Identification of a Novel Membrane-Associated Gene Product That Suppresses Toxicity of a TrfA Peptide from Plasmid RK2 and Its Relationship to the DnaA Host Initiation Protein

Abstract

The toxicity of a peptide derived from the amino-terminal portion of 33-kDa TrfA, one of the initiation proteins encoded by the broad-host-range plasmid RK2, was suppressed by a host protein related to DnaA, the initiation protein of Escherichia coli. The newly identified 28.4-kDa protein, termed a DnaA paralog (Dp) because it is similar to a region of DnaA but likely has a different function in initiation of plasmid RK2 replication, interacts physically with the 33-kDa TrfA initiation protein, including the initiation-active monomeric form. The Dp has a cellular distribution similar to that of the 33-kDa TrfA initiation protein, being found primarily in the inner membrane fraction, with lesser amounts detected in the outer membrane fraction and almost none in the soluble fraction of E. coli. Maintenance and inner membrane-associated replication of plasmid RK2 were enhanced in a Dp knockout strain and inhibited in strains containing extra copies of the Dp gene or in membrane extracts to which a tagged form of Dp was added. Recently, the Dp was independently shown to help prevent overinitiation in E. coli and was termed Hda (S. Kato and T. Katayama, EMBO J. 20:4253-4262, 2001).

The broad-host-range plasmid RK2 is capable of replication and stable maintenance within a wide range of gram-negative bacterial hosts. Carrying resistance genes to three antibiotics (ampicillin, kanamycin, and tetracycline), it has been isolated from a number of different medical environments (4). Despite the diversity of host cells, only two plasmid-carried loci are necessary to initiate unidirectional replication, the cis-acting origin of replication (oriV) and the trans-acting initiator gene trfA, which codes for two proteins of 44 and 33 kDa, the latter of which is expressed from an internal translational start within trfA (35). Recently, specific binding of TrfA and host initiation protein DnaA of Escherichia coli to oriV was found to be required for initiation of plasmid replication, consistent with the presence of four DnaA boxes within the oriV sequence (15). It appears that the DnaA protein cannot by itself form an open complex in oriV but rather enhances formation of the complex by TrfA. Nevertheless, numerous experiments have failed to demonstrate a direct physical interaction between TrfA and the DnaA protein although a physical association between TrfA and the ClpX chaperone protein has been observed (17). The latter interaction is important in converting the dimer form of TrfA, which is inactive in initiating DNA replication, to the active monomer form (16, 38).

RK2 replication is associated with the inner membranes of a number of gram-negative hosts (1). The TrfA initiation proteins were detected in both inner and outer membrane fractions of the hosts (1, 25), but further studies of E. coli revealed that oriV, the TrfA initiator proteins, and replication inhibited by specific anti-TrfA antibody were enriched in a specific subdomain found in the inner membrane fraction, representing less than 10% of the total membrane (13, 23). Finally, Pogliano et al. (29) have localized RK2 to clusters in their E. coli host that are targeted to specific locations which could involve membrane receptors (D. R. Helinski, personal communication).

Numerous parallel studies with E. coli have also suggested that in vivo replication is membrane associated. The DnaA protein is located at the cell membrane (19, 26) and is activated by anionic phospholipids (32, 41), and the origin of replication (oriC) has been found to bind to a sub-inner membrane domain whose density is the same as that of the one which binds oriV (3).

Previous studies in our laboratory have examined the structure of the 33-kDa TrfA protein in relation to its function (14). Two membrane binding domains were detected, one of which was related to TrfA's ability to function as an initiation protein, while the second was part of a peptide that exhibited a severe toxicity for host cells. One possible explanation for this toxicity was that it was the consequence of an interaction between the peptide and a necessary host protein. If this were the case, it should be possible to overcome the toxicity by transformation with a vector containing the gene that encodes the required host protein. An E. coli library was used to select for host sequences which, when placed in trans of the toxic TrfA peptide, would be capable of rescuing the host cells.

This strategy of multisuppression analysis resulted in the isolation of a single open reading frame (ORF) that codes for a protein possessing extensive similarity with the DnaA protein (but that is one-half its size), a DnaA paralog (Dp). The Dp isidentical to a protein referred to as Hda, which is involved in preventing overinitiation of E. coli replication (11, 12). Here, a tagged version of Dp was observed to physically interact in vitro with the monomer and dimer forms of TrfA, unlike the DnaA protein, although the tagged Dp was found enriched in the inner membrane fraction similarly to the TrfA and DnaA proteins (1, 19). Additionally, an E. coli strain in which the gene for Dp had been deleted demonstrated greater maintenance for an RK2 miniplasmid than its isogenic control strain. Conversely, plasmid maintenance was inhibited in a strain that contained extra copies of Dp. Parallel experiments in vitro revealed that membrane-associated RK2 miniplasmid DNA replication was enhanced when membrane extracts prepared from the knockout strain were used and that such replication was inhibited when reactions included the purified tagged Dp protein.

