Evolution of compatible replicons of the related IncQ-like plasmids, pTC-F14 and pTF-FC2

Murray N. Gardner and Douglas E. Rawlings

Department of Microbiology, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa

Correspondence
Douglas E. Rawlings
der{at}sun.ac.za


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Two closely related but compatible plasmids of the IncQ-2{alpha} and IncQ-2{beta} groups, pTF-FC2 and pTC-F14, were discovered in two acidiphilic chemolithotrophic bacteria. Cross-complementation and cross-regulation experiments by the replication proteins were carried out to discover what changes were necessary when the plasmids evolved to produce two incompatibility groups. The requirement of a pTC-F14 oriV for a RepC DNA-binding protein was plasmid specific, whereas the requirement for the RepA helicase and RepB primase was less specific and could be complemented by the IncQ-2{alpha} plasmid pTC-FC2, and the IncQ-1{beta} plasmid pIE1108. None of the IncQ-1{alpha} plasmid replication proteins could complement the pTC-F14 oriV, and pTC-F14 and RSF1010 were incompatible. This incompatibility was associated with the RepC replication protein and was not due to iteron incompatibility. Replication of pTC-F14 took place from a 5·7 kb transcript that originated upstream of the mobB gene located within the region required for mobilization. A pTC-F14 mobB–lacZ fusion was regulated by the pTC-F14 repB gene product and was plasmid specific, as it was not regulated by the RepB proteins of pTF-FC2 or the IncQ-1{alpha} and IncQ-1{beta} plasmids. Plasmid pTC-F14 appears to have evolved independently functioning iterons and a plasmid-specific RepC-binding protein; it also has a major replication transcript that is independently regulated from that of pTF-FC2. However, the RepA and RepB proteins have the ability to function with either replicon.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Plasmids of the Escherichia coli incompatibility group Q (IncQ) are characterized by being capable of replication in a very broad host range of bacteria and by being efficiently mobilized by self-transmissible plasmids of the IncP{alpha} (RK2, RP4 and R68) and IncP{beta} (R751) groups (Derbyshire et al., 1987; Frey & Bagdasarian, 1989). These two properties make IncQ-family plasmids highly promiscuous.

IncQ-type plasmid replicons contain three replication genes, repA, repB and repC, and an oriV region (Scholz et al., 1989). In the case of the best-studied IncQ-1{alpha} plasmids, RSF1010 and R1162, the oriV region contains three and one half 20 bp iterons with 2 bp nucleotide spacers. The iterons exert incompatibility and serve as binding sites for the site-specific DNA-binding protein RepC (Persson & Nordström, 1986; Lin et al., 1987; Haring & Scherzinger, 1989). Binding of RepC is essential for replication and is thought to introduce conformational changes leading to DNA unwinding in the adjacent A+T-rich region (Haring & Scherzinger, 1989; Kim & Meyer, 1991). This RepC-induced DNA melting serves as an entry site for RepA, a plasmid-specific helicase that unwinds the DNA in the flanking regions. One of these flanking regions contains a large inverted repeat that has two single-strand DNA initiation sites, ssiA and ssiB (Lin & Meyer, 1987; Haring & Scherzinger, 1989; Honda et al., 1991). The ssiA and ssiB sites initiate priming of single-strand DNA synthesis by RepB primase on opposite strands in the leftward and rightward directions, respectively (Miao et al., 1993).

We discovered two related IncQ-like plasmids in two chemolithotrophic bacteria isolated from mineral bio-oxidation processes in South Africa. Plasmid pTF-FC2 is a 12 184 bp plasmid isolated from the iron- and sulfur-oxidizing bacterium Acidithiobacillus ferrooxidans and plasmid pTC-F14 is a 14 159 bp plasmid from the sulfur-oxidizing bacterium Acidithiobacillus caldus (Rawlings et al., 1984; Gardner et al., 2001). Plasmids pTF-FC2 and pTC-F14 are closely related to each other and have high levels of amino acid sequence identity (72–81 %) in replication, toxin–antitoxin and two out of five mobilization proteins. In addition, the plasmids have oriV regions that contain 22 bp iterons with related nucleotide sequences (Gardner et al., 2001). Although the two plasmids are very similar to each other, they contain origins of replication that are compatible (Gardner et al., 2001). Given their close similarity, it is reasonable to assume that the two plasmids originated from a common ancestral plasmid and would have originally been incompatible. However, they have diverged sufficiently so that they are now able to coexist in the same host in the absence of selection.

The discovery of two closely related plasmids has enabled us to address some aspects of how IncQ-type plasmids have evolved to produce new compatible plasmids. Here we report on several aspects of the molecular interaction between the replicons of plasmids pTF-F14 (a member of the IncQ-2{beta} group) and pTF-FC2 (an IncQ-2{alpha} plasmid). We have also included representatives of the IncQ-1{alpha} (RSF1010, R300B replicons) and IncQ-1{beta} (pIE1108) groups in this plasmid-interaction study.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacterial strains, bacteriophage, media and growth conditions.
The bacterial strains and plasmids used in this study are shown in Table 1. The E. coli strains were all grown at 37 °C in Luria–Bertani medium, and ampicillin (100 µg ml–1), chloramphenicol (20 µg ml–1), kanamycin (30 µg ml–1) and tetracycline (20 µg ml–1) were added as required.


View this table:
[in this window]
[in a new window]
 
Table 1. Strains, plasmids and primers used in this study

Abbreviations used in the Description column are as follows: Ap, ampicillin; Cm, chloramphenicol; Km, kanamycin; St, streptothricin; Tc, tetracycline. Descriptions of constructs into which a PCR-amplified fragment has been cloned include a reference to the position in base pairs (bp) on the GenBank published sequence (RSF1010, M28829; pTC-F14, AF325537) and the primers used. Restriction enzyme sites incorporated into primers are indicated in parentheses and underlined in the primer sequence.

