observed for plasmids pT181 (Novick et al., 1989), pIP501 (Brantl et al., 1993) and pAMβ1 (Le Chatelier et al., 1996) or inhibition of Rep translation. The latter control mechanism can be further divided into three modes of action: (i) inhibition of translation of a leader peptide as in plasmid R1 (Blomberg et al., 1992); (ii) inhibition of both translation of a leader peptide and formation of an activator pseudoknot as in plasmids ColIb-P9 (Asano et al., 1991; Asano and Mizobuchi, 1998), IncB plasmid pMU720 (Siemering et al., 1994; Wilson et al., 1994) and the IncL/M plasmid pMU604 (Athanasopoulos et al., 1999), and (iii) direct inhibition of translation of the essential rep gene as proposed for plasmid pLS1 (del Solar et al., 1997).

Strains of Staphylococcus aureus are a major cause of nosocomial infections worldwide and are commonly resistant to drugs used for patient treatment (Paulsen et al., 1997). Staphylococcal resistance determinants are often associated with transposable elements and reside on plasmids (Lyon and Skurray, 1987; Firth and Skurray, 1998; Kuroda et al., 2001). Staphylococcal plasmids range from small rolling-circle replicating (RC) plasmids, which are usually cryptic or encode a single resistance determinant, to larger multiresistance plasmids that are capable of conjugation and replicate by a theta-mechanism (Firth and Skurray, 2000). Staphylococcal multiresistance plasmids are broadly divided into three groups; the β-lactamase-heavy-metal resistance plasmids, the pSK1 family and conjugative pSK41-like plasmids (Paulsen et al., 1997; Firth and Skurray, 2000). The predicted replication initiation proteins from representatives of each of these groups, i.e. plasmids pI9789::Tn552, pSK1 and pSK41, respectively, were found to share considerable amino acid sequence homology (44% to 70% identity) (Firth et al., 2000). Indeed, nearly all staphylococcal plasmids larger than 15 kb characterized to date, including those identified in the genome sequences of the S. aureus strains N315 and Mu50 (Kuroda et al., 2001), encode homologous replication proteins. In contrast to RC plasmids from this genera, and despite their worldwide impact on the evolution of resistant staphylococci, the replication systems of staphylococcal multiresistance plasmids have not been analysed in detail.

Isolated from S. aureus, pSK41 confers resistance to the aminoglycosides, gentamicin, tobramycin, kanamycin as well as neomycin, and to antiseptics and disinfectants, and is the prototype of a family of structurally similar staphylococcal multiresistance plasmids (Berg et al., 1998). In addition to encoding its own conjugative transfer, pSK41 can mediate the mobilization of other co-resident plasmids (Paulsen et al., 1997; Firth and Skurray, 2000). In this study, we delineate the essential elements required for pSK41 replication and show that the Rep protein binds to tandem direct repeats (Rep boxes) found centrally within the rep coding region. Antisense control of plasmid replication has been highlighted by the identification of a small countertranscript that negatively regulates rep expression.

Results

pSK41 rep is essential for replication and can be complemented in trans

The pSK41 replication region has been localized to a 1.9 kb fragment (Fig. 1), corresponding to nt 12795–14727 of the complete pSK41 genome (GenBank entry AF051917), and was shown to be capable of autonomous replication in S. aureus strain RN4220 (Firth et al., 2000). The pSK41 rep gene is predicted to encode a protein of 319 amino acid residues. Mutations in pSK41 rep were engineered at two individual sites to truncate the Rep protein by introducing premature translational stop codons. The first mutation was located at the 5′ end of the rep gene removing 311 amino acids from the C-terminal end of Rep and the second mutation was located towards the 3′ end of the rep gene removing 38 amino acid residues from the C-terminal end of Rep. Mutations were generated in plasmid pSK5487 (Table 1), which is an E. coli-S. aureus shuttle vector that is dependent upon pSK41 replication functions in S. aureus. The 5′- and 3′-rep mutations resulted in plasmids pSK5488 and pSK5489, respectively, and the integrity of these plasmids was verified by DNA sequencing throughout the entire pSK41 replication region. These plasmids were electroporated into S. aureusRN4220 with pSK5487 used as a positive control. On three attempts neither pSK5488 nor pSK5489 yielded any transformants displaying chloramphenicol resistance, whereas pSK5487 consistently gave rise to high numbers of transformants (>1000 transformants per µg of plasmid DNA), indicating that both of the rep frame-shift mutations successfully truncated the Rep protein, which resulted in non-functional replicons.

Figure 1.

Nucleotide sequence of the pSK41 replication region. The pSK41 replication region was previously shown to be contained within a contiguous 1.9 kb DNA fragment corresponding to nt 12795–14727 of the pSK41 genome (GenBank entry AFO51917; Firth et al., 2000). The restriction sites BamHI and HindIII were incorporated into the fragment by PCR and naturally occurring XbaI restriction sites are also indicated with underlining. The amino acid sequence of the Rep protein is shown below the rep coding region and the potential rep ribosomal binding site (rbs) is underlined. Arrows denote the position of the four Rep boxes as well as other direct repeat and inverted repeat (IR) sequences. The transcriptional start points for rep mRNA and RNAI are indicated (+1) and their respective predicted promoter sequences (−10 and − 35) are boxed. The positions of deletion end-points are indicated by white flags with rightward flags indicating 5′ deletions and leftward flags indicating 3′ deletions. Deletions were generated in plasmid pSK5487 (upright flags), which is a pSK41 E. coli-S. aureus shuttle vector or plasmid pSK5492 (inverted flags; see Fig. 3) in which the pSK41 rep control region is transcriptionally fused to the cat reporter gene.

Table 1. . Bacterial strains and plasmids used in this study.
Strain or plasmidRelevant characteristicsReference or source
  1. ApR CmR NmR TcR EbQaR GmR TmR KmR indicate resistance to the respective antimicrobial agents and CmS indicates sensitivity to Cm; MCS, multiple cloning site; Tra+, conjugative transfer functions. Refer to Fig. 3 for pSK5492 deletion derivatives and to the text for pSK5473 deletion derivatives.

