DnaA protein is not essential for replication of IncFII plasmid NR1

By transformation of dnaA null mutant host cells that are suppressed either by an rnh mutation or by chromosomal integration of a mini-R1 plasmid, it was shown that replication of miniplasmids composed of the NR1 minimal replicon had no absolute dependence upon DnaA protein. In addition, the suppression of the dnaA null mutation by the integrated mini-R1, which is an IncFII relative of NR1, was found to be sensitive to the expression of IncFII-specific plasmid incompatibility. This suggests that the integrative suppression by mini-R1 is under the control of the normal IncFII plasmid replication circuitry. Although NR1 replication had no absolute requirement for DnaA, the copy numbers of NR1-derived miniplasmids were lower in dnaA null mutants, and the plasmids exhibited a much reduced stability of inheritance during subculture in the absence of selection. This suggests that DnaA protein may participate in IncFII plasmid replication in some auxiliary way, such as by increasing the efficiency of formation of an open initiation complex at the plasmid replication origin. Such an auxiliary role for DnaA in IncFII replication would be different from that for replication of most other plasmids examined, for which DnaA has been found to be either essential or unimportant.

mutants, and the plasmids exhibited a much reduced stability of inheritance during subculture in the absence of selection. This suggests that DnaA protein may participate in IncFII plasmid replication in some auxliary way, such as by increasing the efficiency of formation of an open initiation complex at the pasid repicin origin. Such an auxiliary role for DnaA in IncFII replication would be different from that for replcaion of most other plasmids examined, for which DnaA has been found to be either essential or unimportant.
Escherichia coli DnaA protein plays an essential role in the initiation of chromosomal DNA replication by binding to specific sites in the replication origin, oriC (5, 12-14, 18, 25, 26, 52). Many plasmids, including IncFII plasmids such as NR1, also contain DnaA protein-binding sites, referred to as DnaA boxes, in or near their replication origins (1,12,33,37,40,58). However, most of these plasmids encode their own plasmid origin-specific replication initiation proteins (1,33,37,40,45,49). Previously, the abilities of many plasmids to replicate in a dnaA(Ts) mutant host after a shift to the nonpermissive temperature (11,15,43,50,55) or to integratively suppress a dnaA(Ts) mutation (6,7,29,35,50) were taken as evidence for the DnaA independence of these plasmids. One exception was plasmid pSC101, which was incapable of replication at the nonpermissive temperature in a dnaA(Ts) host (10,11,17). Most recently, however, with the use of dnaA null mutants that contain no detectable DnaA protein, it has been shown that plasmids mini-F and mini-Pi have an absolute requirement for DnaA protein for their replication (16,19,32). The ability of the mini-F and mini-Pi plasmids to integratively suppress dnaA(Ts) mutations has therefore been ascribed to the leakiness of the mutant DnaA protein in dnaA(Ts) hosts (21). In contrast, the replication of ColEl-like plasmids was found to be independent of DnaA protein both in vivo (16,19) and in vitro (38).
The survival of dnaA null mutants is dependent on one of two means of suppression: a secondary mutation in the RNase H gene, rnh, or chromosomal integration of a miniplasmid derived from IncFII plasmid Ri (16,19,20,36). The integrative suppression of dnaA null mutants by mini-Rl was taken to indicate that Ri plasmid replication was DnaA independent. However, no direct tests of this hypothesis have been presented. More recently, from studies of the in vitro replication of IncFII plasmid DNA in extracts prepared from dnaA(Ts) mutant cells, it has been suggested that replication of IncFll plasmids has a strict dependence on DnaA protein in the in vitro systems (24,38). It has also been suggested that the ability of the mini-Ri plasmid to integra-* Corresponding author. tively suppress a dnaA null mutant might be explained if the mini-Ri plasmid contained a second, DnaA-independent replicon (24,38).
In this article, we present evidence that replication of the IncFll plasmid NR1 in vivo has no absolute dependence on DnaA protein and that integrative suppression of a dnaA null mutation by mini-Rl results from the normal mechanism of IncFll plasmid replication.

MATERIALS AND METHODS
Bacterial strains and plasmids. The Escherichia coli K-12 strains used in this study are listed in Table 1. The strains form a nearly isogenic set that contains various alleles of dnaA (19,36). Two independent sets of each of the five strains were obtained from B. C. Kline and T. Kogoma. Experimental results obtained with the two sets of strains were indistinguishable.
