Isolation of an Inner Membrane-Derived Subfraction That Supports In Vitro Replication of a Mini-RK2 Plasmid in Escherichia coli

Abstract

Previous results have demonstrated that the inner, but not the outer, membrane fraction of Escherichia coli is the site of membrane-associated DNA replication of plasmid RK2, a broad-host-range plasmid capable of replication in a wide variety of gram-negative hosts (K. Michaels, J. Mei, and W. Firshein, Plasmid 32:19–31, 1994). To resolve the inner membrane replication site further, the procedure of Ishidate et al. (K. Ishidate, E. S. Creeger, J. Zrike, S. Deb, G. Glauner, T. J. MacAlister, and L. I. Rothfield, J. Biol. Chem. 261:428–443, 1986) was used to separate the inner membrane into a number of subfractions, of which only one, a small subfraction containing only 10% of the entire membrane, was found to synthesize DNA inhibited by antibody prepared against the plasmid-encoded initiation protein TrfA. This is the same subfraction that was also found to bind oriV and TrfA to the greatest extent in filter binding assays (J. Mei, S. Benashski, and W. Firshein, J. Bacteriol. 177:6766–6772, 1995).

As cytological techniques have become more advanced, it has become possible to localize partition as well as replication events to specific locations within the cell. A number of studies with the green fluorescent protein have localized daughter chromosomes in dividing cells to specific loci at the 3/4 points of these cells (7). In addition, green fluorescent protein and immunofluorescence studies have revealed that the origin of replication of B. subtilis associates with the cell poles of dividing cells and that replication occurs at the cell envelope (4).

If replication does occur at a specific locus on the cell membrane, there is a possibility that a subfraction of the cell membrane which would contain highly purified membrane-associated replicating complexes could be biochemically isolated. The high degree of success previously obtained in plasmid DNA synthesis studies on crude inner membrane fractions (3) has led us to examine such synthesis in a number of subcomplexes resolved from the inner membrane fraction.

E. coli maxicell mutant CSR603 (thr-1 leuB6 proA2 phr-1 recA argE3 thi-1 uvrA6 ara-14 alcY1 galK2 xyl-5 gyrA98 rsp31 tsx-33 supE44) (6) bearing RK2 miniplasmid pRK2501 (Kan^r Tet^r) was used. Two liters of culture was grown to saturation in M9-CAA medium containing 50 μg of kanamycin/ml for selection purposes and a mixture of ¹⁴C-labeled amino acids (specific activity, 1.74 mCi/ml; ICN Corp.) at a final concentration of 0.1 μCi/ml in order to label cell proteins prior to further processing. M9-CAA medium contains 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, and 2 g of glucose/liter. Cells were harvested by centrifugation in a Sorvall RC-2 refrigerated centrifuge (7,000 rpm for 10 min) and resuspended in M9-CAA medium. Two-hundred-milliliter samples were subjected to irradiation with UV light in order to destroy chromosomal DNA, as described by Sancar et al. (5). The culture was then returned to 37°C for 1 h, at which point d-cycloserine (200 mg/liter) was added to destroy any residual viable cells. Then the nonviable but intact cells were incubated for 15 h at 37°C to allow plasmid enrichment.