Escherichia coli
DH5αFendA hsdR17 supE44 thi-1 λrecA1 gyrA96 relA1φ80 dLacZΔM15Bethesda Research Laboratories
BL21FompT hsdSB (rB mB) dcm galNovagen
Staphylococcus aureus
RN4220Restrictionless derivative of NCTC 8325–4 Kreiswirth et al. (1983)
CYL316RN4220 harbouring plasmid pYL112Δ19 (CmR) Lee et al. (1991)
SK5491TcR, CmS, CYL316 containing the pSK41 rep gene with transcription being driven from the tetA(K) hybrid promoter inserted into the chromosomal geh gene and cured of plasmid pYL112Δ19This study
SK1660MRSA strain isolated in 1976 Gillespie et al. (1987)
Plasmid
pGEM5ZfApR, α-lac/MCS, pUC18 oriPromega
pRB394ApR, NmR pBR322 ori, pUB110 ori, MCS-promoterless cat gene from pUB112Bruckner (1996)
pSK4146.4 kb conjugative plasmid; NmR EbQaR.GmR TmR KmR Tra+ Berg et al. (1998)
pSK4833ApR, pSK1 replicon cloned into BamHI and HindIII sites of plasmid pWE180 Firth et al. (2000)
pSK5299CmR, source of pC194 cat gene Grkovic et al. (2003)
pSK5413ApR., TcR., pSK41 replication region (nt 12795–14727) and tetA(K) cloned into pUC18 Firth et al. (2000)
pSK5472ApR, the 1.4 kb pSK1 replicon from pSK4833 cloned into the EcoRV site of pGEM5ZfThis study
pSK5473ApR, the 1.9 kb pSK41 replication region (nt 12795–14727) cloned into the EcoRV site of pGEM5ZfThis study
pSK5474ApR, pSK41 rep coding region cloned into the BamHI and HindIII sites of pQE30This study
pSK5475ApR, pSK41 rep coding region cloned into the EcoRI and PstI sites of pTTQ18This study
pSK5483ApR, NmR, pRB394 with the pUB110 replicon replaced by the pSK1 repliconThis study
pSK5487ApR, CmR, the cat gene from pSK5299 cloned into the SalI and SacI sites of pSK5473This study
pSK5488ApR, CmR, pSK5487 with 5′rep mutation (Thr7→Ile, Val8→TAG)This study
pSK5489ApR, CmR, pSK5487 with 3′rep mutation (Lys281→Asn, Val282→TAG)This study
pSK5492ApR, CmR, pSK5483 with Prep (nt 12904–13211) cloned into SphI and KpnI sitesThis study
pTTQ18ApR, E. coli expression vector, pUC18 ori, Ptac Stark (1987)
pQE30ApR, E. coli expression vector, pBR322 ori, T5 promoter/lac operatorQiagen

In order to determine if the rep mutations could be complemented in trans, the strain SK5491 was generated. SK5491 contains the pSK41 rep coding region transcribed from a strong tetA(K) hybrid promoter (Simpson et al., 2000), integrated into the chromosome of S. aureus CYL316 (Lee et al., 1991). Electroporation of pSK5488 and pSK5489 into SK5491 gave rise to chloramphenicol resistant transformants at an approximately 10-fold lower frequency per µg of DNA than that observed with plasmid pSK5487, but nevertheless, plasmid isolations from the resulting transformants confirmed that both the 5′ and 3′rep mutations were complemented in SK5491 allowing their autonomous replication. The lower transformation frequency observed may be attributable to the altered level or unregulated synthesis of Rep protein. However, it is unclear at this time why transformation efficiency was reduced as DNA yields in plasmid isolations suggested no significant differences in the copy numbers of plasmids pSK5487, pSK5488 and pSK5489 replicating in SK5491 (data not shown).

Deletion analysis of the replication region

Nested deletions were generated from both the 5′-end and 3′-end of the replication region (in plasmid pSK5473) to map the minimal region required for pSK41 autonomous replication. To begin with, 5′ deletions were generated by digesting pSK5473 with ApaI and NcoI, treating with exonuclease III and S1 nuclease before religating and transforming into E. coli DH5α. The pC194 cat gene was subsequently cloned into the SacI and SalI restriction sites of deleted plasmids and then plasmids were introduced into S. aureus RN4220 electrocompetent cells to determine whether or not they had maintained the ability to replicate. Electroporation of deleted plasmids was repeated on at least three occasions using pSK5487 as a positive control and plasmids were isolated from the resulting transformants at least once to confirm that the deletion derivative was replicating autonomously. Plasmids containing deletions from the 5′ end of the replication region, up to and including nt 12915 (pSK5487Δ31; Fig. 1) maintained their ability to replicate autonomously. However, plasmids with deletions to nt 12927 (pSK5487Δ52) and beyond, such as pSK5487Δ61 deleted to nt 12965 and pSK5487Δ71 deleted to nt 12980, were found to no longer replicate. The region between nt 12915 and 12927, which results in the loss of replicative ability when deleted, contained the left arm of an A + T-rich inverted repeat sequence (IR-I, Fig. 1). At this point it was not certain whether IR-I formed part of the pSK41 replication origin or whether the region may have housed the rep transcriptional promoter, which if deleted, would also inhibit replication by stopping Rep protein synthesis.

Because pSK5487Δ31 was the smallest 5′-deletion plasmid that maintained the ability to replicate, it was used as starting material for the generation of the 3′ deletions. pSK5487Δ31 was digested with PstI and SpeI and deletions were generated and tested for replication efficiency as described above. As demonstrated with plasmid pSK5487Δ31-45, deletions extending up to nt 14164, which removes all sequences downstream of the rep stop codon (Fig. 1) remained replication proficient, whereas a deletion up to nt 14121 (pSK5487Δ31-48), which deleted 43 nt from the 3′-end of  rep coding region resulted in a non-functional replicon.

