INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Miniplasmid pMU720, a derivative of a large, low-copy-number, conjugative plasmid, pMU707, belongs to incompatibility group B (8) and is closely related to the other members of the I complex (IncI₁, IncI, IncK, and IncZ plasmids), with whom it has extensive sequence homology (11, 18, 29, 34). Replication of pMU720 requires the expression of the repA gene, which codes for a protein (RepA) that is rate limiting for replication. Extensive studies of pMU720 and the IncI₁ plasmid ColIb-P9 established that the expression of the rep gene requires the translation and appropriate termination of a leader peptide, which facilitates the formation of an RNA pseudoknot immediately upstream of the Shine-Dalgarno sequence of repA (1-3, 17, 35, 45). Formation of this RNA tertiary structure activates the translation of the repA mRNA (2, 3, 45). Expression of repA, and consequently the copy numbers of pMU720 and ColIb-P9, is regulated by a small, highly structured antisense RNA molecule (RNAI) which is complementary to the leader region of the repA mRNA (33, 39). Binding of RNAI to its target in repA mRNA inhibits repA expression both directly, by sequestering one of the two complementary sequences involved in the formation of the pseudoknot, and indirectly, by inhibiting translation of the leader peptide (1, 4, 40, 41, 45, 46).

Initiation of DNA replication requires the presence of a replication initiator protein and of a cis-acting DNA sequence, the origin of replication (ori). The binding of the initiator protein to ori is the first step in DNA replication. The ori of the ColIb-P9 plasmid was proposed to lie within a 172-bp sequence located 152 bp downstream of the coding sequence of repA (42). This proposal was based on the finding that the designated sequence is essential for the replication of ColIb-P9 and contains such features common to many origins of replication as a recognition sequence for DnaA protein (DnaA box), AT-rich sequences, and repetitive sequences. The intervening region between repA and ori, denoted CIS, was shown to encode transcription termination signals, which were required for efficient replication of ColIb-P9 (27). Regions of CIS lying downstream of these signals appeared to also be required, although no function had been assigned to them (27).

The requirement for CIS, in addition to repA and ori, in DNA replication of ColIb-P9 and the arrangement of these genes vis-à-vis each other resembles the situation described for the IncFII plasmids R1 and R100, which are distantly related to the plasmids of the I complex. The Rep protein of these plasmids is unusual in that it acts preferentially in cis i.e., it preferentially activates the origin of replication of the DNA molecule from which its messenger RNA was transcribed (14, 22, 24). This property of Rep is dependent on the presence of CIS, in its native position and orientation, and on the transcription-terminating activity of CIS (22). The discovery of weak binding sites for Rep in the C terminus of the repA coding sequence has led to the suggestion that the newly synthesized Rep is loaded onto the DNA through these sites and then translocates on the DNA until it reaches its primary binding sites in ori (22). An alternative hypothesis has been put forward by Maas and Wang, on the basis of studies of the FIC replicon of plasmid P307 (21). They proposed that the repA mRNA is not only the messenger RNA for RepA but also a preprimer for initiation of replication and that Rep acts by facilitating the processing of this transcript into the active primer at a site overlapping the rep stop codon. Neither of these two conflicting models for the initiation of replication of IncFII and FIC plasmids has been verified.

