Species and Incompatibility Determination within the P1par Family of Plasmid Partition Elements

Media, enzymes, and materials.Bacterial strains were grown at 37°C in LB broth (26) unless otherwise stated. The following antibiotics were added to media at the indicated concentrations unless otherwise indicated: ampicillin, 100 μg/ml; chloramphenicol, 10 μg/ml; tetracycline, 6.5 μg/ml; and kanamycin, 40 μg/ml. Restriction endonucleases were obtained from New England Biolabs (Beverly, MA), Gibco-BRL (Gaithersburg, MD), or Boehringer Mannheim Biochemicals (Indianapolis, Ind.). T4 DNA ligase, DNA polymerase, and the Klenow fragment of DNA polymerase I were supplied by Roche Molecular Biochemicals (Indianapolis, IN). QuikChange multiple- and single-site-directed mutagenesis kits and XL10-Gold Kan^r competent cells were obtained from Stratagene (La Jolla, CA).

Bacterial strains and bacteriophages.Strain CC4247, which is a ΔargF-lacZ U169 derivative of W3110 (20), was used for chromosomal integration of the plasmid par genes. DHB6521 (5) was used as a lysogenic strain for λInCh1. Strain M1967 is MG1655 zwf<>kan. It has the kanamycin resistance gene of Tn5 substituting for the deleted complete open reading frame of the zwf gene. Bacteriophage λcI₈₅₇-P1:5RΔ1005 was maintained on strain YMC (15). Partition tests used strains CC4902, CC4913, and CC4916, whose chromosomes contain single copies of the P1, P7, and pMT1par genes, respectively.

Strain CC4902 has the P1par operon integrated at attλ expressed constitutively from an attenuated plac promoter (20). The P1par genes were integrated using the λInCh method (5). The strain has a single P1par operon, a selectable bla gene, and a hybrid λatt site (20). Strains CC4913 and CC4916 are similar to CC4902 but contain the par operons of P7 and pMT1, respectively. Plasmids pALA2905 and pALA2906 were the sources of the integrated DNA. The genetic structure of the insertions was confirmed by PCR analysis. Each construct contains a single copy of the appropriate plasmid's par region under the control of an attenuated, constitutively expressed plac promoter.

Plasmids.Plasmids pALA1901 and pALA1931 have been described previously (14). Plasmids pALA1413, pALA1414, and pALA1846 are derivatives of pBR322 that carry the P1, P7, and pMT1parA-parB operons, respectively (24). Plasmid pALA2466 was constructed from pDSW210 (30) and has a cloned PCR product from pALA1413 (20) by use of the following primers: 5′-CCGGAATTCAGTGAAATTCCAGCCAGCTT and 5′-GAGGAAGCCCTTACTCCTCAGTTAGATCTGA. The parA and parB genes were inserted between EcoRI and HindIII sites, deleting the gfp gene of the vector. This insertion changes the N terminus of ParA from Met Ser to Met Glu Phe Ser and places the par operon under the control of an attenuated lac promoter.

Plasmid pALA2900 was derived from pALA2466 by the deletion of lacI^q by BssHII digestion and religation. Plasmid pDSW210 (30) was digested with MluI and NarI, and the ends were filled in by Klenow enzyme and ligated to give pALA2908. The EcoRI-to-HindIII fragment of pALA2908 was then replaced with a PCR product containing the P7par region from linearized pALA1414 to give pALA2905. The primers were 5′-CGTCTTCAAGAATTCCCGCCATTTCTTTA and 5′-CATTAAAGCTTCTAGAGAGCTTTTTTCAT. A PCR fragment carrying the pMT1par genes was produced from pALA1846 with an XbaI extension at the left and a HindIII extension at the right end. This was inserted between the XbaI/HindIII sites of pALA2908 to produce plasmid pALA2906. The primers were as follows: 5′-GGATCCTCTAGATACAAAATATGTTGTACA and 5′-CAGCCAAGCTTCCTTATCCCTTACTCACCTG. The par-deleted lambda-mini-P1 plasmid λc1₈₅₇-P1:5RΔ1005 and plasmids pALA1952, pALA1993, and pALA1843 were as previously described (24). On recombination, they gave rise to λ-P1:5RΔ1005::pALA1952, λ-P1:5RΔ1005::pALA1993, and λ-P1:5RΔ1005::pALA1843 (31).