In order to identify cis-acting elements essential for replication, pSK5487 deletion derivatives were introduced into SK5491, the S. aureus strain that was able to complement the Rep nonsense mutations. As expected, plasmid pSK5487D31 was found to replicate autonomously in SK5491 but electroporation of plasmids pSK5487D52, pSK5487D61, pSK5487D71 failed to produce any transformants. Thus, IR-I was found to be required for efficient plasmid replication in the presence of Rep, indicating that it was an important cis-acting replication element. From the 3¢-end, a deletion removing 388 bp of the rep coding region (pSK5487D540-18 deleted to nt 13776; Fig. 1) maintained replication proficiency, but a deletion that removed another 15 bp (pSK5487D540-14 deleted to nt 13761; Fig. 1) could no longer replicate as indicated by the repeated failure to produce transformants when electroporated into S. aureus SK5491. In this region, there are several types of iterated sequences that are found centrally within the rep coding region (Fig. 1). Among them, the sequence 5¢-ATATGAAT is directly repeated twice and both repeats are present in pSK5487D540-18 whereas seven out of 8 bp of the downstream repeat is deleted in pSK5487D540-14. This implies that the downstream repeat, or possibly both 5¢-ATATGAAT repeats, are required in cis for pSK41 replication.

Rep binds to iterated sequences found centrally within rep

Directly repeated sequences (iterons) found in plasmid replication regions are often binding sites for the cognate Rep proteins (Helinski et al., 1996; del Solar et al., 1998). To undertake pSK41 Rep:DNA binding studies, the pSK41 Rep protein was overexpressed in E. coli and purified by affinity chromatography. Expression of pSK41 Rep with a C-terminal polyhistidine fusion appeared optimal after 3 h of induction with 1 mM IPTG and the majority of the protein was present in the soluble form. In contrast, pSK41 Rep protein with an N-terminal fusion could not be detected by SDS-PAGE under various growth and induction conditions. The apparent molecular weight of Rep with the C-terminal polyhistidine fusion (hereafter referred to as Rep) was 37.7 kDa (data not shown), close to the theoretical size of 39.2 kDa predicted from the deduced sequence.

In pSK41, several sets of direct repeats have been identified within the rep coding region (Fig. 1; Firth et al., 2000); this repeat region was amplified by PCR using primers SK41R10 and SK41R11 yielding a 263 bp DNA probe, designated P1011 (nt 13535–13797). Electrophoretic mobility shift assays (EMSAs) were conducted with end-labelled P1011 DNA and purified Rep in the presence of 2.0 µg of non-specific poly[dI-dC] competitor DNA. The purified Rep protein was found to bind in vitro to P1011 and the amount of the probe that was retarded appeared to be proportional to the amount of Rep added to the binding reactions (Fig. 2A). The addition of increasing amounts of specific competitor DNA (cold P1011 DNA) to the binding reactions titrated the Rep protein, demonstrating that the observed DNA–protein interaction was specific (Fig. 2B). To investigate the possibility that the Rep protein was involved in its own regulation (by binding to its promoter region), probe P5B09 was used in binding reactions and included the entire region upstream of rep (nt 12795–13296). Rep failed to bind in vitro to probe P5B09 when up to 8 µg of Rep protein was used in binding reactions.

Figure 2.

Identification of pSK41 Rep binding sites.
A. EMSAs were performed with [γ32P]-labelled P1011 DNA (pSK41 nt 13535–13797) and increasing amounts of purified Rep protein (0–8 µg per binding reaction).
B. EMSAs were conducted with a fixed amount of Rep protein (8 µg) and increasing amounts of competitor DNA, shown as the ratio of unlabelled to labelled P1011DNA.
C. DNase I footprint of the interaction between Rep protein and P1011 DNA. The G + A Maxam-Gilbert DNA sequencing ladder is shown on the left. DNA binding reactions were performed in duplicate with 0, 1, 5 and 20 µg of Rep protein after which they were subjected to partial DNase I digestion. The DNase I protected region corresponded to four direct repeat sequences (large arrows) designated Rep box 1 to box 4. Small arrows and outlined letters show bases within the binding sites that are not protected from DNase I nuclease activity.

DNase I footprinting was carried out with P1011 DNA to map more precisely the Rep binding site(s). A relatively large region of the P1011 probe was observed to be protected from DNase I nuclease activity when 5 µg or more of Rep protein had been added to the binding reactions (Fig. 2C). Comparison of the DNase I footprint with the co-electrophoresed G + A Maxam-Gilbert sequencing ladder revealed that the DNase I protected region contained 4 tandem direct repeats, which were designated Rep box 1 to Rep box 4 (Figs 1 and 2C). The consensus of the Rep boxes is a 23 bp sequence, 5′-GWCHAGRAATGTCAAAAGGACAC-3′, although box 1 consists of 24 bp with an additional guanine residue.

