INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial conjugation is the mechanism whereby DNA is transferred from a donor to a recipient cell through a complex apparatus that is encoded by a transfer system. One of the most studied conjugative systems is the F plasmid of Escherichia coli K-12, and the entire sequence of its transfer region of 40 genes (tra) is known (reviewed in reference 12). Five classes of genes have been identified so far: those for (i) pilus synthesis, (ii) surface exclusion, (iii) mating pair stabilization (MPS), (iv) regulation, and (v) origin of transfer (oriT) nicking and DNA mobilization. About 20 genes encoded within the tra region are absolutely required for conjugation to occur since mutations in these genes completely abrogate transfer. In other cases, mutations within tra genes cause transfer levels to decrease by several orders of magnitude. In this report, we demonstrate that traN, whose gene product is involved in MPS, is more important for transfer efficiency than previously thought.

The interactions that occur between a donor cell carrying the F plasmid and a recipient cell trigger plasmid transfer and lead to the establishment of an F plasmid in both, resulting in the formation of two donor cells. Very little is known about the molecular basis of the initial events preceding the transfer of DNA, but three distinct steps are thought to occur: (i) pilus attachment to the recipient cell, (ii) pilus retraction, and (iii) eventual stabilization of the contact established between the cells (26). Fifteen proteins are required for pilus assembly; mutations in these genes yield nonpiliated cells, yet a distinct mechanism of pilus assembly or retraction has not been delineated (12). MPS is comparatively simpler and involves two proteins, TraN and TraG (26). Nonpiliated cells do not transfer DNA, showing that the pilus is absolutely required for DNA transfer, while mutations in traN or traG drastically reduce transfer efficiency but do not completely abolish it (12). Together, TraN and TraG are thought to stabilize mating pairs through an as yet unknown mechanism.

TraN is expressed as a protein of 602 amino acids (aa) that is cleaved at the N terminus to the mature form of 584 aa and targeted to the outer membrane, where it apparently resides (25). Its presence in the outer membrane suggests that it might mediate interactions with the recipient cell during conjugation. TraG, an inner membrane protein, was shown to be bifunctional since mutations in the proximal half of the gene affect pilus assembly while mutations in the distal portion of traG affect MPS (26). A stable 487-aa periplasmic product termed TraG*, possibly resulting from proteolysis at a putative peptide cleavage site found in the inner membrane protein TraG, has been observed (11). It is tempting to speculate that the primary function of TraG*, which is the C-terminal segment of TraG, is MPS.

Mutations in traN and traG affect the formation of contact-specific, multicellular mating aggregates, seen in mixtures of donor and recipient cells by electron microscopy (3). These mating aggregates are resistant to shear forces and appear to play a role in conjugation in liquid culture. The amber mutant traN548, as well as some traG mutants that express F pili, mate more efficiently on a solid surface, whereas mutations in other transfer genes do not have this property (3, 26). It has been shown that stable mating aggregates contain electron-dense regions at the site of contact between the outer membranes of the mating cells, and it has been proposed that this electron-dense region is due to the accumulation of proteins at the contact site. TraN, being an outer membrane protein (OMP), is a likely component of this junction, where it associates with recipient cell components known to be required for efficient mating (see below) (10). In addition, stable mating aggregates accumulated at the nonpermissive temperature in studies using a temperature-sensitive mutant of traD, which encodes the purported DNA pump (34). These results imply that DNA transfer proceeds with greater efficiency after MPS has occurred.

By isolating mutants that are poor recipients (Con), two major types of mutations which lower the mating efficiency of the F plasmid have been found. The first type are mutations in genes (rfa) which affects biosynthesis of the inner core of the lipopolysaccharide (LPS), specifically mutations within the gene rfaP, which encodes a protein responsible for pyrophosphorylethanolamine (PPEA) addition to the heptose I residue and extension of the heptose II residue by heptose III in the inner core of LPS (4). The second type are mutations in the gene for one of the major porins, ompA (3, 4). These mutations decrease mating efficiency by 2 to 3 logs when present in the recipient cell but have no detrimental effects on mating efficiency when they are present in the donor cell (4). Requirements for OmpA in the recipient cell are completely alleviated if the mating is allowed to occur on a solid surface (3). LPS and OmpA may thus be contributed by the recipient cell to the electron-dense region mentioned above and are postulated to act as receptors during conjugation.

