Secretion of a soluble colonization factor by the TCP type 4 pilus biogenesis pathway in Vibrio cholerae


E-mail; Tel. (+1) 603 650 1632; Fax (+1) 603 650 1318.


Colonization of the human small intestine by Vibrio cholerae requires the type 4 toxin co-regulated pilus (TCP). Genes encoding the structure and biogenesis functions of TCP are organized within an operon located on the Vibrio Pathogenicity Island (VPI). In an effort to elucidate the functions of proteins involved in TCP biogenesis, in frame deletions of all of the genes within the tcp operon coding for putative pilus biogenesis proteins have been constructed and the resulting mutants characterized with respect to the assembly and function of TCP. As a result of this analysis, we have identified the product of one of these genes, tcpF, as a novel secreted colonization factor. Chromosomal deletion of tcpF yields a mutant that retains in vitro phenotypes associated with the assembly of functional TCP yet is severely attenuated for colonization of the infant mouse intestine. Furthermore, we have determined that the mechanism by which TcpF is translocated across the bacterial outer membrane requires the TCP biogenesis machinery and is independent of the type II extracellular protein secretion (EPS) system. These results suggest a dual role for the TCP biogenesis apparatus in V. cholerae pathogenesis and a novel mechanism of intestinal colonization mediated by a soluble factor.


Vibrio cholerae is the aetiologic agent responsible for the acute gastrointestinal disorder cholera. Cholera manifests clinically as severe diarrhoea that rapidly leads to hypovolemia and shock due to the profuse loss of electrolytes and fluids (reviewed in (Kaper et al., 1995). This clinical syndrome is in large part a result of the ADP-ribosylating activity of the potent AB5 toxin, cholera toxin (CT). Cholera toxin enters the endocytic pathway of epithelial cells after binding the ganglioside GM1 and, through a cascade of intermediates, constitutively alters the permeability of intestinal epithelial cells in a manner that results in luminal accumulation of electrolytes and fluids. To effectively deliver CT to intestinal epithelial cells, it is necessary for V. cholerae to colonize and proliferate within the small bowel. Oral challenge studies conducted in both humans and mice have demonstrated that the V. cholerae type 4 toxin co-regulated pilus (TCP) is required for efficient colonization (Taylor et al., 1987; Herrington et al., 1988). Type 4 toxin co-regulated pilus expression is co-ordinately regulated with cholera toxin production as part of the ToxR virulon by all strains capable of causing epidemic cholera, including both classical and El Tor biotypes of the O1, as well as the recently emerged O139 serogroups (Taylor et al., 1987; Rhine and Taylor, 1994; Tacket et al., 1998). For O1 classical strains, it has been shown that TCP serves to mediate bacterial interactions through pilus–pilus contacts that lead to the formation of microcolonies in vitro and in vivo. Microcolony formation strongly correlates with intestinal colonization in the infant mouse cholera model suggesting a primary role for TCP in promoting or enhancing bacterial interactions (Kirn et al., 2000).

Structurally, TCP appears as large bundles of pilus fibres. These fibres are thought to be homopolymers composed exclusively of TcpA pilin subunits. Eleven genes that code for proteins putatively involved in TCP biogenesis or function have been identified by TnphoA and DNA sequence analysis (Taylor et al., 1987; Peek and Taylor, 1992; Kaufman et al., 1993; Ogierman et al., 1993). Most of these genes are arranged as an operon and located on the vibrio pathogenicity island (VPI) (Peterson and Mekalanos, 1988; Taylor et al., 1988; Ogierman et al., 1993; Everiss et al., 1994; Karaolis et al., 1998). Although some of these gene products share homology with proteins required for the elaboration of type 4 pili in other Gram-negative bacteria, their function is unclear as the mechanism by which type 4 pilus biogenesis occurs is not well understood.