Localization of the rep promoter and detection of a negative control element

For rep promoter (Prep) analysis, the sequences upstream of the rep coding region that appeared essential for replication (nt 12904–13211) were cloned upstream of the promoterless cat gene in the promoter probe vector pSK5483 to generate pSK5492 (Fig. 3). pSK5483 and pSK5492 were transformed into RN4220 and chloramphenicol acetyl transferase (CAT) assays were performed on total soluble protein extracts. Plasmid pSK5492 produced on average 38 units of CAT activity, whereas the vector consistently produced less than 2 units of CAT activity. Nested deletions were generated from the 3′-end of the cloned DNA to locate Prep. Deletion plasmids pSK5492Δ4, pSK5492Δ2 and pSK5492Δ1, which had 3′-deletions to nt 13159, 13147 and 13096, respectively, mediated CAT activities comparable to that of pSK5492 (Fig. 3). Deletion plasmids pSK5492Δ46 and pSK5492Δ57, deleted to nt 13077 and 13032, respectively, showed two- and fourfold increases in CAT activity. However, in plasmids with deletion endpoints between nt 13022 and nt 12995 (pSK5492Δ61, pSK5492Δ58, pSK5492Δ7, pSK5492Δ63 and pSK5492Δ65) the promoter activity was observed to increase over 10-fold (Fig. 3). These results suggested that the effect of a negative regulator of rep transcription was lost in plasmids displaying elevated levels of promoter activity. Because there was no pSK41 Rep protein present in strains that were used to conduct CAT assays and negative regulators of plasmid replication are usually plasmid-encoded, it appeared that the negative regulator itself was at least partly located in the deleted region and because of its relatively small size and position was most likely an antisense RNA; this was designated RNAI. Plasmids pSK5492Δ11 and pSK5492Δ17, which were deleted to nt 12948 and nt 12918, respectively, showed a complete loss of promoter activity. Therefore, sequences important for Prep activity were located between pSK41 coordinates 12948 and 12995. The expected promoter for the putative RNAI antisense gene (PRNAI) would be located on the DNA strand complementary to Prep. We attempted to measure the relative promoter strength of PRNAI through CAT assays, but plasmid construction by inversion of the cloned fragment in pSK5492 resulted in plasmid structural instability when ligations were transformed into E. coli DH5α.

Figure 3.

Reporter gene constructs used in the analysis of the pSK41 rep transcriptional promoter (Prep). Prep (pSK41 nt 12 904–13 211) was cloned into pSK5483 upstream from the promoterless cat gene (hatched arrow) resulting in the plasmid pSK5492. The reporter gene constructs are flanked by the E. coli rrnB (T1) and phage λ (To) transcriptional terminators. Deletions from the 3′-end of Prep were generated and the resulting plasmids were assayed for CAT activity (shown as µmol of chloramphenicol acetylated per mg protein per min). Open bars represent the relative sizes of deletions and coordinates show the deletion endpoints determined through DNA sequencing.

Rep does not repress Prepin vivo

Although Rep did not bind to Prep (P5B09 DNA) in vitro, we performed experiments to test whether Rep influenced Prep activity in vivo. Plasmid pSK5492 and the deletion derivatives listed in Fig. 3 were electroporated into S. aureus SK5491 and CAT assays were performed with the resulting strains. No significant differences in CAT activities were observed between corresponding SK5491 and RN4220 derivatives, when harbouring pSK5492 or any of the deletion clones listed in Fig. 3. This provided evidence that the Rep protein had no influence over Prep transcriptional activity.

Determination of rep mRNA and RNAI transcriptional start points

The precise location of the rep transcriptional start point (TSP) and the RNAI TSP were identified through primer extension analysis. Primers PEX3 and PEX5 were designed to anneal to two different positions of the rep mRNA, whereas primer PEX7 was complementary to the predicted RNAI. Primer extension, conducted using primers PEX3 and PEX5 with total RNA extracted from RN4220 harbouring either pSK41 or pSK5487, did not result in any detectable primer extension products (data not shown). In contrast, a readily identifiable extension product was obtained using primer PEX7 and total RNA isolated from RN4220 harbouring plasmid pSK5487. When compared to a DNA sequencing ladder, also generated using PEX7, the RNAI TSP was determined to be located at the adenine (A) residue at position 13049 on the antisense DNA strand (Fig. 4A).

Figure 4.

Identification of RNAI and rep mRNA transcriptional start points by primer extension. Primers PEX7 and PEX9 were labelled with [γ-32P] and hybridized to total RNA that was isolated from S. aureus RN4220 harbouring plasmids pSK5487 (A) and pSK5492Δ63 (B), respectively. Primer extension was performed with M-MuLV reverse transcriptase and extension products (lane 1; indicated by dark arrow heads) were co-electrophoresed on a denaturing 9% acrylamide sequencing gel along with sequencing reactions (lanes G, A, T and C) that were generated using the same labelled primers. The adenine residues (A) corresponding to the transcript start points (+1) for RNAI and rep mRNA are both boxed with an attached arrow showing the direction of transcription. The predicted transcriptional promoter sequences are also shown with their − 10 and − 35 sequences boxed.

The deletion plasmids pSK5492Δ58, pSK5492Δ61, pSK5492Δ63 and pSK5492Δ65 demonstrated increased levels (10-fold) of CAT activity from Prep and primer extension was performed using total RNA isolated from S. aureus RN4220 carrying each of these plasmids. Primer PEX9, which was complementary to the rep leader-cat mRNA hybrid transcripts produced from each of the deletion plasmids, was used in primer extension reactions. All four of the deletion mutants gave extension products whose difference in size corresponded to the size of the deletion carried by each plasmid (data not shown). Comparison of the pSK5492Δ58- (data not shown) and pSK5492Δ63-derived extension products with their respective DNA sequencing ladders showed that the rep mRNA TSP was located at the adenine residue at position 12965 (Fig. 4B).

Northern hybridization experiments, that employed a double-stranded fragment spanning the region upstream of the rep coding sequence (nt 12904–13211; Fig. 1) to probe RNA isolated from S. aureus cells harbouring the pSK41 minireplicon pSK5413, revealed a transcript of approximately 1.3 kb and a more abundant species of less than 150 nt (data not shown). The first of these is in good agreement with the 1259 nt rep message calculated using the TSP determined here and a predicted transcription terminator located 28 nt downstream of the rep stop codon (Fig. 1, IR-V). Similarly, the smaller RNA molecule detected is consistent with the 82 nt size expected of RNAI if IR-II (Fig. 1) acts as a transcriptional terminator for this transcript. A strand-specific probe complementary to RNAI (primer PEX7) was found to hybridize only to the small transcript, confirming that it corresponds to RNAI (data not shown).