R100-1, also known as NR1, is a closely related plasmid that belongs to the incompatibility group FII and encodes a number of antibiotic resistance determinants (reviewed in reference 47). Transfer of R100-1 is not affected by the recipient cell mutations in rfa or ompA, suggesting that R100-1 requires an alternate component of the recipient cell (4). Initially, it was thought that OmpA and LPS were receptors for the pilus tip during recipient cell recognition, but experiments demonstrated that both F traA and R100-1 traA complemented a null mutant of the pilin gene, traA, to equal levels, irrespective of the recipient cell used, indicating that the pilin subunit is not the donor component that interacts with OmpA or LPS (4).

In this study, we examined the MPS gene, traN, of F and R100-1. A null mutation of traN was constructed by inserting a chloramphenicol resistance (Cm^r) cassette into the coding region of traN to generate pOX38N1::CAT. We ascertained the potential effect of Con recipient cells by complementing this traN null mutation with either F or R100-1 traN in the donor cell. Assays for mating efficiencies to LPS-deficient or ompA recipient cells, as well as to recipient cells expressing F TraT, a protein involved in surface exclusion, were done to demonstrate plasmid or allelic specificity. Finally, the sequence of R100-1 TraN is reported for comparison to that of F.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and general reagents. Bacterial strains and plasmids used in this study are described in Table 1. Cells were grown in Luria-Bertani media (LB; 1% Difco tryptone, 0.5% Difco yeast extract, 1% NaCl [BDH Inc.]) at 37°C on a tube roller or shaker. Glucose was added to cultures to a final concentration of 100 mM to repress expression from an isopropyl--D-thiogalactopyranoside (IPTG)-inducible promoter or as a general growth enhancer. Antibiotics (Sigma Chemical Co.) were added to final concentrations as follows: ampicillin, 50 µg/ml; chloramphenicol, 25 µg/ml; kanamycin, 25 µg/ml; spectinomycin, 100 µg/ml; streptomycin, 200 µg/ml; tetracycline, 10 µg/ml; and nalidixic acid, 16 µg/ml. IPTG (Sigma) was added as an inducer, and 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (Sigma) was added as an indicator where needed.

Cloning. General techniques were performed as described elsewhere (6) unless otherwise indicated. All DNA modification enzymes were from Boehringer Mannheim. Vent polymerase (proofreading⁺) was from New England Biolabs. Large-scale plasmid preparations were done as specified by Qiagen (36a) except that after resuspension of the DNA pellet in Tris-EDTA buffer, the DNA was subjected to two phenol extractions, and the aqueous phase was collected and ethanol precipitated as usual. Small-scale plasmid preparations were performed as previously described (8) or by using QIAquick miniprep kits.

Construction of pOX38N1::CAT. A null mutant of traN was constructed by inserting the Cm^r (chloramphenicol acetyltransferase [CAT]) cassette from pUC4C into traN. To isolate the cassette, pUC4C was digested with BamHI to give the desired fragment of 0.86 kb, which was excised from a 1.5% agarose gel and purified by the GlassMAX DNA isolation matrix system (Gibco BRL). The ends were filled in with Klenow enzyme to blunt the fragment, and the entire cassette (978 bp) was ligated to pKI375 digested with EcoRV and BbrPI (which removes 860 bp and creates blunt ends) to generate pKI375N1::CAT. To generate pOX38N1::CAT, a triparental mating was conducted (32). Donor cells containing pOX38 (RD17/pOX38) and recipient cells containing pKI375N1::CAT (XK100/pKI375N1::CAT) were grown to mid-log phase in LB with appropriate antibiotics, the cells were pelleted and washed to remove antibiotics, resuspended in fresh LB, then mixed at equal volumes, and incubated for 1 h at 37°C to allow plasmid transfer and recombination. Recipient HB101 cells at mid-log phase were added at a fourfold excess volume and incubated for a further hour at 37°C to allow the newly generated pOX38N1::CAT to be transferred. Desired recombinants were Nal^r Cm^r and Amp^s. PCR amplification using primers SRNF and SRNR, and subsequent digestion with EcoRI, demonstrated that pKI375N1::CAT had correctly recombined with the traN sequence on pOX38.