IncFll plasmid NR1 (8,34,42,57) is a transmissible, multiple-antibiotic-resistant (R) plasmid with a size of 94.5 kilobase pairs (kb) and a low copy number of about 2 per bacterial chromosome (43,44,59). Plasmid pRR12 (31) is an Inc-Copmutant of NR1 that contains a single-base-pair substitution in the incRNA gene that regulates the translation of the mRNA for the repAl gene (40,41,44,53,54), which encodes the replication initiation protein of the plasmid (37,40,45). Plasmids pRR933 and pRR942 (9, 27) are 4.9-kb miniplasmids derived from NR1 and pRR12, respectively, that contain the minimal replicons of the plasmids. Each is composed of three PstI restriction fragments: a 1.1-kb fragment that encodes plasmid replication control and expresses plasmid-specific incompatibility, a 1.6-kb fragment that contains the plasmid replication origin, and a 2.2-kb fragment that contains the chloramphenicol resistance gene (cat). Plasmid pRR714 (9,27) contains this cat fragment inserted at the PstI site of the ColEl-like cloning vector pBR322 (3). Plasmids pRR775 and pRR790 (58) contain the 1.1-kb PstI inc fragment from NR1 and pRR12, respectively, inserted at the PstI site of pUC8, a cloning vector derived from pBR322 (51). Plasmid pRR720 (39,49) is an NR1-derived miniplasmid that contains the wild-type  NRi minimal replicon, the cat gene, and the stb locus responsible for stable plasmid inheritance (27,49). Plasmid pFZY1 (22) is a cloning vector derived from mini-F that confers resistance to ampicillin and contains a polylinker sequence and a promoterless lacZY operon. Plasmid pKN500 (30) is a miniplasmid composed of EcoRI fragments B and F of Ri, another member of the IncFII group of R plasmids, and confers resistance to kanamycin.
Culture media and conditions. Strains were cultured at 37°C in LB medium (28) supplemented with thymine (20 mg/liter) or in M9 minimal medium (28) containing 0.2% glucose and supplemented with thymine (20 mg/liter) and appropriate amino acids. Minimal medium was used to culture rnh mutant strains owing to their sensitivity to growth in rich medium (20,36). Agar media contained 15 g of Bacto agar (Difco Laboratories) per liter. When appropriate, antibiotics were also included in some cultures: chloramphenicol, 25 mg/liter; tetracycline hydrochloride, 5 mg/liter; sodium ampicillin, 20 mg/liter; and kanamycin sulfate, 50 mg/liter. Cell growth was monitored by turbidity at 650 nm with a Gilford model 260 spectrophotometer.
Purification and manipulation of plasmid DNA. Plasmid DNA isolation, restriction endonuclease digestion, gel electrophoresis, and transformation of E. coli cells with plasmid DNA were performed by standard methods as described before (9,27). Plasmid identification and cellular plasmid DNA content in the transformants were monitored by the alkaline minilysate method (2).
Plasmid copy number measurements. The relative copy numbers of plasmids that carry the cat gene were estimated from gene dosage effects by measuring chloramphenicol acetyltransferase (CAT) specific activity in cell extracts prepared from exponential-phase cultures as described before (9,47), with total protein content measured by the method of Bradford (4). Plasmid DNA content was also monitored by the minilysate method (2).
Plasmid stability test. The stability of plasmid inheritance was determined by measuring the decrease in the fraction of antibiotic-resistant cells during a period of nonselective growth (27). To begin the experiments with a uniform population of plasmid-bearing cells, the cells were initially cultured in medium containing the antibiotic to which the plasmid conferred resistance. The cells were then repeatedly subcultured by 106-fold dilution into drug-free medium, followed by overnight incubation. After each subculture, appropriate dilutions were spread on drug-free agar plates, and the antibiotic resistance of at least 50 of the resulting colonies was tested by replica plating. The plasmid DNA content of representative antibiotic-resistant and -sensitive colonies was examined by the minilysate method (2).

RESULTS AND DISCUSSION
Transformation of dnaA mutant strains with plasmid DNA. To test the requirement of NR1-derived miniplasmids for DnaA protein, several different E. coli strains were transformed with various plasmid derivatives ( Table 2). Recipient strain AQ3563 is a dnaA::TnJO (null) mutant suppressed by a copy of a mini-Ri plasmid integrated into the chromosome (chr::pKN500), whereas recipient strain AQ3585 is a dnaA:: TnJO null mutant suppressed by an rnh mutation. Recipient strain AQ634 served as the wild-type dnaA+ control. The other control strains were AQ3553, which is a dnaA(Ts) mutant that also contains an integrated copy of mini-Ri, and AQ666, which is dnaA+ rnh. The NR1-derived tester plas- The wild-type control strain AQ634 was plated on both rich and minimal media, with identical results. Plasmid transformants were selected with either chloramphenicol (Cm) or ampicillin (Ap). Host strains with integrated pKN500 were also selected with kanamycin, and those with TnlO were selected with tetracycline. Symbols: +, in multiple experiments, colonies of normal size and frequency, i.e., several hundred to several thousand colonies per plate, were present after 18 h (rich medium) or 40 h (minimal medium) of incubation at 37°C; -, normal-sized colonies were not present after the standard period of incubation.