Discussion

This study represents the first detailed description concerning the replication of a large multiresistance plasmid from S. aureus. Related replication systems are evident in the plasmids pSK1, pI9789::Tn552 (Firth et al., 2000), pN315, VRSAp (Kuroda et al., 2001), pIP680 (Allignet and El Solh, 1999) and pSR1 (GenBank accession AAF99572) from S. aureus, and in the plasmids of coagulase-negative species pSX267 from S. xylosis (Gering et al., 1996), pNVH97A (Anthonisen et al., 2002) and pNVH96 (GenBank accession AJ302698) from S. haemolyticus, and pSE-12228–05 from S. epidermidis (GenBank accession AE015934). This type of replication system appears to be quite widely distributed among Gram-positive bacteria as plasmids encoding homologous replication proteins have been identified in enterococci, lactococci, streptococci, bacilli and lactobacilli (Firth et al., 2000).

The essential components of the pSK41 replication region were shown to be contained within a contiguous 1.25 kb segment (Fig. 5; nt 12916–14164). The identified loci include the rep gene, that gives rise to the polypeptide product Rep, and a 0.3 kb upstream region that contains the rep transcriptional promoter (Prep), a regulatory antisense RNA,  RNAI  (and  its promoter, PRNAI), and an A + T-rich inverted repeat sequence, designated IR-I (Fig. 5). Also, found centrally within the rep coding region are four binding sites (Rep boxes) for the Rep protein (Fig. 5). The location of the Rep binding sites in the rep coding region appears to be a common feature among this family of replicons. The replication origins of pSX267, and pLS32 from Bacillus natto, have been shown to reside within their respective rep gene coding regions (Gering et al., 1996; Tanaka and Ogura, 1998), and analogously located direct repeat sequences are evident in these and all other plasmids that possess a rep gene homologous to that of pSK41. The number, length and arrangement of these repeats were found to be variable but conceivably they may also function as binding sites for their respective Rep proteins.

Figure 5.

Overview of pSK41 replication. The genetic map of the 1.9 kb pSK41 replication region flanked by BamHI and HindIII restriction sites with coordinates corresponding to the complete pSK41 genome (GenBank entry AF051917). The rep coding region (long open box) gives rise to the essential Rep protein (shaded oval). Inverted repeat sequences (IR-I, IR-II, IR-III, IR-IV and IR-V) are represented by a stem-loop structure and direct repeats are shown as dark arrow heads. Large arrowheads represent the four Rep boxes and small arrowheads represent the two direct repeats 5′- ATATGAAT that also appear to be essential for replication. The positions of transcriptional promoters (PRNAI and Prep) directing synthesis of the RNAI and rep mRNA transcripts (thick arrows) are also indicated. Shown below are the minimal regions required for replication and required for replication in cis determined by deletion analysis.

Several other types of direct repeats are also present in the region immediately downstream of the pSK41 Rep boxes. For example, the sequence 5′-ATATGAAT-3′ is repeated twice 52 nt downstream of Rep box 4 and deletion of one of these sequences (in plasmid pSK5487Δ540-14) resulted in a non-functional replicon when Rep was provided in trans from the chromosome. Interestingly, the sequence 5′-TGATACTGATTTTAT-3′, which is found 5 nt downstream of Rep box 4 and which is repeated two and a half times in pSK41, may function as a recognition site for a host replication factor. This suggestion is supported by the fact that similar sequences, conforming to the consensus 5′-TGATACTGABTWWAD-3′, are found immediately downstream of the potential Rep boxes in all of the staphylococcal plasmids that carry a pSK41-like replication region. However, in most of these replicons the sequence is present in only one copy.

Although the Rep boxes and the adjacent repeats are likely to constitute the pSK41 replication origin, IR-I may also be involved in replisome formation as it was also shown to be essential for replication when Rep was provided in trans. The sequences which comprise IR-I and Prep overlap, so it was initially thought possible that deletion of IR-I in plasmids pSK5487Δ52, pSK5487Δ61 and pSK5487Δ71 may have simultaneously deleted sequences important for rep promoter activity, which could have been required to supplement the level of Rep produced from the chromosome. However, plasmids that produced no functional Rep protein at all, such as the mutants with truncated Rep proteins (encoded by pSK5488 and pSK5489) or those with 3′-Rep deletions (encoded by pSK5487Δ31-48, pSK5487Δ540-14 and pSK5487Δ540-18), were able to replicate proficiently in the Rep-producing strain SK5491. Furthermore, the deletion in plasmid pSK5487Δ52 removes the left-arm repeat of IR-I but leaves the − 35 promoter sequence of Prep intact (Fig. 1). These observations imply that the A + T-rich IR-I sequence is genuinely an essential replication component required in cis. A + T-rich regions are often found in plasmid replicons and are associated with strand melting of the origin. Possibly IR-I is recruited to the nucleoprotein complex formed at the Rep boxes by Rep itself or with the help of DNA bending proteins analogous to the E. coli IHF or FIS proteins.

The abundance of RNAI in strains containing the pSK41 replicon, in comparison to rep mRNA, detected by Northern analysis, probably explains why an extension product for RNAI was readily obtained in transcript mapping experiments, whereas no product was detected using primers that were complementary to the rep mRNA. The high levels of the longer RNAI would be expected to competitively inhibit annealing of the 20–23 nt primers to their target rep mRNA. Thus, in the presence of RNAI very little of the rep primer extension product would be expected to be synthesized. Indeed, extension products were only detected for rep when RNA templates from strains harbouring plasmids with deletions extending into the RNAI 5′-region were used. In plasmid replication control systems that use regulatory antisense RNAs, the rate of synthesis of the antisense RNA needs to be much higher than that of the essential RNA for efficient regulation because the antisense RNAs are expected to have a short half-life (Brantl and Wagner, 2000; del Solar and Espinosa, 2000). In addition, the antisense RNAs are synthesized from a constitutive promoter so that their abundance reflects the copy number of the plasmid on which they are carried (Wagner and Brantl, 1998).