 
DNA techniques, sequencing and analysis.
Plasmid preparation, restriction endonuclease digestions, gel electrophoresis, and cloning were carried out using standard methods (Sambrook et al., 1989; Ausubel et al., 1993). Where no suitable restriction sites were present, DNA fragments to be cloned for incompatibility and complementation assays were amplified by PCR. An initial denaturation step of 60 s at 94 °C was followed by 25 cycles of denaturation (30 s at 94 °C), a variable annealing step and a standard elongation step (90 s at 72 °C). Annealing temperatures were based on the mean primer annealing temperature, and extension times were altered as required (Table 1). PCR was performed in a PCR Sprint temperature cycling system (Hybaid) using the Expand High Fidelity PCR system DNA polymerase (Roche Molecular Biochemicals). The sequences of all constructs that required a PCR step were confirmed by DNA sequencing, using the dideoxy chain-termination method and an ABI PRISM 3100 genetic analyser.

Incompatibility assay.
The ability of an incoming plasmid to displace a resident plasmid was used as the test for incompatibility. The incompatibility assay was performed as reported previously (Gardner et al., 2001).

Complementation assay.
Complementation of a plasmid oriV was detected by the transformation with a ColE1-oriVincQ test construct of an E. coli GW125a (polA mutant) strain containing various coresident plasmids. This was followed by selection for the presence of the oriV-containing construct by growth of the E. coli transformant on solid media containing an appropriate antibiotic.

Northern blotting.
Total RNA was isolated by an adaptation of the protocol of Laing & Pretorius (1992). An overnight E. coli DH5{alpha} culture was diluted into pre-warmed Luria–Bertani broth, with antibiotic selection as required, and grown to an OD600 of 0·7–0·8 at 37 °C, with constant shaking. The cells were harvested and resuspended in 300 µl cold saline/Tris/EDTA (STE) buffer (Ausubel et al., 1993). The suspension was added to 300 µl phenol (pH 6·7; Sigma), containing 0·3 g glass beads, and vortexed for 30 s before being placed on ice. SDS (4 µl of 10 % solution) was added and the solution incubated on ice for 15 min. Chloroform/isoamyl alcohol (24 : 1, v/v) was added and the solution centrifuged at high speed for 10 min. The aqueous phase was removed and extracted with 1 vol. phenol/chloroform/isoamyl alcohol (24 : 24 : 1, by vol.). To remove contaminating DNA, the precipitated nucleic acids were pelleted, resuspended in 200 µl DNase buffer (20 mM MgCl2, 2 mM dithioerythritol) and incubated at 37 °C for 15 min in the presence of 5 units DNase I (RNase-free; Roche Molecular Biochemicals). Extracted RNA (30 µg) was separated in a 1 % (w/v) agarose/18 % (v/v) formaldehyde denaturing gel in 1x MOPS buffer (Ausubel et al., 1993) and capillary-blotted onto a Hybond-N+ (AEC-Amersham) nylon membrane. The pTC-F14 replicon probes were labelled with [{alpha}-32P]dATP (AEC-Amersham) using a Random Primed DNA labelling kit (Roche Molecular Biochemicals) and the probes were hybridized overnight to the RNA at 60 °C in hybridization buffer (7 % SDS, 1 % BSA, 1 mM EDTA, pH 8·0, 0·25 M Na2HPO4). After hybridization, the membrane was washed at 60 °C in 1x SSC (Ausubel et al., 1993) containing 0·1 % SDS, and again in 0·1x SSC containing 0·1 % SDS. Bound probe was detected by autoradiography using MG-SR X-ray film (Konica).

RT-PCR.
Total RNA was isolated using the SV Total RNA isolation system (Promega). The RNA was resuspended in 200 µl DNase buffer (18 mM Tris/HCl, pH 7·3, 9 mM MnCl2, 0·9 mM NaCl) and incubated for 90 min at 28 °C in the presence of 100 units DNase I (RNase-free; Roche Molecular Biochemicals). The DNase treatment was stopped by the addition of 1 vol. phenol/chloroform/isoamyl alcohol (24 : 24 : 1, by vol.), and the RNA precipitated from the aqueous phase overnight at –20 °C by the addition of 0·1 vol. 5 M NaCl and 2 vols absolute ethanol. A two-step RT-PCR protocol was used for cDNA synthesis and cDNA product detection. The protocol of the manufacturer of the 1st Strand cDNA synthesis kit for RT-PCR (AMV; Roche Molecular Biochemicals) was used for the reverse-transcriptase reaction. The PCR was performed in a PCR Sprint temperature cycling system (Hybaid), using Taq DNA polymerase (Promega). For standard PCR reactions, a protocol of an initial denaturation of 60 s at 94 °C, 25 cycles of 30 s at 94 °C, 30 s at 54 °C and 90 s at 72 °C was used. A final extension step of 120 s at 72 °C, before cooling to 4 °C, completed the reaction. Extension times were altered as required for the primer pairs (Table 1), and 2 µl of the 20 µl (total volume) reverse-transcriptase reaction was used in each PCR reaction.

{beta}-Galactosidase assays.
The putative promoter regions of the pTC-F14 replicon region were PCR amplified using primer pairs LACF14MOBBF/LACF14MOBBR (mobB promoter), LACF14MOBAF/LACF14MOBAR (mobA promoter), LACF14REPBF/LACF14REPBR (repB promoter), primer5/primer6 (pas operon promoter), LACF14REPAF/LACF14REPAR (repA promoter) and LACF14REPCF/LACF14REPCR (repC promoter) (Table 1). A PCR Sprint temperature cycling system (Hybaid) and Expand High Fidelity PCR system DNA polymerase (Roche Molecular Biochemicals) were used to amplify the putative promoter regions. After an initial denaturation of 60 s at 94 °C, 25 cycles of 30 s at 94 °C, 30 s at 61 °C and 90 s at 72 °C were performed. A final extension step of 120 s at 72 °C, before cooling to 4 °C, completed the reaction. The PCR products were digested with EcoRI and BamHI and cloned into pMC1403 (Table 1) to construct a promoter–lacZ reporter gene fusion. Since these recombinant plasmids were translational fusions requiring in-frame ligation of the promoter-associated ORF to the reporter gene (lacZ), all constructs were sequenced with primer LACZPRI (Table 1) to ensure the promoter fusions were correct. These constructs were transformed into E. coli CSH50Iq, and the {beta}-galactosidase activity measured using the method of Miller (1972). Overnight cultures were diluted 1 : 100 into fresh pre-warmed Luria–Bertani medium containing the appropriate antibiotic selection and grown at 37 °C for 4 h. After incubation, OD600 was recorded, and the culture diluted 1 : 5 into Z-buffer (Miller, 1972). The exception was E. coli CSH50Iq(pMCF14PAS), which was initially diluted 1 : 500 into fresh medium with appropriate selection, grown for 3 h at 37 °C, and diluted 1 : 20 into Z-buffer. Following dilution in Z-buffer, the culture suspensions were vortexed briefly in the presence of toluene (1 %, v/v), and held at 37 °C for 45 min, before being placed at 28 °C for the remainder of the assay. The assay was started with the addition of ONPG to a final concentration of 0·67 mg ml–1. After 30 min incubation [6 min for E. coli CSH50Iq(pMCF14PAS)] at 28 °C, the reaction was stopped with the addition of 0·42 vol. 1 M Na2CO3. The reaction was then centrifuged at high speed for 5 min to remove the cells, and the supernatant absorbance was measured at both 420 nm and 550 nm. The {beta}-galactosidase activity was calculated according to the equation of Miller (1972).