For the negative plates from AQ3563 with pRR933 and pFZY1, a few colonies were often observed after several days of incubation (see text). For the negative plates from AQ3585 with pFZY1, incubation for additional time usually failed to give even a few colonies. b The AQ3553 transformants that contained pUC8 and pRR790 also formed colonies during incubation at 30 and 42°C, whereas those that contained pRR775 formed colonies at 30   a Measured by CAT enzyme activity in comparison with that of plasmid pRR933 in the wild-type host, AQ634. The copy numbers for strains cultured in rich medium were compared with those of pRR933 in AQ634 cultured in rich medium, whereas the copy numbers for strains cultured in minimal medium were compared with those of pRR933 in AQ634 cultured in minimal medium. The results are presented as the averages of two or more determinations. b This strain, AQ3563(pRR933), was obtained from one of the rare transformant colonies that arose after several days of incubation. mids were pRR933, which is composed of the wild-type NRl minimal replicon, and pRR942, which is the equivalent construct from pRR12, a high-copy-number mutant of NRl. Various pBR322 derivatives served as DnaA-independent control plasmids, and the mini-F derivative pFZY1 served as the DnaA-dependent control plasmid. Cells were cultured at 37°C, a temperature at which the mutant dnaA(Ts) protein in the AQ3553 control strain is still active. All plasmids were able to transform the three control strains, AQ634, AQ3553, and AQ666, at similar frequencies, with the formation of normal-sized colonies under standard incubation conditions. This indicated that the integrated copy of pKN500 in AQ3553 and the rnh mutation in AQ666 did not interfere with the establishment of any of these plasmids. Although pRR933 and the integrated mini-Ri might express mutual incompatibility in AQ3553, they were apparently capable of coexisting under the conditions of antibiotic selection used here. Also as expected, the mini-F derivative pFZY1 was unable to transform AQ3563 (dnaA::TnlO chr::pKN500) or AQ3585 (dnaA::TnJO rnh), since mini-F has an absolute requirement for DnaA. In contrast, the pBR322 derivatives pRR714 and pUC8 were able to transform both AQ3563 and AQ3585 owing to their DnaA independence. Because both pRR933 and pRR942 were also able to transform the dnaA null mutant AQ3585 at normal frequency, it appears that NRl-derived miniplasmids have no absolute requirement for DnaA protein for replication in vivo. An interesting difference in the abilities of pRR933 (wildtype) and pRR942 (Inc-Cop-) to transform AQ3563 (dnaA:: TnOO chr::pKN500) was observed. The failure of pRR933 to transform AQ3563 might be explained by the mutual incompatibility of pRR933 and the integrated mini-Ri, upon which strain AQ3563 has an absolute dependence for survival. The Incmutant pRR942, however, is compatible with NR1 and Ri and therefore does not cause any interference with replication of the integrated mini-Ri plasmid.
When the strains were transformed by various mixtures of ptasmid DNAs, results consistent with those obtained with pure plasmid preparations were observed ( Table 2). When both plasmids in the mixture were capable of transforming a given strain, some cotransformation was also detected (data not shown). When the plates that gave negative results for pRR933 and pFZY1 with recipient AQ3563 (dnaA: :TnlO chr::pKN500) were incubated for 2 to 3 additional days, formation of a few small colonies was sometimes observed. This was not observed for the plasmid-free transformation controls (data not shown). As will be discussed in more detail below, the formation of these rare transformant colo-nies after prolonged incubation could result from integration of the plasmids into the chromosome.
To test the hypothesis that the failure of the wild-type miniplasmid pRR933 to transform AQ3563 (dnaA::TnJO chr::pKN500) was a result of the mutual incompatibility between pRR933 and the suppressing integrated mini-Ri, pUC8 derivatives that contain the cloned inc+ region of NRi (pRR775) or the inc region of pRR12 (pRR790) were transformed into each of these strains. It is well established that the incompatibility expressed by the cloned NRi plasmid DNA in these derivatives is the result of specific inhibition of IncFII plasmid replication and that the Inc-Cop-phenotypes of the pRR12 mutant result from a single-base-pair substitution that changes both the strength and specificity of this inhibition (9,23,27,44,54,56). All three strains tested were transformed by both pRR775 and pRR790 except the dnaA null mutant AQ3563, which was not transformed by the inc+ pRR775 (Table 2). This was consistent with the idea that the IncFll-specific plasmid incompatibility expressed by pRR775 was interfering with the replication of the chromosome in AQ3563, which is absolutely dependent on initiation of replication from the integrated mini-Ri plasmid.
