Mechanisms of Strand Replacement Synthesis for Plasmid DNA Transferred by Conjugation

Abstract

A single strand of plasmid DNA is transferred during conjugation. We examined the mechanism of complementary strand synthesis in recipient cells following conjugative mobilization of derivatives of the IncQ plasmid R1162. A system for electroporation of donor cells, followed by immediate mating, was used to eliminate plasmid-specific replicative functions. Under these conditions, Escherichia coli recipients provided a robust mechanism for initiation of complementary strand synthesis on transferred DNA. In contrast, plasmid functions were important for efficient strand replacement in recipient cells of Salmonella enterica serovar Typhimurium. The mobilizing vector for R1162 transfer, the IncP1 plasmid R751, encodes a DNA primase with low specificity for initiation. This protein increased the frequency of transfer of R751 into Salmonella, but despite its low specificity, it was inactive on the R1162 derivatives. The R751 primase was slightly inhibitory for the transfer of both R751 and R1162 into E. coli. The results show that there is a chromosomally encoded mechanism for complementary strand synthesis of incoming transferred DNA in E. coli, while plasmid-specific mechanisms for this synthesis are important in Salmonella.

It has been known for some time that only one of the two plasmid DNA strands is transferred during conjugation (9). As a result, the complementary strand must be synthesized prior to establishment of the plasmid in the new host. In the donor cell, the departing strand is also replaced, although this is not essential for transfer (17). The mechanisms used for replacement strand synthesis are not understood. In particular, it is not known to what extent components of the mechanism for vegetative replication of the plasmid, or other plasmid-encoded proteins, are recruited for this process. The IncQ plasmid R1162, which is efficiently comobilized during conjugative transfer of IncP-1 plasmids, such as RK2 and R751, is particularly suitable for investigating this question. R1162 encodes the proteins required for the several steps in initiating its own replication, rather than relying on host counterparts, a feature that no doubt contributes to its broad host range. This independence could be extended to the important process of strand replacement synthesis in conjugation.

Iterons within the origin of replication (oriV) of R1162 are the target of a DNA-binding protein, encoded by the plasmid, which locally distorts the helix and allows entry of a DNA helicase, also plasmid encoded (16, 31). The helicase activates oriL and oriR, two opposite-facing sites in the origin where complementary strand synthesis is initiated. These sites are recognized by a third protein, the plasmid-encoded primase (30). The primase is synthesized both as a free protein and as the C-terminal domain of the R1162 relaxase, a protein essential for conjugative transfer (32). It is the relaxase that cleaves one of the DNA strands at the origin of transfer (oriT), with the protein becoming covalently linked to the 5′ end of the DNA. The cleaved strand is unwound from its complement, and the DNA protein intermediate is transported into a new cell by a type IV secretory machine. Displacement of the protein by the trailing 3′ end of the transported intermediate forms a circular, single-stranded molecule, and the complementary strand is then synthesized. The linkage of the primase to the relaxase suggests that this replication protein is also involved in transfer. However, when the R1162 mob genes are cloned into a vector, such as pBR322, the resulting molecule is mobilizable even when the primase domain of the relaxase is deleted (4). Thus, the primase is not absolutely required for transfer, perhaps because it can be replaced by the priming systems active on the vector. Evidence that the R1162 priming system is nevertheless active during transfer was obtained under conditions where mobilization is inefficient. In this case, the transfer frequency was increased by the presence of the primase and its cognate initiation sites (12).

Assessing the contribution of the R1162 replication proteins to strand replacement synthesis during conjugation has been difficult due to the ongoing vegetative replication in the cell. We recently developed a system where plasmid DNA is first introduced by electroporation into potential donor cells, containing a complete set of proteins for transfer but not for plasmid replication, and then immediately mating these cells with recipients encoding the λ integrase (29). Since the transferred DNA contains a λ att site, the DNA is captured after transfer by integration at attB. Furthermore, parental and daughter transferred strands could be marked with different restriction sites, by using DNA containing a mismatched oligonucleotide for electroporation. We found that the R1162 primase and its appropriately oriented initiation site are important for strand regeneration in the donor after a round of transfer, and the repaired molecules are then able to participate in subsequent rounds of transfer.