Most commonly, antisense RNAs act at the post-transcriptional level and modulate Rep protein synthesis by inhibiting translation (del Solar and Espinosa, 2000). This is typified by plasmids, such as R1, that rely on the translation of a short leader peptide, which allows access of the ribosomes to the translation initiation signals of the replication initiator (Blomberg et al., 1992). The initiator transcripts of plasmids ColIb-P9, pMU604 and pMU720 are also to varying degrees dependent upon translation of a leader peptide (Asano et al., 1991; Praszkier et al., 1992; Wilson et al., 1994) but also contain distal sequences that form a long-range RNA tertiary structure (pseudoknot) that actively enhances rep translation. Binding of antisense RNAs to the rep leader predominantly regulates translation of Rep by sequestering bases involved in pseudoknot formation (Siemering et al., 1993; Asano and Mizobuchi, 1998) but also sterically hinders access of the ribosomes to the translation initiation region of the leader peptide. However, in the staphylococcal plasmid pT181, the countertranscripts act at the transcriptional level to induce premature termination (attenuation) of the initiator mRNA by promoting formation of a terminator hairpin just 5′ to the initiator start codon (Novick et al., 1989). A similar transcriptional attenuation mechanism is used to inhibit transcription of the replication initiator in the related streptococcal plasmids pIP501 (Brantl et al., 1993) and pAMβ1 (Le Chatelier et al., 1996). Using transcriptional fusions, we observed a two- to fourfold increase in transcription in plasmids that had PRNAI deleted (pSK5492Δ46 and pSK5492Δ57) and a 10-fold increase in plasmids that had both PRNAI and IR-III deleted (Fig. 3). RNAI is therefore expected to modulate rep expression, and IR-III appears to do likewise, at least in the absence of RNAI. It is conceivable that RNAI is a transcriptional regulator of rep, or acts post-transcriptionally by targeting rep mRNA for degradation. Alternatively, Prep may be subject to transcriptional interference caused by convergent transcription from PRNAI. In plasmids where rep is regulated by an attenuation mechanism, an inverted repeat that functions as a transcriptional terminator in the presence of the antisense RNA is found just 5′ to the rep translational start codon. IR-IV of pSK41 appears in a similar position and is followed by a polyT(5) sequence that may have allowed it to function as a transcriptional terminator. However, this notion was not supported by the observation that deletion of IR-IV did not result in an increase in transcription from Prep (Fig. 3).

The level of RNAI-mediated repression of pSK41 rep expression described here is modest in comparison to that determined for other antisense RNA-regulated replication control systems. In this regard it should be noted that the transcriptional fusions employed in this study would not have allowed detection of RNAI-mediated inhibition of rep translation initiation. Current studies are aimed at addressing this possibility which, if demonstrated, would imply a novel combination of two mechanisms of antisense control in the regulation of pSK41 rep expression.

Experimental procedures

Bacterial strains, plasmids and growth conditions

The relevant characteristics of bacterial strains and plasmids used in this study are listed in Table 1. Bacterial cultures were grown at 37°C in Luria–Bertani (LB) media with aeration or on LB plates containing agar (1.5% w/v). When required, antibiotics were added to the media at the following concentrations: ampicillin 100 µg ml−1, chloramphenicol 7 µg ml−1 for E. coli and 10 µg ml−1 for S. aureus, tetracycline 10 µg ml−1, neomycin 15 µg ml−1.

Staphylococcus aureus strain SK5491 was engineered using the integration vector pCL84 (Lee et al., 1991). pCL84 contains the staphylococcal L54a viral attachment site, attP, and can integrate into the attB site on the host chromosome by site-specific recombination when the L54a integrase is provided from plasmid pYL112Δ19 (CmR). CYL316, which is S. aureus RN4220 harbouring plasmid pYL112Δ19, has an attB site located in the lipase structural gene (geh), therefore L54a lysogens are lipase-negative (Lee and Iandolo, 1985). The promoterless rep coding region was amplified using primers SK41R13 (5′-AACTGCAGTTAAAGGAGTTTTTAT CATG-3′, PstI) and SK41R08 (5′-GCGGATCCAACTTTG CAACAGAACCG-3′, BamHI) such that the resulting PCR product retained the putative rep ribosomal binding site and putative transcriptional terminator sequences. The tetA(K) hybrid promoter (Simpson et al., 2000) was PCR amplified from SK1660 genomic DNA using primers PDFRA-F-(5′-CCCAAGCTTGAATTCATACAGAAGACTCC-3′, EcoRI) and PHYB-R (5′-AACTGCAGGTGACTAAAGTTTATAGG-3′, PstI). The tetA(K) promoter was ligated to the promoterless rep fragment after both had been digested with PstI and the ligation mix was subjected to a second PCR amplification using primers PDFRA-F and SK41R08. The resulting PCR product contained rep downstream of the tetA(K) hybrid promoter and flanked by EcoRI and BamHI restriction sites. This fragment was cloned into pCL84 and verified by sequencing before the plasmid was electroporated into CYL316. Single transformants were then replica plated onto Sierra media (Atlas, 1993) to identify lipase mutants resulting from site-specific recombination of the plasmid into the chromosomal geh gene. The resulting strain was subsequently cured of plasmid pYL112Δ19 by growing the culture for 10 generations with only tetracycline selection and identifying chloramphenicol sensitive colonies, one of which was designated SK5491.