Figure 1 shows an outline of the procedure used for obtaining four subfractions derived from the inner membrane. Cells were treated with lysozyme (2 mg/ml) and then lysed with 6,000 lb/in² of French pressure in order to minimize shearing of plasmid DNA. The lysate was subjected to ultracentrifugation in a Beckman L7 ultracentrifuge for 45 min at 42,000 rpm after an initial low-speed (8,000 rpm for 5 min) centrifugation (Sorvall) to remove large debris. The insoluble pellet was dispersed in HEPES buffer (0.01 M, pH 7.6) containing 30 mM KCl, 1 mM EDTA, and 1 mM dithiothreitol (Sigma) and adjusted to 2 to 3 mg of protein/ml. After dispersal, the suspension was layered on top of a two-step sucrose gradient (70% [1 ml]-53% [2.5 ml] [wt/vol]), and inner and outer membrane fractions were collected by visualization (SGO gradient). After isolation of these crude inner and outer membrane fractions, the outer membrane fraction was discarded while the inner membrane fraction was diluted with HEPES buffer to a refractive index of 1.365 to 1.370. The inner membrane fraction was then loaded onto a seven-step sucrose gradient (60% [0.4 ml]-55% [0.8 ml]-50% [1.95 ml]-45% [1.95 ml]-40% [1.95 ml]-35% [1.14 ml]-30% [0.8 ml], wt/wt) and centrifuged for 16 h at 36,000 rpm (SG1 gradient). Fractions of 400 μl were taken from the top of the gradient. Fifty microliters was removed from each fraction for precipitation with 7.5% trichloroacetic acid and analysis by the Rackbeta (LKB) scintillation counter. Two resulting peaks determined by the radioactivity profile were saved, and sucrose was added to each one until a refractive index of 1.428 was obtained. These fractions were then loaded at the bottom of a seven-step flotation sucrose gradient (66% [0.4 ml]-sample [2 ml]-50% [2.7 ml]-45% [2.7 ml]-40% [1.8 ml]-35% [0.9 ml]-30% [0.4 ml]). The gradient was subjected to ultracentrifugation for 72 h at 36,000 rpm in the Beckman ultracentrifuge (SG2 gradient). The outer membrane subfractions were not examined due to the previous observation that membrane-associated in vitro DNA synthesis is associated with inner membrane extracts only (3). The procedure is adapted from that of Ishidate et al. (1) except that a much lower French pressure was used to break open the cells, as the high pressure used previously would shear the plasmid DNA (2). In addition, a maxicell mutant of Escherichia coli was chosen for study so that the bacterial chromosome could be degraded by irradiation with UV light and to allow plasmid enrichment within the nonviable cells upon further incubation (5, 6).

FIG. 1.

Growth conditions for maxicell mutant containing the RK2 miniplasmid and procedure for fractionation of membrane subfractions. E. coli maxicell mutant CSR603 cultures were prepared as described in the text. To separate the various membrane subfractions, the procedure of Ishidate et al. (1) as modified by Mei et al. (2) was used. Sup., supernatant.

Figure 2a and b show a typical fractionation profile obtained using this method. After the second sucrose gradient (SG1), two fractions were observed (Fig. 2a), although fraction II was often observed as a shoulder of fraction III (fractions Ia and Ib from the original procedure are not present because they are derived from the outer membrane). The accuracy of these fractions was confirmed when the third sucrose gradient, SG2, was performed (Fig. 2b) and fraction II was resolved into subfractions B and F at predicted densities, while fraction III fractionated into subfractions I and Z as expected (1, 2).

FIG. 2.

Profiles of SG1 and SG2 sucrose gradients performed on the inner membrane fraction of maxicell mutant CSR603(pRK2501). (a) SG1 gradient. The crude radioactive inner membrane fraction obtained from the SG0 gradient as described in the text was centrifuged in a seven-step sucrose gradient and processed as described in the text. (b) SG2 gradient. Fractions II and III from the SG1 gradient were centrifuged separately on a seven-step flotation gradient and processed as described in the text.

According to Ishidate et al. (1) and our own previous studies (2), fractions III and II from the SG1 gradient represent inner-membrane-derived crude fractions while subfractions B and I from the SG2 gradient represent a small inner membrane subfraction comprising approximately 10% of the entire membrane fraction (1) and the classic inner membrane fraction, respectively. Subfraction F contains flagellum fragments, while subfraction Z has a high buoyant density and does not contain any characteristic markers from either the inner or outer membrane subfractions (1, 2).

The four highly purified inner-membrane-derived subfractions from the SG2 gradient were then assayed for their ability to synthesize DNA that was sensitive to antibody against the TrfA proteins. Membrane extracts containing 20 mg of total protein were added to a HEPES-buffered solution (30 mM HEPES, 30 mM KCl, 7.5 mM Mg acetate [pH 7.6]) to which an energy source and other cofactors were added (0.1 mM NAD, 0.1 mM cyclic AMP, 7.5 mM creatine phosphate, 0.1 mg of creatine phosphokinase/ml, 2 mM ATP) as well as nucleoside and deoxynucleoside triphosphate precursors (0.5 mM [each] GTP, CTP, UTP; 0.04 mM [each] dATP, dCTP, and dGTP). [³H]dTTP (0.04 mM) was added at 2 μCi/0.1 ml (20 Ci/mmol; ICN Corp.) To each reaction mixture either 1 or 2 μl of anti-TrfA or rabbit preimmune serum (representing an equal amount of protein) was added. Finally, 2 mg of tRNA/ml was added to absorb nucleases. The total volume of the reaction mixture was 100 μl. All components were purchased from Sigma.