Mating assays. Mating assays were performed as previously described (4). Briefly, cultures were grown to mid-log to late log phase in LB with appropriate antibiotics, and glucose was added to all donor cultures to a final concentration of 100 mM. Cells were pelleted and washed once with cold LB to remove antibiotics. One hundred microliters each of donors and recipients was mixed with 800 µl of cold LB, vortexed briefly, and allowed to mate for 30 min at 37°C. Matings were terminated by vortexing vigorously and putting the cells on ice to prevent further mating events. After serial dilutions in cold LB, 10 µl of each dilution was plated separately on selective plates for donors and for transconjugants. Colonies were counted after overnight incubation, and the number of transconjugants per 100 donor cells was calculated. Typically 10⁷ to 10⁸ donor cells/ml were used in each mating assay. Multiple experiments were performed with separate and multiple cultures in each experiment; each culture was started from a single isolated colony. Plasmids containing mutations were always compared to the wild-type plasmid within the same experiment. Surface exclusion indices were as previously described (4). Cold minimal saline solution caused loss of donor cell viability when used as a diluent; therefore, cold LB was used instead. The donor cultures required the addition of glucose to optimize the growth rates. Since pKI375 and related plasmids were not stably maintained in cells, only fresh transformants were used for all mating assays, and glycerol stocks were not used except for E. coli ED24 containing pOX38::Tc, pOX38::Km, or FlacN548::Tc, E. coli JE2571-1 or XK1200 containing R100-1, and all recipient strains.

DNA synthesis and sequencing. DNA synthesis was performed on an Applied Biosystems 391 or 381A DNA synthesizer (PCR-MATE). DNA sequencing was performed by the Sanger dideoxy method, using an Applied Biosystems 373 DNA Sequencer (STRETCH) or a Sequenase 2.0 kit (United States Biochemical). Oligonucleotides used in this study were SRNF (5'-GATCCGCAGGTGGCTCATGATTTAC-3'), SRNR (5'-GAATTCTTACCGTATCCCACGACC-3'), LFR53 (5'-ACCGGAACAACCATC-3'), LFR55 (5'-CACAGGAGAGTATCC-3'), LFR56 (5'-TTCATCAGGTCTTCG-3'), LFR59 (5'-ACTCCGACACTGACGGTCAG-3'), LFR60 (5'-AGTACTGCACACGGACTGCC-3'), and LFR61 (5'-CAGACAGGTTCCCTCAGCGG-3'). Sequence analysis was performed with the Wisconsin Package sequence analysis software (Genetics Computer Group Inc. [GCG], Madison, Wis.).

Site-directed mutagenesis. Single-stranded DNA was produced as described elsewhere (44). Site-directed mutagenesis was performed as specified by Zoller and Smith (48), using the two-primer method and primers LFR53 and SRNF. Heteroduplex DNA was then transformed into E. coli DH5, and the colonies were screened by colony hybridization. Colony lifts were done as described previously (39) on nitrocellulose membranes. The DNA was cross-linked to the membrane by using a Bio-Rad GS Gene Linker UV chamber, and -³²P-labeled LFR53 was hybridized in 5× Denhardt's solution (2.5× SSC [20× SSC is 3 M NaCl plus 0.3 M sodium citrate · 2H₂O, pH 7.0], 0.5% sodium dodecyl sulfate [SDS], 90 mM Tris [pH 7.5], 0.9 M NaCl, 6 mM EDTA, 100 µg of heat-denatured calf thymus DNA [Sigma] per ml) overnight at 37°C. Membranes were washed in 50 ml of 6× SSC-0.1% SDS for 15-min intervals initially at room temperature and then with increasingly stringent conditions (increasing temperature) until signal from potentially positive colonies could be distinguished above background. The membranes were autoradiographed with Kodak X-Omat AR film at 70°C, using an intensifying screen. DNA was extracted from potentially positive colonies, and the presence of the mutation was confirmed by automated sequencing.