The extracellular protein secretion (EPS) system is a type II secretion pathway responsible for secretion of CT by V. cholerae (reviewed in Sandkvist, 2001). Type II secretion, also referred to as the main terminal branch of the general secretory pathway, is a two step mechanism for extracellular localization of proteins by Gram-negative bacteria (Pugsley et al., 1997). Proteins transiting both membranes of Gram-negative bacteria utilizing this secretion system are first secreted to the periplasm via the Sec pathway and subsequently translocated across the outer membrane by a conserved machinery composed of at least 13 different proteins, which together appear to functionally span the cytoplasmic membrane, periplasm and outer membrane (Sandkvist et al., 1997; 1999; 2000; Russel, 1998). Key components of the type II system and representative examples from the EPS system include: (i) several proteins sharing homology with type 4 prepilins and therefore referred to as pseudopilins (Eps G, H, I, J); (ii) an ATP binding protein peripherally associated with the inner membrane components (EpsE); (iii) a type 4 prepilin peptidase capable of cleaving the N-terminal signal sequence of type 4 prepilins (VcpD for EPS pathway); and (iv) an outer membrane secretin which multimerizes and acts as a gated channel for translocated proteins (EpsD). Similar proteins encoded by genes located in the tcp operon include: (i) tcpA, which encodes the type 4 prepilin and tcpB which encodes a protein with an N-terminal type 4 leader sequence but otherwise is structurally unique among the type 4 prepilins; (ii) tcpT, which codes for an ATP binding protein that putatively associates with the inner membrane; (iii) tcpJ, which encodes a type 4 prepilin peptidase required for N-terminal processing of TcpA and TcpB; and (iv) tcpC, which codes for an outer membrane lipoprotein that likely functions as the secretin for the TCP biogenesis apparatus (Parsot et al., 1991; Iredell and Manning, 1997; Sandkvist et al., 1999; LaPointe and Taylor, 2000). This extensive functional and structural homology, generally present among type 4 pilus biogenesis and type II secretory systems, suggests a common evolutionary origin and perhaps a common functional mechanism.

Based on these similarities and evolutionary relationships, it has been hypothesized that a pseudopilus structure, composed of the prepilin-like proteins associated with type II secretion systems, may assemble in the bacterial membrane(s) and serve as a base from which protein secretion occurs (Py et al., 2001). Likewise, type 4-pilus biogenesis may represent an extension of this platform, into a large, extracellular structure. It might then be predicted that the pilus structure would be able to translocate proteins secreted to the periplasm via a Sec-dependent system. This has recently been demonstrated by overexpression of the genes required for pullulanase secretion in E. coli (Sauvonnet et al., 2000). A pilus-like structure, composed at least partially of PulG (pilin-like protein) monomers, is present on the surface of these bacteria when grown on solid media. Pullulanase secretion is dependent on the assembly of this structure and the two processes are non-competitive. Additionally, it has been suggested that the tip located adhesin (PilC) associated with the Neisseria gonorrhoeae type 4 pilus, which is synthesized with a Sec pathway-dependent signal sequence, transits the outer membrane in a manner similar to type II secreted soluble factors but instead is incorporated into the pilus structure (Rudel et al., 1995; Lory, 1998). Further evidence of a common evolutionary precursor is garnered by observing that proteins involved in the function or structure of type II secretion and type IV pilus biogenesis systems overlap in some species. For example, it has been shown that the type 4 pilin subunit PilA that comprises the adhesive type 4 pilus of P. aeruginosa is also required for efficient secretion of exoenzymes (Lu et al., 1997). This group demonstrated that the molecular basis for this overlap is the dual role for PilA in the general secretory apparatus (where it is part of the secretion machinery) and the Type 4 pilus (where it is the structural subunit). Results from our current study suggest further parallels between these systems by demonstrating that the TCP apparatus is responsible for the terminal secretion step of the soluble TcpF colonization factor.


Construction and analysis of in frame tcp deletions

All of the genes known to code for the structure and biogenesis functions of TCP, with the exception of dsbA, are located within an operon on the VPI Fig. 1). Based mainly on the analysis of tcp operon genes harbouring Tn::phoA insertional mutations, it has been suggested that all of the gene products derived from the tcpA through tcpJ region are essential for TCP expression, function or biogenesis (Taylor et al., 1987; 1988; Kaufman et al., 1993; Brown and Taylor, 1995). Whereas tcpA through tcpF and tcpJ are believed to encode the structural and biogenesis functions of TCP, toxT encodes a transcriptional activator that positively regulates the tcp operon promoter (Hulbert and Taylor, 2002). Because of the potential for the phenotypes of insertional mutations to be misinterpreted because of polar effects, we have assembled a panel of mutants harbouring non-polar, in frame deletions of tcpA through tcpF (see Experimental procedures). Verification of the non-polar nature of each of these mutations was confirmed by trans complementation of each deletion with the corresponding gene (data not shown). The construction of these mutants allowed us to unambiguously assess the contribution of each tcp gene product to TCP biogenesis and function.

Figure 1.

Organization of the tcp operon. The predicted molecular weight of mature proteins encoded by each gene is indicated and asterisks designate genes coding for proteins with signal sequences. The orfH gene codes for a hypothetical conserved protein homologous to proline rich antigen (Heidelberg et al., 2000).