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Replication of the IncQ and IncQ-like plasmids is polA independent, whereas the replication of cloning vectors based on the ColE1 plasmid replicon is polA dependent. We had previously shown that when the oriV regions alone of either pTF-F14 (pTCF109) or pTF-FC2 (pTV4164) were cloned into a ColE1-based plasmid, they could not replicate in the E. coli polA mutant (GW125a) unless there was a coresident plasmid containing the complete replicon of the parent plasmid (Gardner et al., 2001). The ability of the oriV-containing clones (pTCF109 and pTV4164) to replicate in the polA mutant was plasmid specific, i.e. a coresident plasmid containing the replicon of pTF-FC2 could not complement a plasmid containing the pTC-F14 oriV and vice versa. Similarly, the replication proteins of the IncQ-1{alpha} plasmid pKE462 (R300B replicon) and the IncQ-1{beta} plasmid pIE1108 were unable to support the replication of the pTC-F14 oriV (pTF-F109) in the E. coli GW125a polA mutant (Gardner et al., 2001). As these experiments had shown that the combination of all three replication proteins could not complement a heterologous pTF-F14 or pTC-FC2 oriV-containing construct, we wished to investigate whether all three replication proteins, RepA, RepB and RepC, were plasmid specific.

Ability of individual IncQ-plasmid replication proteins to support replication from the pTC-F14 oriV
The ability of pTC-F109 (ColE1-oriVpTC-F14) to replicate in E. coli GW125a in the presence of the closely related pTF-FC2 as well as the less related pIE1108 and R300B replicons was retested, except that, in addition, one or more of the replication proteins of pTC-F14 was supplied in trans (Table 2, rows 3–9). To provide the pTC-F14 replication proteins in trans, fragments of pTC-F14 containing various replication genes were amplified by PCR and cloned in plasmid pGL10 (RK2/RP4 replicon vector). When pTC-F109 (ColE1-oriVpTC-F14) was transformed into these heteroplasmid-containing recipients, E. coli GW125a(pDER412; pTF-FC2 replicon) and E. coli GW125a(pIE1108) would permit replication of the pTC-F14 oriV only when pTC-F14 RepC was supplied in trans (Table 2). This indicated that the RepC binding-protein was plasmid specific, but that the RepA and RepB of pTF-FC2 and pIE1108 were able to substitute for the equivalent replication proteins of pTC-F14.


View this table:
[in this window]
[in a new window]
 
Table 2. Complementation of pTC-F14 replication proteins by IncQ and IncQ-like plasmids

The oriV–ColE1 construct was transformed into E. coli GW125a containing the coresident plasmids indicated. Results are shown as growth (+) or no growth (–) on medium selecting only for the incoming oriV–ColE1 construct. Plasmid isolations followed by restriction enzyme analysis to generate diagnostic banding patterns and restreaking of transformants on selective medium was used to confirm strains (not shown).

 
The replication proteins of the IncQ (RSF1010/R300B) replicon were also unable to complement the replication proteins of pTC-F14. Furthermore, in the positive control, where all of the pTC-F14 replicon proteins were supplied in trans (pGL10–BAC), the presence of a coresident R300B plasmid resulted in a few slow-growing transformants. This suggested that replication interference between the pTC-F14 oriV (pTC-F109) and R300B (pKE462) had occurred (Table 2). To further investigate this interference, an E. coli GW125a(pKE462+pGL10–BAC+pTC-F109) transformant was streaked onto three solid medium plates containing antibiotic selection for one of the resident plasmids. Individual colonies were restreaked twice, before being tested for the presence of the other two plasmids by plating on solid media containing the appropriate antibiotics. When pKE462 (R300B replicon) was selected, displacement of pTC-F109 (ColE1-oriVpTC-F14) was demonstrated, but not pGL10–BAC. When pTC-F109 was selected (pGL10–BAC selected automatically, as pTC-F109 was RepBAC dependent), very weak growth was observed when testing for the presence of pKE462. This suggested that pKE462 interfered with pTC-F109 replication and that pTC-F109 or the pTC-F14 replication proteins supplied in trans by pGL10–BAC interfered with pKE462 replication.

Identifying the IncQ plasmid locus expressing incompatibility to pTC-F14
We previously reported that pTC-F14 was incompatible with RSF1010 (Gardner et al., 2001). To determine which region of RSF1010 expressed incompatibility towards the pTC-F14 replicon (pTC-F101), a number of subclones were constructed from RSF1010 (Fig. 1). Surprisingly, plasmid incompatibility was not associated with the iteron-containing oriV region of RSF1010, but rather with construct pGEMRSFREP2, which contained a 2·4 kb fragment incorporating the region spanning OrfE to repC (Fig. 1). Since pGEMRSFCAC, containing OrfE and the cac gene, including the region encoding the 75 bp anti-sense ctRNA species (Kim & Meyer, 1986), did not exert incompatibility, this suggested that the incompatibility determinant could be localized to the RSF1010 repAC operon (Fig. 1). Furthermore, it was found that pTC-F109 (ColE1-oriVpTC-F14) was compatible with a resident RSF1010 replicon (RSF1010K), whereas a fragment containing the pTC-F14 repAC operon cloned into the ColE1-vector pGEM-T (pGEM-TCAC) displaced a resident RSF1010 replicon (data not shown). As the repA, repC and repAC genes of pTC-F14 had been cloned behind an IPTG-inducible tac promoter in vector pKK223-3, we investigated whether the overexpression of the individual repA and repC gene products affected incompatibility with RSF1010. Plasmid pKK223-3, containing either repAC or repC cloned behind a tac promoter, displaced approximately 90 % of coresident RSF1010K plasmids, and the level of displacement was not affected by the presence of 1 mM IPTG. In contrast, plasmid pKK223-3 containing the pTC-F14 repA gene was fully compatible with RSF1010K. These results suggested that the product of the repC gene interacts with the oriV region of the competing plasmid. Overexpression of the pTC-F14 repA and repC did not displace pDER412 (pTF-FC2 replicon), which suggests that only certain combinations of RepC proteins and oriVs are incompatible. Incompatibility could be through recognition and binding of the RepC proteins to the competing oriV region and subsequent inhibition of replication as a consequence of this binding. Alternatively, expression of repC could be affecting the regulation of the competing plasmid's repAC operon.