Plasmids with hybrid parS sites were obtained by multiple- and single-site-directed mutagenesis of the P1, P7, and pMT1parS plasmids pALA1952, pALA1993, and pALA1843. The mutagenic primers that introduced point mutations into both B boxes of the parS sites were as follows: (i) 5′-CCTTTTTTGTATGTTTTTCGCCACGCCAATTTCATGG (places a P1 base in pMT1 BoxB1) and 5′-CTTTCACACTGAAATCGCCACGCTTTTCAACCTC (places a P1 base in pMT1 BoxB2), (ii) 5′-CCTTTTTTGTATGTTTTGTCCCACGCCAATTTCATGG (places P7 bases in pMT1 BoxB1) and 5′-CTTTCACACTGAAATTCCCACGCTTTTCAACCTC (places P7 bases in pMT1 BoxB2), (iii) 5′-CCGGATCCAAACTTTCACCATTCAAATTTCAC (places a pMT1 base in P1 BoxB1) and 5′-CAAGGTGAAATCACCACGATTTCACCTTGG (places a pMT1 base in P1 BoxB2), and (iv) 5′-CCGGATCCTAAAATTCACCGCGCCTATTTCATG (places pMT1 bases in P7 BoxB1) and 5′-CAGGCTGAAATCACCACGGTTTCACGCCTG (places pMT1 bases in P7 BoxB2).

The following pairs of primers were used to introduce changes in a single BoxB: (i) 5′-AATAAGTGTCCGGATCCAAACTTTCACCATTCAAATTTCACTATTAAC and 5′-GTTAATAGTGAAATTTGAATGGTGAAAGTTTGGATCCGGACACTTATT (places a P1 base in pMT1parS BoxB1), (ii) 5′-CTCTAAAATTTCAAGGTGAAATCACCACGATTTCACCTTGGATCG and 5′-CGATCCAAGGTGAAATCGTGGTGATTTCACCTTGAAATTTTAGAG (places a P1 base in pMT1parS BoxB2), and (iii) 5′-CCTTTTTTGTATGTTTTTCGCCACGCCAATTTCATGG and 5′-CCATGAAATTGGCGTGGCGAAAAACATACAAAAAAGG (places a pMT1 base in P1parS BoxB1).

The primers were annealed to the template plasmid, digested with DpnI, and transformed into XL10-Gold competent cells (Stratagene). The DNA sequences of the mutant plasmids were determined. The resulting plasmids were recombined with λc1₈₅₇-P1:5RΔ1005 and used in partition tests as previously described (20).

Plasmids pALA2920, pALA2921, and pALA2922 are derivatives of pACYC184 carrying the P1parS, P7parS, and pMT1parS sites, respectively. They were derived from pALA1849, pALA1850, and pALA1851 (31) by having the tet gene replaced by a kan (kanamycin resistance) cassette. The kan cassette was amplified from strain M1967 genomic DNA (a gift from Lynn Thomason), using primers 5′-GCTTATCATCGATAAGCTTTATGGACAGCAAGCGAACCG and 5′-CGCCGAAACAAGCGCTCATGATCAGAAGAACTCGTCAAGAAGto give a fragment with HindIII and BspHI extensions. This was placed between HindIII and BspHI sites of pALA1849 and pALA1851 to give pALA2920 and pALA2922, respectively. To obtain pALA2921, the kan cassette was recombined with pALA1850 to replace part of the tet gene with kan by λred-promoted recombination (32). The recombining fragment was produced by amplification of the kan cassette using the primers 5′-CTTATCATCGATAAGCTTTATGGACAGCAAGCGAACCG and 5′-CCGAAACAAGCGCTCATGATCAGAAGAACTCGTCAAG.

DNA procedures.Plasmid DNA purification, DNA sequencing analysis, oligonucleotide preparation, and other DNA techniques were carried out as previously described (25).

Incompatibility tests.Incompatibility tests were carried out essentially as previously described (31). The retention of the pACYC 184 derivatives throughout the growth of the strains on the plates was ensured by 30 μg/ml of kanamycin.

Partition tests.Each parS site was tested as a λ-P1 miniplasmid construct (λ-P1:5RΔ1005::pALA1952, etc.) using the pickup partition assay (15). The constructs were introduced by infection into a strain supplying the relevant Par proteins in trans. They establish as a low-copy-number plasmid. Their stability was estimated by measuring loss of the plasmid after 25 generations of unselected growth (15).

