The Cytology of Bacterial Conjugation

MATTHEW W. GILMOUR,^† TREVOR D. LAWLEY,^‡ AND DIANE E. TAYLOR*

[SECTION EDITOR: BRETT FINLAY]

Posted November 15, 2004

Department of Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta, Canada T6G 2R3

^*Corresponding author: Mailing address: Department of Medical Microbiology and Immunology, University of Alberta, 1-28 Medical Sciences Building, Edmonton, Alberta, Canada T6G 2R3. Phone: (780) 492-4777, Fax: (780) 492-7512, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

^†Present address: National Microbiology Laboratory, 1015 Arlington Street, Winnipeg, Manitoba, Canada R3E 3R2. Phone: (204) 784-5920, Fax: (204) 789-2018, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

^‡Present address: Department of Microbiology and Immunology, Stanford University Medical Center, 299 Campus Drive, Stanford, CA 94305. Phone: (650) 723-2671, Fax: (650) 723-1837, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

In order that various genes may have the opportunity to recombine, a cell fusion would be required.

J. Lederberg and E. L. Tatum, 1946 (47)

HORIZONTAL GENE TRANSFER AFTER CELLULAR FUSION

During the analysis of mixed nutritional mutants, Lederberg and Tatum were the first to observe conjugative transfer between bacterial cells (47). Their conjecture that bacteria are capable of a "sexual process" that requires cell-cell contact also surmises the basic principles of bacterial conjugation. Donor bacteria, harboring a transmissible plasmid, must intimately interact with recipient bacteria to transfer that DNA substrate through the cell envelope of both cells. The proteins that facilitate this process are principally plasmid encoded, and, logically, most are associated with (or are located between) the cellular and outer membranes of the donor cell.

The ability of the donor cell to mediate cell-cell contacts for conjugation is a critical aspect in the evolution of bacterial genomes. Prokaryotic cells, which are otherwise limited by their unicellular nature and reproductive strategy, use such horizontal gene transfer processes to achieve genetic diversity (14, 45). Eukaryotes propagate in most instances through sexual reproduction, wherein two distinct meiotic genomes are combined to create progeny distinct from (but representative of) the parental genomes. Conversely, prokaryotic cells generate progeny through binary fission after faithful replication of their own chromosomal and extrachromosomal genetic elements, without the requirement of genetic input from a second parental cell. In addition, cell fusion, as required during eukaryotic reproduction, is prohibited by the peptidoglycan exoskeleton possessed by bacterial cells (9). In the absence of gene transfer, the only diversity introduced into a prokaryotic genome would be errors (i.e., point mutations, insertions, and deletions) introduced during DNA replication or intrachromosomal rearrangements. The evolutionary force of point mutations to create new species would eventually be seen over the course of millennia, but the immediate effect would be genomic stasis (45). Alternatively, the integration of "foreign" genetic determinants (possibly encoding novel functions) into the genome of a prokaryote would result in the inheritance of those determinants after a single generation. For a multicellular organism, germ line cells would be the only recipients that would perpetuate transferred genetic information.

The evolutionary significance of horizontal transfer was highlighted by observations that antibiotic resistance determinants were spread through the bacterial gene pool by plasmid DNA molecules rather than by de novo synthesis of these features in each resistant isolate (for reviews, see references 12 and 75). These extrachromosomal genetic elements, capable of intercellular transfer, may confer upon the host cell (and recipients after a transfer event) several resistance phenotypes, including resistance to antibiotics, toxic heavy metals, colicins, and bacteriophage. Notably, most pre-antibiotic era bacteria were not found to contain transferable resistance plasmids (34); therefore, the spread of resistance determinants to pathogens by conjugative plasmids has occurred mainly in response to human activity.

This chapter will focus on the membrane-associated structures present at cell-cell contact sites during bacterial conjugation. These transfer proteins/structures have roles in the formation and stabilization of mating contacts and ultimately the passage of substrate across the cell envelope between two bacterial cells. Using both electron and fluorescence micrographs, we present evidence for the dynamic interaction between donor and recipient cells, including the assembly of a transmembrane protein complex, and conclude with a refined model for the mechanism of bacterial conjugation.

BACTERIAL CONJUGATION

The most obvious physical barrier that must be overcome during horizontal gene transfer is the bacterial envelope. The three known mechanisms of horizontal genetic exchange are transduction (by bacteriophage), transformation (by naked DNA), and conjugation (by plasmids), each capable of transmitting DNA across the bacterial cell envelope (17). During transduction, the cell envelope is breached by bacteriophage, or during natural transformation, DNA is sampled from the environment passively or actively through an envelope-spanning protein apparatus, with each of these mechanisms potentially allowing acquisition of a range of novel genetic determinants by the recipient cells (46).