In the experiments described here, we used the electroporation and mating system to ask if the R1162 priming system is similarly involved in the essential step of strand replacement in the recipient. Although newly created transconjugant cells do not themselves provide a source of the R1162-encoded primase specific for oriL and oriR, the incoming strand could still be covalently linked to the relaxase, nonetheless, with the primase as the C-terminal domain. Priming could be initiated by the linked protein at the neighboring oriL and oriR or by the released protein after recircularization. In addition, relaxase is transported independently by the type IV transport machine at levels sufficient for detection (21), and these could be the source of priming complementary strand synthesis on the entering plasmid DNA. Our results show that for Escherichia coli there is a robust cellular system for complementary strand synthesis on the transferred strand. In Salmonella, where this system is less active, the R1162 priming system can be deployed to initiate complementary strand synthesis. The results parallel, in a broad sense, those obtained for the larger, self-transmissible plasmids, which encode a primase for strand replacement synthesis that is dispensable for E. coli.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The E. coli K-12 donor strains used for bacterial matings were derived from MC1061 (7) because it is highly transformable by electroporation (6). Salmonella enterica serovar Typhimurium donor strains were all derived from strain LT2 (22). Donor strains generally contained either pUT1559, encoding the R1162 Mob proteins and described previously (11), or pUT1795, consisting of a 1.8-kb fragment of DNA containing the pSC101 Mob genes mobA and mobX (26) cloned into pBR322 (3). The relevant features of the different Mob helper plasmids used in this study are shown as part of Fig. 1. Donor strains also contained either R751 (33) or a traC deletion derivative. The 1,627-bp deletion (base pairs 40,036 to 41,664 on the R751 map) (33) was obtained by generating two 500-bp fragments of R751 that flank the majority of the traC4 gene, including the catalytic domain, by PCR amplification. The fragments were then cloned into pUC18K (23) so that they flanked the promoterless kanamycin resistance (aphA-3) cassette preceded by nonsense codons in all three forward reading frames. This cassette was then introduced into R751 by homologous recombination. An R751 derivative that contained the integrated plasmid, due to a single crossover, was obtained by mating with selection on medium containing kanamycin. A second crossover in the other duplicated segment of R751 DNA forms a plasmid retaining the cassette but lacking the rest of pUC18K. Cells containing this plasmid were identified by screening for sensitivity to ampicillin. Translation of traC is initiated from several locations (28), but in all cases there is an early truncation of the protein due to the nonsense codons in the cassette.

FIG. 1.

Top left, general scheme for electroporation and mating of plasmid DNA. On the top right are shown the general features of the different test plasmids used for electroporation. At the bottom are the Mob helper plasmids in the different donor strains. Mob_R1162 and Mob_pSC101, contains genes for mobilization of plasmid R1162 and plasmid pSC101, respectively; Rep_R1162, encodes plasmid genes for replication of R1162; ApR and CmR, encodes resistance to ampicillin or chloramphenicol, respectively.

Donor cells containing plasmids selected from the above were used for electroporation and mating with mobilizable test plasmids. The properties of the different plasmids used are also shown in Fig. 1. The mobilizable plasmid pUT1735 (29) contains the R1162 oriT and oriV; lambda attP, to allow integration of the plasmid DNA into the chromosome in the presence of lambda integrase; and a gene for chloramphenicol resistance derived from pACYC184 (8). The oriV DNA had been modified so that oriL is flanked by unique EcoO109I and SmaI restriction sites, and oriR by SmaI and EcoRV sites (35). The plasmid was routinely maintained in cells by the helper plasmid pMS94 (27), which encodes the R1162 replication proteins.