To characterize each of these mutants, they were independently assessed for the presence of pili by in vitro autoagglutination, transduction by the CTX-Knφ and transmission electron microscopy. Autoagglutination and CTX-Knφ transduction are reliable indicators of the presence of functional surface associated TCP and therefore a reasonable assay for the biogenesis of a functional pilus. Direct examination by electron microscopy allows the detection of non-functional pili. As shown in Fig. 2, each of the tcp mutants, with the exception of the tcpF mutant, failed to autoagglutinate and was not transduced by the CTX-Knφ. Additionally, each mutant with the exception of ΔtcpF failed to elaborate a TCP structure visible by TEM (data not shown) although they all expressed levels of TcpA similar to wild type as determined by Western blot analysis. In the case of the tcpF mutant, the level of CTX-Knφ transduction was approximately 70% of wild type, autoagglutination was nearly normal and direct observation of TCP elaborated by this strain revealed bundled, laterally associated pilus fibres morphologically similar to the wild-type pilus (Fig. 3). These results unambiguously confirm a role for the tcpA through tcpE genes in pilus biogenesis but, contrary to previous suggestions, indicate that TcpF does not appear to play an important role in TCP biogenesis (Taylor et al., 1988; Manning, 1997).

Figure 2.

TCP biogenesis and TcpA stability in tcp deletion mutants. Autoagglutination and CTX-Knφ transduction phenotypes are reported for each tcp deletion mutant. The ΔtcpF mutant exhibited an autoagglutination phenotype slightly reduced from that observed for the wild type and a transduction frequency 70% of wild-type levels (see Fig. 8 for a detailed analysis of these phenotypes exhibited by the tcpF mutant). All of the mutants (except for ΔtcpA) produced near normal levels of TcpA as judged by immunoblot analysis of proteins present in whole cell extracts prepared from each strain. The αTcpA6 serum was used as the primary antibody (Experimental procedures).

Figure 3.

ΔtcpF elaborates a TCP structure morphologically similar to wild-type TCP. Transmission electron micrographs of TCP elaborated by wild-type (A) and tcpF mutant strains (B). Images are at 55 000 × original magnification and arrows indicate TCP.

Analysis of crude protein fractions by protein gel electrophoresis and subsequent Coomassie blue staining revealed that a 36 kDa protein accumulated in the periplasmic fraction from each of the tcp mutants except ΔtcpF and the wild-type strain. Furthermore, a protein with a similar electrophoretic mobility was observed in Coomassie stained gels loaded with culture supernatant from wild type, but not from culture supernatants of any of the tcp mutants. A representative example of this phenotype is shown for ΔtcpA in Fig. 4A (all other mutants exhibited an identical phenotype, see below). Because the ΔtcpF mutant was unique in that the 36 kDa protein was absent from both the culture supernatant and periplasmic extracts we hypothesized that the protein accumulating in the periplasm of tcp mutants and secreted by the wild-type strain was TcpF. Indeed, N-terminal sequencing of the suspect bands corresponding to the 36 kDa protein from both the wild-type culture supernatants and periplasmic fractions from ΔtcpA were consistent with the TcpF protein sequence (Fig. 4B and C). The N-terminal sequence obtained from the periplasmic form of TcpF was consistent with a + 20 processing site, suggesting that TcpF is secreted to the periplasmic space in a manner that involves cleavage of the Sec-dependent N-terminal signal sequence. This is consistent with the two step secretion mechanism attributed to type II secretion systems.

Figure 4.

TcpF is secreted by wild-type V. cholerae and accumulates in the periplasm of ΔtcpA mutants.
A. Proteins in whole cell (WC), spheroplast (S), periplasm (P) and culture supernatant (CS) fractions from wild-type, ΔtcpA and ΔtcpF strains were analysed by SDS-PAGE followed by staining with Coomassie blue. A prominent band at 36 kDa was observed in the periplasmic fraction of the ΔtcpA strain but was absent from the periplasmic fractions of the wild-type and ΔtcpF strains. Thirty-six and 33 kDa bands were observed in the wild type and ΔctxA CS lanes but not in the ΔtcpA and ΔtcpF CS lanes. A band corresponding to CT-A was visible in the CS fractions from all strains except the ΔctxA mutant.
B. Protein sequence of the N-terminus of TcpF indicating the predicted signal sequence (underlined) and the amino acids determined by Edmund degradation sequencing (bold) of the 36 kDa band accumulating in the periplasmic fraction of ΔtcpA.
C. Protein sequence of TcpF indicating the amino acids determined by tryptic digest and subsequent Edmund degradation sequencing of the 36 kDa protein secreted by the wild type and isolated in the CS. Underlined amino acids indicate predicted trypsin cleavage sites in TcpF (DNASTAR PROTEAN).