The Rep protein of I-complex plasmids is also thought to act preferentially in cis (27, 28, 30), but little is known about the system that keeps the newly synthesized Rep tethered to its DNA template and then loads it onto ori. In this paper we describe mutational analyses of the CIS and ori sequences of pMU720. We find that (i) the RepA protein shows specificity for ori but not for CIS, (ii) deletion of the DnaA box reduces the efficiency of replication of pMU720, (iii) deletion of the sequence immediately downstream of the DnaA box produces a plasmid that is unable to replicate, (iv) the repA-proximal portion of CIS is required for efficient replication but can be replaced by the corresponding sequences from other I-complex plasmids or from IncL/M plasmid, (v) the repA-distal portion of CIS is a spacer which can be replaced by an unrelated sequence of correct length, (vi) the role of this spacer is to position sequence(s) in ori on the correct face of the DNA helix vis-à-vis the repA-proximal portion of CIS, and (vii) the length of the CIS spacer can be reduced by one or increased by four helical turns without affecting the efficiency of replication. A model describing the role of CIS in loading the RepA protein onto ori is presented.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and phages. The strains of Escherichia coli K-12 used in this study are given below. JM101 [(lac-proAB) supE thi F'(traD36 proA⁺B⁺ lacI^qZM15)] (23) was used for cloning and propagating M13 derivatives. XL1 Blue MRF' [(mcrA)183 (mcrCB-hsdSMR-mrr)173 endA1 supE44 thi-1 recA1 gyrA96 relA1 lac [F' proAB lacI^qZM15 Tn10 (Tet^r)] (Stratagene) was used to grow M13 derivatives which had undergone mutagenesis as described by Vandeyar et al. (43). JP3438 (thr-1 leuB6 thi-1 lacY1 gal-351 supE44 tonA21 hsdR4 rpoB364 recA56) was used for propagating pMU720 derivatives and for all copy number determinations.

Media, enzymes, and chemicals. The minimal medium used was half-strength buffer 56 (26) supplemented with 0.2% glucose, thiamine (10 µg/ml), and necessary growth factors. Enzymes and chemicals of a suitable grade were purchased commercially and not purified further. ³⁵S-dATPS (>1,000 Ci/mmol) for use in sequencing was obtained from Amersham Corporation. Ampicillin was used at a final concentration of 50 µg/ml, chloramphenicol was used at 10 µg/ml, isopropylthiogalactoside (IPTG) was used at 1 mM, and 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) was used at 25 µg/ml.

Recombinant DNA techniques. Plasmid and bacteriophage DNA were isolated and manipulated as described by Sambrook et al. (36). DNA was sequenced with a model 373 DNA sequencer and ABI PRISM dye terminator kits (Perkin-Elmer Corporation) or by the method of Sanger et al. (37), modified in that T7 DNA polymerase was used instead of the Klenow fragment and terminated chains were uniformly labeled with ³⁵S-dATPS. Oligonucleotide-directed in vitro mutagenesis reactions were performed on single-stranded M13 templates with a kit from United States Biochemical Corp. Oligonucleotides were purchased from Bresatec Ltd. or Gibco BRL. DNA sequencing was used to screen for and confirm the presence of mutations.

Construction of plasmids for use in copy number determinations. The plasmids for copy number determinations were derived from pMU1598 (Fig. 1) and pMU1599 by replacement of the HindIII-SacI and SacI-SacII fragments carrying wild-type CIS and ori, respectively, by appropriate fragments containing the mutations to be tested. Plasmids pMU1598 and pMU1599 contain both the IncB replicon from pMU720 (nucleotides [nt] 1 to 2170) and the replicon from pAM34 (15). The latter is a modified pMB1 replicon in which the essential preprimer RNA is transcribed from the lacZ promoter operator. Since pMU1598 and pMU1599 contain the lacI^q gene, replication of their pAM34 replicons requires the presence of a lac inducer, such as IPTG. Thus, in the absence of IPTG, replication of pMU1598 and pMU1599 is dependent on the IncB replicon. As well as allowing the rescue of replication-defective IncB replicons, these plasmids permit determination of relative copy numbers by making use of a chloramphenicol acetyl transferase (CAT) reporter gene (derived from pACYC184), which is expressed constitutively. Plasmid pMU1598 differs from pMU1599 in that it carries the substitution T642C in the rnaI promoter, which, by reducing the expression of this gene, increases the synthesis of RepA and hence the copy number of the IncB replicon.

