TrfA-Dependent Inner Membrane-Associated Plasmid RK2 DNA Synthesis and Association of TrfA with Membranes of Different Gram-Negative Hosts

Abstract

TrfA, the replication initiator protein of broad-host-range plasmid RK2, was tested for its ability to bind to the membrane of four different gram-negative hosts in addition to Escherichia coli: Pseudomonas aeruginosa, Pseudomonas putida, Salmonella enterica serovar Typhimurium, and Rhodobacter sphaeroides. Cells harboring TrfA-encoding plasmids were fractionated into soluble, inner membrane, and outer membrane fractions. The fractions were subjected to Western blotting, and the blots were probed with antibody to the TrfA proteins. TrfA was found to fractionate with the cell membranes of all species tested. When the two membrane fractions of these species were tested for their ability to synthesize plasmid DNA endogenously (i.e., without added template or enzymes), only the inner membrane fraction was capable of extensive synthesis that was inhibited by anti-TrfA antibody in a manner similar to that of the original host species, E. coli. In addition, although DNA synthesis did occur in the outer membrane fraction, it was much less extensive than that exhibited by the inner membrane fraction and only slightly affected by anti-TrfA antibody. Plasmid DNA synthesized by the inner membrane fraction of one representative species, P. aeruginosa, was characteristic of supercoil and intermediate forms of the plasmid. Extensive DNA synthesis was observed in the soluble fraction of another representative species, R. sphaeroides, but it was completely unaffected by anti-TrfA antibody, suggesting that such synthesis was due to repair and/or nonspecific chain extension of plasmid DNA fragments.

Over the past 20 years, numerous studies have implicated the cell membrane as the site of replication in the prokaryotic cell. Of significance has been work done with the broad-host-range plasmid RK2 as the model system (for reviews, see references 3 and 4). RK2 is a medically important (resistant to three antibiotics), naturally occurring plasmid which is capable of replication and stable maintenance within a wide range of gram-negative bacteria (2, 14). An advantage of this system is the ease with which it allows for the study of replication within these many species.

In addition to the origin of replication (oriV), one plasmid-encoded element, trfA, is necessary for replication to occur in all species tested (1). The trfA gene codes for two polypeptides (TrfA-44 and TrfA-33), the smaller of which is the result of an internal translational start site. Both proteins are found associated with the membranes of Escherichia coli (8, 9, 11, 12). The same small hydrophobic region located toward the C termini of both proteins has been implicated in this association (8), and the association appears to be necessary for plasmid viability. If a necessary function of TrfA is to sequester replication of the plasmid within the membrane of the host cell, it would be reasonable to predict that TrfA would be present in membrane fractions of other hosts; however, studies localizing TrfA within other host cells have not previously been undertaken.

In the present study, it was found not only that TrfA was associated with both the inner and outer membrane fraction of each species examined but also that plasmid-specific synthesis resided in the inner membrane fraction of these species, as was observed with the original host, E. coli.

MATERIALS AND METHODS

Bacterial and plasmid strains, growth conditions, plasmid transformation, and mating.

Four gram-negative species, Pseudomonas aeruginosa, Pseudomonas putida, Salmonella enterica serovar Typhimurium, and Rhodobacter sphaeroides, were kindly supplied by D. Oliver and D. Figurski. A modified E. coli strain, MV10 (thr-1 leu-6 thi-1 lacY1 tonA21 supE44 trpE5 λ⁻) was provided by D. Figurski. RK2-derived plasmids (either pRK21382, a full-sized RK2 derivative resistant to ampicillin, 50 μg/ml, or pRK2501, a miniderivative of RK2 resistant to kanamycin and tetracycline, 50 μg/ml and 10 μg/ml, respectively) were used in this investigation. The Pseudomonas strains containing the RK2 plasmids (see below) were cultured at 37°C for 12 h in Luria-Bertani (LB) medium (10 g of tryptone, 5 g of yeast extract [both Difco], and 10 g of NaCl/liter) supplemented with 50 μg of kanamycin/ml. Salmonella and Rhodobacter strains containing the RK2 plasmid were cultured at 37 or 30°C, respectively, for 18 h in LB medium containing 50 μg of kanamycin/ml for Salmonella and 2 μg of tetracycline/ml for Rhodobacter. After growth, all cultures were centrifuged in a Sorvall RC2 centrifuge for 10 min at 7,000 rpm (4°C) prior to the extraction of the various membrane and cell fractions (see below).