It was noted that the secreted TcpF appeared to be represented by bands with two electrophoretic mobilities on the Coomassie stained gels and by immunoblot analysis (see below). When the tcpF mutant was complemented in trans by a plasmid bearing a TcpF-6His fusion (pTK43), both bands were detected in the culture supernatant utilizing a monoclonal antibody generated against tetra-histidine (Qiagen, data not shown). This suggests that the two bands representing TcpF differ either by further N-terminal cleavage, degradation, or by another modification.

All tcp biogenesis mutants fail to secrete TcpF or colonize the infant mouse intestine

Each of the above described in frame deletions was systematically examined for the ability to secrete TcpF by immunoblot. As shown in Fig. 5, all of the mutants were defective for secretion of TcpF into the culture supernatant. For ΔtcpA through ΔtcpE this severe defect for TcpF secretion was consistently linked to the inability of these strains to elaborate TCP and colonize the infant mouse intestine. The colonization defect of these mutants was therefore not unexpected as they were all deficient for TCP production. In contrast, the in vivo phenotype exhibited by the tcpF mutant was quite surprising. This mutant was capable of making functional TCP as judged by in vitro studies (see above) yet exhibited a colonization defect similar to that of the tcpA mutant. Because ΔtcpF is specifically defective in TcpF secretion, and not TCP biogenesis, these results suggest a significant TCP independent role for TcpF in colonization. To further assess the role of TcpF in virulence, LD50 experiments were conducted. The LD50 of the tcpF mutant was found to be 1000-fold higher than that of the wild-type parental strain. This is similar to results obtained for tcpA mutants (data not shown and Taylor et al., 1987). Taken together, these data indicate that deleting tcpF results in a major attenuation of virulence via a colonization defect and therefore demonstrate a significant role for TcpF in colonization.

Figure 5.

A ΔtcpF strain is defective for colonization and all other tcp mutants are defective for TcpF secretion and colonization. Competitive indices and TcpF expression and secretion were determined for wild type and tcp deletion mutants. Immunoblotting of proteins in periplasmic (P), spheroplast (S) and culture supernatant (CS) fractions from wild type and tcp mutants using anti-TcpF antibodies indicates that TcpF is stably produced in all tcp mutants (except ΔtcpF) but is efficiently secreted into the culture supernatant only by the wild-type strain. The ΔtcpR and ΔtcpS mutants appeared to be completely defective for secretion of TcpF. In vivo competition experiments indicated that each of the tcp mutants was severely attenuated for colonization of the infant mouse intestine. Surprisingly the ΔtcpF mutant was attenuated nearly as much as the ΔtcpA strain. In vitro competitive indices were all close to 1 (data not shown).

TcpF secretion occurs in a Sec-dependent manner

Based on the above data, we hypothesized that TcpF secretion occurs via a mechanism much like Type II secretion, utilizing the TCP biogenesis apparatus to transit the outer membrane. All current models of Type II secretion indicate that the protein to be translocated across the outer membrane is first secreted to the periplasm via a Sec-dependent mechanism and then in a second step translocated across the outer membrane. This type of mechanism is supported for TcpF in that deletion of components of the secretion apparatus results in accumulation of a periplasmic intermediate (Fig. 5). The primary components of the Sec translocon are the heterotrimeric SecYEG membrane complex which acts as a channel and the SecA protein, an ATPase that drives protein movement into and across the membrane. Additionally, a leader peptidase cleaves the signal sequence from the N-terminus of many proteins that utilize this pathway before their entry into the periplasm. To determine if the mechanism by which TcpF gains entry to the periplasm involves the Sec machinery, temperature-sensitive alleles of secY and secA were used in a continuous labelling assay. Because Sec-dependent secretion is characterized by cleavage of the N-terminal signal sequence of substrate proteins, this was exploited as a marker for utilization of this pathway and the relative amount of unprocessed, labelled TcpF was determined at several time points after shifting the temperature sensitive mutants to the non-permissive temperature. As seen in Fig. 6, only mature TcpF was present in the bacterial lysates at the permissive temperature (30°C). When cultures were shifted to 42°C, however, the unprocessed form of TcpF began to accumulate at 60 min for the secAts mutant and at 240 min for the secYts mutant. This result clearly demonstrates the requirement for both SecA and SecY in the secretion of TcpF to the periplasmic space.

Figure 6.