Plasmid pSK5473 contains the 1.9 kb pSK41 replicon flanked by restriction sites suitable for the generation of both 5′- and 3′-nested deletions. The 1.9 kb pSK41 replication region was obtained by digesting plasmid pSK5413 (Firth et al., 2000) with BamHI and HindIII. The gel-purified fragment was then treated with S1 nuclease and ligated to pGEM5Zf that had been cleaved with EcoRV resulting in plasmid pSK5473. One of the plasmid clones that gave the expected restriction profile was sequenced using the universal M13FWD (5′-GTAAAACGACGGCCAGT-3′) and M13REV (5′-GATAACAATTTCACACAGGA-3′) primers. The M13FWD primer was found to sequence into the 5′-end of the pSK41 replication region and the M13REV primer was found to sequence into the 3′-end. To provide a means of plasmid selection in S. aureus, the cat gene from pC194 (Horinouchi and Weisblum, 1982) was cloned between the SacI and SalI restriction sites of pSK5473. The resulting plasmid, pSK5487, could be electroporated into S. aureus RN4220 with relatively high efficiency (>1000 transformants per µg of plasmid DNA) and plasmid isolations demonstrated that pSK5487 existed in the RN4220 transformants as an autonomous replicon. pSK5472 is the same as pSK5473 except it contains the pSK1 replication region instead. The 1.4 kb pSK1 replicon was cleaved from pSK4833 with BamHI and HindIII, blunt-ended and cloned into the EcoRV site of pGEM5Zf.

Plasmid pSK5483 was constructed for promoter analysis in S. aureus and is based on the promoter probe vector pRB394 (Bruckner, 1992). pRB394 contains the promoterless cat gene from plasmid pUB112 downstream from the pUC18 multiple cloning site and these two elements are flanked by transcriptional terminators. pRB394 replicates via a pUB110 replicon in Gram-positive bacteria but exhibits low segregational stability in Bacillus subtilus (Bruckner, 1992). The segregational stability of pRB394 in S. aureus RN4220 was found to be low, with plasmid loss occurring in greater than 90% of cells after 10 generations of growth with neomycin selection. Therefore, the pUB110 replicon was replaced with the pSK1 replicon. Primers AMPR01 (5′-ACTCAAC CAAGTCATTCTGAG-3′) and KANF01 (5′-CATGCCATG GCAATGTGGAATTGGGAACGG-3′, NcoI) were used in PCR with pRB394 DNA and the resulting product was digested with NcoI and ligated to pSK5472 that had been digested with NcoI and ScaI resulting in plasmid pSK5483. In RN4220, pSK5483 exhibited better segregational stability than pRB394, although plasmids were still lost in 30–40% of the population after 10 generations of growth with neomycin selection.

DNA manipulations

Plasmid DNA was purified from E. coli by the alkaline lysis method (Birnboim and Doly, 1979) or with the Quantum Prep plasmid miniprep kit (Bio-Rad). Cloning in E. coli was performed by standard procedures (Sambrook et al., 1989). Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs. Plasmids were introduced into S. aureus strains by electroporation using a Bio-Rad gene pulser (1.3 kV) as described previously (Schenk and Laddaga, 1992). Staphylococcus aureus cells were lysed with 0.2 mg ml−1 lysostaphin (Sigma) in TES buffer (50 mM Tris-Cl, 5 mM EDTA, 50 mM NaCl, pH 8.0) and plasmids were isolated by the method of Lyon et al. (1983) or by using the Quantum Prep kit (Bio-Rad). Genomic DNA was prepared with the AquaPure Genomic DNA kit (Bio-Rad). Nested deletions were generated with exonuclease III and S1 nuclease purchased from AP Biotech. Polymerase chain reaction amplifications and in vitro mutagenesis were carried out with Pfu Turbo DNA polymerase (Stratagene) according to the manufacturer's instructions. Translational stop codons were introduced into pSK41 rep using the complementary mutagenic primer pairs SK41mut01 (5′-GTCTAAACAATTTTTTATCTAGAAGAAAATTATAAAG-3′) with SK41mut02 (5′-CTTTATAATTTTCTTCTAGATAAAAAAT TGTTTAGAC-3′) for the 5′-rep mutation and SK41mut03 (5′-GTTAAAAAACAAGAAAACTAGTAGCTAATATGGAAG-3′) with SK41mut04 (5′-CTTCCATATTAGCTACTAGTTTTCT TGTTTTTTAAC-3′) for the 3′-rep mutation. The 5′-rep mutation converted Thr7 (ACA) of Rep to Ile (ATC) and Val8 (GTA) to an amber stop codon (TAG) and at the same time introduced an XbaI restriction site. The 3′-rep mutation converted Lys281 (AAA) to Asn (AAC) and both Val282 (GTA) and Ala283 (GCT) to amber stop codons (TAG), simultaneously introducing an SpeI restriction site. The resulting plasmids were screened with the appropriate restriction enzymes and then one of each mutant type was chosen for DNA sequencing throughout the entire pSK41 replication region. Plasmids containing the correct mutations were designated pSK5488 for the 5′rep mutation and pSK5489 for the 3′ mutation.

Expression and purification of recombinant proteins

The pSK41 rep coding region was amplified using the primer pairs SK41R01 (5′-GCGGATCCAAACAATTTTTTACAGTAG 3′, BamHI) with SK41R02 (5′-CCCAAGCTTCTAATTAAATA GAGATAATGG-3′, HindIII) and SK41R03 (5-GGAATTC TAAACAATTTTTTACAGTA-3′, EcoRI) with SK41R04 (5′-AACTGCAGGATTAAATAGAGATAATGG-3′, PstI). The former PCR product was cloned into pQE30 (Qiagen) that had been digested with BamHI and HindIII, giving rise to plasmid pSK5474, whereas the latter PCR was cloned into pTTQ18 (Stark, 1987) that was digested with EcoRI and PstI, yielding plasmid pSK5475. pQE30 is a mid-range copy number vector that employs a phage T5 promoter/lac operator for expression of cloned genes whereas pTTQ18 is a high copy number replicon employing a Ptac hybrid promoter. Both expression vectors incorporate a polyhistidine tag into recombinant proteins for purification purposes, with pQE30 producing a N-terminal polyhistidine fusion and pTTQ18 a C-terminal polyhistidine fusion. The integrity of pSK5474 and pSK5475 were verified by DNA sequencing before being transformed into E. coli for analysis of protein expression. pSK5474 was transformed into E. coli BL21 harbouring plasmid pREP4, which represses the T5 promoter/lac operator in the absence of IPTG and pSK5475 was transformed into BL21. Protein expression and purification was performed as previously described (Kwong et al., 2001) and proteins visualized by SDS-PAGE (Sambrook et al., 1989).