MATERIALS AND METHODS

Bacterial strains, plasmids and growth conditions.

E. coli strains used were DH5α (39); BL21(DE3)/pLysS, which contains an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible T7 RNA polymerase gene (37); MC4100 (2); SCS1, a supercompetent strain from Stratagene; JE202, a strain in which the Dp coding sequence was replaced with a Tet cassette (J. E. Camara, K. Skarstad, and E. Crooke, submitted for publication); and MG1655, the isogenic wild-type parent of JE202. The plasmids used in this study include the IncQ plasmid pJAK13 (J. A. Kornacki and D. Figurski, unpublished results), containing an inducible Ptac promoter and the streptomycin (SM) resistance (Sm^r) gene; pACYC184, a vector containing oriV from plasmid p15A and two genes encoding resistance to chloramphenicol (Cm^r) and tetracycline (Tet^r) (5); pBluescript II KS, a phagemid vector from Invitrogen encoding resistance to ampicillin (Amp^r); another phagemid vector, pCR2.1, from Invitrogen with an inducible Ptac promoter encoding resistance to antibiotics chloramphenicol and kanamycin; pRK2501, an RK2 miniplasmid containing oriV, the trfA gene, a maintenance region (trfB), and genes encoding resistance to kanamycin and tetracycline (10); and five plasmids constructed for this study, pJAK13-T1, containing the coding sequence for the toxic fragment (termed T1); pMC54 and pMC58, derived from the p15A library containing 10-kb chromosomal inserts, pPK100, a pCR2.1 derivative containing the Dp coding sequence from the p15A insert, and pPK101, which expresses an N-terminal T7 epitope-tagged Dp in a pET17b vector (Novagen).

Plasmid-containing strains were constructed by CaCl₂ transformation (9). The growth medium was Luria-Bertani (LB) broth (10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl/liter). For selection of antibiotic resistance, the medium was supplemented either with kanamycin (50 μg/ml), penicillin (150 μg/ml), ampicillin (15 μg/ml), SM (10 to 20 μg/ml), or chloramphenicol (30 to 50 μg/ml). Strains expressing various alleles constitutively were grown to saturation before harvesting by centrifugation. Strains carrying inducible alleles were grown to mid-log phase and induced for 3 h, usually with 1 mM IPTG.

Plasmid DNA was isolated by the miniprep alkaline lysis procedure (22), by the Qiagen Midi kit method, or, when larger quantities of pure plasmid DNA were desired, by equilibrium centrifugation in CsCl-ethidium bromide density gradients (22).

Construction of Dp deletion.

A Tet^r cassette was introduced into the gene coding for Dp of strain MG1655 by using a lambda recombination system (40) such that 647 nucleotides were deleted and only the first 50 and last 50 nucleotides remained (strain JE202). Additional details will be published elsewhere (J. E. Camara, K. Skarstad, and E. Crooke, submitted for publication).

Preparation of E. coli library pACYC184.

Chromosomal DNA was prepared from E. coli MC4100 (2) and partially digested with Sau3A and TaqI, resulting in a DNA fragment size range of approximately 0.5 to 20 kb based on sizing by agarose gel electrophoresis. The DNA was cloned into pACYC184 (5) that had been singly digested with AccI and BamHI (which are compatible with TaqI and Sau3 ends, respectively). AccI- and BamHI-cleaved vector DNA that had been cleaved solely within the Tet^r cassette region was isolated by agarose gel electrophoresis before use. After ligation, the DNA was transformed into SCS1 supercompetent cells (Stratagene), which were then plated on LB plates containing chloramphenicol (30 μg/ml). Sufficient numbers of transformants were harvested and pooled to create a library of approximately 35 genome equivalents. The approximate size range of the DNA inserts in the library was estimated by analysis of plasmid DNA from 20 transformants selected at random.

Preparation of T-peptide fragments from TrfA, cloning of fragments, and identification of sequences from isolated plasmids.

Various 80-amino-acid-residue fragments of the 33-kDa TrfA protein (T1 to T4) were constructed as described by Kim et al. (14) by PCR amplification of specific portions of trfA such that an XbaI site was added to the amino terminus-encoding end, and a TGA stop codon and XhoI site were added to the carboxyl terminus-encoding end, of each trfA fragment. The coding sequence for T1 or those for the other T fragments, as controls, were first cloned into pBluescript KS by utilizing XbaI and XhoI sites. From the resulting plasmids, DNA was excised by cutting with enzymes XbaI and KpnI, the KpnI site being present on the polylinker of pKS. These DNA fragments were then directionally ligated into the IncQ plasmid pJAK13 and oriented so that each fragment could be expressed from the lac promoter.