Conjugative transfer is achieved in part by a macromolecular assembly of proteins that span the bacterial envelope and terminate in the extracellular space with a filamentous pilus (43). The pilus, which is responsible for contact with recipient cells, then undergoes retraction to permit formation of the mating pair, in which there is close contact between the cell walls of donor and recipient bacteria. Different pilus morphologies have been observed (discussed below), each possibly having different roles for initiating conjugation. In all circumstances, the pilus is produced by the mating pair formation (Mpf) complex (49), a transmembrane assembly of transfer proteins. This complex may also be involved in pore formation and the subsequent translocation of DNA. Plasmids are transferred to recipient cells as a single-stranded DNA molecule, a product generated by the cytoplasmic relaxosome complex, which includes DNA-binding proteins and a relaxase responsible for DNA cleavage at the origin of transfer (oriT) (65). The Mpf complex and the relaxosome are linked via an integral inner membrane-coupling protein (20, 51), completing the conjugative apparatus.

Despite decades of research, few molecular details are known about the mechanism of bacterial conjugation. One of the principal hindrances has been the absence of visible superstructures associated with the conjugative apparatus, with the exceptions being the pilus and in vitro relaxosome assemblies, each visualized under transmission electron microscopy (Fig. 1 and 2). Using purified preparations of plasmid RP4 DNA, the RP4-encoded oriT-specific nickase/relaxase TraI, and the relaxosomal accessory proteins TraH and TraJ, Pansegrau et al. (64) observed protein complexes associated with the RP4 oriT which induced a sharp bend (Fig. 1). The enzymatic action of TraI at oriT is, however, one of the better characterized aspects of conjugation (65), with studies aided by structural determination of a relaxase in association with oriT (28). In general, the oriT is an AT-rich DNA segment represented as a tandem of repeats containing a nic site where the relaxase cleaves a single strand of the plasmid DNA duplex. The nucleophilic tyrosine of the relaxase remains covalently bound to the 5' end of the cleaved strand (which will serve as the transferred strand), and an equilibrium exists between cleaved and uncleaved states, allowing the plasmid to retain superhelicity (28, 65). If a transfer event is initiated, the remaining free 3'-OH of the cleaved strand serves to initiate replacement strand synthesis using the host replication machinery and the uncleaved strand as a template (66). After transfer of the single DNA strand to the recipient is complete, both the replacement strand (in the donor) and the transferred strand are recircularized by the relaxase (81). In the recipient cell, the transferred strand is converted into a double-stranded molecule after synthesis of nucleotide primers complementary to the transferred strand and containing a free 3'-OH from which complementary strand synthesis is initiated (5, 65). The processing of the conjugal plasmid into transfer intermediates was proposed to be initiated by a mating signal, indirect evidence for which was provided by transfer genes which are required for this conjugal DNA synthesis (37). The molecular details of plasmid processing are, from a research standpoint, more clearly understood than plasmid transport through the cell envelope.

Fig. 1Transmission electron microscopy of in vitro-assembled relaxosomes. Plasmids encoding the RP4 oriT were incubated with RP4-encoded relaxosomal proteins TraH, TraI, and TraJ. Arrows indicate bound Tra proteins. See reference 64 for additional details.

Source: Reproduced with permission from E. Lanka.

Fig. 2Transmission electron microscopy of donor cell producing an H-type flexible pilus with Hgal RNA phage attached to the distal end (indicated by the upper three arrows). The lower arrow indicates a newly synthesized pilus lacking absorbed Hgal phage. See reference 52 for additional details.

Source: Reproduced with permission from the American Society for Microbiology.

Bacterial conjugation, in addition to being a mechanism for genome evolution, can be considered as a mechanism for macromolecular secretion. In particular, plasmid-conjugative transfer is classified as a type IV secretion (T4S) system and represents the only known bacterial system for secretion of DNA (10). Bacteria not only transfer genetic information, such as determinants encoding antibiotic and heavy metal resistance using T4S systems; but the transfer of virulence proteins and possibly other factors into eukaryotic cells is also accomplished using these machines (57). Despite being encoded by a diverse range of bacterial genomes, and each transferring a different protein or nucleoprotein substrate, these multicomponent modules are each classified as T4S systems, because there is conservation of both transfer gene sequences and organization (10, 57). In particular, components of the membrane-associated conjugative Mpf apparatus required for plasmid DNA transfer are similar to proteins responsible for the intercellular translocation of virulence factors from pathogenic gram-negative bacteria (4).

THE CONJUGATIVE PILUS

The pilus is a filamentous extracellular appendage that is essential for bacterial conjugation. In all known conjugative transfer systems, a multitude of proteins are required for both plasmid transfer and pilus production. These two roles may be synonymous, because the pilus could participate in the recognition and capture of recipient cells, the stabilization of contact points between donor and recipient cells, and, in addition, by acting in a needle-like fashion, the transfer of plasmid DNA to the recipient.

Pili are classified in part by their morphology, because three different types have been observed (7): thin-flexible (6-nm outer diameter), thick-flexible (9-nm diameter, see Fig. 2), and rigid pili (variable diameter). Cells producing rigid pili are more effective at plasmid transfer when mating mixtures are assembled on a solid surface, whereas donor cells with flexible pili are generally most effective in liquid culture, although many are equally effective with either media type (7). The plasmids that will be discussed in this chapter include the F factor (encoding a thick-flexible pilus); the P group of plasmids, including RP4 and R751 (rigid); and the H plasmid group, including R27 (also thick-flexible). The central lumen of the F pili is a 2- to 2.5-nm channel (56), which notably is sufficient to potentially allow passage of a single strand of DNA.