The P1par family of plasmid partition systems.There are currently five clearly distinct sequences closely related to the P1par operon (6, 22, 31, 23, 27). They are present in a variety of different plasmid types from five different gram-negative bacterial genera. The six resulting P1par family members form a compact clade, with both Par proteins showing high degrees of similarity to their P1 counterparts (12). Each of the six P1par family members has adjacent open reading frames for the ParA and ParB proteins of similar sizes and in the same order as those of P1. Previously described members have functional parS sites placed closely downstream of the parB gene (24, 31). A search of the regions downstream of parB in the remaining members showed that these also have candidate parS sites. An alignment of the six sequences is shown in Fig. 1. This alignment is in close agreement with that recently proposed by Funnell and Slavcev (12). In the case of Rts1parS, the proposed site does not correspond to the potential IHF binding site and partition site mentioned in the annotated sequence AP004237.1 (23).

• Open in new tab
• Download powerpoint

FIG. 1.

Alignments of key sequences from the par regions of P1par family members. Top: The regions corresponding to the P1parS site are aligned. Shading identifies bases that are identical to P1parS. Center: The BoxB sequences within parS are compared. Bottom: Portions of the ParB protein sequences near the carboxy termini are aligned. Numbers indicate the amino acid positions in the ParB open reading frame. Shading identifies amino acids that are identical or similar to those of P1 ParB. Similarities are shown as white letters. The bold, underlined sequences contain the DRS regions shown to be involved in the ParB-BoxB contacts in P1 and P7par.

The specificity difference between the P1 and P7parS sites resides in the sequences of the two BoxB repeats (15). We speculated that the BoxB sequences might determine the species specificities of the parS sites of the other family members.

In the P1 and P7 cases, the region of the ParB protein that contacts BoxB, the DRS, has been mapped. In both cases, it lies near the C terminus of the protein (25) (Fig. 1). If BoxB contacts the equivalent ParB regions in all family members, there should be a correlation between the similarities of the BoxB sequences and the similarities in the contacting ParB DRSs. Alignment of the regions surrounding the putative DRSs of the six ParB proteins (Fig. 1) shows that the three members with identical or very similar BoxB sequences (P1, pSLT, pWR100) also have similar ParB contact regions. The equivalent regions in the three members with unique BoxB sequences are dissimilar from those of the P1 group and from each other. This correlation fits with the currently known species specificities of the systems: P1 and pWR100 have identical specificities (27), whereas those of P7 and pMT1 are different from P1 and each other (31). Based on the analysis in Fig. 1, we predict that, of the two members remaining uncharacterized, pSLTpar will prove to have identical specificity to P1, whereas Rts1par will show a unique specificity.

A specificity test for partition with Par proteins produced from the E. coli chromosome.We have recently shown that a P1 miniplasmid carrying P1parS is properly maintained when the P1 ParA and ParB proteins are supplied from an insert in the E. coli host chromosome (20). The P1par operon was integrated at attλ and was expressed from an attenuated plac promoter (20). This type of construct is preferable for genetic studies of partition because the par operons and parS sites can be tested in different combinations without the necessity of maintaining more than one plasmid in the cell. Also, the test system is free from complications imposed by par operon autoregulation. We constructed similar strains carrying the par operons from pMT1 and P7 integrated into the chromosome (see Materials and Methods). The maintenance of the appropriate parS-containing plasmids in these strains is shown in Table 1. Each strain maintained the cognate parS plasmid when the appropriate ParA and ParB proteins were supplied from the chromosome. This confirms our previous findings, made with a plasmid-borne par operon, that the Y. pestis pMT1 par system works well in E. coli and that Par proteins work in trans to the partitioning plasmid (31). By introducing the parS plasmids into strains with different par operons, the species specificities of the parS sites were also readily demonstrated. Thus, as found previously using plasmid-borne par operons, the Par proteins of P1, P7, and pMT1 are not interchangeable. They worked only with the cognate P1parS, P7parS, and pMT1parS sites, respectively (Table 1).

Substitution of the BoxB sequences can switch specificity completely.The BoxB repeat sequences of P1 are both TCGCCA. The P7 BoxB1 and BoxB2 differ somewhat and are GTCCCG and TTCCCA, respectively (Fig. 1). Changing the five bases necessary to convert the otherwise-intact P7parS site to give TCGCCA in both BoxB1 and BoxB2 resulted in a complete change in specificity from the P7 to the P1 type (15). For pMT1, the BoxB repeats are both TCACCA; that is, they differ by one base from the P1 boxes and by four and two bases, respectively, from the two P7 BoxB repeats (Fig. 1). These differences involve the same second and third base positions that are important in distinguishing the P1 and P7 sites from each other. Using site-directed mutagenesis, we modified the critical nucleotides in each BoxB of the pMT1 parS site, changing both B boxes to their P1 or P7 equivalents. The sequence of each hybrid parS was confirmed by DNA sequencing. Note that the BoxB sequences are by no means the only differences between the parS sites of pMT1 and those of P1 or P7. The P1parS site has 27 of 82 base pairs, and P7parS has 35 of 82 base pairs, that differ from those of pMT1parS (Fig. 1).