Reaction mixtures in duplicate were incubated at 30°C for 15 min, after which the reactions were stopped by the addition of 1 ml of cold 5% trichloroacetic acid containing 1% sodium pyrophosphate. Mixtures were then allowed to precipitate on ice for 1 h, after which they were passed through Whatman 24 mm GF/C filters and washed with ice cold 5% trichloroacetic acid-pyrophosphate and 95% ethanol. Filters were dried, and radioactivity was assayed in the scintillation counter (Rackbeta; LKB). Residual ¹⁴C-protein levels from the submembrane fractions were reduced to negligible levels due to the efficient discriminatory program in the counter which distinguished between such levels and [³H]DNA.

Figure 3 shows the results of these studies. It can be seen first that although all of the subfractions were capable of synthesizing DNA in the absence of anti-TrfA antibody, only subfraction B exhibited a complete inhibition of synthetic activity in the presence of the antibody. In contrast, synthesis was hardly influenced by anti-TrfA antibody in all of the other subfractions. Second, subfraction B exhibited the greatest level of such synthesis, followed by subfractions F and Z. Interestingly, the classic inner membrane subfraction I exhibited the lowest level of synthesis, suggesting that plasmid-synthetic activity originally detected in this fraction (3) was, in fact, due to subfraction B (a significant purification). This probability was supported further by the results shown in Fig. 4, in which a kinetic comparison between the synthetic activity of subfraction B (from the SG2 gradient) and the crude inner membrane fraction from which it is derived (the SG1 gradient) was made in the presence and absence of anti-TrfA antibody. It can be seen that the difference between anti-TrfA-inhibited DNA synthesis and the control (preimmune serum) in subfraction B was much greater than in the crude inner membrane fraction over a 30-min incubation period.

FIG. 3.

In vitro synthesis by various inner membrane subfractions derived from the SG2 flotation gradient in the presence and absence of anti-TrfA antiserum. Reactions were carried out in duplicate as described in the text, and values are the means and standards error of the means (error bars) of two replicate experiments.

FIG. 4.

Comparison of in vitro synthesis by subfraction B and the crude inner membrane fraction in the presence and absence of anti-TrfA antiserum. Subfraction B was extracted as described in the text from the SG2 sucrose gradient shown in Fig. 2, while the inner membrane fraction was extracted as described in Fig. 1 from the SG1 sucrose gradient shown in Fig. 2. DNA synthesis assays were performed as described in the text, except that several SG1 or SG2 extracts were pooled separately to provide sufficient material to assay the periods shown.

An important consideration is whether the DNA synthesized by subfraction B is RK2 DNA that is a result of de novo synthetic events and not repair. Besides the aforementioned fact that the inner membrane fraction from which subfraction B is derived synthesizes an RK2 supercoil plasmid product (3), other lines of evidence supporting this supposition include the following. This is the only subfraction that binds oriV significantly and the TrfA initiation proteins (2). Additionally, it is difficult to envision what other DNA could be synthesized that is completely inhibited by antibody against the initiation proteins required for its synthesis. Because of this complete inhibition, such synthesis most certainly cannot be explained by repair which requires DNA polymerase I and would not be inhibited by anti-TrfA antibody. Finally, most if not all of the host DNA has been degraded to fragments by prior treatment of the maxicell E. coli mutant with UV light. This latter phenomenon (degradation of host DNA in the UV-treated maxicells) probably explains the type of synthesis which is largely unaffected by anti-TrfA antibody and which occurs in the other subfractions. It can be attributed either to nonspecific chain extension or repair of host (and possibly) plasmid DNA fragments that survived UV treatment.

These results are significant in that they provide further confirmation of the ability of the inner membrane to support TrfA-initiated synthesis of RK2 DNA. When highly purified subtractions derived from the inner membrane of E. coli are used, the chance of residual proteins not associated with the membrane in vivo is significantly less than in experiments using crude membrane fractions. Furthermore, these results bring together for the first time synthetic capability, presence of the TrfA initiation proteins, and binding of oriV into one relatively small membrane domain, suggesting that this domain is the site of plasmid DNA replication in the bacterial cell.