Plasmid pUT1735 was used to derive a series of additional test plasmids (Fig. 1), deleted for oriL, oriR, or both sites, by digesting DNA with combinations of EcoO109, EcoRV, and SmaI. The resulting test plasmids, which have defective oriVs, were maintained by ligating the DNA to pBR322 at the BamHI site. We also constructed a test plasmid identical to pUT1735 but containing a pSC101 oriT in the place of the R1162 oriT. This plasmid was made by reverse PCR (10) with pUT1735 as the template and with oligonucleotides designed to replace the R1162 oriT with the pSC101 oriT. These oligonucleotides each contained one-half of the pSC101 oriT, so ligating the linear PCR product reconstituted a test plasmid with the complete pSC101 oriT. The oriV deletion derivative was constructed by restriction fragment exchange with the plasmid containing the R1162 oriT.

In addition to the strains for electroporation and mating, a set of donor strains that contained pUT1735 as an established, replicating plasmid was prepared. These strains also contained either R751 or R751ΔtraC and a plasmid identical to pUT1559 but containing the R1162 replication proteins as well as the Mob proteins (Fig. 1).

The E. coli recipient in the mating experiments was an MV12 derivative resistant to nalidixic acid and containing a plasmid encoding the λ integrase, as described previously (11). The serovar Typhimurium recipients were either LT2 or the restrictionless strain LB5000 (5). In both cases, we prepared strains resistant to nalidixic acid and containing the integrase plasmid.

Electroporation and mating assay.

Mobilizable test plasmids capable of replication were maintained in strains containing the helper plasmid pMS94. Test plasmid DNA was purified from these strains by a commercial procedure (QIAGEN), and the helper plasmid was then cleaved with XmnI. Test plasmids having a defective oriV were maintained as fusions with pBR322. Test plasmid DNA was prepared by digesting the purified DNA with BamHI, purifying the test plasmid portion by agarose gel electrophoresis and gel extraction, and then recircularizing the plasmid by ligation. The DNA was further digested with BstZ17I to inactivate for transformation any residual pBR322 DNA. The electroporation and mating assay was performed as outlined in Fig. 1 and previously described (11), with the following modifications. After electroporation of the donor cells with 100 ng of prepared test plasmid, the cells were resuspended in 1 ml medium containing 2% tryptone, 0.5% yeast extract, 20 mM glucose, 10 mM NaCl, 10 mM MgCl₂, 10 mM MgSO₄, and 2.5 mM KCl. A 450-μl sample of these cells was combined with a 2.5-fold excess of recipient cells grown to approximately 2 × 10⁸ to 5 × 10⁸ cells/ml in the same medium. This mixture was then applied to a 0.45-micron, 25-mm-diameter filter by gentle vacuum. The filter was placed on TYE (1% tryptone, 0.5% yeast extract, 0.5% NaCl)-1.5% agar medium for 90 min. The mated cells were then washed from the filter with 1 ml of fresh medium and plated for transconjugants.

Filter matings of donors with established test plasmids.

Donor and recipient cells were grown overnight and then diluted 1:10 in TYE and grown for an additional 90 min. Mating mixtures consisted of 2.5 ml recipient cells and 0.25 ml donor cells, which were applied to a filter and subsequently treated as described above. To determine the transfer frequency of R751 or R751ΔtraC, donor cells were counted by plating on TYE agar plates containing trimethoprim (200 μg/ml) and transconjugants were selected on TYE containing the same antibiotic and nalidixic acid (25 μg/ml). Transfer frequencies of pUT1735 were determined by plating on medium containing chloramphenicol (25 μg/ml) and chloramphenicol and nalidixic acid.

Other procedures.

M13ΔE101 bacteriophage (15) was propagated in JM103 (25) grown in 1.5% tryptone-1% yeast extract-0.5% NaCl (pH 7.4). Plaques are on soft agar (0.6%) lawns containing this medium. M13ΔE101-pas (see Fig. 5) contained pBR322 DNA (base pairs 2348 to 2415), which includes a primosome assembly site (36). This DNA was flanked by additional DNA from R1162, which does not contain any efficiently utilized primosome assembly sites (2, 13).

FIG. 5.

Bar graph showing the relative mating frequencies for donor strains containing R751 or R751ΔtraC. Actual average mating frequencies (transconjugants per donor cell; three trials) are shown below.

RESULTS

Mobilization of R1162 derivatives lacking the origin of vegetative replication.