Cells harboring the toxic T1-expressing IncQ plasmid were plated at increasing amounts of IPTG to determine minimal toxicity. Without IPTG, growth of the cells containing the low-copy-number IncQ plasmid was not inhibited. At 0.02 mM IPTG (for 3 h at 37°C) cells grew slowly, and at 0.05 mM IPTG the cells did not grow at all. A concentration of 0.05 mM IPTG was taken to be the minimal toxic concentration and was used to screen the p15A library. It is important to emphasize that the IPTG concentration necessary to induce a level of T1 peptide sufficient for complete inhibition of growth is at least 10-fold lower than typical IPTG levels used for induction of target proteins. In effect, the peptide is highly lethal for cells.

Sequences obtained from isolated plasmids were compared against GenBank by BLAST searches (http://www.ncbi/nlm.nih.govI). Sequences with 100% homology were found on section EA336 of the E. coli chromosome (see Results). Protein sequences and Kyte-Doolittle (20) hydropathy plots were determined by using DNA Strider.

Membrane binding of Dp.

Inner and outer membrane fractions were extracted from BL21(DE3)/pLysS containing plasmid pPK101 (see below) by using a French pressure cell followed by ultracentrifugation of the disrupted cells in neutral sucrose gradients as described previously (25). In this method, inclusion bodies (if they form) are removed from the total membrane fraction by a low-speed centrifugation prior to separating the inner and outer membrane fractions. The soluble fraction is obtained from the supernatant fluid remaining after high-speed centrifugation to pellet the membrane fraction.

To selectively label the Dp, plasmid pPK101 was constructed such that, when expressed, Dp was tagged with an N-terminal epitope (T7). Paralog DNA from pPK100 was amplified by PCR. Primers were designed with restriction sites as well as an extra G base on the upstream primer to allow for in-frame preservation when being cloned into the pET17b vector from Novagen. The resulting plasmid was transformed into BL21(DE3)/pLysS (strain 1921) with no induction imposed to avoid any deleterious effects possibly associated with overexpression of the Dp. The sequencing of the gene encoding the tagged Dp revealed that, in addition to the fusion of the amino-terminal T7 epitope, a single nucleotide change in the second residue resulted in a Val-to-Ala conserved substitution and that the loss of a nucleotide near the 3′ end resulted in a frameshift such that the carboxy Leu of Dp was changed to a Cys followed by seven additional amino acids (Ser Ser Ser Arg Ser Gly Gly). Cells seem not to be capable of tolerating more than one copy of the gene encoding Dp unless the resulting protein is part of a large fusion complex (12) or, as observed here, a slightly altered protein is produced. Western blotting with a monoclonal antibody against the T7 tag was performed on each membrane or soluble fraction after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using the enhanced chemiluminescence (ECL) kit of Amersham.

Immunoaffinity assays.

Whole-cell extracts of an induced (1 mM IPTG for 3 h at 37°C) log phase culture of strain 1921 or 1110 (pET17b vector without the cloned Dp gene; see above) were obtained by sonication. The induced cultures were centrifuged at 6,000 rpm in an SS34 rotor for 10 min, and the pellet was stored at −70°C. The pellet was thawed and resuspended in 300 μl of sonication buffer (100 mM NaCl, 50 mM NaH₂ PO₄, 20 mM Tris-HCl [pH 8.0], 1 mM phenylmethylsulfonyl fluoride). The suspension was sonicated on ice with two 20-s pulses (550 sonic disruptor; Fisher Scientific; setting 6). The resulting lysate was probed for T7-Dp by using Western blots as described above. Isogenic vector (pET17b) extracts and induced tagged TrfA fragment extracts (14) were used as controls, the latter to test for nonspecific interaction of the TrfA protein with T7-tagged peptides. A monoclonal antibody against the T7 tag was bound to protein A-Sepharose beads (Pharmacia) by adding 2 μl of antibody to 10 μl of protein A-Sepharose beads and 180 μl of adsorption buffer (20 mM NaHPO₄ [pH 7.0], 10 μl of 10-mg/ml bovine serum albumin [BSA]). The antibody-bead mixture was incubated on a rocker for 1.5 h at 4°C and washed twice with 200 μl of adsorption buffer by using an Eppendorf microcentrifuge. To this mixture, induced T7-tagged Dp extracts or vector extracts or one of two T7 TrfA fragment extracts (75 μg/ml each; 10 μl) and either purified dimer or monomer forms of the TrfA protein (kindly supplied by A. Toukdarian, University of California, San Diego; 50 μl; 2.5 to 5.0 μg/ml) were added and brought to a total volume of 200 μl of binding buffer (20 mM HEPES, pH 8.0). These mixtures were incubated with gentle shaking at 4°C for 1.5 h and washed four times with 200 μl of binding buffer. The precipitated bead complex was then brought to a volume of 200 μl of binding buffer (20 mM HEPES, pH 8.0), resuspended in SDS-PAGE loading buffer, boiled for 10 min, and loaded onto the gels. After electrophoresis, Western blots were probed for TrfA by an anti-TrfA antibody.