View larger version (11K):
[in this window]
[in a new window]
 
Fig. 1. IncQ-1{alpha} plasmid (RSF1010K) subclones used to challenge a resident pTC-F14 replicon (pTC-F101) for the identification of the IncQ-1{alpha} plasmid locus expressing incompatibility to pTC-F14. The cac gene is a regulator of repA and repC.

 
Numbers and sizes of pTC-F14 replicon transcripts
No studies on the regulation of the replication of either pTF-F14 or pTC-FC2 have been reported. We wished to investigate how many different-sized transcripts are made from the pTC-F14 replicon and whether the synthesis of these transcripts is cross-regulated by pTF-FC2 or RSF1010. Experiments to measure the size of the pTC-F14 transcripts were carried out using the Northern blot technique. Total RNA from E. coli DH5{alpha}(pTC-F14Cm) was isolated and probed with DNA fragments representing the pTC-F14 repC, repA, repB, repAC, repBA, mobA–repB and mobB gene fragments. Several signals were detected when E. coli DH5{alpha}(pTC-F14Cm) RNA was hybridized to the above DNA probes. A signal for a high-molecular-mass mRNA species of 5·1–5·7 kb, two smaller signals of approximately 2·4 kb and 1·6 kb, as well as a smear in the region below 1 kb were detected for each of the DNA probes tested (Fig. 2). The 5·1–5·7 kb hybridization signal is approximately the size predicted (~5·7 kb) from a large polycistronic mRNA species that was initiated upstream of mobB (in the oriTpTC-F14 region) and terminated downstream of repC (Fig. 2).



View larger version (33K):
[in this window]
[in a new window]
 
Fig. 2. Northern blot and RT-PCR analysis of the pTC-F14 replicon. (A) Example of Northern blot, showing approximate sizes of transcripts detected with a [{alpha}-32P]dATP-labelled replicon probe (see text). (B) Primers used in RT-PCR and their location. Bracketed numbers represent the primers and their location (base pair position) relative to the pTC-F14 DNA sequence (GenBank accession no. AF325537). The reverse primers used in the RT reaction were as follows: (2499) RTF14MOBBR, (3069) RTF14MOBAR, (4474) RTF14REPBR, (5607) RTF14PASBR, (6452) RTF14REPAR and (7371) RTF14REPCR. Roman numerals in square brackets refer to the primer pairs used in the PCR reactions following the RT reaction. (C) RT-PCR reaction products on ethidium-bromide-stained agarose gels. The forward primers used in the PCR reaction were as follows: (1795) RTF14MOBCJMOBB, (2217) RTF14MOBBF, (2480) RTF14MOBBJMOBAF, (3987) RTF14MOBAJREPBF, (4930) RTF14REPBJPASF, (5325) RTF14PASAF, (6043) RTF14REPAF, and (6937) RTF14REPCF. The RT-PCR reaction products shown are grouped in sets of three, with the same primer pair used in each reaction of the set. Lanes 1, 4, 7, 10, 13, 16, 19, 22, 25 and 28 are RT-PCR reactions performed on extracted E. coli(pTC-F14Cm) total RNA. Lanes 2, 5, 8, 11, 14, 17, 20, 23, 26 and 29 are PCR reactions performed on purified pTC-F14Cm DNA as a PCR control. Lanes 3, 6, 9, 12, 15, 18, 21, 24, 27 and 30 are exactly the same RT-PCR reactions as those performed in the first lane of each set, except that AMV reverse-transcriptase was not added to the reaction mix.

 
Signals in the approximate size range of 2·4 kb and 1·6 kb, and a smear below 1 kb of varying intensity, were also detected when E. coli DH5{alpha}(pTC-F14Cm) RNA was hybridized to any of the above DNA probes (Fig. 2). These signals were plasmid specific, as they were absent from all pTC-F14-free control blots; however, the origin of these signals was difficult to interpret. The signals were of similar size to the predicted sizes of the E. coli 23S, 16S and 5S rRNA species. Therefore, the ~2·4 kb, ~1·6 kb and smaller smear signals could have been due to small intact (or degradation) products of plasmid mRNA that had been entrapped by the substantial quantity of E. coli rRNA present in this part of the blot. Because of these signals, it was not possible to detect whether smaller transcripts from the replication region were produced.

RT-PCR analysis of the pTC-F14 replicon
RT-PCR was used to confirm the 5·1–5·7 kb transcript detected by Northern blot analysis. The RT-PCR experimental design was such that by using the primer combinations shown in Fig. 2, the PCR products obtained would confirm which gene junctions were transcriptionally linked, allowing for the identification of polycistronic transcripts. If a large polycistronic mRNA encoding the entire region from pTC-F14 mobB to repC was indeed transcribed, this method would not allow one to detect whether additional smaller mRNA transcripts were also present.

The RT-PCR reactions (Fig. 2) were each performed on three RNA extracts to confirm the result obtained. To detect DNA contamination in the RNA extracts, PCR reactions were performed with each primer pair, using a threefold higher concentration of the E. coli DH5{alpha} (pTC-F14Cm) total RNA extract than that used in each RT-PCR reaction, but without AMV reverse transcriptase. No amplification products were detected. Occasionally, following PCR of the cDNA, a second, smaller PCR product was observed (e.g. Fig. 2, lane 25), but as this did not correspond to the product size expected (Fig. 2, middle lane of each set), it was deemed to be the result of mispriming.