View larger version (65K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the dual-origin plasmid used to construct mutants carrying deletions and substitutions in the CIS and ori regions of the IncB replicon and to determine their relative copy numbers. Features important for the use of this plasmid and its derivatives are shown. RNAI, antisense RNA regulating the expression of repA of the pMU720 replicon; RNAII, repA-mRNA; CAT, gene for CAT conferring resistance to chloramphenicol; ApR, gene conferring resistance to ampicillin (bla). The triangles denote transcription terminators; , T4 gene 32 terminator; , pheR terminator. Unique restriction endonuclease sites are shown in boldface.

Measurement of CAT activity. The CAT activity of mid-log-phase cultures, grown in minimal medium containing 0.4% glucose, thiamine, leucine, threonine, ampicillin, and chloramphenicol, was assayed as described by Shaw (38). The cells were disrupted by sonication with a Braun Labsonic 2000 sonicator, and cellular debris was removed by centrifugation before the assays were carried out. Each assay was performed at least six times. CAT activity was expressed as units per milligram of protein.

Protein assay. The concentration of protein in cleared cell lysates was determined by the method of Bradford (9), using bovine serum albumin as a standard.

Measurement of -galactosidase activity. The -galactosidase activity of mid-log-phase cultures was assayed as described by Miller (25). Each sample was done in duplicate, and each assay was performed at least three times.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Characterization of sequences downstream of repA that are essential for replication of the IncB plasmid. In order to facilitate the analysis of sequences lying downstream of repA, we constructed a dual-origin plasmid, pMU1598 (Fig. 1), which contains both the IncB replicon and a fully repressible pMB1 replicon (15). The pMB1 replicon is inactive in the absence of a lac inducer but can be switched on by the addition of IPTG to rescue mutations which inactivate the IncB replicon. The IncB replicon of pMU1598 differs from that of pMU720 in that (i) nt 2171 to 3251 are deleted; (ii) it has a number of unique restriction endonuclease recognition sites, which have been created to facilitate the exchange of DNA fragments; and (iii) its rnaI gene carries a mutation (underlined) in the 10 region (RNAI.1; TATACT to TGTACT), which reduces its level of expression and consequently increases the level of expression of repA and the plasmid copy number (44). pMU1598 carries a gene encoding constitutively expressed CAT, which can be used as a reporter to assess the plasmid copy number. By using this reporter, it was found that the copy number of pMU1598 was not significantly different from that of pACYC184 (data not shown), which had been reported to be approximately 20 per E. coli chromosomal equivalent (10). Creation of the BglII site at positions 737 to 742 (which changed codons 6 and 7 of repA), of a HindIII site at positions 1759 to 1764 (5 nt downstream of the repA stop codon); and of a SacI site at positions 1912 to 1917 (2 nt upstream of the DnaA box) had no significant effect on the copy number (data not shown). Replacing the mutant rnaI (RNAI.1) by the wild-type gene, by exchanging the BamHI-SphI fragment of pMU1598 for the BamHI-SphI fragment from pMU720, reduced the copy number of the IncB replicon ~4-fold (Fig. 2), which is in reasonable agreement with data obtained previously (44).

View larger version (50K): [in this window] [in a new window]   FIG. 2.   Effects of deletions in CIS and ori and of reversing the polarity of ori on the relative copy number of the IncB replicons of pMU1598 and pMU1599. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598 or its derivatives in the absence of IPTG and is expressed relative to the value obtained for pMU1598. The values shown are the averages of at least six independent determinations. The dashed lines represent deleted sequences, and the open boxes represent the DnaA boxes. The map coordinates shown correspond to those of the published sequence of pMU720 (34). ND, not done; NR, plasmid unable to initiate replication from the IncB replicon.

RepA does not have base-specific interactions with CIS. The RepA protein of the IncB plasmid shows ~95% sequence identity with the RepA proteins of the closely related I-complex plasmids R144-3 (IncI₁), R64-11 (IncI₁), R621a (IncI), and R387 (IncK) but has no significant homology with the RepA protein of the IncZ plasmid pIE545 (34). The CIS and ori sequences of these I-complex plasmids show a level of homology with each other similar to that exhibited by their RepA proteins (34).