Maintenance studies with the mini-RK2 plasmid.

The ability of the mini-RK2 plasmid to maintain itself under nonselective conditions in different host strains containing different levels of the Dp protein was analyzed. These levels ranged from extra copies of the paralog (as exemplified by strain 1921, which contained an inducible plasmid with a Dp gene insert) to a Dp knockout strain (strain JE202). Their isogenic parent cells (strain 1110, containing the inducible vector without the Dp gene insert, and strain MG1655, respectively) were also assayed. After transformation with the mini-RK2 plasmid, growth in LB broth was monitored by measuring turbidity until the mid-log phase for strains 1921 and 1110, after which IPTG (0.5 mM) was added and the cultures were incubated further until saturation (approximately 3 h). Penicillin and chloramphenicol were present throughout growth to maintain selection of the vector plasmid and the host strain (BL21), respectively. However, no selection for the mini-RK2 plasmid was imposed. After saturation, cultures were diluted 10⁵-fold in fresh LB broth. This procedure was repeated until the strains had grown for approximately 60 generations, calculated as described in the legend of Fig. 5. Colonies were screened for the presence of the mini-RK2 plasmid on LB agar plates containing kanamycin and compared to similar plates without the antibiotic. The same procedure was followed for strains JE202 and MG1655 but without addition of IPTG or antibiotics to the LB broth.

FIG. 5.

Stability of the mini-RK2 plasmid in different hosts. (a) Strain 1921, Dp coding sequence inserted into a pET17b vector transformed into BL21(DE3)/pLysS with IPTG-inducible T7 RNA polymerase (○); strain 1110, pET17b vector without paralog in BL21(DE3)/pLysS (▵). (b) Strain JE202 (knockout of Dp gene (○); MG1655, isogenic parent strain of JE202 (▵). Procedures are described in Materials and Methods. The number of generations (n) in each overnight culture was calculated by the formula n = (log₁₀ y/x)/0.301, where y is the number of organisms at the end of a specific time period, x is the number of organisms at the beginning, and 0.301 is log₁₀ 2.

In vitro plasmid DNA synthesis by the inner membrane fraction.

From previous results (1, 13, 25), the inner membrane fraction of E. coli (or a subdomain from this fraction) has been identified as the site of plasmid RK2 DNA synthesis. Such synthesis occurred in an endogenous manner (i.e., without addition of exogenous template or replication enzymes) and was inhibited by an anti-TrfA antibody. In these studies, inner membrane-associated plasmid synthesis was assayed as described in detail by Banack et al. (1) by using Qiagen anionic-exchange resin (Midi kit; steps 8 to 13) to separate newly synthesized plasmid DNA from bacterial DNA. An antibody prepared against the TrfA proteins was used as a screen to detect specific RK2 synthesis, together with rabbit preimmune serum as a control. The N-terminal epitope T7-tagged Dp protein was purified from an induced culture of strain 1921 by using the T7 tag affinity purification kit of Novagen, as described by the manufacturer.

The membrane fraction was first preincubated with either antibody or preimmune serum (1 to 2 μl at 10 to 20 μg/ml) for 20 min at 4°C before addition to the assay mixture.

RESULTS

Detection of suppressor for toxic T1 fragment of 33-kDa TrfA.