The pilus is assembled from pilin monomers, forming a helical tubular structure. The pilin subunits encoded within the different transfer systems are not conserved at the primary sequence level; however, there is conservation of proteolytic cleavage sites and transmembrane domains (43). These features target pilin subunits to the cytoplasmic membrane; and the pilin protein has been visualized in this compartment and in the outer membrane through immunogold electron microscopy (62). Extensive posttranslational modification of pilin subunits occurs prior to pilus assembly: the F pilin subunit TraA is N-terminally processed by leader peptidase B (LepB) and is subsequently acetylated by TraX (53, 54). The P-type pilin TrbC is cleaved at the C terminus by TraF and an unknown peptidase, and at the N terminus by LepB, permitting cyclization of these periplasm-located termini by TraF (16, 29). The primary sequence of the H-type pilin subunit TrhA more closely resembles TrbC than TraA, and the prototypical H plasmid R27 also encodes a putative protease that is homologous to TraF (41). These properties suggest that the TrhA may also undergo cyclization.

Retraction of recipient cells toward a donor cell via flexible pili has been observed (15), but the subsequent activities of flexible pili during conjugation are uncertain. It is tempting to speculate that the transferring DNA strand passes through the conjugative pilus, possibly penetrating the recipient envelope and depositing the DNA directly within the recipient cytoplasm. Two studies present evidence for the flexible pilus as the route of DNA transmission (31, 61), but no studies convincingly demonstrate this premise. By using a porous solid support to separate donor and recipient cells in liquid media, transfer was still observed, implying that cell-cell contact is not a requisite for conjugation (31). This study, however, does not account for passive diffusion of DNA across the barrier and for the possibility that the barrier is nonuniform and contains pores large enough to permit cell contact, nor was there any direct observation that pili extended through the pores to contact and transmit DNA to recipient cells. The results from micromanipulation of live cells under microscopic examination also suggested that intimate contact between cells is not required for conjugative transfer, but rather that the pilus achieves plasmid transfer between physically separated cells (61). The conclusions from this latter study are also difficult to accept because, in addition to all problems associated with the micromanipulation of live motile cells, the differentiation of donor and recipient cells was ambiguous, and the pilus was "invisible" under the microscopic conditions used by the authors (61). Although attempts were made to prove that the pilus serves as the conduit between mating cells in these studies (31, 61), it is possible to draw different conclusions. In both studies, the physical separation of donor and recipient cells resulted in a dramatic decrease in mating frequency; therefore, close association of cells is a prerequisite for bacterial conjugation. Furthermore, the "brittle and loose attachment" of rigid pili to donor cells makes it unlikely that this type of pilus plays a role in either recipient recognition, retraction, or substrate transfer (36).

FORMATION OF MATING AGGREGATES

In culture, the term "mating pairs" is likely an underrepresentation, because multicomponent "mating aggregates" have frequently been observed (1, 2, 52). Achtman provided the first evidence for aggregation (of 2 to 20 cells) by three methods: light microscopy, examination of particle numbers with a Coulter counter, and sucrose gradient centrifugation (1). In subsequent studies using electron microscopy, mating aggregates of 2 to 50 cells were observed (2). Maher et al. (52) provided the first electron microscopic study of mating aggregates in which there was clear differentiation between donor and recipient strains (Fig. 3). Examination of mating cultures revealed that 50% of cells were involved in aggregates, but, notably, homomeric associations between donor or recipient cells were limited to 20% of the aggregated population, and these donor-only or recipient-only aggregates were composed of only two or three cells (52). In contrast, aggregates containing both donor and recipient cells constituted 80% of the aggregated population, and these cell clusters contained between 2 and 120 cells (52).

Fig. 3Transmission electron microscopy of donor and recipient cells present as mating aggregates. Black arrows indicate recipient cells producing common pili and flagella, whereas donor cells are "bald" and indicated with white arrows. Donor cells contain a derepressed derivate of R27 and would produce few (0 to 4) pili under the conditions used. See reference 52 for additional details.

Source: Reproduced with permission from the American Society for Microbiology.

The role of pili in mating aggregation has been addressed through various treatments. Mating aggregates once formed are resistant to shear forces (60), but treatment of donor cells with Zn²⁺, pilus-specific bacteriophage, sodium dodecyl sulfate (SDS), or CN^– prevents aggregation of donor and recipient cells (2, 59, 60). Each of these treatments likely destabilizes, inactivates, or induces retraction of the conjugative pilus. The donor cells examined by Maher et al. (52) contained a derivative of the conjugative plasmid R27, for which expression and function of the R27 transfer system (including the H pilus) is optimal at 27^oC but inhibited at 37^oC (21). When mating cultures grown at the elevated temperature (37°C) were examined under the electron microscope, cell aggregation was dramatically decreased and no large cell aggregates were observed (52). The absence of pili therefore resulted in the loss of mating aggregate assembly.