Secretion of TcpF to the periplasm occurs via a Sec-mediated pathway. 35S methionine-labelled TcpF was present in a mature form only in secA and secY temperature sensitive mutants at the permissive temperature. At 42°C, however, unprocessed TcpF began to accumulate in the secAts strain at 60 min and in the secY ts strain at 240 min demonstrating the requirement for these proteins in TcpF secretion to the periplasm.

TcpF secretion is not dependent on EpsD

It has been shown that EpsD, the secretin associated with the EPS type II secretion system of V. cholerae, is absolutely essential for the translocation of cholera toxin across the outer membrane (Sandkvist et al., 1997). We postulated that TcpF was secreted via the TCP biogenesis apparatus independently of the EPS system. To test this hypothesis, the extracellular localization of TcpF was evaluated in an epsD mutant background (Davis et al., 2000). In the epsD mutant, TcpF was present in the culture supernatant at levels consistent with wild type whereas cholera toxin levels were significantly decreased (Fig. 7). Conversely, in the tcpA mutant TcpF levels in the culture supernatant were diminished whereas cholera toxin levels were unaffected. This experiment indicates that TcpF secretion is independent of EpsD and supports the hypothesis that the TCP biogenesis apparatus secretes TcpF via a mechanism independent of the EPS system.

Figure 7.

TcpF secretion is independent of EpsD. TcpF secretion was assessed by separating proteins from culture supernatants (CS) produced by wild type and tcpA, tcpF and epsD mutants by SDS-PAGE and subsequent Coomassie staining. Whole cell (WC) protein preparations were loaded as controls for total protein content. Bands corresponding to TcpF and CT-A from the culture supernatant preparations are indicated. The wild type secretes both CT-A and TcpF. The tcpA and tcpF mutants secrete CT-A but not TcpF whereas the epsD mutant secretes TcpF but not CT-A.

Complementation of ΔtcpF with TcpF-His

Because TcpF is located directly upstream of the toxT promoter and the deletion of TcpF leads to slightly reduced levels of TCP as judged by CTX-Knφ transduction, we wanted to ensure that this phenotype was indeed a result of the absence of TcpF and not a result of any polarity on toxT. Plasmid pTK43 expressing a TcpF-His fusion was activated with increasing concentrations of arabinose and the level of functional, surface associated TCP was assessed by the degree of autoagglutination. This was determined by letting overnight cultures stand on the bench for 30 min in cuvettes and then measuring the OD550 of the resulting partially cleared cultures. Additionally, a second assessment of TCP levels was made by measuring transduction frequency using the CTX-Knφ. As expected, the wild-type strain yielded a very low OD550 because most of the bacteria had settled to the bottom of the cuvette in large aggregates typical of autoagglutination assays with this strain (Kirn et al., 2000). A high transduction frequency reflecting the presence of surface associated TCP was also noted for this strain. The negative control, ΔtcpA, exhibited a high OD550 and null transduction frequency, both because of the absence of TCP. The OD550 of the ΔtcpF culture was slightly higher than that of the wild type but considerably lower than the ΔtcpA measurement, reflecting the slight autoagglutination defect described above. Additionally, the transduction frequency, as reported above, was approximately 70% of that measured for the wild type.

As increasing levels of arabinose were added to the ΔtcpF strain bearing pTK43, the OD550 decreased to a level near that of wild type (Fig. 8). Interestingly, however, the response of the strain to arabinose was biphasic; at concentrations of arabinose above 0.001% the ability of the strain to autoagglutinate began to decrease. CTX-Knφ transduction frequency displayed a similar biphasic relationship with TcpF-His expression; transduction frequency increased to near wild-type levels at 0.001% arabinose and decreased to levels close to those measured for ΔtcpA at higher concentrations. Each of these experiments was conducted in parallel with wild type that had been grown in concentrations of arabinose paralleling those of the mutants and transduction and OD550 was calculated as a percentage relative to wild type. It is therefore unlikely that the arabinose alone was responsible for this phenomenon. These results are consistent with a model where TcpA and TcpF secretion are competitive processes although other explanations are equally plausible.

Figure 8.

Trans-complementation of the ΔtcpF mutation by a tcpF-his construct. Overexpression of TcpF-His resulted in a biphasic phenotype; initially autoagglutination and phage transduction levels were complemented, increasing to near wild-type levels. However, at high TcpF levels both of these indicators of TCP production significantly declined.
A. OD550 of culture supernatants from ΔtcpA (pTK43) and ΔtcpF (pTK43) compared to wild type with increasing levels of arabinose.
B. CTX-Knφ transduction as a percentage of wild type for ΔtcpF (pTK43) with increasing concentrations of arabinose.