Transformation of pRK2501 into Salmonella serovar Typhimurium and R. sphaeroides was carried out by the method of Hanahan (5). The full-length RK2 plasmid (pRK 21382) was introduced into P. aeruginosa and P. putida by batch mating according to the following procedure modified from a previous study (1). E. coli strain MV10 (unable to synthesize tryptophan) (see above) containing the RK2 plasmid was used as the donor strain and was selected against by plating on M9-CAA medium (6 g of Na₂HPO₄, 3 g of KH₂PO₄, 0.5 g of NaCl, 1 g of NH₄Cl, 1 mM MgSO₄ · 7H₂O, 0.1 mM CaCl₂ · 2H₂O, 5 g of Difco Casamino Acids [which lacks tryptophan], and 2 g of glucose/liter), while the Pseudomonas strains without plasmids were selected against with ampicillin. Donor and recipient cells were cultured to saturation (18 h at 37°C) in LB broth, and 1-ml samples were mixed at equal concentrations, plated on nonselective LB medium, and incubated for 2 h at 37°C. The cells were scraped off the plates, resuspended in LB broth, and plated in serial dilutions on M9-CAA containing 50 μg of ampicillin/ml. Plasmid DNA was extracted using an anionic exchange column protocol described by Qiagen (see below) to confirm that the plasmid had been introduced, and Western blotting with polyclonal anti-TrfA antibody (kindly provided by D. R. Helinski) was performed on whole-cell extracts of the transconjugants by using an ECL kit from Amersham to confirm that TrfA protein was expressed in the cells.

Membrane and soluble fraction isolation.

Inner and outer membrane fractions were extracted from all of the species by low French pressure to minimize the shearing of membrane-bound DNA using the basic protocols originally described for S. enterica serovar Typhimurium by Osborn et al. (13) and modified by Michaels et al. for E. coli (12). The supernatant fluid remaining after the initial high-speed centrifugation of the whole-cell extract was designated as the soluble (cytoplasmic) fraction. Western blotting was performed on all fractions (in which protein levels detected with a kit from Bio-Rad were normalized) by using anti-TrfA antibody with the Amersham ECL kit to detect the presence of the TrfA proteins.

In vitro synthesis of DNA by membrane and soluble fractions and detection of various types of plasmid DNA.

The two membrane fractions and soluble fraction were extracted from the various species as described above. Each of them was adjusted to a working concentration of 20 to 30 μg/ml (Bio-Rad) as part of the following assay solution (final volume of 100 μl): 30 mM HEPES (pH 8.0), 30 mM KCl, 7.5 mM magnesium acetate, 0.1 mM NAD, 7.5 mM creatine phosphate, 0.1 mg of creatine phosphokinase/ml, tRNA (10 μl of a 2-mg/ml solution), 0.1 mM cyclic AMP, 2 mM ATP, 0.5 mM concentrations each of GTP, CTP, and UTP, 0.04 mM concentrations each of dATP, dCTP, dGTP, and [³H]dTTP (2 μCi/0.1 ml, 20 Ci/mmol; ICN Corp.). All nonradioactive components were obtained from Sigma. Either 1 or 2 μl of anti-TrfA antibody (17.2 μg/ml) or rabbit preimmune serum (representing an equal amount of protein) was preincubated first with the fractions at 4°C for 20 min before being added to the assay mixture. After incubation of three to four replicate samples at 30°C at each time period, they were pooled and the plasmid DNA was separated from bacterial DNA by using the Qiagen anionic exchange resin according to their detailed specifications (Midi kit, steps 8 through 13). The procedure involves equilibrating the column with a buffer containing Triton X-100, which effectively dissociates the membrane-bound, newly synthesized DNA. Little or no contaminating bacterial DNA was present in the final eluate as shown by numerous controls in which such DNA was used only during the purification procedure (data not shown). Each sample was then precipitated with trichloroacetic acid (final concentration, 5% containing 1% sodium pyrophosphate), washed with ethanol (95%), and assayed for radioactive DNA in an LKB Rackbeta scintillation counter.

To determine the types of plasmid DNA synthesized, the assay mixture described above was scaled up fivefold, and one time period (30 min) was chosen. After incubation, plasmid DNA was separated from bacterial DNA by the Qiagen anionic exchange resin, concentrated (see Fig. 2), and loaded onto a 1.0% agarose gel with marker supercoil plasmid DNA and open circular or nicked linear molecules. The DNA molecules were separated by electrophoresis in a Tris-borate-EDTA buffer (0.08, 0.089, and 0.0025 M, respectively; pH 8.0). After electrophoresis, the gel was treated with ethidium bromide (0.8 mg/ml) to visualize the various bands and sliced into 0.5-cm pieces, the DNA was extracted by the Geneclean technique of Qiagen, and radioactivity was determined in the scintillation counter.

FIG. 2.

In vitro synthesis of DNA by the inner and outer membrane fractions of various Gram-negative species in the presence and absence of anti-TrfA antibody. Incubation and reactions were carried out as described in Materials and Methods. ⊡, preimmune serum; ●, anti-TrfA antibody.

RESULTS

TrfA in membrane and soluble fractions.

To determine the distribution of TrfA in the various fractions of each of the gram-negative species, Western blotting was performed on the fractions as described in Materials and Methods. Figure 1 shows that TrfA was present in the membrane fractions of each of the species tested. The relative levels of TrfA in the various fractions were not constant throughout the species, but in each species the majority of the TrfA was found to fractionate with the membrane and only slightly or not at all in the soluble fraction. In the case of P. putida as well as S. enterica serovar Typhimurium, detectable levels of TrfA were present exclusively in membrane fractions. In P. aeruginosa and R. sphaeroides significant levels of both species were exhibited in both membrane fractions of TrfA. In Salmonella fractions only a faint band of TrfA-33 was observed, while in P. putida fractions only TrfA-33 was observed.