Furthermore, when the AQ3553 (dnaA(Ts) chr::pKN500) transformants were plated at 42°C as well as at 37°C, those containing pRR775 failed to form colonies at 42°C. Transformants that contained pRR790 or pUC8 formed colonies at both temperatures. This indicated that chromosomal replication from the integrated mini-RI in AQ3553 was being "switched off" (29) by the inc+ pRR775 at 42°C in this temperature-sensitive dnaA mutant. Together, these results indicated that integrative suppression of both the dnaA null mutation in AQ3563 and the dnaA(Ts) mutation in AQ3553 by the integrated copy of mini-Ri was under control of the normal IncFII replication circuitry. We cannot rule out the possibility that replication of the IncFII miniplasmids in the dnaA null mutants occurs through some secondary mechanism. However, because the miniplasmids contain only the minimal IncFII replicon, and because their replication in the dnaA null mutants is specifically sensitive to the normal IncFII replication control elements, we consider this unlikely.
Measurements of plasmid copy number in dnaA mutant strains. The copy number of NRi and its Cop' miniplasmid derivatives is approximately 2 per chromosome or approximately 4 per average E. coli cell (44,56). The relative copy numbers of plasmids pRR933 (wild type), pRR942 (Inc-Cop-), and pRR714 (pBR322 cat) in the various tester strains were determined (Table 3) Analysis of plasmid DNA content in minilysates of dnaA mutant strains. Minilysate DNA was prepared from strains AQ634 (dnaA+) (lanes 7 to 9), AQ3553 [dnaA(Ts) chr::pKN500] (lanes 4 to 6), and AQ3563 (dnaA::TnJO chr::pKN500) (lanes 1 to 3) harboring either pRR933, pRR942, or pRR714, as indicated above each lane. The minilysate DNA was then electrophoresed in 1% agarose and stained with ethidium bromide. Marker DNA was electrophoresed in the leftmost lane. The positions of the supercoiled forms of plasmids pRR714 (6.5 kb) and pRR933 and pRR942 (4.9 kb) are indicated on the right. that the copy numbers of the NRi-derived plasmids pRR933 and pRR942 were lower in the dnaA null mutant strains. Although the relative copy number of pBR322 derivative pRR714 was higher in minimal medium than in LB, the absence of DnaA was inconsequential. This suggests that the replication of pRR933 and pRR942 is less efficient in the absence of DnaA protein. The copy number of pRR933, but not of pRR942 or pRR714, was also reduced in strain AQ3553 (dnaA(Ts) chr::pKN500). This most likely reflects the incompatibility interaction between pRR933 and the integrated mini-Ri in this strain, in which the total number of copies of pRR933 plus mini-Ri may have been approximately equivalent to that of pRR933 alone in the wild-type strain. In a test of one of the pRR933 transformant colonies of strain AQ3563 (dnaA::TnJO chr::pKN500) that had appeared after prolonged incubation, as described above, the apparent copy number of pRR933 was also quite low (Table  3). Although this might be consistent with the combined effects of plasmid incompatibility and the absence of DnaA, an alternative explanation could be chromosomal integration of the plasmid, as discussed below.
The plasmid DNA contents of strains AQ634 (wild type), AQ3553 [dnaA(Ts) chr::pKN500], and AQ3563 (dnaA::TnJO chr::pKN500), visualized by the minilysate method (Fig. 1), were consistent with the CAT assay data in Table 3. That is, the copy numbers of pRR942 and pRR714 were high in both AQ634 and AQ3553, whereas the copy number of pRR933 was low in both strains. The copy number of pRR714 was also high in AQ3563 (lane 3). However, the copy number of pRR942 (lane 1) was lower in this dnaA null mutant strain. In one of the dnaA null AQ3563 transformants that had appeared after prolonged incubation, no pRR933 plasmid DNA was detected at all (lane 2). One possibility is that this represents a case of integration of pRR933 into the chromosome. pRR933 shows homology with the integrated copy of mini-Ri in this strain and also has some homology from segments of ISJ on either side of the cat gene with IS1 elements located in the E. coli chromosome. A similar explanation might apply to the rare colonies that arose after transformation and prolonged incubation of the dnaA null mutants with the pFZY1 plasmid, which shares lac homology with the chromosome (22).