The hybrid parS sites were incorporated into a λminiP1 plasmid and supplied with the appropriate Par proteins expressed from the host chromosome. Their stabilities were determined in a partition test by measuring the loss of the composite plasmid after 25 generations of unselected growth. The wild-type pMT1parS site responded only to the pMT1 Par proteins (Table 2). Substitutions of both BoxB sequences of the pMT1parS site gave a complete switch in specificity. When the P1 BoxB sequences were substituted for their equivalents in the pMT1parS site, the plasmid was stabilized by the P1 Par proteins but not by the pMT1 proteins. When the P7 BoxB sequences were introduced, the plasmid was stabilized by the P7 Par proteins but not by the pMT1 proteins (Table 2).

Reciprocal constructs that put the pMT1 bases into P1 and P7 sites were also made but were not informative: these sites were inactive no matter which proteins were provided. It appears that one or more of the many base differences that have accumulated between these parS sites interfere with site function when combined with certain BoxB changes.

A single modification in the pMT1parS site can change species specificity.To examine further the pMT1parS species specificity, we obtained single mutations in the left or right BoxB sequence. The resulting hybrid pMT1parS plasmids with P1 nucleotide substitutions were tested with the pMT1 and P1 Par proteins (Table 3). Substitution of just BoxB2 of pMT1 with the equivalent P1 sequence was sufficient to change the specificity of the site completely from pMT1 to the P1 type. This substitution involves the change of only a single base. Substitution of the equivalent base in BoxB1 also caused a specificity switch: the hybrid site lost all activity with the pMT1 proteins but responded to the P1 proteins, albeit with a much-reduced activity (Table 3). Thus, a single base change in either of the two BoxB sequences is sufficient to change the species specificity. There are 27 base pairs that differ between the 82-bp P1parS and pMT1parS sites (Fig. 1). All but the one critical base pair in BoxB appear to be uninformative as far as species specificity is concerned.

Species and incompatibility specificities switch in concert.Partition-mediated incompatibility determines whether two related plasmids will be able to coexist in the same cell. The P1, P7, and pMT1parS sites each have a unique incompatibility specificity. For example, supernumerary copies of pMT1parS interfere with the function of pMT1par but not with P1 or P7par (31). P1 and P7parS also have their own unique incompatibility specificities (2, 16). These specificities were previously determined using strains which have the Par proteins supplied from plasmids (31). Unique incompatibility specificities were also seen for all three species by use of the system in which the Par proteins are constitutively supplied from the chromosome (Table 4).

When the species specificity of the pMT1parS site is changed by mutation, does its incompatibility specificity change also? The wild-type pMT1parS plasmid, maintained in the presence of the pMT1 Par proteins, is rapidly lost when a second plasmid carrying pMT1parS is introduced into the cell (Table 4). Table 5 shows that the hybrid pMT1parS sites adopt the incompatibility of their newly acquired species: the hybrid sites with P1 partition specificities were displaced by supernumerary P1parS sites but not by pMT1parS sites, despite the fact that the bulk of the parS bases correspond to those in the pMT1 sequence. Note that incompatibility specificity can be switched by as little as one parS base change. Likewise, the hybrid sites based on the P1parS sequence, but with BoxB modifications giving pMT1 species specificity, were displaced by supernumerary pMT1parS sites but not by P1parS sites. Thus, species specificity and incompatibility switched in concert. This was true for sites with changes in both copies or in only one copy of BoxB (Table 5).

We have previously described the construction of P1 and P7parS sites that have BoxB sequences altered to that of the other species. These changes cause a switch from P1 to P7 species specificity and vice versa (15). Table 5 shows the results of incompatibility tests using these hybrid parS sequences. In both cases, the species specificity switches were accompanied by corresponding switches in incompatibility specificity (Table 5). For example, an otherwise-normal P1parS site with changes in the BoxB sequences that alter them to the P7 sequence now shows P7-type incompatibility: a plasmid carrying it is displaced by another plasmid carrying the P7parS site but not by one carrying the P1parS site (Table 5).