The orientation of oriL and oriR, the initiation sites for vegetative replication of R1162, with respect to the direction of conjugative transfer, is shown in Fig. 2. The top strand in the figure is transferred, in the 5′-to-3′ direction (14), so that any strand replacement in the recipient due to the R1162 primase would be initiated at oriR. In order to determine whether the R1162-encoded priming system is required for recovery of transconjugants in recipients, we separately electroporated into donor cells different plasmid DNAs either containing the complete origin of replication (oriV⁺) or lacking one or both of the priming sites oriL and oriR. The transformed cells, which contained R751 and the R1162 Mob proteins (including the relaxase with the attached primase), were then mated with a recipient encoding the λ integrase. In each case, the transforming DNA was cut out from a carrier plasmid and ligated to join the ends, and the amount of product was quantified so that similar amounts of DNA were used in the electroporations. We found that there was a significant frequency of transfer in every case, with about 50% fewer transconjugants if oriV was incomplete. This indicated that the priming sites in the R1162 oriV were not required for strand replacement synthesis in the recipient, even when there was no other obvious priming system on the transferring plasmid.

FIG. 2.

Relative orientations of oriT and initiation sites for vegetative replication (oriL and oriR) in plasmid R1162 and derivatives used in this work. The top strand is transferred in the 5′-to-3′ direction shown by the arrow. The arrows with dashed lines show the direction of DNA synthesis from each initiation site for replication.

We initially considered two possible explanations for these results. The R1162 primase might be active in strand replacement, but at sites other than oriL and oriR. Alternatively, R751 itself encodes a primase (20), which might be initiating strand synthesis on the R1162 derivative. To test the first of these possibilities, we took advantage of the fact that the plasmid pSC101 has a mobilization system similar to that of R1162 (26). However, the mechanisms of replication of the two plasmids are unrelated, and the relaxase of pSC101, although otherwise similar to the R1162 homolog, is not linked to a primase. In addition, both the pSC101 and the R1162 Mob proteins are active on the pSC101 oriT (26). We chose the pSC101 relaxase rather than an R1162 relaxase with a mutation in the primase domain to avoid the possibility of a pleiotropic effect on strand cleavage. Indeed, we found that some insertion mutations in the primase domain do affect nicking (C. Parker and R. Meyer, unpublished results).

In our mating system, the pSC101 Mob proteins were able to transfer a test plasmid containing the pSC101 oriT, although at a lower frequency than the R1162 Mob proteins (Table 1, upper grouping), showing that the relaxase-linked R1162 primase was not required for complementary strand synthesis of transferred plasmid DNA. We had shown previously that the pSC101 Mob proteins are inactive on the R1162 oriT under normal mating conditions (26). This remained true for the electroporation and mating system used here (Table 1, upper grouping), verifying that the colonies were the result of mating and that the normal specificity of the relaxase was retained.

TABLE 1.

Transfer of test plasmid DNA

Property of transferred plasmid	Protein provided by donor cell	Avg no. of transconjugants/100 μl mated cells
oriTpSC101oriVR1162	MobpSC101, R751	333
oriTpSC101oriVR1162	MobR1162, R751	1,490
oriTR1162oriVR1162	MobR1162, R751	707
oriTR1162oriVR1162	MobpSC101, R751	0
oriTpSC101oriVR1162	MobpSC101, R751traC+	281
oriTpSC101oriVR1162	MobpSC101, R751ΔtraC	1,049
oriTpSC101oriV-negative	MobpSC101, R751traC+	297
oriTpSC101oriV-negative	MobpSC101, R751ΔtraC	1,103

To evaluate the role of the R751 primase, we introduced a deletion into traC, which encodes the primase of R751, and tested for the effect of transfer (Table 1, lower grouping). In this case, we used the test plasmid with the pSC101 oriT, and the pSC101 Mob proteins were provided in the donor, so that R1162-specific primase was not available. Deletion of traC, in fact, increased the number of transconjugants, with the presence or absence of oriV having no effect. Thus, neither the primase of R751 nor the Mob primase of R1162 is essential for recovery of the transferred test plasmid. Instead, the evidence pointed to an endogenous mechanism for strand replacement, independent of the plasmid-priming proteins.