Type 4 pili are surface exposed organelles essential for the pathogenesis of a wide variety of bacterial species. The mechanisms by which they function are quite varied; some mediate adhesion directly, others facilitate motility through a twitching mechanism and yet others initiate or maintain bacterial interactions. One or more of these functions may play a role in the pathogenic lifestyle of a single organism. Based on structural and functional homology, it is likely that the mechanism by which these structures are assembled is conserved. Unlike the well-characterized mechanism of assembly determined for P pili, type 4 pilus biogenesis is not well understood. In our efforts to understand type 4 pilus biogenesis with respect to V. cholerae TCP we have generated a series of in frame deletion mutations in genes within the tcp operon previously implicated as having a role in TCP biogenesis based on the phenotypes of TnphoA insertion mutations (Taylor et al., 1988). The results of the in frame deletion analysis confirmed the previous findings that the majority of the genes encode products necessary for TCP biogenesis. However, an unexpected result of this analysis was the finding that TcpF, previously assigned as having a function in pilus biogenesis, is not required for this process based on the phenotype of an in frame tcpF deletion mutation.

Further analysis of wild-type and tcpF mutant strains revealed that tcpF encodes a protein that is localized to the culture supernatant during growth in vitro. Furthermore, all of the deletion mutations in genes required for pilus formation also prevented secretion of TcpF, implicating the importance of the TCP biogenesis machinery for the secretion of TcpF and suggesting that, like TcpA, TcpF is a substrate for this apparatus. Interestingly, TcpF was determined to be a very abundant protein in the culture supernatant, present in quantities similar to CT (see Fig. 7).

TCP was still elaborated by the tcpF deletion mutant and it was tested in the infant mouse cholera model to determine if there was any defect in colonization. Surprisingly, although TcpF is a soluble secreted protein, it apparently plays a major role in colonization because the colonization index of the mutant is nearly five logs lower than wild type. This defect is comparable to a tcpA mutant. Possibilities for the mechanism associated with this function include a direct role for TcpF in mediating bacteria–host interactions, TcpF-mediated modification of host cells, or co-operation between TcpF and TCP for inter–bacterial interactions. It should be noted that in the competitive index experiments conducted in infant mice where both the wild-type and ΔtcpF strains are inoculated together the ΔtcpF mutant is still cleared from the intestine. This demonstrates that soluble TcpF secreted from one bacterium (the wild-type strain) cannot compensate for the inability of a mutant to produce its own TcpF. This suggests that localized secretion of this protein is critical for its function. Studies currently underway designed to address the precise function of TcpF in colonization suggest that TcpF may be a reasonable target for vaccine or therapeutic design (T. J. Kirn and R. K. Taylor, in preparation).

In addition to the role for TcpF in colonization, an equally interesting aspect of TcpF is the mechanism by which it is secreted. Several investigators have pointed out the common features of type 4 pilus biogenesis and type II protein secretion (Sandkvist, 2001). This work, however, represents the first demonstration of a bifunctional apparatus known to assemble type 4 pili yet also able to secrete a soluble protein via an apparent type II-like secretion mechanism associated with the pilus biogenesis apparatus. This may be likened to the recent observation that pili have been observed to be associated with the pullulanase type II secretion system when it is overexpressed (Sauvonnet et al., 2000). Interestingly, overexpression of TcpF leads to decreased TCP production that is likely the result of decreased secretion of TcpA. This suggests that at least some parts of the secretion pathway utilized by these two very different secreted factors are shared. These observations further support the notion that similar mechanisms operate to form type 4 pili and to secrete proteins by the type II pathway, suggesting that findings in either system may be applicable to the other. Importantly, we found the secretion of TcpF is independent of the EPS type II secretion pathway of V. cholerae demonstrating that each of these parallel systems can function independently from the other. Experiments we are currently conducting to answer with respect to TcpF secretion include identifying proteins that TcpF interacts with during transit and defining determinants that mediate entry to the secretion pathway.

Experimental procedures

Bacterial stains and plasmids

Bacterial strains and plasmids used in this study are listed in Table 1. With one exception, bacteria were either grown in Luria–Bertani (LB) liquid broth at 37°C (standard conditions) or in LB broth adjusted to a starting pH of 6.5 at 30°C for 16 h (TCP inducing conditions). When necessary, ampicillin was used at 100 µg ml−1 and kanamycin was used at 45 µg ml−1. For induction of pTK43 and pTK55, bacteria were grown with appropriate antibiotics overnight and then diluted 1 : 100 in fresh LB plus antibiotics and various concentrations of arabinose and incubated for an additional 5 h at 37°C. Amplifying TcpF::6His using primers FHIS5 and FHIS3 and cloning into the expression vector pBAD22 generated plasmid pTK43. Plasmids pTK50 and pTK55 were generated in a similar fashion using appropriate primer sets and cloning into pBAD22 and pGEX5X-1 respectively.