The ionic detergent SDS has proven particularly useful in identifying the stage(s) of conjugation at which the flexible pilus acts. The F pilus is sensitive to SDS treatment; therefore, treatment of mating mixtures at time = 0 prevented transfer, whereas DNA transfer proceeded during SDS treatment of cells already present as aggregates, presumably because transfer is independent of the pilus at this stage (2). These results were extended by using donors that were temperature sensitive for DNA transfer but retained the ability to enter into mating aggregates because the harbored traD^ts F plasmid mutant still encoded a functional Mpf system (63). Once mating aggregates had formed, DNA transfer was restored by reactivating the traD product at a lower temperature, and transfer occurred even during treatment of the aggregates with SDS to remove pili (63). These results also suggest that the pilus was not required for DNA transport once cells had aggregated.

Nonmating cultures such as donor- or recipient-only cultures still form small aggregates, but these are not as stable as mating aggregates, as revealed by SDS exposure (2). The close association observed between donor and recipient cells must be actively stabilized; however, the exact identity of the factor(s) involved in mating aggregate stabilization has not been resolved. For F plasmid transfer there is anecdotal evidence for the transfer protein TraN and a derivative of the pilus in mating aggregate stabilization (38, 72). Resistance to SDS by mating aggregates seemingly abrogates the SDS-sensitive pilus as the stabilization factor; however, it is possible that the pili are inaccessible or resistant to SDS when associated in a mating aggregate. Alternatively, the outer membrane-associated protein TraN (55) has been implicated in the interaction of OmpA and lipopolysaccharide of recipient bacteria (38). This protein may therefore function as an adhesin between donor and recipient cells.

Examination of the contact point between mating pairs by transmission electron microscopy of thin section samples has revealed that there is an electron-dense interface between cells (15, 68). These contact zones were termed conjugative junctions (Fig. 4), because there was a clear distinction between contacts made between mating pairs versus random cell juxtapositions. Durrenberger et al. (15), examining mating mixtures with F⁺ donors, demonstrated that formation of the conjugative junctions with recipient cells depended on the presence of the F plasmid but did not depend on the relaxase TraI. Mpf functions therefore seem sufficient for the creation of conjugative junctions. Samuels et al. (68), who examined mating aggregates with donor cells harboring RP4, observed a similar electron-dense layer between cells, which extended for a mean length of 193 ± 41 nm. Samuels et al. (68) also observed that the distance between the outer membranes of donor and recipient bacteria was decreased to 14 nm in conjugative junctions, whereas a distance of 35 nm was observed in nonmating associations. Neither study observed cellular superstructures or fusion of the donor and recipient cytoplasm at these junctions, signifying that the pore between the cells is subtle and difficult to detect.

Fig. 4Copper block frozen/freeze-substituted E. coli during (A) no-plasmid control "mating" and (B) RP4-mediated conjugation on solid substrate (filter matings). Straight, closely appressed outer membranes are typical of junctions between bacteria in RP4-mediated mating with electron-dense material between outer membranes (arrow). The outer membrane (OM), periplasm (P), and cytoplasmic membrane (CM) are indicated. Bar, 25 nm. See reference 68 for additional details.

Source: Reproduced with permission from the American Society for Microbiology.

In the absence of the RP4-encoded pilin subunit TrbC, conjugative junctions of a length similar to those resulting from the full complement of transfer proteins were still observed (68). The pilus is essential for RP4 transfer (30), but these data suggest that the rigid P pilus is dispensable for both the formation of mating aggregates and stabilized junctions. In the F plasmid system the outer membrane protein TraN was implicated as the material existing in conjugative junctions (15), but since RP4 does not encode a TraN homologue, transfer protein(s) other than TraN and pilin must act to "rivet" donor and recipient cells together (68).

The studies reviewed above suggest that transfer occurs as a result of formation of stable mating aggregates; however, the presence of a tubular structure such as the pilus, containing a central channel of sufficient diameter to move a single strand of DNA, is an alluring candidate for DNA translocation. If tubular pili are capable of transporting a single strand of DNA, then why is the close association of a mating aggregate required? Why are no pilus substructures observed at cell-cell contact points under electron or immunogold microscopy? In light of these questions, our preferred model is an apparatus composed of coupling and Mpf proteins which forms the pore for bacterial conjugation.

THE CONJUGATIVE APPARATUS

The localization of the Mpf proteins to the cell envelope implies that they assemble into and function as a transmembrane apparatus for the transport of substrate across this barrier (27). These substrates are presumably pilin subunits used in the assembly of the conjugative pilus and a nucleoprotein complex consisting of the DNA transfer strand and covalently linked relaxase. The exact nature of these substrates is not known, nor is the architecture of the Mpf apparatus and its relationship to the outer and cytoplasmic membranes (of both donor and recipient cells). Assembly of the pilus occurs from the base, rather than the tip (52), whereas assembly of bacterial flagella occurs at the tip (35). Pilin scaffolds may therefore be transported by the Mpf apparatus to the pilus base, and the conjugative apparatus may serve as basal structure that secures the pilus. Whether or not the relaxase is cotransported with the transferred strand (possibly serving as a pilot protein) still remains to be determined (50).