Strand replacement in E. coli and S. enterica serovar Typhimurium during mating in the absence of TraC and R1162 primases.

In an early study, Lanka and Barth (18) showed that plasmid RP4 derivatives containing mutations in the plasmid primase gene transferred well into E. coli but not into S. enterica serovar Typhimurium. This led them to suggest that any host-encoded mechanism for strand replacement is probably less efficient in Salmonella recipient cells. We therefore carried out our electroporation and mating protocol, but with serovar Typhimurium LT2 recipients. To avoid problems with host restriction, we used cells of serovar Typhimurium as donors as well and, in that case, DNA was isolated from this species prior to electroporation.

We carried out sets of electroporation matings, using R751ΔtraC as the mobilizing vector to avoid any effect of the IncP primase, to determine if we could now detect a contribution of the R1162 priming system in plasmid establishment after transfer. We compared the transfer frequencies of oriV⁺ and ΔoriV test plasmids under three conditions for both E. coli-E. coli and serovar Typhimurium-serovar Typhimurium matings: a transferring plasmid containing the R1162 oriT with the R1162 Mob proteins in the donor and a transferring plasmid containing the pSC101 oriT with either the R1162 or the pSC101 Mob proteins in the donor. If the R1162 priming system can contribute to plasmid recovery in Salmonella recipients, then we would expect to see an increase in the number of transconjugants for the oriV⁺ plasmids, but only when the donor contains the R1162 Mob proteins. The results of this analysis, shown in Fig. 3, indicate first that the presence of components of the R1162 priming system, the relaxase-linked primase and oriV, had little effect on transfer in E. coli, in agreement with the result described above for TraC⁺ donors. The number of transconjugants was lower overall in Salmonella matings. However, the R1162 primase did increase the transfer frequency in these matings, provided that the transferring plasmid was oriV⁺. This was true regardless of which oriT was used for transfer.

FIG. 3.

Bar graph showing the ratio of the number of transconjugants for oriV⁺ and ΔoriV test plasmids for the indicated Mob systems and oriTs. Given below the graphs are the average numbers of transconjugants used to construct the graphs. The averages are from two independent trials. The error bars in the graph indicate the standard deviations.

We tested the effect of the primase in the recipient in matings where the donors containing the pSC101 Mob proteins. The test plasmid contained the pSC101 oriT and the R1162 oriV. In E. coli-E. coli matings, the primase had no effect: the ratio of the number of transconjugant colonies for Pri⁺/Pri⁻ recipients was 1.1 ± 0.05 (two trials). In crosses involving Salmonella donors and recipients, the ratio was 3.3 ± 0.7 (four trials).

If in Salmonella the R1162 primase increases the number of transconjugants because it initiates complementary strand synthesis in the recipient, then oriR would be utilized and deletion of this priming site should have a greater effect on the transfer frequency than deletion of oriL. We therefore carried out a separate experiment to compare the transfer frequencies of oriV⁺ plasmids lacking one or the other of these sites. The results (Fig. 4) indicate that oriR is responsible for the increase in transfer into Salmonella.

FIG. 4.

Bar graph showing relative number of transconjugants obtained for test plasmids deleted for one or both initiation sites for replication compared to plasmids containing the complete oriV. The average numbers of transconjugants used to construct the graph are shown below. Data are derived from four or five independent trials.

Contribution of the R751 primase to plasmid recovery in recipient cells.

Our results showed that in Salmonella recipients, the R1162 priming system was involved in synthesis of the complement to the transferred strand. In fact, without this system, the number of transconjugants was extremely low (pSC101 oriT and pSC101 Mob [see Fig. 3] and ΔoriR plasmids [see Fig. 4]). Does the presence of the R751-encoded priming system, which is absent from the donors used for Fig. 3 and 4, increase the transfer frequency of the test plasmid into Salmonella? This seemed likely, since Lanka and Barth (18) showed that the RP4 primase, which is related to the R751 primase, increased not only the transfer frequency of this plasmid into Salmonella, but also the mobilization frequency of plasmid R300B (nearly identical to R1162). We compared the number of transconjugants in Salmonella-Salmonella matings for TraC⁺ and TraC⁻ donors. We used the electroporation and mating protocol, with the R1162 Mob proteins in the donors and with the transferring plasmid containing the R1162 oriT. We found that, as for the E. coli matings shown in Table 1 (lower grouping), the R751 primase appeared to reduce the number of transconjugants (ratio of the number of TraC⁺ to TraC⁻ transconjugants, 0.10 ± 0.10 for three independent trials). These results showed that, contrary to expectation, the TraC protein inhibited transfer of the test plasmid.