Cells (DH5α) containing the potentially toxic T1 trfA gene fragment insert in an IncQ plasmid (Sm^r) under the control of the lac promoter were transformed with an E. coli chromosomal DNA library on a p15A plasmid (Cm^r) as described in Materials and Methods. Transformants were plated in the absence of IPTG in order to calculate transformation efficiency as well as in the presence of the inducer IPTG at 0.05 mM, which had been determined to be the minimal toxic level for T1 induction (see Materials and Methods). Twelve of approximately 10⁵ transformants were selected for their ability to grow at 0.05 mM IPTG. DNA was isolated from each of these strains and used to retransform DH5α cells which were cultured in the absence of SM but in the presence of chloramphenicol. The lack of SM resulted in a rapid loss of the low-copy-number IncQ-T1-encoding plasmid (Sm^r) and insured that DNA of the p15A library plasmid (Cm^r) was isolated. This DNA was then used to retransform a fresh culture of IncQ-T1 plasmid-containing cells, and transformants were reexamined for their ability to grow at 0.05 mM IPTG. Of the original 12 isolates, 2 were deemed to be true plasmid suppressors on the basis of their ability to grow at 0.05 mM IPTG (the remaining 10 presumably possessed secondary suppressors located on the chromosome or had alterations in the peptide itself). Plasmid DNA from these two strains was sequenced. The plasmids, pMC54 and pMC58, were each found to contain different but overlapping fragments of E. coli DNA, consisting of approximately 10 kb. However, pMC54 contained a heterologous piece of DNA, while the other contained a continuous insert. Comparison of the predicted sequences showed that the sequences of the two inserts overlapped by about 8 kb, and this region of DNA contained all or part of six predicted ORFs, as summarized in Table 1. All of the predicted ORFs have been previously characterized with the exception of ORF b2495, encoding a putative oxidoreductase, and ORF b2496, which is listed as encoding a predicted replication initiator protein based on its similarity to the ORF encoding the DnaA host initiation protein.

TABLE 1.

Genes and ORFs present in two overlapping E. coli chromosomal inserts from a p15A library plasmid that suppress the toxicity of a TrfA peptide extract

ORF (gene)	Function of product	Product
b2495	Putative enzyme; not classified	Putative oxidoreductase
b2496	Putative factor; not classified	Putative DNA replication factor
b2497 (uraA)	Transport of small molecules	Uracil transport
b2498 (upp)	Salvage of nucleosides and nucleotides	Uracil phosphoribosyltransferase
b2499 (purM)	Salvage of nucleosides and nucleotides	Phosphoribosylaminoimidazole transferase
b2500 (purN)	Purine ribonucleotide biosynthesis	Phosphoribosylglycinamide formyltransferase 1

To determine the specific source of suppression (Fig. 1), pMC58 was found to have two PstI restriction sites, both in the region of the chromosomal insert, spaced 2,897 bp apart. When pMC58 was treated with PstI, isolated from the insert, and then religated, it was found that the resulting DNA was unable to suppress T1 toxicity. The excised insert contained b2496. However, because portions of two additional ORFs besides that encoding the putative initiation factor were present in this 2,897-bp region, the initiation factor ORF was cloned alone into the high-copy-number plasmid pCR2.1 (Invitrogen) by PCR as described in Materials and Methods. The primers were designed to hybridize to the end of the ORF as well as about 40 bases before its beginning. This recombinant plasmid was found specifically to relieve toxicity of the IncQ-T1-encoding plasmid. As controls, the vector pCR2.1 plasmid by itself and two plasmids containing other inserts were tested for their ability to suppress T1 toxicity and were found to have no effect (results not shown).

FIG. 1.

Detection of the source of repression of a toxic 33-kDa TrfA peptide from a cloned chromosomal insert in a p15A library. Methods for detecting suppression of the toxic T1 peptide and cloning the suppressor are described in Materials and Methods. Each line depicts the size of the insert (in kilobases) and whether or not it suppresses the toxicity.

It was therefore concluded that the putative E. coli initiator gene b2496 suppresses the toxicity of the T1-TrfA peptide from RK2 in trans. It is identical to a gene that codes for a recently identified protein named Hda, which may help regulate the E. coli replication cycle by controlling overinitiation (12). Hda appears responsible for the DnaA-inactivating activity previously attributed to partially purified fraction IdaB, which works in conjunction with the β clamp of DNA polymerase III (11) although the Hda protein has not successfully been purified from an IdaB fraction (12).

Independently, we have named the protein a Dp because it has a high degree of similarity to a region within domain III (24) of the DnaA protein but lacks the amino-terminal protein-protein interaction domains I and II and the carboxyl-terminal DNA binding domain IV of DnaA (30). A Kyte-Doolittle (20) plot of the predicted 248 amino acids of Dp revealed a stretch of hydrophobic residues starting at amino acid 100, as well as the existence of a possible amphipathic α-helix. suggesting an interactive membrane domain (Fig. 2), perhaps similar to the one likely to exist in DnaA (8, 34).

FIG. 2.

Characterization of Dp by hydropathy plot (a) and possible amphipathic α-helix determination (b). The letters represent the amino acids making up the possible amphipathic helix starting at amino acid residue 100 in Dp

Distribution of Dp in various cellular fractions.

Immunoblot analysis for T7-tagged Dp in cellular fractions demonstrated that Dp is membrane associated (Fig. 3), with a large majority of the protein present in the inner membrane fraction, a smaller amount in the outer membrane fraction, and hardly detectable levels in the soluble fraction. This is similar to the distribution of the TrfA initiation proteins (1) and partially similar to that of the DnaA protein, which is found in the inner but not outer membrane fraction (19) and in the cytosol (33).