A structural assembly correlating to the large module of Mpf proteins has never been directly visualized at the cell envelope; nevertheless, there are studies that are indicative of transfer protein assemblies. Cellular fractionation of cells producing the entire array of RP4-encoded Mpf proteins revealed that these transfer proteins cofractionate to a membrane fraction of intermediate density between the cytoplasmic and outer membranes (27). Conversely, when the RP4 transfer proteins were produced individually, they resided in the cytoplasmic and/or outer membrane fractions. These results suggested that an assembly of Mpf proteins is capable of connecting the two membranes of the cell envelope (27). In another study, fluorescent labeling of the R27-encoded Mpf protein TrhC, a cytoplasmic membrane-associated ATPase, permitted visualization of a transfer protein complex (21). The TrhC-green fluorescent protein (GFP) fusion protein was observed as foci present at the cell periphery (Fig. 5), and focus formation depended on 12 of the other 18 R27-encoded proteins (21, 22). These results represented an interaction network for the conjugative proteins, wherein the fluorescent foci likely were a large assembly of both Mpf and non-Mpf proteins, rather than homomultimers of TrhC-GFP (22).

Fig. 5The R27 transfer apparatus visualized through a TrhC-GFP fusion produced from R27. Fluorescent microscopy of live E. coli cells harboring R27::trhC-gfp. Bacteria were grown exclusively at 27°C and visualized under UV illumination. See references 21 and 22 for additional details.

Source: Reproduced with permission from the American Society for Microbiology.

The observation of the intermediary RP4-dependent membrane fraction and the R27 fluorescent protein complex both suggest that a large transfer protein interaction network exists. To date only a few interactions among individual conjugal transfer proteins have been made, either through biochemical or two-hybrid techniques. Yeast two-hybrid studies have cumulatively indicated that the F plasmid-encoded proteins TraB, TraV, and TraK form a tripartite interaction network, possibly forming the core of the transfer apparatus, initiating in the cytoplasmic membrane with the integral protein TraB and spanning to the outer membrane with the secretin-like protein TraK (32, 41). Notably, orthologues of TraB have been found to participate in an interaction with the coupling protein, also an integral cytoplasmic membrane-associated protein (20, 51). This interaction provides a mechanism to target the relaxase-associated transferring strand to the envelope-spanning Mpf proteins, as the coupling protein also interacts with the relaxase (70, 73). Furthermore, the coupling protein can interact with noncognate TraB orthologues (51), but relaxase-coupling protein interactions are generally limited to cognate proteins encoded by the same plasmid (8).

The coupling protein is not required for plasmid DNA processing, pilus formation, or pilus function; however, the act of linking the relaxosome to the Mpf proteins offers an implicit use of the transmembrane apparatus for transport of substrate. This role is also suggested by the structure of TrwB, the coupling protein encoded by the Escherichia coli W group conjugative plasmid R388 (25). TrwB exists as a hexamer with a 20-Å central channel, and the overall quaternary structure resembles DNA ring helicases such as FtsK (25). TrwB, as well as the RP4 and F plasmid-encoded coupling proteins (TraG and TraD, respectively), have also each been shown to have nonspecific DNA-binding activity (70); and transfer of DNA from the cytosol through the central channel may be achieved after structural rearrangements to the coupling protein induced by ATP hydrolysis (23, 24).

A model for the assembly of the envelope-spanning conjugative apparatus can be constructed from the interaction studies above (Fig. 6). By using the nomenclature of the F plasmid transfer system, the foundation of the transmembrane apparatus is likely the hexameric coupling protein TraD, which interacts at the inner membrane with TraB. This latter protein contains proline-rich and coiled-coil domains that are presumed to participate in periplasmic extension to TraK and TraB multimerization, respectively (3, 20). The apparatus would culminate in the outer membrane with the TraK, which may be a large cyclic oligomer with a central channel similar to related secretins (11, 41, 43). The TraB molecules linking TraD and TraK might also then be present as a ring-type structure.

Fig. 6Model for the assembly of the core transfer apparatus that traverses the cell envelope. (A) The predicted transmembrane domains of the coupling protein TraD and Mpf protein TraB monomers are represented as cylinders, with N and C termini specified. The rigid conformation illustrated for TrhB is predicted from the high proline content in the primary sequence. (B) The multimeric state of the coupling protein (modeled after TrwB [see reference 25]) and the minimal multimeric state of TraB are shown. TraK, predicted to be the secretin molecule (41, 43), is represented as a multimeric ring structure, modeled after the three-dimensional structure of Neisseria meningitidis dodecameric secretin PilQ (11). (C) A possible assembly of these proteins into a cell envelope-spanning pore-like superstructure is presented. Specific interactions are proposed or known to occur between the relaxase-coupling protein, coupling protein-TraB, TraB-TraK (see text). The outer membrane (OM), periplasm/peptidoglycan (P), and cytoplasmic membrane (CM) are indicated.