The electroporation and transfer matings might not reflect the role of TraC under normal mating conditions. In particular, we wanted to test whether TraC increased the transfer frequency of R751 and then determine how this protein was affecting the transfer of our test plasmid under the same conditions. The transfer systems of R751 and RP4 are very closely related, so these two IncP-1 plasmids should behave similarly.

We carried out three sets of crosses, from E. coli into E. coli, E. coli into Salmonella, and Salmonella into Salmonella. For the interspecific crosses, the recipient Salmonella strain was restrictionless (5). The donor cells contained R751 or R751ΔtraC and the R1162-derived test plasmid, used in the electroporation experiments, having the R1162 oriT and oriV. This test plasmid was maintained in the donor cells by a third, nonmobilizable plasmid that provided the R1162 proteins for mobilization and replication. As before, the cells were rescued in the recipient by λ integrase-mediated recombination into the chromosome. The results of this experiment are summarized in Fig. 5.

For the matings involving only E. coli, the R751 primase inhibited transfer slightly for R751 itself and, in agreement with the data in Table 1 (lower grouping), for the mobilizable test plasmid. When the restrictionless strain of Salmonella was the recipient, the primase stimulated transfer of R751, but transfer of the test plasmid continued to be inhibited slightly. This failure of the R751 primase to stimulate transfer of the test plasmid is not due to a peculiarity of the R1162 derivative being used. When donor strains contained instead R1162 itself, unchanged from the naturally occurring plasmid except for a cloned chloramphenicol resistance gene (1) and either R751 or R751ΔtraC, the results were the same. The TraC⁺/ΔTraC transfer frequency ratio was 4.3 ± 0.8 for the R751 pair and 0.3 ± 0.05 for R1162 (three trials). Thus, the effect of TraC was specific for the IncP1 plasmid, even though it initiates synthesis on DNA at positions that do not have a high degree of base sequence specificity (20). In this way, it mimics the more specific R1162 primase, which has no effect on the transfer of R751ΔtraC into Salmonella (data not shown).

The plasmid-specific effect of TraC on transfer of R751 suggested that the primase might recognize particular sites on the plasmid. Yakobson et al. (34) showed that DNA containing the oriT of RK2, another IncP plasmid, partially complemented in cis an oriC-defective M13 phage. The complementation required TraC in the cell, provided either by RK2 or by R751. One of the complementing sites, furthermore, was oriented in the plasmid so that it could potentially initiate replacement strand synthesis on the transferred strand. We planned to test similarly whether fragments of R1162 DNA could complement for growth the ΔoriC phage M13ΔE101 (15). However, we found that the phage by itself, without any cloned DNA, formed large plaques on JM103 (R751), but not on either JM103 (R751ΔtraC) or JM103 alone (Fig. 6A to C). The plaques on the R751-containing strain were similar in size to those produced by phage containing a primosome assembly site able to complement the oriC defect (Fig. 6E). In contrast, only small plaques were obtained on cells containing RK2 (Fig. 6D).

FIG. 6.

Complementation of ΔoriC phage M13ΔE101, indicated by large plaque size. Plaques of M13ΔE101 on JM103 (A), M13ΔE101 on JM103(R751) (B), M13ΔE101 on JM103(R751ΔtraC) (C), and M13ΔE101 on JM103(RK2) (D). As a positive control, M13ΔE101 containing a primosome assembly site able to complement the oriC defect was plated on JM103 (panel E).