Generation of 3' deletions of pBK8. Deletion of the 3' end of pBK8 was done with the Erase-a-Base system (Promega) (36). Approximately 10 µg of plasmid DNA (pBK8) was digested with BamHI and SacI, and deletions were generated by using exonuclease III. The first three time points were screened for deletions of the appropriate size by EcoRI restriction digests of small-scale plasmid preparations. The exact endpoint of the deletion was determined by sequencing using Sequenase and the reverse sequencing primer of plasmid pBluescript (pBS) SK+.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of pOX38N1::CAT. The amber mutant FlacN548::Tc (4) was initially used for complementation experiments; however, a high level of reversion resulted in loss of the amber mutation, possibly due to recombination between the wild-type traN sequences on pKI375 and the N548 amber mutation on Flac. Therefore, a promotorless CAT cassette was inserted in traN on the pOX38 F plasmid derivative to generate a null mutation. This would have allowed the Cm^r marker to be followed easily during mating assays as well as enabling a determination of the relative expression levels of traN. Unfortunately, Cm^r was never observed, suggesting that insufficient traN expression was generated from the main promoter of the F plasmid, P_Y, which is approximately 11 kb upstream of the traN start codon. To ensure that Cm^r was expressed, a CAT cassette with its own promoter was isolated from plasmid pUC4C and ligated into BbrPI/EcoRV-digested pKI375, resulting in replacement of approximately one-half of the coding sequence of traN with the CAT cassette to generate pKI375N1::CAT (Fig. 1). A triparental mating was used to cross this traN null mutation into pOX38, and the desired recombinants were selected on the basis of resistance to nalidixic acid and chloramphenicol and sensitivity to ampicillin. Sensitivity to phage f1 infection demonstrated proper pilus assembly and therefore uninterrupted expression of the tra genes downstream of traN which are required for pilus synthesis.

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Physical maps of the various F constructs used in this study. (A) F plasmid derivatives. (B) R100-1 plasmid derivatives. The region corresponding to the F or R100-1 plasmid from which each of these constructs is generated is shown; sizes are marked at the top. Restriction enzymes: As, Asp7001; Bb, BbrPI; Ns, NsiI; R, EcoRI; V, EcoRV. tra and trb genes are boxed. The variable region in TraN is also boxed, and the range of amino acids that this region represents in F and R100-1 TraN is shown. Each construct is shown approximately to scale, and the length of the insert in pBS is shown along with the number of expected amino acids remaining in TraN. The F plasmid deletions are all derivatives of the initial clone of F traN, pKI375, which is a 3-kb Asp7001 fragment of the F plasmid cloned into pBS KS+/EcoRV (25). The R100-1 traN clone is a derivative of a 3.5-kb NsiI fragment of R100-1 cloned into pBS SK+/PstI and subsequently treated with exonuclease to generate a deletion. pS82 is an in-frame deletion of 18 aa. T7 RNA polymerase-directed transcription proceeds from left to right in both diagrams. The region replaced by the CAT cassette in pKI375N1::CAT is boxed. The K90T mutation is a Lys-to-Thr mutation at residue 90 in TraN.

Cloning and sequencing of R100-1 traN. R100-1 and ColB2 are plasmids closely related to F; their transfer proteins are almost identical in sequence and function to those of F (4, 5, 20). The differences in R100-1 and ColB2 have been exploited as reservoirs of naturally occurring alleles of transfer genes. PCR amplification of traN using primers SRNF and SRNR demonstrated that a band of the same size could be generated by using F, R100-1, or ColB2 template DNA, indicating that a coding region corresponding to traN exists in all of these plasmids (data not shown). traN of R100-1 was sequenced with primers SRNF and SRNR after PCR amplification. Primers were generated as needed to sequence across the entire R100-1 traN open reading frame on both strands. The predicted amino acid sequence yielded a protein that was very similar to F TraN (73.6% identity; 82.3% similarity [Fig. 2]). The central region of R100-1 TraN from aa 162 to 348 is remarkably dissimilar to that of the corresponding region of F TraN (aa 162 to 333) and includes 15 extra amino acids. This sequence divergence in the central portion of TraN may indicate specificity of receptor interaction, while the conserved N and C termini might imply preserved functional domains. Preliminary sequence analysis of ColB2 TraN showed that it was virtually identical to F TraN in the central region and was not studied further.

View larger version (46K): [in this window] [in a new window]   FIG. 2.   Comparison of the protein sequences of F and R100-1 TraN. F TraN sequence is derived from nucleotides 12997 to 14802 of the F tra operon and corrected from a previously published sequence (12 [GenBank accession no. U01159]). R100-1 traN was sequenced as described in Materials and Methods. The sequence of R100-1 traN is contained within a larger fragment of R100-1 sequence (nucleotides 8469 to 10322; GenBank accession no. AF005044). The sequences were aligned with the GCG Pileup program and shaded with the Genedoc program (http://www.cris.com/~ketchup/genedoc.shtml). Single-letter amino acid codes are used; conserved amino acids are shaded in black, with amino acid residues corresponding to either F or R100-1 TraN numbered on the right. Dashes are used to indicate gaps in the sequence generated during computer alignment.