Source: Reproduced with permission from the American Society for Microbiology.

Transmembrane structures having stacked ring symmetry have previously been observed for Salmonella enterica type III secretion systems (39) and flagellar basal bodies (74). It is perplexing that cellular modules such as the conjugative transfer systems, which encode numerous (~10) membrane-associated proteins (many of which are known or hypothesized to form homo- and heteromultimers), are not detected as a similar superstructure under electron microscopy. The generalized goal of these apparatuses (type III and IV secretion systems), however, is similar: transport of substrate through the cell envelope to a recipient cell. It may therefore be possible that the generalized organization of these apparatuses (stacked rings forming a transmembrane pore) is also similar.

PLASMID LOCALIZATION DOMAINS

The behavior of plasmids prior to and after conjugative transfer is referred to as the vegetative cycle, which includes both plasmid replication and partitioning (reviewed in references 13 and 33). These processes ensure the vertical transmission of plasmids from parental to daughter cells, whereas conjugative transfer ensures horizontal transmission. Replication is initiated by plasmid-encoded proteins, but DNA synthesis is performed by the host replication machinery. For low-copy-number plasmids, such as the conjugative plasmids described in this chapter, partitioning of these replication products to each daughter cell at cell division is achieved by a module of ATPase and DNA-binding proteins (19). Plasmid replication, partitioning, and conjugative transfer were once viewed as separate stages, but it is now becoming increasingly clear that these plasmid functions represent continuous and overlapping phases of the plasmid life cycle (6).

The understanding of DNA segregation in prokaryotes has been drastically enhanced by cell biology techniques adapted for visualizing the localizations and movements of DNA in living cells. Central to these observations was the development of a method of tagging individual DNA molecules with GFP (LacI-GFP/lacO system), where a lacO cassette, consisting of 256 tandem repeats of the lactose operator, is introduced into the target DNA (26, 76). Production of LacI-GFP results in LacI-GFP binding to the tandem operators and causes the hybrid repressor molecules to cluster as a fluorescent focus. The foci can be visualized by fluorescence microscopy and represent the location of the DNA, either chromosome or plasmid (Fig. 7). The accuracy of the localized patterns obtained with this method has been verified with fluorescence in situ hybridization (FISH), a highly sensitive DNA-labeling technique (67).

Fig. 7Subcellular localization of R751::lacO in E. coli donor cells producing GFP-LacI (from pSG20). GFP-LacI clusters on lacO and results in a discrete GFP signal correlating to the position of the lacO-labeled plasmid when live cells are viewed with fluorescence microscopy. Fluorescence micrographs show cells harboring (A) R751 and pSG20, (B) R751 only, (C–F) R751::lacO and pSG20 with foci located at the mid-cell (one focus), ¼ and ¾ (two foci), mid- and quarter-cell (three foci), and 1/5, 2/5, 3/5, and 4/5 (four foci) cellular positions. Experimental details have previously been described (42).

Source: Reproduced with permission from the American Society for Microbiology.

The localization and movement determined by replication and partitioning functions are sufficient for plasmid maintenance and therefore central to the vertical transmission of plasmid molecules from mother to daughter cells at cell division. With the LacI-GFP/lacO system, the F, P, and H plasmids were observed to reside at well-defined positions located at the mid- and quarter-cell positions of E. coli throughout the vegetative cycle (Fig. 7) (26, 42, 58). An increase in plasmid clusters results from cluster duplication at these sites, followed by the diametrically opposed movement of clusters toward the cell poles, a reflection of the vegetative cycle (Fig. 8) (42, 43). Plasmids likely replicate within the clusters, because the host replisomes are present as fixed assemblies at these same subcellular regions (48), and plasmid cluster movement is directed by the partitioning apparatus (58). In summary, if a single fluorescent cluster is observed, it is present at the mid cell, but after replication the resulting clusters are partitioned to the ¼ and ¾ cell positions. After cell division, these sites will serve as the mid-cell positions in the daughter cells, each containing a sole fluorescent cluster to be replicated and partitioned again. As plasmids are undergoing this vegetative cycle, each plasmid molecule may exist in association with a relaxosome in anticipation of the proposed "mating signal" that would trigger a transfer event, thereby initiating the conjugative cycle.

Fig. 8Time-lapse fluorescence microscopy of R27 plasmid foci (R27::lacO labeled with LacI-GFP) duplicating at either the mid- or quarter-cell positions of live E. coli. The time (in minutes) at which the image was collected is indicated in the top left-hand corner. See reference 44 for additional details.

Source: Reproduced with permission from the American Society for Microbiology.