Table 1. . Bacterial strains and plasmids used in this study.
Strain or plasmidRelevant characteristicsSource or reference
E. coli
TOP10FmcrAΔ(mrr-hsdRMS-mcrBC) φ80lacZΔM15ΔlacX74 recA1 deoR araD139 D(ara-leu)7697 galU galK rpsL endA1 nupGStratagene
S17–1λpirthi pro recA hsdR[RP4–2Tc::Mu-Km::Tn7]λpir TpRSmRde Lorenzo and Timmis (1994)
BL21F- ompT hsdS (rB, mB) galStudier and Moffatt (1986)
CJ105HJM114, secAts, leu82::Tn10Wolfe et al. (1985)
CJ107HJM114, secYtsWolfe et al. (1985)
V. cholerae
O395SmClassical, Ogawa SmR derivativeTaylor et al. (1987)
RT4031O395Sm ΔtcpA11Kirn et al. (2000)
RT4368O395Sm ΔtcpB1This study
RT4369O395Sm ΔtcpC1This study
RT4370O395Sm ΔtcpD1This study
RT4371O395Sm ΔtcpE1This study
RT4372O395Sm ΔtcpF1This study
RT4373O395Sm ΔtcpQ1This study
RT4374O395Sm ΔtcpR1This study
RT4375O395Sm ΔtcpS1This study
RT4376O395Sm ΔtcpT1This study
CG842O395Sm ΔlacZGardel and Mekalanos (1996)
EL30O395Sm epsD::KanDavis et al. (2000)
O395N1O395Sm ΔctxAMekalanos et al. (1983)
CL101O395, Ctx-KnφLaboratory collection
pKAS32pGP704, rpsL ApRSkorupski and Taylor (1996)
pTK2pKAS32, ΔtcpB ApRThis study
pNB6pKAS32, ΔtcpC ApRThis study
pCW2pKAS32, ΔtcpD ApRThis study
pYLT2pKAS32, ΔtcpE ApRThis study
pTK7pKAS32, ΔtcpF ApRThis study
pYY8pKAS32, ΔtcpQ ApRThis study
pNB2pKAS32, ΔtcpR ApRThis study
pTK25pKAS32, ΔtcpS ApRThis study
pBAD22pMB1, Para promoter, araC, ApRLaboratory collection
pTK41pKAS32, ΔtcpT ApRThis study
pTK43pBAD22, TcpF-6His ApRThis study
pTK55pBAD22, TcpB-6His ApRThis study
pGEX5X-1pBR322, lacI q, Ptac promoter, GST ApRAmersham pharmacia biotech
PTK50pGEX5X-1, GST-TcpF(Δ1-20) ApRThis study

Cell fractions

For isolation of periplasmic contents, cells grown under TCP inducing conditions were collected by centrifugation, resuspended in PBS plus polymyxin B sulphate (10 mg ml−1) and incubated on ice for 10 min (Hirst and Holmgren, 1987; Peterson and Mekalanos, 1988; Peek and Taylor, 1992). Spheroplasts (S) were pelleted and resuspended in 1 × SDS-PAGE loading buffer and the supernatant, containing periplasmic (P) contents was diluted with 2 × SDS-PAGE loading buffer. Culture supernatants were prepared by centrifugation of overnight cultures followed by passage through a 0.2 µm filter. The resulting culture supernatant (CS) was then diluted with 6 × SDS-PAGE loading buffer.

Purification of GST-TcpF(Δ1-20) and generation of antibodies

BL21 bearing pTK50 was grown overnight under standard conditions and then diluted 1:100 into fresh media containing ampicillin and 0.1 mM IPTG. After incubation at 30° for 6 h, bacteria were harvested by centrifugation and resuspended in PBS. A French press was used for cell lysis and after pelleting cell debris the lysate was passed over a GSTrap column (Amersham). The fusion protein was then eluted with reduced glutathione and collected in multiple fractions by hand. Antibodies were generated in rabbits using proteins contained in the fractions eluted from the column (Pocono Rabbit Farm).