View larger version (38K): [in this window] [in a new window]   FIG. 3.   Effects of substitutions of CIS and ori on the relative copy numbers of the IncB replicons of pMU1598 and pMU1599. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598 or its derivatives in the absence of IPTG and is expressed relative to the value obtained for pMU1598. The values shown are the averages of at least six independent determinations. trpA ter is a 28-bp trpA transcription terminator flanked by linker sequences so that the overall length of 5'CIS is maintained. ND, not done; NR, plasmid unable to initiate replication from the IncB replicon.

Role of 3'CIS. The finding that 3'CIS (nt 1844 to 1917) could be separated from 5'CIS (the repA-proximal portion of CIS; nt 1759 to 1843) with little loss of replication activity and that it could be deleted from the RNAI.1 mutant, provided the polarity of the ori region was reversed, raised the possibility that 3'CIS was a spacer, whose sole role was to keep sequence(s) within ori at the appropriate position with respect to repA and/or 5'CIS. To test this possibility, the entire CIS (nt 1759 to 1917) and 3'CIS were replaced by sequences of equivalent lengths, which were amplified from the coding region of the aminoglycoside 3'-phosphotransferase of Tn903. Replacement of 3'CIS by the spacer fragment (spacer 1) had no significant effect on the efficiency of replication of the IncB replicon of the dual-origin plasmid, even in the presence of the wild-type rnaI gene (Fig. 4). Surprisingly, replacement of the entire CIS by a spacer fragment (spacer 2) did not inactivate the IncB replicon, although its ability to replicate was severely reduced (Fig. 4). Once again, the increase in synthesis of RepA, caused by the reduction in the expression of rnaI, compensated in part for the loss of CIS, so that the copy number was reduced only fourfold. In contrast, the copy number of the plasmid carrying wild-type rnaI was reduced to the point that its transformants exhibited severe slowing in growth rate, producing very small colonies even after prolonged incubation, and their CAT activity could not be measured. Insertion of spacer fragment 2 between the promoter and the lacZ of pMU1597 had no significant effect on repA-lacZ expression, confirming that this fragment does not contain a transcription terminator. These data suggest that simply separating the RepA coding region from ori may be sufficient to permit replication to occur when repA expression is elevated. However, the spacer sequence must be of sufficient length, as replacing CIS by a 28-bp linker produced a plasmid that was unable to replicate (Fig. 2).

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Effect of replacement of the entire CIS and 3'CIS by spacer fragments of equivalent length, amplified from the coding sequence of aminoglycoside 3'-phosphotransferase, on the relative copy number of the IncB replicon of pMU1598 and pMU1599. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598 or its derivatives in the absence of IPTG and is expressed relative to the value obtained for pMU1598. The values shown are the averages of at least six independent determinations. NA, not able to be assayed.

Effect of moving ori sequences along the face of the DNA helix. To determine whether it was important for sequences within ori to be at a particular position on the face of the DNA helix, the length of 3'CIS was increased by 1 to 9 bp, in small increments, and the effects of these changes on the copy number of the IncB plasmid were measured. Because data (presented in this paper) indicated that increased expression of repA can compensate for decreased efficiency of RepA-ori interactions, these experiments were carried out on both pMU1598 and its derivative carrying the wild-type rnaI gene. The changes in the copy number, relative to those of the two parental plasmids, are presented in Fig. 5. These data show that increasing the length of 3'CIS by 1 bp resulted in a slight but significant increase in the copy number of the plasmid carrying the wild-type rnaI gene but had no effect on pMU1598. Increasing the length of 3'CIS by 2 or 9 bp had no significant effect, but adding 3, 4, 5, or 7 bp reduced the copy number, with the plasmid carrying the wild-type rnaI gene being much more severely affected than pMU1598. The greatest reduction in copy number was seen with plasmids whose 3'CIS was increased by 4 to 7 bp, which moved sequences within ori approximately half a helical turn relative to sequences within repA and 5'CIS.