SITES OF PLASMID TRANSFER

The conjugative cycle begins with conjugal DNA synthesis by the relaxosome and terminates with the successful transfer and recircularization of the transferred strand in the recipient cell. The spatial arrangement of cells during mating aggregation and the subcellular location of the conjugative apparatus may depend on or be relative to the synchronized localization of plasmids and host replication machinery during the vegetative cycle. Alternatively, the conjugative apparatus may not colocalize to these domains, as suggested by the position of transfer protein complexes visualized through an Mpf-GFP fusion. A TrhC-GFP fusion revealed the ubiquitous, random nature of the transfer apparatus throughout the donor cell envelope, without an observable bias for either the polar and lateral positions (Fig. 5) (21, 22). The arrangement of donor and recipient cells in mating aggregates also may correlate to the location of the plasmid or conjugative apparatus.

The application of similar cell biology techniques, specifically the combination of the GFP-LacI probe and the membrane stain FM4-64, allowed the differentiation and orientation of donor, recipients, and notably transconjugants. This approach enabled the first visualization for the sites of transfer between successful, individual mating pairs, as defined as a donor and transconjugant shortly after transfer (42). A survey of successful mating pairs suggested that close physical contact between donor and recipient bacteria is required for DNA transfer (Fig. 9). In addition, regions of intimate contact representing functional conjugative junctions can occur at any location on the donor or recipient cell membrane. This observation demonstrates that the conjugative pore can be located at any region of the donor membrane, which is consistent with the location of the Mpf proteins (21, 27). These observations also imply that the DNA entry point can be at any region of the recipient membrane. Concomitantly, conjugative junctions were observed at cell poles and along the cell lengths (15, 68). Because bacteria may encounter each other in a variety of spatial relationships, the presence of multiple, randomly situated pores increases the likelihood of a conjugative pore encountering the envelope of a recipient cell.

Fig. 9Immobilized live donor and recipients identify sites of plasmid transfer. Combined fluorescence/DIC micrographs of mating pairs: donor cells carrying R751::lacO are nonfluorescent, whereas recipient cells were producing GFP-LacI (uniformly green) and stained in the red membrane dye FM 4-64. Cultures were mixed and immediately immobilized in a nutrient agarose slab. Transconjugants have a red membrane signal but, instead of confluent green fluorescence, contain discrete GFP signals representing transferred R751::lacO molecules. Regions of direct contact between donors and transconjugants represent locations of mating pores through which DNA is transferred. Representative mating pairs of an unsuccessful mating pair (A) and successful mating pairs that are aligned with the lateral wall of donor and transconjugants in direct contact (B), with the pole of the donor in direct contact with the lateral wall of transconjugants (C), and with the lateral wall of the donor in direct contact with the pole of transconjugants (D). See reference 42 for additional details.

Transfer of R751::lacO also provided a unique opportunity to monitor plasmid establishment in recipient cells producing GFP-LacI (Fig. 9). The transferred DNA (observed as a fluorescent cluster) is first visualized at the characteristic mid- and quarter-cell positions of the recipient (correlating to the position of host replisomes), presumably after being converted to a double-stranded molecule (42). Initial duplication of plasmids often results in an asymmetric distribution of fluorescent foci near a single quarter-cell position, which is indicative of partitioning-deficient plasmids (44, 58). Symmetric localization (either at the center or at ¼ and ¾ cell lengths) occurs only after a significant lag, presumably reflecting the time required to synthesize the plasmid-encoded partitioning proteins (42). At this time the plasmid is partitioning proficient and has re-entered the vegetative cycle. The conjugative cycle would therefore be complete.

MODEL FOR CONJUGATIVE TRANSFER

In this chapter, recent observations based on bacterial cell biology techniques, including visualization of plasmid DNA and proteins at the subcellular level, have been combined with electron and light microscopy studies of mating cells to create an integrated overview of gram-negative bacterial conjugation, a concept we refer to as the conjugative cycle. This model accounts for the plasmid DNA movements and associations with replication, partitioning, and conjugative apparatuses. During vegetative growth, plasmids reside at the mid- or quarter-cell regions of donor cells, ensuring their vertical transmission. The partitioning apparatus appears to position the plasmids near the replisome to provide plasmids with continuous access to this machinery. Upon contact of the conjugative pilus with a suitable recipient, the vegetative cycle is interrupted by a proposed mating signal that triggers conjugal DNA synthesis of a transferring strand via the relaxosome. This event would be the initial event of the conjugative cycle (Fig. 10). Contact between flexible pili and recipient cells is subsequently followed by the formation of mating pairs and/or aggregates, and the width of material between the cytoplasmic space of the donor and recipient gram-negative bacteria is approximately ~500 nm (36).

Fig. 10Animation of the conjugative cycle. See text for details. Duration: approximately 51 seconds.

The region of intimate contact between a donor and a recipient cell is the conjugative junction. Since conjugative junctions can extend for ~200 nm (68), it is likely that multiple conjugative apparatuses could fit within one junction. Stabilization of mating aggregates is possibly achieved by a matrix of TraN, observed as the electron-dense material observed at conjugative junctions mediated by the F plasmid (15). Silverman (72) contended that the pilus (or pilus subassemblies) may act in this capacity, an argument strengthened by observations that TraN is not essential for conjugative transfer of the F factor (18), is only required in liquid matings (38), and is not encoded in conjugation systems specifying rigid pili. Furthermore, the hydrophobic nature of both flexible and rigid pili and the existence of pilin in the outer membrane (associated with pilus intermediates or subassemblies) may aid in cell-cell stabilization (36).