The RT-PCR analysis (Fig. 2) confirmed the result from the Northern blot that a large transcript, which extends beyond the repC gene, is initiated upstream of the pTC-F14 mobB gene. As expected, the RT-PCR reaction did not yield a reaction product when using primers to the divergent mobC (RTF14MOBCJMOBB, 1795 forward) and mobB genes (RTF14MOBBR, 2499 reverse) (Fig. 2, lane 1 [i]). A putative promoter, 202 bp upstream of the mobB start codon, was identified that differs by one nucleotide from the E. coli {sigma}70 consensus sequence in both its –35 region (TTGACT) and –10 region (TACAAT), and with a spacer of N16 (Harley & Reynolds, 1987). Should a transcript be initiated from this promoter that terminates downstream of repC, a transcript of approximately 5·7 kb is predicted.

Regulation of putative promoter regions by reporter-gene studies
Although evidence for a 5·7 kb polycistronic mobB–repC mRNA transcript is strong, evidence for the existence of smaller constructs within this region was unclear. We therefore decided to create lacZ-translation reporter gene fusions to all ORFs within the mobB–repC region, with the exception of the gene for the PasB toxin. Reporter fusions also permitted regulation studies of the genes of this region. To allow sufficient DNA fragment length to accommodate putative promoter regions, the following PCR product size for each upstream gene region was cloned into the pMC1403 vector: 364 bp (mobB), 395 bp (mobA), 419 bp (repB), 307 bp (pasA), 397 bp (repA) and 400 bp (repC). To ensure that no host-cell background {beta}-galactosidase activity interfered with the assays, and that any Ptac-controlled genes added in trans were repressed in the absence of IPTG, E. coli CSH50Iq was used as the host cell, as this strain contains lacIq on an F' plasmid (Table 2). The {beta}-galactosidase activity for each putative promoter fusion and how this activity was affected when the parental pTC-F14 (pTC-F14Cm) plasmid and the related IncQ-type plasmids RSF1010 (RSF1010K) and pTF-FC2 (pDER412) were placed in trans is shown in Fig. 3.



View larger version (31K):
[in this window]
[in a new window]
 
Fig. 3. Cross-regulation of putative pTC-F14 promoter regions by related IncQ plasmids present in trans, as determined by {beta}-galactosidase activity. The plasmid present in trans to the promoter–reporter gene construct (given below each set of assays) is indicated by the key. The {beta}-galactosidase activity for each construct was calculated from the mean of three different assays, and within each assay the {beta}-galactosidase activity was measured from three samples Vertical error bars represent the SD of three different assays.

 
The strongest {beta}-galactosidase activity measured for the pTC-F14 replicon was for the plasmid addiction system (pas) gene translational fusion pMCF14PAS (Fig. 3). The {beta}-galactosidase activity of 14426±2943 units obtained for E. coli CSH50Iq(pMCF14PAS) was very much greater than the substantially lower {beta}-galactosidase activities measured for the other pTC-F14 promoter–lacZ fusions. The mobB promoter region (pMCF14MOBB) gave the second highest activity of 138±8 units (Fig. 3). The negligible {beta}-galactosidase activities (less than 5 units) measured for the putative promoter regions of pTC-F14 mobA and repC discounted these regions as containing promoter sequences active in E. coli, while repB and repA promoter–lacZ constructs had {beta}-galactosidase activities of 14±2 units and 28±2 units, respectively (Fig. 3). When pTC-F14 and related IncQ-family plasmids were placed in trans with the repB and repA promoter–lacZ fusions, {beta}-galactosidase activities were unaffected (Fig. 3), suggesting that the low level of activity was spurious.

Repression of the pTC-F14 mobB and pas operon putative promoters by pTC-F14Cm was evident from the reduction in {beta}-galactosidase activity when this plasmid was present in trans (Fig. 3). The {beta}-galactosidase activities of the pTC-F14 mobB (pMCF14MOBB) and pas operon (pMCF14PAS) promoter–lacZ constructs were reduced by 52 % and 65 %, respectively, when pTC-F14Cm was present in trans (Fig. 3). The return to 77 % of E. coli CSH50Iq(pMCF14PAS) {beta}-galactosidase activity when pTC-F101{Delta}AB (pasAB deletion) was present in trans suggested that the pas operon putative promoter is subject to regulation by products of the pas operon (Fig. 3). The putative promoter region of the pas operon also appears to be the only promoter region which responds to cross-regulation by a related IncQ-like plasmid, as the {beta}-galactosidase activity of the pTC-F14 pas operon promoter–lacZ fusion (pMCF14PAS) was reduced by 88 % when pTF-FC2 (pDER412) was present in trans (Fig. 3). Regulation of the pTC-F14 pas operon promoter and repression of the pas operon promoter by pTF-FC2 are being investigated.

The repressor of the pTC-F14 mobB–lacZ fusion was localized to the region of the pTC-F14 replicon upstream of the pas operon (Fig. 4). This was determined by {beta}-galactosidase assays of E. coli CSH50Iq(pMCF14MOBB) into which constructs containing various regions of the pTC-F14 replicon had been transformed. Placement in trans of a pTC-F14 replicon containing a deletion of the pasAB genes (pTC-F101{Delta}AB) did not relieve the repression of the mobB promoter (41±1 units; not shown), while pGL10–AC (Table 1) did not repress {beta}-galactosidase activity (139±13 units; not shown). This indicated that the repressor of this promoter was located upstream of the pTC-F14 pas operon. The regulation of the mobB promoter–lacZ fusion was plasmid specific, as neither RSF1010 nor pDER412 reduced expression of the fusion construct (Fig. 3). The 73 % reduction of {beta}-galactosidase activity for the assay with pTC-F14Cm placed in trans (Fig. 4), compared with the 52 % reduction shown in Fig. 3, is thought to be due to variable stability of the mobB promoter fusion construct when pTC-F14Cm was present in trans. No other plasmid when placed in trans to the mobB promoter fusion construct caused promoter–reporter gene-construct instability.