View larger version (38K): [in this window] [in a new window]   FIG. 5.   Effect of increasing the length of 3'CIS, by adding 1 to 9 nt at map position 1912, on the copy number of the IncB replicon of pMU1598 and pMU1599. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598, pMU1599, or their derivatives in the absence of IPTG and is expressed relative to the value obtained for the parental plasmid (i.e., either pMU1598 or pMU1599, as appropriate). The values shown are the averages (± standard deviations) of at least six independent determinations. WT, wild type.

What are the limits to the length of 3'CIS for efficient plasmid replication? To determine the permissible limits for the spacing between the RepA coding sequence or 5'CIS and ori, the effects on the copy number of decreasing and increasing the length of 3'CIS by multiples of DNA helical turns (10.5 bp) were measured. As seen from the data in Fig. 6, shortening 3'CIS by 32 bp, which meant that it was only 44 bp long, reduced the copy number of the plasmid carrying wild-type rnaI ~12-fold, and CIS could not be shortened further without the total loss of the ability to replicate. The plasmid carrying the RNAI.1 mutation could still replicate when 3'CIS was shortened by 42 and 53 bp, but its copy number was reduced ~6- and 30-fold, respectively. Shortening 3'CIS by 21 bp reduced the copy number of the plasmid carrying wild-type rnaI twofold, but deleting 11 bp or inserting 11 or 42 bp had little effect. Insertions of 63 bp or more caused progressive decreases in the copy number. These data show that efficient replication of the IncB replicon requires that ori be separated from 5'CIS by no less than 55 and no more than 118 bp.

View larger version (43K): [in this window] [in a new window]   FIG. 6.   Effect of increasing or decreasing the length of 3'CIS by full turns of the helix (i.e., ~10.5 bp) on the copy number of the IncB replicon of pMU1598 and pMU1599. The length of 3'CIS was altered by replacing nt 1844 to 1911 by spacers of appropriate length, except for the +42 mutant, which was constructed by insertion of a 42-bp polylinker at map position 1844. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598, pMU1599, or their derivatives in the absence of IPTG and is expressed relative to the value obtained for the parental plasmid (i.e., either pMU1598 or pMU1599, as appropriate). The values shown are the averages (± standard deviations) of at least six independent determinations. WT, wild type.

Spacing between 5'CIS and ori is critical for efficient replication. To determine whether the effects observed when the length of 3'CIS was altered were due to changes in the spacing between ori and 5'CIS or between ori and the end of the RepA coding sequence, insertions and deletions were introduced in 3'CIS or immediately downstream of the repA stop codon. The effects of these mutations on the copy numbers of plasmids expressing wild-type or reduced levels of RNAI were determined, and they are presented in Fig. 7. The data show that insertion of 4 bp immediately downstream of the repA stop codon had no effect on the copy number, which is in stark contrast to the severe reduction in copy number seen when the 4 bp is inserted into 3'CIS. Similarly, the reduction in copy number caused by deleting 14 bp from 3'CIS was completely reversed by the introduction of 4 bp into 3'CIS but not by the introduction of 4 bp between the RepA coding sequence and 5'CIS. These data show that it is the spacing between ori and 5'CIS that is crucial for efficient replication and that the helical phasing of 3'CIS must be preserved. The introduction of 63 bp immediately downstream of the repA stop codon had no effect on the copy number, but insertion of 244 bp decreased the copy number ~2.6- and 6-fold in the plasmid carrying the mutant and wild type rnaI, respectively. Thus, the spacing between the RepA coding sequence and CIS and/or between the RepA coding sequence and ori is also important, as only relatively short insertions (longer than 63 but shorter than 244 bp) are tolerated.