In the donor cell cytosol, conjugal DNA synthesis is initiated by the relaxase and proceeds using the host replisome, and a specific interaction between the relaxase and the coupling protein targets the single-stranded DNA (ssDNA) to the membrane. It is unknown, however, whether the interaction between the relaxosome and coupling protein is long-lived, placing the relaxosome at the conjugative pore, or transient, interacting when the relaxosome migrates to the conjugative pore for a transfer event. The model shown in Fig. 10 predicts relaxosome transfer into the recipient, although it is not known if any part of the relaxosome transfers. Because the relaxase interacts with the coupling protein, it could serve as a "pilot protein" to guide the transferring strand to the recipient and subsequently rejoin the free ends after complementary strand synthesis (50). The single-stranded transfer strand (~1 nm in width) could potentially fit through the coupling protein pore (2 nm); however, if the relaxase is to fit through this pore it must be 2 nm or less in diameter. Given the structural similarities between the plasmid R388-encoded coupling protein TrwB and DNA ring helicases, it has been proposed that the coupling protein uses ATP hydrolysis to energize pumping of ssDNA through the transenvelope assembly of Mpf proteins, possibly through a conformation change in the coupling protein hexamer (25, 50). If the relaxase involved in initializing conjugal DNA synthesis is transferred, a second relaxase would be required to cleave the reconstituted oriT of the transferred strand in the donor after replacement strand synthesis and to re-establish the superhelical nature of the template plasmid.

As the ssDNA is entering the recipient, possibly even before termination, it is targeted to the host replisome by an unidentified process to begin complementary strand synthesis. The maximum distance between the mating pore and the host replisome is ~0.5 μm (cell pole to quarter-cell position of a 2-μm cell, therefore 0.5 μm), which is estimated to be approximately 1.5 kb of DNA (based on the formula in reference 40). It is therefore physically possible for DNA entry and complementary strand synthesis to be coupled. After both replacement and complementary strand synthesis are completed, and the reconstituted oriT is cleaved by the relaxase, the conjugative cycle is complete. The newly acquired plasmid molecule can now replicate but is partitioning deficient, as observed by an asymmetric localization of plasmid clusters (42). After production of the partitioning module, the transconjugant plasmids enter a normal vegetative cycle and these cells can themselves serve as donors. Notably, there is disaggregation of cells after transfer (2), and this may be similarly achieved by production of the transfer apparatus in the transconjugant cell.

FUTURE DIRECTIONS

The interaction between bacterial cells is fundamental for the genetic exchange of plasmid DNA molecules. The use of light, electron, and fluorescence microscopy has made significant contributions in determining membrane-associated features at the cell envelope and movements of plasmids during conjugative transfer. These methodologies will also likely continue to provide insights into the cell biology of bacterial conjugation.

In 1993, Willets (77) predicted that the next frontier in the field of bacterial conjugation was the elucidation of the transmembrane apparatus architecture, and we are still searching for it more than a decade later. Large homomultimeric structures are known to be present, but these are not of substantial density to be visualized with electron microscopy. Alternatively, two-hybrid and affinity-capture techniques have successfully revealed binary interactions, which cumulatively allow the architecture of core structures to be proposed (see above). The limitations of these techniques to study membrane-associated proteins, however, necessitate the use of additional technologies.

Crystallographic data have provided the most promising advancement for exposition of the architecture and assembly of the conjugative apparatus (79). Currently, T4S proteins with structural data include the relaxase TrwC (28), the coupling protein TrwB (25), the cytoplasmic membrane-associated ATPases Hp0525 and VirB11 (69, 78), and the pKM101-encoded transfer protein TraC (80). Each of these structures has provided significant insight into molecular aspects of conjugation and other T4S processes, and the structural determination of heteromeric complexes may be on the horizon. Of critical importance would be the structure of the putative conjugative secretin, possibly confirming the bioinformatic observation of H plasmid-encoded TrhK and F plasmid-encoded TraK having primary sequence similarity to known secretin proteins (41, 43).

The architecture of the conjugative apparatus will undoubtedly indicate the structural role of individual transfer proteins during plasmid DNA transport through the cell envelope but may not precisely identify the mechanism of transport. The exact enzymatic role and the precise contacts made between the transfer proteins and the transferring substrate will also require study. Last, the true function of the pilus also still remains ambiguous. This problem is compounded because the different pilus morphologies have different roles in mating aggregation. Both rigid and flexible pili are required for transfer of their cognate plasmids, but rigid pili are not required for mating aggregation, whereas flexible pili are essential for cell-cell interaction. As old questions about conjugative transfer continue to linger, we are slowly providing answers through novel cytological and structural techniques.