DISCUSSION

Restoration of a transferred plasmid strand to the duplex DNA form is an essential step for successful conjugative transfer. The broad host-range plasmids examined here, and perhaps other plasmids as well, can use more than one mechanism to initiate synthesis of the complementary strand. When the recipient cell is E. coli, there is an endogenous mechanism for replacement strand synthesis in the cell; in Salmonella, the corresponding system is considerably less robust, and plasmid-specific priming mechanisms become important. For R1162, this is the priming system normally used for vegetative replication of the plasmid and active also in strand replacement in the donor.

While looking at the role of the R1162-encoded primase in transfer, we also wanted to examine the effect of the R751 primase, since this protein and the related primases of other IncP1 plasmids are frequently present in donor cells containing IncQ group plasmids. Lanka and Barth (18) showed that primase of the IncP1 plasmid RP4 increased the transfer frequency of this plasmid from E. coli into S. enterica serovar Typhimurium, a result we also observe here for the R751 primase (Fig. 5). However, the two primases appear to differ in their effect on the comobilization of R1162 and its derivatives. Whereas the RP4 primase seemed to increase the mobilization of the IncQ plasmid R300B (nearly identical to R1162) in the Barth and Lanka study, in our work the R751 primase did not increase the transfer of our test plasmids, and in fact inhibited slightly their mobilization (Table 1 and Fig. 5). In the Lanka and Barth study, the mutations in the primase gene were generated by Tn7 mutagenesis, and crosses into Salmonella were from E. coli donors. Under these conditions, restriction reduces the transfer frequency, possibly revealing slight polar effects invisible at higher frequencies.

It was shown earlier that the RP4 primase in donor cells cannot complement for transfer a ColI plasmid lacking its own primase gene (24). Taken with the failure of the R751 primase to substitute for the R1162 primase in our experiments, it appears that in general the primases encoded by IncP1 plasmids show a surprising degree of specificity during conjugative transfer (19). The inability of the R751 primase to complement the R1162 primase is all the more remarkable in view of the vigorous activity of the TraC protein in trans on foreign substrates. While both the R751 and RP4 primases effectively suppress a temperature-sensitive dnaG mutation, the R751 primase is better than the RP4 version in this regard (20). Moreover, the R751 primase effectively complements the oriC defect in M13ΔE101 (Fig. 6). These observations suggest that the failure of the R751 primase to complement R1162 for transfer is not because it is weakly active or present in small amounts in the cell. The specificity during transfer could reflect the fact that single-stranded DNA is unwound from the complementary strand and transported through the type IV secretory pore. There might be only a short time when proteins have access the single-stranded DNA prior to transport, so that only primases made in cis are likely to find this DNA. An argument against this is that a TraC⁻ plasmid is complemented for transfer by traC cloned on another plasmid (24). Possibly, an overproduction of primase increases the concentration of the protein sufficiently so that it can find the transferring DNA; interestingly, complementation never fully restores the frequency of transfer to normal levels (24).

A ΔtraC, IncP1 plasmid is also complemented by primase in the recipient (24). In contrast, transfer of our test plasmids from donors lacking the R1162 primase is not greatly increased if the recipient contains this protein (reference 11 and results not shown). The difference might reflect the smaller size of the test plasmid and the high site specificity of its primase; both of these would reduce the chances of the protein making an effective contact with the incoming DNA. We detect an apparent threefold increase in the number of transconjugants when a Salmonella recipient contains the primase; it would be interesting to see if that number increases when a larger plasmid is being transferred.

What is the nature of the endogenous system for strand replacement in E. coli recipients? There has been speculation for some time that DNA entering a cell by conjugation is acted upon by a cellular priming system (18), but identifying the mechanism of replacement strand synthesis has proven difficult. We used without success the M13 derivative with a defective complementary strand origin to search for single-strand initiation sites on our mobilized test plasmids (data not shown). Synthesis on an entering strand is probably much less demanding, in terms of required initiation rate, than phage DNA replication, and it thus might not be identified in this assay. Other experiments also suggest that R1162 contains no efficient, single-strand initiation sites other than oriL and oriR. When either of these sites is deleted, the plasmid is unstable but is still replicated (2, 35). It remains to be determined whether the mechanism allowing replication of an R1162 derivative lacking oriL or oriR is also responsible for recovery of transferred DNA.