View larger version (38K): [in this window] [in a new window]   FIG. 7.   Effect of deleting and inserting sequences immediately downstream of the repA stop codon, at the boundary between 5'CIS and 3'CIS, or at the end of 3'CIS on the relative copy numbers of the IncB replicons of pMU1598 and pMU1599. The copy number was determined by assaying the CAT activity of E. coli JP3438 carrying pMU1598 or its derivatives in the absence of IPTG and is expressed relative to the value obtained for pMU1598. The values shown are the averages of at least six independent determinations. NA, not able to be assayed.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have constructed an IncB plasmid carrying a ColE1-type replicon that is inactive in the absence of a lac inducer but whose induction permits the rescue of replication-defective mutants of the IncB replicon. This plasmid also carries a constitutively expressed CAT gene, so that its CAT activity is directly proportional to its copy number. This dual-origin plasmid was used to construct mutants carrying deletions, insertions, and substitutions within the CIS and ori regions of the IncB replicon and to accurately measure the effects of these mutations on the efficiency of replication of this replicon.

The minimal origin of replication of pMU720 was mapped to a sequence lying immediately downstream of the DnaA box. This sequence has base-specific interactions with RepA, as indicated by the finding that the RepA protein of pMU720 showed a preference for its own ori. The RepA protein of pMU720 was able to activate the ori of an IncL/M plasmid, pMU407.1, but not that of an IncZ plasmid. Examination of these two ori sequences revealed that the first contained a motif (5'A/TANCNGCAAA/T3') which is also present in the ori of pMU720. This motif, which was not found in the ori of the IncZ plasmid, was repeated four and two times in the ori of the IncB and IncL/M plasmids, respectively. It is thus tempting to speculate that this motif represents the binding site of RepA. In both pMU720 and pMU407.1 a copy of this motif is present 2 bp downstream of the DnaA box, and preliminary data indicate that in vitro the RepA protein of pMU720 binds immediately downstream of this box (7). Although the DnaA box is not essential, its deletion reduced the copy number of the IncB replicon threefold. Increased expression of RepA compensated in part for the loss of the DnaA box. The minimal ori of IncFII plasmids also lies immediately downstream of a DnaA box (31). Moreover, an R1 plasmid whose DnaA box has been inactivated by mutations so that it no longer binds the DnaA protein in vitro replicates very inefficiently unless its RepA protein is overexpressed (32). Similarly, the E. coli chromosome of a dnaA null mutant, whose replication was placed under the direction of an integrated R1 plasmid, was underreplicated unless the synthesis of RepA was increased (6).

The CIS of pMU720 can be separated into two domains, which have different functions and can be moved apart from each other. The 3' domain is a spacer whose sole function is to place a sequence(s) within ori at an appropriate distance and helical phasing, relative to sequences in 5'CIS. 5'CIS possesses strong transcription termination activity, but it is unlikely that terminating the repA mRNA is its only function. Replacing 5'CIS by a fragment of equal length, carrying the trpA transcription terminator, led to a severe reduction in replication activity. Since transcription termination in this mutant is predicted to occur at the same point, relative to the repA stop codon and ori, as the predominant termination site mapped in ColIb-P9 (27), it is unlikely that the replication defect of this mutant is due to an inappropriate site of termination. However, it is possible that the system for the loading of RepA onto ori requires that termination of transcription of repA be Rho dependent, as has been suggested for plasmid R1 (22). The proposed reason for this requirement was that Rho-dependent termination is likely to take longer to complete, allowing the ribosome translating repA mRNA time to reach the stop codon before this transcript is released from the template (22). Transcription termination in ColIb-P9 was shown to be only partly Rho dependent (27), and it is unclear whether this dependence was important for efficient replication, because the ColIb-P9 derivative used in the study was a mutant whose copy number was elevated ninefold due to increased expression of repA (4). In pMU720, the replication activity of the mutant carrying the trpA terminator could be largely restored by increasing the synthesis of RepA, suggesting that under these conditions the native system for loading RepA onto ori may be bypassed. However, it should be pointed out that replication initiation in the mutants expressing repA at elevated levels requires the presence of an intervening sequence between ori and the repA coding sequence, indicating that their RepA proteins are still acting in cis.