CAT assays

Chloramphenicol acetyltransferase (CAT) assays were performed essentially as described previously (Shaw, 1975); however, several modifications were used to adapt the assay to a microplate. Staphylococcus aureus RN4220 containing the plasmid to be tested was grown overnight in 10 ml of LB supplemented with the appropriate antibiotics. The cells from 3 ml of culture were harvested by centrifugation, resuspended in 300 µl of cold WL buffer (25 mM Tris-Cl, 10 mM EDTA, pH 8.0) containing 0.3 mg ml−1 lysostaphin (Sigma) and incubated at 37°C for 30 min. The viscous cell lysates were briefly sonicated on ice (1 × 5 s burst) and insoluble material was pelleted by centrifugation (13 000 r.p.m., 4°C, 30 min). The supernatant was transferred to a fresh microcentrifuge tube and stored on ice. For each sample, 186 µl of CAT assay buffer (100 mM Tris-Cl, pH 7.8, 0.1 mM acetyl CoA and 1 mM 5,5′-dithio-bis[2-nitrobenzoic acid]) and 10 µl of cell extract were allowed to equilibrate for 2 min at 37°C in the plate reader (Bio-Rad, Benchmark) before the assay was started by the addition of 4.0 µl of 5 mM chloramphenicol. The change in absorbance at 415 nm was monitored and CAT activity was calculated as described previously (Shaw, 1975).

DNA-binding experiments

Primers were end-labelled using [γ32P]ATP (Perkin-Elmer) and T4 polynucleotide kinase (NEB) and used directly in PCR amplifications to generate probes, which were purified with Microcon PCR centrifugal filters (Millipore), eluted in water and stored at − 20°C. The P1011 probe (nt 13 535–13 797) was amplified by PCR using primers SK41R10 (5′-GGAATTCTTAAAACCAGCCATAAC-3′, EcoRI) and SK41R11 (5′-CCCAAGCTTTGAATGATTTGAATG-3′, HindIII) and the P5B09 probe (nt 12795–13296) was generated with primers rep 5′BamHI (5′-AATGGATCC ATATAGTTTTTGTATACGGTATTC-3′, BamHI) and SK41R09 (5′-CCCAAGCTTCTTTATAATTAGGATTAG-3′, HindIII).

Electrophoretic mobility shift assays were performed by incubating end-labelled DNA (2000 c.p.m.) with 2 µg of poly[dI-dC] and increasing amounts of purified Rep protein in 1 × binding buffer (10 mM Tris-Cl, pH 7.5, 10 mM MgCl2, 100 mM NaCl2, 0.2 mM DTT, 10% glycerol) and a final reaction volume of 50 µl. Reactions were incubated at 22°C for 20 min. Binding reactions were separated on 4.0% polyacrylamide gels at 22°C using a 0.25 × TBE buffering system.

For DNase I footprinting, DNA was labelled at one end and incubated with increasing amounts of purified Rep protein under the same conditions used in EMSAs. The volume of the reaction was brought to 200 µl with DNase I assay buffer (10 mM Tris-Cl pH 8.0, 5 mM MgCl, 1 mM CaCl, 100 mM KCl, 2 mM DTT, 50 µg ml−1 BSA, 2 µg ml−1 salmon sperm DNA). DNase I (Sigma) was added (to a concentration that had been predetermined to nick approximately 50% of DNA once) and the reaction was mixed and incubated at room temperature for 2 min. The reaction was stopped by adding 700 µl ice cold DNase I stop buffer (92% ethanol, 3 M sodium acetate, 10 µg ml−1 salmon sperm DNA). Samples were analysed on denaturing 8% polyacrylamide sequencing gels. The G + A sequencing ladder was generated by a rapid method of Maxam and Gilbert sequencing (Sambrook et al., 1989).

Primer extension

Plasmids were electroporated into S. aureus RN4220 and total RNA was extracted from log-phase cultures (OD600 0.9) using the hot phenol method (Miller, 1972). Primer extension was performed essentially as described by Ausubel et al. (1987) using M-MuLV reverse transcriptase (NEB). Sequencing ladders were prepared with the SequiTherm EXCEL II DNA sequencing kit (Epicentre Technologies). Primers PEX3 (5′-CTTGTAATGTATCGAGG-3′) and PEX5 (5′-CTCTTTAGA CATCTAACAAC-3′) were designed to anneal to the rep mRNA whereas PEX7 (5′-GGCGGTTGTTAGATGTC-3′) was complementary to RNAI. PEX9 (5′-TTTTAAATCCAAGCT CGG-3′) is complementary to the rep-cat transcripts produced from deletion derivatives of plasmid pSK5492 and was used to identify the rep transcriptional start point.

Northern blot analysis

Total RNA was separated by formaldehyde agarose (2.0%) gel electrophoresis and transferred to a Hybond N+ nylon membrane (AP Biotech) as previously described (Sambrook et al., 1989). An RNA molecular weight marker (0.16–1.77 kb, Invitrogen) was also run on the same agarose gel and marker bands were visualized after ethidium bromide staining. Double-stranded probes were labelled with the Alkphos Direct Labelling Kit (AP Biotech) and strand-specific probes were labelled using the Gene Images 3′-oligolabelling module (AP Biotech) and hybridisation and detection performed with CDP-Star detection module (AP Biotech) according to the supplier's instructions.

Acknowledgements

This work was supported by Project Grant 153816 from the National Health and Medical Research Council (Australia).

Ancillary