FIG. 3.

Dp distribution in whole-cell extracts (Total), soluble fraction (Sol), and inner and outer membrane fractions of strain 1921 [BL21(DE3)/pLys transformed with pPK101, the pET17b vector containing the Dp insert]. Growth conditions, transformation procedures for preparing an N-terminal (T7)-tagged Dp, extraction of membrane fractions, and detection of the T7-tagged Dp are described in Materials and Methods. The intense bands migrating at the 30,000-molecular-weight marker (30K marker) represent the T7-tagged Dp. The lighter bands migrating near the 66K marker could represent a possible dimer form of the Dp. A new monoclonal T7 antibody from Novagen which shows no cross-reaction to nontarget E. coli proteins and no background on Western blots was used for these assays. Absence of cross-reaction was confirmed by a lack of any reaction in the soluble fraction since, if such cross-reactions to nontarget proteins occurred, they should have been detected there as well.

Interaction of the Dp with TrfA.

Immunoaffinity assays were used to determine whether a physical interaction between the TrfA and Dp proteins occurred. Initially, interaction with the dimer (wild-type) form of 33-kDa TrfA was assessed. When the purified 33-kDa TrfA protein was added to extracts derived from cells expressing or not expressing T7-tagged Dp, an anti-TrfA antibody reacted to the greatest extent when probing mixtures containing the Dp extract and TrfA protein, but not when added to control vector extracts or extracts containing various TrfA peptides (Fig. 4a). When increasing concentrations of either the dimer or monomer form of the 33-kDa TrfA protein were added to a constant amount of Dp or control extracts (Fig. 4b), there was a proportional increase in the amount of TrfA (monomer or dimer) bound to Dp after SDS-PAGE, with greater binding levels observed with the dimer form, presumably because there are two copies of the T1 region in the dimer compared to one in the monomer form. Little or no binding was observed with either the vector control or T2 fragment extracts. Thus, it can be concluded that, unlike the DnaA protein, Dp does physically interact with the TrfA initiation protein.

FIG. 4.

Interaction of the Dp with 33-kDa TrfA. Immunoaffinity assays and Western blotting using an anti-TrfA antibody were used to detect the physical binding of Dp with the 33-kDa TrfA initiation protein as described in Materials and Methods. (a) Interaction of the wild-type dimer form of 33-kDa TrfA with Dp. Strain 1110 contains the pET17b vector without the Dp insert transformed into BL21(DE3)/pLysS. Strain 1921 contains the pET17b vector with the Dp insert (pPK101; constructed as described in Materials and Methods) transformed into BL21(DE3)/pLysS. T2 and T3 represent T7-tagged peptides derived from 33-kDa TrfA (14). (b) Interaction of various concentrations of both the dimer and monomer forms of TrfA with Dp. See the panel a legend for a description of strains and extracts.

In vivo and in vitro effects of the Dp. (i) In vivo.

To determine the biological relevance of the Dp, different levels of the Dp protein were assayed for their ability to affect mini-RK2 maintenance as described in Materials and Methods. It can be seen first (Fig. 5a) that, although the miniplasmid is unstable without selection in both control cells and cells containing extra copies of the Dp, the degree of instability is much greater in the latter: more than 85% of the population lost the plasmid after 30 generations compared to only 50% in the control. After 45 and 60 generations, the loss is greater than 95% in “Dp-extra” cells in comparison to 20 and 16%, respectively, in controls. In contrast, in knockout and control cells (Fig. 5b) the loss of the mini-RK2 plasmid is not as severe as that in cells containing extra copies of the Dp but clearly the degree of loss is less after 45 and 60 generations in knockout cells than in their isogenic control cells. The relative difference in maintenance of the mini-RK2 plasmid in control cells from each strain (1110 and MG1655; see Materials and Methods) is probably due to their different genotypes. Nevertheless, the inverse correlation between in vivo levels of the Dp and survivability of the mini-RK2 plasmid is consistent with in vitro results shown in Fig. 6 and 7 and the model of Dp action proposed in Discussion.

FIG. 6.

Inner membrane-associated mini-RK2 plasmid replication in vitro in a Dp knockout strain and its isogenic control. Specific plasmid DNA synthesis was assayed by using inner membrane fractions of E. coli as described in detail by Banack et al. (1). Membrane fractions were first preincubated with either the anti-TrfA antibody or preimmune serum, as described in Materials and Methods, to further identify RK2 replication. Each data point represents an average value derived from four pooled replicate samples.

FIG. 7.