The 5'CIS sequences of the IncZ and IncL/M plasmids were found to be adequate replacements for the 5'CIS of the IncB plasmid, despite the lack of any discernible homology between them, indicating that they contain functional features lacking in the trpA terminator. The IncZ replicon is a naturally occurring chimera in which the sequences upstream of the repA coding sequence are highly homologous to those of the IncB and other I-complex plasmids, whereas the repA coding sequence, CIS, and ori are highly homologous with those of the IncFII plasmids R1 and R100 (34). The RepA protein of the IncL/M replicon shows 40% identity to RepA of pMU720 (5), the ori sequences of these two plasmids show strong homology at their 5' ends, and the RepA of pMU720 is able to initiate replication at the ori of the IncL/M replicon. Thus, it is reasonable to assume that the IncL/M, IncFII, and I-complex plasmids use very similar mechanisms to load their Rep proteins onto ori.

A recent model postulates that replication of the IncFIC replicon of P307 initiates at the stop codon of repA and involves the processing of the repA mRNA at that site to produce an RNA primer for leading-strand synthesis (21). The IncFIC replicon of P307 is a naturally occurring chimera which is the mirror image of the IncZ replicon in that its sequences lying upstream of the repA coding sequence are highly homologous to those of the IncFII plasmids R1 and R100, whereas the repA coding sequence, CIS, and ori are highly homologous with those of the IncB plasmid (20, 21, 34). Given these similarities between their sequences, it is reasonable to assume that the IncFIC replicons of P307 and pMU720 use the same strategy for initiation of replication. However, it is difficult to envisage how our data, showing the necessity for maintaining correct distance and native helical phasing between 5'CIS and ori, can be reconciled with the model put forward by Maas and Wang (21).

There are two lines of evidence suggesting that the mechanism for loading Rep onto ori is unlikely to involve loading onto secondary binding sites in the C terminus of the rep coding sequence followed by translocation to ori. Firstly, the requirement for preserving the helical phasing between 5'CIS and ori suggests that the newly synthesized RepA finds its initial target within 5'CIS. Secondly, the secondary binding sites detected in R1 are not conserved in the IncZ replicon, despite the strong homology between the RepA proteins and the CIS and the ori sequences of these two plasmids. If the secondary binding sites were an integral part of the replication machinery of these plasmids, there would have been a strong selective pressure for their retention. It is noteworthy that the sequence lying immediately downstream of the DnaA box, which is thought to be the primary binding site for RepA (16), has been conserved.

The requirement that the spacing between the repA coding sequence and 5'CIS be kept relatively short is consistent with the notion that the nascent protein has to interact with nucleotide sequences or with the RNA polymerase transcribing its messenger RNA in order to be efficiently loaded onto ori. If RepA is recognizing a specific nucleotide motif, then this motif must be present in the 5'CIS of all the plasmids tested in this work. Although some potential motifs can be found in 5'CIS, they are relatively short and not well conserved, and most importantly, the helical phasing between these sequences and ori is not always maintained. For these reasons we favor the notion that the first step in the loading of RepA onto ori involves an interaction between the nascent RepA and the RNA polymerase transcribing its messenger RNA. RepA is then translocated onto ori by a process which does not involve passive sliding along the DNA until its primary binding site in ori is encountered but may involve bending of the DNA. Although at present there is no information regarding DNA bending near the ori of I-complex plasmids, static DNA bending was found to occur within the CIS of the IncFII plasmid R100 (13). This model for the loading of RepA onto ori accommodates the data presented in this paper and explains the importance of transcription terminating at the appropriate position in 5'CIS, as well as the cis action of RepA.