FIG. 1.

TrfA proteins in whole-cell (W.C.), soluble (Sol.), inner membrane, and outer membrane fractions of various Gram-negative species. Growth conditions, transformation, mating, and procedures for extraction of the membrane fractions and identification of the TrfA proteins are described in Materials and Methods.

Membrane-associated plasmid DNA replication.

To ascertain whether TrfA association with the membrane is functional for plasmid DNA replication in these other species, and whether there is specificity for either the inner or outer membrane fraction, a screen for endogenous DNA replication (i.e., replication in the absence of exogenous template or enzymes) inhibited specifically by anti-TrfA antibody was undertaken. The inner and outer membrane fractions were extracted from the various species as described in Materials and Methods and assayed as described in Materials and Methods. The results (Fig. 2) showed that a significant inhibitory effect by the antibody occurred when the inner membrane fraction was used as the source of the replicating complex. However, when the outer membrane fraction was tested there was much less total synthetic activity observed, and although there was a difference between anti-TrfA-antibody-treated samples and controls, the effects were minimal.

Synthesis of various types of plasmid DNA.

To confirm that the DNA synthesized by the inner membrane fraction represented complete plasmid DNA molecules and not simply fragments of plasmid DNA, the assay mixture used as described for Fig. 2 was used for one representative species, P. aeruginosa, except that the reaction was scaled up as described in Materials and Methods. Figure 3 shows first that a broad band of radioactive DNA migrated to the area containing the open circular and supercoil plasmid DNA forms. Second, another narrower band was detected that coincided with the linear plasmid species. It was not possible to separate the open circular from the supercoil species in the experimental lane because the size of the excised gel fragments overlapped both DNA forms and because a broad band of template DNA was present in the experimental lane that tended to smear the open circular and supercoil forms even when low concentrations were used for gel separation. Nevertheless, the conclusions are supported by previous results obtained with RK2-containing E. coli in which all three forms were also detected after synthesis by the inner membrane fraction in the gel separation system (9) and in which supercoil DNA was detected after CsCl-ethidium bromide density gradient centrifugation (12).

FIG. 3.

Characterization of newly synthesized plasmid DNA by the inner membrane fraction of P. aeruginosa. The arrows point to the marker RK2 (pRK21382) plasmid DNA molecules: top, open circular; middle, supercoil; bottom, linear. The intense band in the experimental lane represents template as well as newly synthesized open circular and supercoil plasmid DNA. M, marker lane; E, experimental lane.

Synthesis of DNA by the soluble fraction.

Although levels of TrfA varied in the soluble fraction of each of the species from low (compared to the membrane fractions) to almost undetectable, it still was important to ascertain whether plasmid-specific DNA was synthesized endogenously as observed for the membrane fractions. The results with one representative species, R. sphaeroides (Fig. 4), showed that extensive synthesis did occur in the soluble fraction, but such synthesis was completely unaffected by anti-TrfA antibody, suggesting that it represented nonspecific chain extension and/or repair of plasmid DNA fragments.

FIG. 4.

In vitro synthesis of DNA by the soluble fraction of R. sphaeroides. Incubation and reactions were carried out as described in Materials and Methods. □, preimmune serum; ●, anti-TrfA antibody.

DISCUSSION

The data presented in this paper are consistent with the model in which TrfA functions as both a membrane anchor and initiator for plasmid DNA replication. It is remarkable that the TrfA proteins which contain no apparent signal sequence and only one small hydrophobic region that would not span the membrane (although it could be embedded within the lipid protein bilayer) would be present and synthetically active in membrane fractions of five distinct gram-negative species. Particularly important is the fact that extensive synthesis and inhibition by anti-TrfA antibody occurs primarily with the inner membrane fraction of all these species and only slightly in the outer membrane fraction, suggesting strongly that there is selectivity and consistency with the results of E. coli. It remains to be seen, however, whether the same subfraction derived from the inner membrane of E. coli (subfraction B containing approximately 10% of the total membrane) that has been shown to contain the plasmid replicon (7) will also be shown to contain the replicon in these other species. Confirmation of the importance of the inner membrane as the site of plasmid DNA synthesis is also based upon the lack of anti-TrfA-antibody-inhibited synthesis in the soluble fraction (Fig. 4). Somewhat surprising, however, was the relatively extensive level of synthesis in this fraction; the source of the synthesis is speculative, but it is probably representative of plasmid DNA fragments which are dissociated from the membrane during extraction by shearing and which are extended nonspecifically by DNA polymerase. Nevertheless, this study, coupled with previous results (7) concerning the synthetic capability of the subcomplex derived from the inner membrane of E. coli, reinforces the universality of the inner membrane as the site of the plasmid replicon in the bacterial cell. Recent critical cytological evidence by others observed with both E. coli (6) and Bacillus subtilis (10) has also lent strong support to the idea that components of the replication machinery are specifically localized, probably at the membrane. Whether the inner membrane represents that site in bacteria is unknown.