Inner membrane-associated mini-RK2 plasmid replication in vitro in the presence of different concentrations of the purified Dp. Both the miniplasmid-containing knockout strain and its isogenic control were used to extract the inner membrane fraction and assay it as described in Materials and Methods. Each data point represents an average value derived from four pooled replicate samples.

(ii) In vitro.

Membrane-associated RK2 DNA synthesis was investigated by using membrane extracts prepared from cells that lack or had Dp (knockout and control strains). The effect that purified T7-tagged Dp had on membrane-associated RK2 replication derived from these two strains was also examined.

In the knockout strain there is a significant enhancement of inner membrane-associated plasmid DNA synthesis compared to that for the isogenic control (Fig. 6), with the degree of inhibition of such synthesis by the anti-TrfA antibody the same for the knockout and control samples after 15 min (approximately 10 to 11 pmol). However after 30 min there is a significant difference in the degree of inhibition by the anti-TrfA antibody between the two inner membrane extracts (50% greater in knockout extracts), indicating that the initiation potential of the knockout extract accelerates. Whereas 29 pmol was incorporated into plasmid DNA in the knockout extract, only 14 pmol was incorporated after anti-TrfA antibody treatment (a difference of 15 pmol). In contrast, the relevant figures for the control extract were 19 and 9 pmol, respectively (a difference of 10 pmol). Since overall replication continues at an increasing rate in untreated knockout extracts in comparison to that in control extracts and in comparison to that in both extracts treated with the anti-TrfA antibody, the projected initiation potential would be greater than 50% at longer incubation times.

When different amounts of purified Dp protein (4 and 16 μg) were added to an inner membrane extract prepared from the knockout strain, mini-RK2 plasmid DNA synthesis was significantly inhibited (Fig. 7). However, the higher concentration (16 μg) yielded only a slightly greater inhibition, suggesting that saturation levels for exogenous Dp had nearly been achieved with 4 μg of Dp, which is what might be expected when a missing component is restored to a physiologically reasonable level. To a lesser extent, a similar inhibitory effect by T7-tagged Dp on wild-type inner membrane miniplasmid replication was observed, presumably because there already was a significant level of endogenous Dp present. Any additional Dp would be mostly refractory.

It should be also be noted, as in Fig. 6, that plasmid synthesis is greater when using the knockout inner membrane extract than the control membrane extract. The difference between the two is greater in Fig. 7. The reason for this may be related to differences in control supplements in the different experiments, preimmune serum as a control for the anti-TrfA antibody (Fig. 6) and BSA as a protein control for added Dp (Fig. 7). Preimmune serum is obviously a much more complex supplement than BSA, with unknown components that could affect plasmid replication.

DISCUSSION

With such an important process as DNA replication, there are bound to be numerous (perhaps overlapping) regulatory controls on all phases of such synthesis including (and in particular) initiation. The central role of the DnaA initiation protein acting alone or in tandem with other initiation proteins to denature AT-rich regions at the origin has been recognized for some time (18). In plasmid RK2 this protein cooperates with the plasmid-encoded TrfA initiation proteins to open the AT-rich region so that elongation can begin. DnaA by itself cannot form an open complex with oriV but can when working in conjunction with TrfA. Another aspect of DnaA function is DnaA activation by bound ATP and its inactivation when converted to the ADP-DnaA form. In E. coli this conversion is crucial so that initiation occurs once, and only once, at a fixed time in the cell cycle; otherwise the presence of continuously active DnaA could promote extra initiation events, resulting ultimately in abortive elongation and nonviable cells. Whether overinitiation represents a critical problem for a plasmid such as RK2 is uncertain since it could result simply in a higher copy number. Nevertheless, it would be prudent for any replicon that requires DnaA for initiation to be able to control the open “window” of activation. Thus far, at least three mechanisms in E. coli to shut the window have been detected. The first is sequestration of oriC by the membrane for a short time after initiation. Sequestration involves the hemimethylated state of newly synthesized DNA (31) and the SeqA protein (21) to prevent DnaA from binding to oriC. The second mechanism is the titration of the DnaA protein by a cluster of high-affinity DnaA binding sites (at the datA locus), thus limiting the levels of available DnaA to act at oriC (28). Third is the action of the Hda protein acting with the β clamp subunit of DNA polymerase III to accelerate the conversion of ATP-DnaA to ADP-DnaA (11, 12). None of these mechanisms, which negatively regulate initiation of replication from oriC, is absolutely required. For example, a knockout strain of the Hda protein proliferates as well as its parent (J. E. Camara, K. Skarstad, and E. Crooke, submitted for publication). Nevertheless, quantitative differences between these processes may be operative depending on the type of replicon involved, including the RK2 plasmid. The Hda protein (Dp) was shown in this investigation to suppress the toxicity of a TrfA peptide. The paralog was found physically to interact with the TrfA protein (unlike the DnaA protein), including the active monomer form of the TrfA protein. The role of Dp in regulating plasmid RK2 initiation may be different from that for initiation at oriC, although it is possible that its participation in accelerating the inactivation of the DnaA protein may still be relevant. In the present studies, it appears that Dp has a negative regulatory function (the same result as envisioned for initiation of host chromosomal replication). For the plasmid, deletion of Dp results in enhanced maintenance and replication, whereas increasing the levels of Dp causes an inhibition of such effects. Thus, even with the overinitiation of replication at oriC in Dp-depleted host cells (12), there does not seem to be any deleterious effects on plasmid replication. It is possible that the physical interaction between the TrfA initiation protein and Dp provides the main clue as to its mechanism of action. Since both the TrfA and DnaA proteins cooperate in open complex formation of oriV but do not stably interact, any interference of this cooperative effect could be inhibitory, such as a steric hindrance produced by protein-to-protein binding of Dp to the TrfA protein. There are several ways this could occur. For example, the delivery of the DnaB helicase to the open complex of oriV requires a physical interaction between DnaA and DnaB proteins, but helicase activation cannot occur unless both DnaA and TrfA proteins are present to reposition the helicase so that an appropriate prepriming complex can form (16). Such a repositioning could be inhibited by the binding of the Dp to the TrfA protein. Similarly, although the DnaA protein binds to four DnaA boxes within oriV that potentially can form an open complex in an AT-rich region, this open complex does not, in fact, form unless the TrfA and HU proteins are also present (15). The DnaA protein enhances or stabilizes the open complex. The binding of Dp to TrfA could sterically interfere with this cooperative activity. However, it is obvious that some type of regulatory system must exist to prevent Dp from exerting its inhibitory effects on TrfA during the short window of initiation activity. It must act only after initiation has been completed. What this timing control system might consist of is unknown, but it could involve a membrane component (see below). Such possible steric-hindrance effects are under investigation at present.