View larger version (33K):
[in this window]
[in a new window]
 
Fig. 4. Regulation of the pTC-F14 mobB promoter region, as determined by {beta}-galactosidase assays. The {beta}-galactosidase activity for each construct was calculated from the mean of three different assays, and within each assay the {beta}-galactosidase activity was measured from three samples. Vertical error bars represent the SD of three different assays.

 
The pGL10–mob1 construct, which contains the pTC-F14 mobilization region from mobE to repB, encodes a fully functional MobA–RepB fusion protein. When pGL10–mob1 was transformed into E. coli CSH50Iq(pMCF14MOBB), {beta}-galactosidase activity was reduced to 33 % of the activity measured for E. coli CSH50Iq(pMCF14MOBB) (Fig. 4). To investigate which of the pGL10–mob1 proteins was the repressor, the pGL10–mob5 construct, which encodes a truncated MobA–RepB protein (primase domain removed), was transformed into E. coli CSH50Iq(pMCF14MOBB). With pGL10–mob5 in trans, the {beta}-galactosidase activity of the mobB promoter–lacZ construct was 92 % of the activity measured for E. coli CSH50Iq(pMCF14MOBB), suggesting that the RepB protein was the repressor of the mobB promoter (Fig. 4).

To confirm that the RepB primase portion of pTC-F14 could repress the mobB promoter, the construct pTC–KmMO was transformed into E. coli CSH50Iq(pMCF14MOBB). The relaxase domain of the MobA–RepB fusion protein had been removed in the construction of pTC–KmMO by ligation of PCR-amplified pTC-F14 repBAC+oriV to a kanamycin cassette. With pTC–KmMO present in trans, the {beta}-galactosidase activity was reduced to 43 % of that of E. coli CSH50Iq(pMCF14MOBB) (Fig. 4). This allowed for the identification of RepB (the primase domain of the MobA–RepB fusion protein) as the repressor of the mobB promoter. Regulation of the transcript beginning at the mobB gene was plasmid specific, as only the RepB of pTF-F14 affected expression of mobB–lacZ.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
We are interested in the question of how two closely related replicons, such as those of the IncQ-2{alpha} and IncQ-2{beta} plasmids pTF-FC2 and pTC-F14, which must have originated from a common ancestral plasmid, have evolved so as to become compatible. A model for the diversification of plasmid incompatibility has been proposed by Sykora (1992) and could explain how the two replicons of pTF-FC2 and pTC-F14 became compatible. This model assumes that an ancestral IncQ plasmid dimer was formed either through recombination of plasmid progeny monomers or as a result of defective replication termination (Snyder & Champness, 2003). Pressure for at least one of the dimer replicons to acquire mutations leading to a sufficiently different replicon to the original replicon could be provided by the dimer being challenged by an identical plasmid attempting to displace the plasmid dimer from the host. Sufficient mutations would accumulate to allow for an unrelated but functional replicon to evolve, so that the second replicon of the dimer would be compatible with the ancestral plasmid. This would ensure maintenance of the dimer in the presence of the ancestral plasmid. Resolution of the plasmid dimer after mutation of one of the replicons to an unrelated, but functional, replicon would lead to two compatible sister plasmids, such as pTF-FC2 and pTC-F14. Irrespective of whether compatible plasmids can arise through the mechanism described above or as the consequence of the independent evolution of a new replicon by gradual accumulation of mutations, the driving force for the acquisition of mutations that lead to new IncQ replicon compatibility is the same, that is, the attempted displacement by a competitive plasmid.

Irrespective of the model chosen, sufficient concomitant evolution of an IncQ plasmid iteron sequence and its RepC protein would be required to produce compatible replicons. Evidence for this part of pTC-F14 and pTF-FC2 speciation has been obtained, as the two oriV regions are compatible (Gardner et al., 2001), and in this work we showed that replication from the pTC-F14 oriV was RepC-specific. That is, neither the pTF-FC2 RepC protein nor any other IncQ-type plasmid which was RepC tested could substitute for the requirement of the pTC-F14 oriV for its own RepC. However, the RepA and RepB proteins were less plasmid specific, as the RepA and RepB proteins of pTF-FC2 and pIE1108 could substitute for the pTC-F14 proteins provided that the pTC-F14 RepC was present. This suggests that the sites in the oriV region to which RepA and RepB bind are less plasmid specific. As all IncQ plasmids have at least three identical 20–22 bp iterons, an as yet unsolved mystery is how mutations in one iteron are spread to all functional iteron copies. In some cases, such as pTC-F14 and pIE1130 (Smalla et al., 2000), extra copies of iterons with non-identical sequences exist, but, where tested, mutated iterons have been found to be non-functional (Miao et al., 1995).

It is interesting that the replicons of the closely related plasmids pTC-F14 and pTF-FC2 have become compatible, whereas the more dissimilar plasmids pTC-F14 and RSF1010 are incompatible. One can speculate that this is because pTC-F14 and pTF-FC2 are promiscuous plasmids that were isolated from acidiphilic chemolithotrophic bacteria that share the same habitat and the plasmids may encounter each other. In contrast, RSF1010 (and other IncQ-1{alpha} plasmids) are typically found in neutrophilic heterotrophic bacteria; pTF-F14 and RSF1010 are therefore unlikely to encounter each other and so have not evolved to accommodate each other. Incompatibility appears to be associated with the RepC proteins rather than with the iterons, and it is only certain combinations of RepC proteins and iterons that are incompatible. The most likely explanation is that RepC-associated incompatibility is due to non-productive binding of the heterologous RepC to the iterons of an incompatible plasmid, and that this reduces the ability of the homologous RepC to initiate replication. The replication interference was reciprocal, as RSF1010 could inhibit the replication of pTC-F14 and vice versa. No purified RepC protein was available, and as this was a fairly minor aspect of this work, DNA-binding studies were not carried out. Incompatibility due to repAC is similar to that reported by Tietze (1998) for the plasmid pairs p95L28 and pDER412. The plasmid p95L28 is a ColE1-based vector (pUCBM20), into which the oriVb and the repAC operon of pIE1107 were cloned. Cloned on its own into this vector, oriVbpIE1107 was compatible with pDER412 (pTF-FC2 replicon), but when the pIE1107 repAC operon was present (p95L28), the construct was incompatible with the pTF-FC2 replicon (Tietze, 1998).