Complementation with R100-1 traN. R100-1 traN was initially cloned on a SacI/SalI fragment in pBS SK+ (pBK7, 6.2 kb). A smaller NsiI fragment of pBK7 was subcloned into pBS SK+/PstI (pBK8, 3.5 kb). A further deletions in pBK8 resulted in the elimination of R100-1 traF, a gene which in the F plasmid is involved in pilus assembly and is essential for transfer. Exonuclease III and S1 nuclease were used to generate a 3' deletion of pBK8, which was characterized by DNA sequencing. This plasmid, pBK8-2818, carries a 2,818-bp fragment of R100-1 which encodes traN and trbE but not traF. It was assumed that R100-1 trbE would not interfere in complementation assays since a null mutation of F trbE had no effect on mating efficiency, F-pilus expression, or phage infection (25).

Interactions of F TraN and OmpA. Since TraN is an OMP in the donor cell and appears to be involved in MPS, it might interact with exposed moieties on the surface of the recipient cell such as OmpA or LPS which have previously been implicated in F conjugation (4). CC277 is a strain which carries a null mutation in ompA. As shown in Table 2, depending on the nature of the recipient cell (CC277 versus MC4100; ompA versus ompA⁺), pOX38N1::CAT transfer is complemented to different levels with plasmids carrying F or R100-1 traN. When pOX38N1::CAT was complemented with F traN, mating efficiency decreased to 0.025% of wild-type levels with CC277 compared to MC4100 (ompA versus wild type; 0.04 versus 160 transconjugants per 100 donor cells), while R100-1 traN mating efficiency decreased to 56% of wild-type levels (55 versus 99 transconjugants per 100 donors). This finding suggests that F TraN, but not R100-1 TraN, may interact with OmpA during conjugation to bring about MPS. These results have been confirmed with strains CC650 and CC651, both of which carry ompA point mutations that are known to affect conjugation (data not shown and reference 27). Since the two TraN proteins exhibit dissimilar sequences for at least a third of their lengths, it is possible that this region is important for specific interactions with OMPs in the recipient cell during conjugation.

Effects of LPS-deficient recipient cells on traN complementation. It has been demonstrated that F and R100-1 mating efficiencies are differentially affected by various mutations in biosynthetic genes for LPS (4). Similar to the assays with the ompA mutations described above, complementation was assayed by using rfa recipients, which are deficient in various moieties in the inner core of the LPS, and donor cells carrying pOX38N1::CAT and plasmids expressing either of the two traN alleles.

Surface exclusion. Complementation assays were again performed with pOX38N1::CAT and the two traN alleles to determine if TraN is important for the surface exclusion function of TraT, an OMP. Surface exclusion can be measured as a ratio of the mating efficiency of a recipient cell expressing TraT compared to a wild-type recipient cell (no TraT expressed), which is called the surface exclusion index (Sfx) (15). A positive Sfx, therefore, indicates surface exclusion and, since surface exclusion is plasmid specific, would indicate specificity of interactions of TraT with donor cell components. As shown in Table 4, the Sfx is independent of traN. pOX38N1::CAT in the presence of F traN exhibited an Sfx of 245, while R100-1 traN had an Sfx of 284. The wild-type plasmids pOX38::Tc and R100-1 had Sfxs of 77 and 1.5 respectively, indicating that pOX38N1::CAT is surface excluded to approximately the same degree as pOX38::Tc, regardless of which TraN protein is expressed. Thus, TraT appears not to interfere with TraN function during mating pair formation, and surface exclusion as mediated by TraT is not specific to TraN but may be due to some other plasmid-specific protein.

Complementation with F traN 3' deletions. To further examine the function of TraN, we obtained a number of 3' deletions of F traN (25) and tested them for the ability to complement pOX38N1::CAT (Table 5). All of these constructs express a stable form of TraN from the T7 promoter in XK100 cells, as determined previously (25) (Fig. 1).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Previously, OmpA and LPS were thought to be the receptors for the F pilus during the initial contact between the donor and recipient cells (4). Our work demonstrates that the outer membrane protein TraN is responsible for interactions with OmpA and LPS, and that this interaction enables stable mating pair formation which leads to a much more efficient transfer of DNA. The receptor in the recipient cell for the pilus tip, as well as for R100-1 TraN, remains unknown.