Another dimension of the regulatory effects of Dp in the initiation process is its probable location in the cell, namely, the cell membrane as shown in Fig. 3, like the TrfA and DnaA proteins (1, 19, 26, 33). In addition, the fact that different cellular Dp levels (from “0” to “extra”) can affect membrane-associated replication is also suggestive of the cellular site in which it functions (Fig. 6 and 7). However, while the membrane may act as a positive factor in the function of the TrfA and DnaA proteins, it may act as a negative factor for Dp, sequestering it at a specific site to prevent it from binding to the TrfA protein in a manner similar to that of the SeqA-hemimethylated oriC membrane negative regulatory system (21, 31). Although there is compelling evidence from a variety of studies, including our own, for the membrane site of plasmid (and bacterial) replication (for reviews, see references 6 and 7), a new analysis of Bacillus subtilis replication has expanded this evidence significantly to include a linkage between a number of membrane-bound respiratory enzyme complexes and DNA replication (27), including a subunit (E2) of the pyruvate dehydrogenase complex which Stein and Firshein had identified previously (36) as a membrane-associated regulator of B. subtilis replication. However, in addition, Noirot-Gros et al. (27) identified another protein in B. subtilis (YabA) which may also play a role in initiation similar to that of the Dp (or Hda) protein although there is no similarity of protein composition and therefore no homology with DnaA. Nevertheless, their analysis of YabA function as a negative regulator of initiation in B. subtilis is similar to the analysis of Hda protein function and supports our own conclusions that Dp represents a class of proteins that play accessory or redundant roles in prokaryotic DNA replication. Their results also point to interactions of YabA with a variety of other proteins in B. subtilis, including certain methyl acceptor transmembrane receptors. Although we have no evidence for such an interaction with the Dp protein in E. coli or RK2, it is not surprising that Dp can interact with other initiation proteins such as TrfA, encoded by plasmid RK2.

Much remains to be elucidated regarding Dp control of plasmid RK2 initiation. It would be important, for example, to determine why the T1 fragment is so toxic to cells. The reasons might include an interference with the activity of the Dp protein or other as yet unrecognized host proteins, especially because the T1 fragment is overexpressed. From previous results (14), toxicity was partially eliminated when a fusion between the T1 and T2 peptides was constructed, and, of course, no toxicity is detected when the peptide is part of the entire TrfA protein. Such results could suggest that the conformation of the T1 peptide is altered when it is removed from its natural region in the full protein. Other considerations include which amino acids are involved in the binding of Dp to the TrfA protein and which residues in the T1 peptide are similarly involved in the interaction with Dp. Does Dp interact physically with other replication (or nonreplication) proteins besides TrfA, as does YabA,? Regardless of the answers to these questions, the versatility of Dp in affecting diverse replicons represents an intriguing aspect of its function.