Tests of plasmid stability in dnaA mutant strains. Miniplasmids pRR933 and pRR942 contain the complete minimal replicon necessary for proper control of replication, but they lack the stb locus required for proper partitioning (27). Therefore, even in a wild-type host strain, pRR933 was not inherited stably in the absence of continuous antibiotic selection (Table 4). However, owing to its elevated copy number, the inheritance of pRR942 was quite stable under the same conditions (Table 4). Whereas pRR942 was stable in the dnaA+ and dnaA(Ts) strains, it was very unstable in the dnaA (null) strains. The stability of pRR933 was even lower than normal in AQ3585 (dnaA::TnJO rnh). These results are consistent with a reduced efficiency of plasmid replication for NR1 derivatives in the absence of DnaA protein, as reflected in a reduction of plasmid copy number, which would cause a decrease in stable inheritance. The very low stability of pRR933 in strain AQ3553 [dnaA(Ts) chr: :pKN500] compared with the stable inheritance of pRR942 in the same strain can be explained by the incompatibility between pRR933 and the integrated mini-Ri. pRR720 is a miniplasmid that contains the stb locus responsible for stable inheritance, i.e., for plasmid partitioning at cell division (27,49), in addition to the wild-type NRi minimal replicon. The results of transformation and copy number measurements of the various tester strains with pRR720 DNA were similar to those with pRR933 (data not shown). Whereas pRR720 was inherited stably in the wildtype host, AQ634, and in the rnh mutant host, AQ666, it was not inherited stably in the dnaA null mutant AQ3585 (Table  4). Although pRR720 was unstable in the dnaA null mutant, it was much more stable than pRR933 stb in the same host strain (Table 4, and data not shown). This suggested that the stb locus was able to function in a dnaA null mutant and that the instability of pRR720 in this host was caused by its lower-than-normal copy number resulting from the absence of DnaA protein.
Together, the above results indicate that replication of plasmid NR1 does not have an absolute dependence upon DnaA protein in vivo. In this respect, replication of NR1 is similar to that of the ColEl-like plasmids, such as pBR322.
In contrast, the replication of mini-F and mini-Pl appears to have a strict dependence upon DnaA protein (16,19,32). It is interesting that the plasmids that require DnaA are those whose replication is regulated by repeated DNA sequences present in their origins, whereas the plasmids that do not require DnaA are those whose replication is regulated by the interaction of two complementary RNA transcripts (46).
Although replication of NR1 does not have an absolute requirement for DnaA protein, the copy number and stability of NR1 derivatives were lower in the dnaA null mutant strains. This suggests that DnaA protein participates in an auxiliary role in initiation of NR1 replication. However, it could be argued that such a role is "essential" to the overall survival of NR1, because any reduction in copy number of this already low-copy-number plasmid could seriously reduce its stable maintenance, as observed (Table 4). Other plasmids that have been tested for DnaA dependence had either an absolute requirement for DnaA, such as mini-F, or were indifferent, such as ColEl. The auxiliary role played by DnaA in IncFII replication therefore may be different from that in the replication of other plasmids.
Other studies have shown that NR1 miniplasmids can survive and replicate in vivo after deletion of the single DnaA box that lies next to the replication origin (37; unpublished data), whereas deletions that proceeded further into the origin region caused inactivation of origin function (37). However, pRR942 derivatives that lacked the DnaA box had a lower copy number (unpublished data). Therefore, this is also consistent with a nonessential auxiliary role for DnaA protein in NR1 replication. However, those studies were carried out with miniplasmids derived from the high-copynumber mutant pRR12. Therefore, the importance of the DnaA box for the stable replication of low-copy-number NR1-derived miniplasmids has not been tested.
Based on the replication of IncFll plasmid DNA in vitro in cellular extracts prepared from dnaA(Ts) mutant E. coli strains, it was suggested that replication of these plasmids has an absolute requirement for DnaA protein (24,38). However, in vitro replication reactions have not been tested in extracts prepared from dnaA null mutant cells, nor have plasmids from which the DnaA box has been deleted been tested for replication in vitro in the dnaA(Ts) extracts. Therefore, an alternative interpretation might be that the mutant DnaA protein present in the extracts prepared from the dnaA(Ts) mutant cells was inhibitory to IncFll plasmid replication. This would be more consistent with the results in this paper that suggest that DnaA protein plays an auxiliary but nonessential role in the replication of NR1 in vivo. From the results of our transformation experiments, it can also be concluded that the integrative suppression of the dnaA null mutant by mini-RI was under the control of the normal (i.e., incompatibility sensitive) IncFII plasmid replication mechanism of pKN500.