An additional possibility for replication interference by two related plasmids is that the proteins involved in the control of transcription of the replication genes cross-regulate each other, and hence interfere with the replication of the heterologous plasmid. No regulation studies have been carried out on the expression of replication genes of the IncQ-2 group plasmids, and these plasmids lack the cac regulator (control of repA and repC), shown to be involved in RSF1010 replication (Rawlings & Tietze, 2001). Although the number of different-sized transcripts involved in the expression of the rep genes of pTC-F14 is uncertain, evidence from Northern hybridization and RT-PCR suggested that a large ~5·7 kb mRNA transcript is synthesized. Studies of reporter gene fusions to all of the ORFs in the region of the replicon suggested that the mobB and, by implication, the downstream mobA–repB genes are expressed at low levels relative to the pasAB genes. Futhermore, evidence was obtained that the pTC-F14 RepB gene product was responsible for repression of the transcript that begins at the mobB gene. As neither pTF-FC2 nor RSF1010, when provided in trans, reduced expression of the mobB–lacZ reporter gene, this indicated that regulation of the large pTC-F14 replication transcript was RepB specific.

The evidence obtained in this work is therefore consistent with a model in which pTC-F14 and pTF-FC2 have evolved such that the oriV of each plasmid is recognized only by its own RepC. However, each RepC protein is able to induce changes in the oriV that allows the less specific heterologous RepA helicase and RepB single-strand primase proteins to function. Besides being a primase, the RepB also appears to be a regulator of a large replication transcript. RepB, therefore, presumably has single-strand DNA-binding activity required for its primase function that is not plasmid specific and double-strand DNA-binding repressor activity that has mutated to become plasmid specific. This plasmid specificity would appear to be a requisite for pTC-F14 and pTF-FC2 to be compatible. Although some of the details of this model are speculative, it nevertheless provides a basis for future work on plasmid evolution.

Besides the large transcript, the high levels of pasA–lacZ expression suggested that a second transcript within the replication region is produced. A putative promoter region has been identified upstream of pasA that has one nucleotide difference in its –35 region (TTCACA) and a two-nucleotide difference in its –10 region (TATATC) from a consensus E. coli {sigma}70 promoter sequence plus a consensus N17 spacer region (Harley & Reynolds, 1987). This transcript was not unequivocally detectable by Northern blot analysis, as the predicted size of the product fell within the region of the blot where the rRNA species were present. Internal short transcripts were also not detectable using RT-PCR, since a long transcript would serve as template for shorter transcripts. Unlike the pTC-F14 mobB–lacZ fusion, the pasA–lacZ fusion was regulated by pTF-FC2 when placed in trans. However, as the RepB primase is essential for IncQ-type plasmid replication, and expression of the repB gene is from the mobB promoter, pTC-F14 is likely to have retained control over its own replication. The role of the pas in the replication of pTC-F14 and pTC-FC2 is unknown and puzzling, as the entire pas of both pTF-FC2 and pTC-F14 (Smith & Rawlings 1997; Deane & Rawlings, 2004) could be deleted without a detectable effect on plasmid replication or copy number. This pasA cross-regulation is the subject of a recent study (Deane & Rawlings, 2004).

This study has highlighted a difficulty with what may be understood by plasmid incompatibility. A formal definition of plasmid incompatibility is that it is the failure of two coresident plasmids to be stably inherited in the absence of external selection (Novick, 1987). However, what may be frequently inferred is that two plasmids that are incompatible belong to the same incompatibility group because they have replicons or partitioning systems that are closely related. This does not hold in the case of pTC-F14 and RSF1010, which clearly belong to two very different plasmid homology groups. In addition, their iterons have substantially different sequences, and so might be expected to be compatible; however, they are incompatible. In contrast, plasmids pTF-FC2 and pTC-F14 are clearly more closely related to each other than RSF1010 is to pTC-F14, but the more closely related plasmids are compatible. This is a clear instance where replicon incompatibility does not reflect the sequence relatedness of plasmid replicons.


   ACKNOWLEDGEMENTS
 
We thank Erhard Tietze for plamid pIE1108, Gunther Ziegelin for RSF1010K and Aresa Toukdarian for cloning vector pGL10. This work was supported by grants from the National Research Foundation, the Technology for Human Resource Development Programme (Pretoria, South Africa) and BHP Billiton Minerals Technology (Randburg, South Africa).


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K. (1993). Current Protocols in Molecular Biology. New York: Wiley Interscience.

Casadaban, M. J., Martinez-Arias, A., Shapira, S. K. & Chou, J. (1983). {beta}-Galactosidase gene fusions for analyzing gene expression in Escherichia coli and yeast. Methods Enzymol 100, 293–308.[Medline]

Deane, S. M. & Rawlings, D. E. (2004). Plasmid evolution and interaction between the plasmid addiction stability systems of two related broad-host-range IncQ-like plasmids. J Bacteriol 186, 2123–2133.[Abstract/Free Full Text]

Derbyshire, K. M., Hatfull, G. & Willetts, N. S. (1987). Mobilization of the non-conjugative plasmid RSF1010: a genetic analysis of its origin of transfer. Mol Gen Genet 206, 154–160.[Medline]

Dorrington, R. A. & Rawlings, D. E. (1989). Identification and sequence of the basic replication region of a broad-host-range plasmid isolated from Thiobacillus ferrooxidans. J Bacteriol 171, 2735–2739.[Medline]

Dorrington, R. A., Bardien, S. & Rawlings, D. E. (1991). The broad-host-range plasmid pTF-FC2 requires a primase-like protein for autonomous replication in Escherichia coli. Gene 108, 7–14.[CrossRef][Medline]

Frey, J. & Bagdasarian, M. (1989). The molecular biology of IncQ plasmids. In Promiscuous Plasmids of Gram-Negative Bacteria, pp. 79–93. Edited by C. M. Thomas. London: Academic Press.