TraN receptor in the recipient cell. Our results indicate that F TraN interacts either directly or indirectly with OmpA and LPS to enable MPS. The LPS component is potentially a PPEA moiety in the inner core (4). In contrast, R100-1 TraN probably recognizes a distinct moiety during conjugation, as it gives wild-type levels of plasmid transfer when either Con mutation (rfaP or ompA) exists in the recipient cell. Two previous R100-1 Con mutants have been previously isolated; however, the nature of the mutation which specifically decreased R100-1 transfer efficiency but not F transfer efficiency is unknown (16). Sequence analysis of R100-1 TraN revealed a high degree of similarity to F TraN at the N and C termini, while a central dissimilar region indicated significant sequence mosaicism which might reflect allele-specific receptor interactions. Radioactive labeling indicated that a protein of the expected size was expressed from pBK8-2818, and we concluded that this was R100-1 TraN. Preliminary sequence analysis showed that the ColB2 traN central region was identical to that of F traN (data not shown). We expected sequence conservation between F and ColB2 traN since ColB2 and F have similar requirements for moieties in the recipient cell during conjugation (4).

Critical region important for receptor interactions. In an attempt to determine exactly which region of TraN is important for its function, we used a number of deletion derivatives of F traN and assayed for their ability to complement pOX38N1::CAT transfer to wild-type and ompA recipients. The minimal traN sequence that gave reasonable levels of complementation was a truncated TraN of 399 aa (pS84) that presumably contains sufficient information for TraN function. Since complementation by pS84 was affected by an ompA mutation in the recipient cell, the possibility exists that the recognition domain for OmpA is in the first 399 aa.

Surface exclusion. Redundant transfer of plasmid DNA from a donor to a recipient cell is prevented by a process known as surface exclusion, which is mediated by two proteins, TraT in the outer membrane and TraS in the inner membrane. In conjunction, they inhibit the level of F⁺ to F⁺ transfer by several hundred-fold. TraT is plasmid specific; that is, F TraT inhibits F transfer, while R100-1 TraT inhibits R100-1 transfer (reviewed in reference 42). This specificity has been correlated with a single amino acid change in TraT (G120A) which changes the specificity from F to R100-1 (15). It has been reported that TraT interferes with infection of E. coli by bacteriophages which are specific for OmpA, suggesting that TraT blocks binding of F plasmid-encoded products to OmpA in a similar manner (38). We tested the possibility that TraT inhibits TraN-mediated MPS during donor-to-donor conjugation and found that TraT-dependent surface exclusion was independent of the allele of traN expressed in the donor cell. Therefore, TraT cannot block TraN-OmpA interactions and might interfere with some other aspect of transfer, possibly F pilus recognition of the recipient cell.

Two functions for TraN? The results presented here suggest that TraN has two functions: (i) recognition of a receptor and (ii) another function, possibly interactions with other plasmid-specific components. This is most apparent when one compares the effects on transfer efficiency when either ompA or traN is mutated. The lack of OmpA in the recipient cell leads to a 100- to 1,000-fold decrease in transfer efficiency. When traN is mutated in the donor cell, however, transfer efficiency is decreased 1 million-fold, suggesting that other conjugative functions are dependent on TraN. Similarly, mating on solid surfaces relieves the requirement for OmpA in the recipient cell, while traN mutations are only partially suppressed on a solid support. The MPS pathway may, therefore, be more complex and involve interactions with other transfer proteins as a prelude to DNA transport. TraG is potentially a partner in MPS (11, 26). Since MPS, a process that stabilizes two cells during conjugation, requires both an OMP and an inner membrane or periplasmic protein, this process might involve formation of a complex in the donor cell envelope.

Computer analysis of TraN. An analysis of the topology of TraN within the outer membrane could provide clues as to which portions of TraN are responsible for interactions with OmpA. Sequence comparison of TraN to the porin superfamily showed no similarity of TraN to the conserved transmembrane segments (17). Membrane topology experiments using epitope insertions are currently being done to identify those regions of TraN which are exposed on the cell surface and those which reside within the periplasm.