Gardner, M. N., Deane, S. M. & Rawlings, D. E. (2001). Isolation of a new broad-host-range IncQ-like plasmid, pTC-F14, from the acidophilic bacterium Acidithiobacillus caldus and analysis of the plasmid replicon. J Bacteriol 183, 3303–3309.[Abstract/Free Full Text]

Haring, V. & Scherzinger, E. (1989). Replication proteins of IncQ plasmid RSF1010. In Promiscuous Plasmids of Gram-Negative Bacteria, pp. 95–124. Edited by C. M. Thomas. London: Academic Press.

Harley, C. B. & Reynolds, R. P. (1987). Analysis of E. coli promoter sequences. Nucleic Acids Res 15, 2343–2361.[Abstract]

Honda, Y., Sakai, H., Hiasa, H., Tanaka, K., Komano, T. & Bagdasarian, M. (1991). Functional division and reconstruction of a plasmid replication origin: molecular dissection of the oriV of the broad-host-range plasmid RSF1010. Proc Natl Acad Sci U S A 88, 179–183.[Abstract]

Kim, K. & Meyer, R. J. (1986). Copy-number of broad host-range plasmid R1162 is regulated by a small RNA. Nucleic Acids Res 14, 8027–8046.[Abstract]

Kim, Y.-J. & Meyer, R. J. (1991). An essential iteron-binding protein required for plasmid R1162 replication induces localized melting within the origin at a specific site in AT-rich DNA. J Bacteriol 173, 5539–5545.[Medline]

Laing, E. & Pretorius, I. S. (1992). Synthesis and secretion of an Erwinia chrysanthemi pectate lyase in Saccharomyces cerevisiae regulated by different combinations of bacterial and yeast promoter and signal sequences. Gene 121, 35–45.[CrossRef][Medline]

Lin, L.-S. & Meyer, R. J. (1987). DNA synthesis is initiated at two positions within the origin of replication of plasmid R1162. Nucleic Acids Res 15, 8319–8331.[Abstract]

Lin, L.-S., Kim, Y.-J. & Meyer, R. J. (1987). The 20 bp, directly repeated DNA sequence of broad host range plasmid R1162 exerts incompatibility in vivo and inhibits R1162 DNA replication in vitro. Mol Gen Genet 208, 390–397.[Medline]

Miao, D.-M., Honda, Y., Tanaka, K., Higashi, A., Nakamura, T., Taguchi, Y., Sakai, H., Komano, T. & Bagdasarian, M. (1993). A base-paired hairpin structure essential for the functional priming signal for DNA replication of the broad host range plasmid RSF1010. Nucleic Acids Res 21, 4900–4903.[Abstract]

Miao, D.-M., Sakai, H., Okamoto, S., Tanaka, K., Okuda, M., Honda, Y., Komano, T. & Bagdasarian, M. (1995). The interaction of RepC initiator with iterons in the replication of the broad host-range plasmid RSF1010. Nucleic Acids Res 23, 3295–3300.[Abstract]

Miller, J. H. (1972). Experiments in Molecular Genetics. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.

Novick, R. P. (1987). Plasmid incompatibility. Microbiol Rev 51, 381–395.[Medline]

Persson, C. & Nordström, K. (1986). Control of replication of the broad host range plasmid RSF1010: the incompatibility determinant consists of directly repeated sequences. Mol Gen Genet 203, 189–192.[Medline]

Rawlings, D. E. & Tietze, E. (2001). Comparative biology of IncQ and IncQ-like plasmids. Microbiol Mol Biol Rev 65, 481–496.[Abstract/Free Full Text]

Rawlings, D. E. & Woods, D. R. (1985). Mobilization of Thiobacillus ferrooxidans plasmids among Escherichia coli strains. Appl Environ Microbiol 49, 1323–1325.[Medline]

Rawlings, D. E., Pretorius, I. & Woods, D. R. (1984). Expression of a Thiobacillus ferrooxidans origin of replication in Escherichia coli. J Bacteriol 158, 737–738.[Medline]

Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.

Scholz, P., Haring, V., Wittmann-Liebold, B., Ashman, K., Bagdasarian, M. & Scherzinger, E. (1989). Complete nucleotide sequence and gene organization of the broad-host-range plasmid RSF1010. Gene 75, 271–288.[CrossRef][Medline]

Smalla, K., Heuer, H., Gotz, A., Niemeyer, D., Krogerrecklenfort, E. & Tietze, E. (2000). Exogenous isolation of antibiotic resistance plasmids from piggery manure slurries reveals a high prevalence and diversity of IncQ-like plasmids. Appl Environ Microbiol 66, 4854–4862.[Abstract/Free Full Text]

Smith, A. S. & Rawlings, D. E. (1997). The poison-antidote stability system of the broad-host-range Thiobacillus ferrooxidans plasmid pTF-FC2. Mol Microbiol 26, 961–970.[CrossRef][Medline]

Smith, A. S. & Rawlings, D. E. (1998). Efficiency of the pTF-FC2 pas poison-antidote stability system in Escherichia coli is affected by the host strain, and antidote degradation requires the Lon protease. J Bacteriol 180, 5458–5462.[Abstract/Free Full Text]

Snyder, L. & Champness, W. (2003). Molecular Genetics of Bacteria. Washington, DC: American Society for Microbiology

Sykora, P. (1992). Macroevolution of plasmids: a model for plasmid speciation. J Theor Biol 159, 53–65.[Medline]

Tietze, E. (1998). Nucleotide sequence and genetic characterization of the novel IncQ-like plasmid pIE1107. Plasmid 39, 165–181.[CrossRef][Medline]

Yanisch-Perron, C., Vieira, J. & Messing, J. (1985). Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13mp18 and pUC19 vectors. Gene 33, 103–119.[CrossRef][Medline]

Received 1 December 2003; revised 9 February 2004; accepted 11 February 2004.



This Article
Abstract
Full Text (PDF)
Alert me when this article is cited
Alert me if a correction is posted
Citation Map
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Download to citation manager
Google Scholar
Articles by Gardner, M. N.
Articles by Rawlings, D. E.
Articles citing this Article
PubMed
PubMed Citation
Articles by Gardner, M. N.
Articles by Rawlings, D. E.
Agricola
Articles by Gardner, M. N.
Articles by Rawlings, D. E.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
INT J SYST EVOL MICROBIOL MICROBIOLOGY J GEN VIROL
J MED MICROBIOL ALL SGM JOURNALS
Copyright © 2004 Society for General Microbiology.