Construction of tcp deletion mutants

Primer pairs were used to amplify ∼ 800 bp chromosomal regions flanking the intended site of deletion. Each primer pair was designed such that the resulting PCR products would contain unique restriction sites distal to the intended deletion and a common restriction site joining the two products. These fragments were then cloned, stepwise, into the allelic exchange vector pKAS32. The pKAS32 derivatives were propagated in S17–1λpir and the mutations were introduced to the O395Sm chromosome via allelic exchange as previously described (Skorupski and Taylor, 1996). Appropriate regions of the chromosome of each mutant were then sequenced on both strands to confirm that the desired mutation was introduced and that no unintentional mutations occurred.

Autoagglutination assay

Vibrio cholerae strains were grown under TCP inducing conditions, removed from the shaker and allowed to stand on the bench for 30 min. The cultures were then inspected without disturbing any cell pellet that may have formed. For qualitative assessments of autoagglutination (Fig. 2), a +, – or ± was assigned to each culture based on the degree of supernatant clearing observed where + represented a significant degree of clearing. For quantitative assessments (Fig. 8), the cultures were transferred to cuvettes upon removal from the incubator, allowed to stand for 30 min and the OD550 of the cleared supernatant was measured.

SDS-PAGE and Western blotting

Proteins from whole cell lysates, culture supernatants and the two fractions obtained from periplasmic fractionation were separated on 12.5% polyacrylamide gels containing SDS and visualized with Coomassie brilliant blue (Bio-Rad). For immunodetection, proteins were transferred to nitrocellulose at 4°C in transfer buffer (25 mM Tris; 192 mM glycine; 20% methanol; pH 8.3) using a wet transfer apparatus (Bio-Rad (Towbin et al., 1979). TcpA was detected using polyclonal anti-TcpA antiserum generated in rabbits with a synthetic peptide antigen corresponding to the C-terminus of TcpA (TcpA6 (Sun et al., 1997). TcpF was detected using a polyclonal antiserum prepared as described above.

Peptide sequencing

Proteins were separated on a 12.5% polyacrylamide gel as described above, electrophoretically transferred to PVDF (Bio-Rad) and stained with 0.1% Coomassie brilliant blue in 40% methanol for 1 min. After destaining with 50% methanol, appropriate bands were excised from the membrane, washed extensively in water and retained for sequencing or chromatography. For samples digested with trypsin, liquid chromatography was performed on a Vydac protein-peptide C18 column in a Hewlett-Packard 1090 Series HPLC using a diode array detector and fractions were collected manually. N-terminal sequencing was done by Edmund degradation chemistry using an Applied Biosystems 476 A protein sequencer.

CTX-Knφ transduction assay

CTX-Knφ-containing supernatants were generated by growing strain CL101 under standard conditions and filtering the culture supernatant through a 0.2 µm syringe filter. Transductions were conducted as previously described (Waldor and Mekalanos, 1996). Briefly, each strain to be tested was grown under TCP inducing conditions and mixed with an equal volume of CTX-Knφ containing supernatant. After a 30 min incubation at RT, appropriate dilutions were made and the bacteria were plated on solid medium containing kanamycin. Transduction frequency was calculated as the number of Kanr colonies recovered divided by the number of input colony forming units. All transductions were performed on the same day with a single phage stock.

LD50 and competitive index analysis

The LD50 of each strain was assessed by orally inoculating groups of four, 5–7-day-old mice with dilutions of bacteria grown under TCP inducing conditions. The concentrations of inoculums tested were 10-fold dilutions from 5 × 108 to 5 × 105 bacteria per dose. The LD50 is based on the extrapolated dose that would result in a mean survival rate of 50% after 48 h.

For the in vivo competitive index determinations, strains to be tested were grown under TCP inducing conditions and then mixed with equal numbers of the isogenic ΔlacZ O395 derivative CG842. Five- to seven-day-old CD−1 mice from mixed litters were orally inoculated with 50 µl of a 1 × 10−2 dilution of the mixture and incubated at 30°C for 24 h. The bacteria were then recovered by homogenizing harvested intestines with an 18-Ga syringe needle in 5 ml of PBS. The homogenate was appropriately diluted and plated on solid medium containing streptomycin and X-GAL. The competitive index was calculated by comparing the ratio of test strain bacteria recovered from the intestine to test strain input to the input/output ratio of CG842. The values reported are the average of at least five mice.

In vitro competitive index values were obtained by inoculating 2 ml of LB liquid media with equal numbers of the test strain and CG842, growing overnight at 37°C and plating dilutions on solid media containing streptomycin and X-Gal. The change in ratio of test strain to CG842 was then determined.

Electron microscopy

Strains to be visualized by EM were grown under TCP inducing conditions, pelleted in a microfuge and resuspended in PBS. Formvar coated copper EM grids were then floated on top of drops of the bacterial suspensions at room temperature for 3 min. The grid was then transferred to a drop of 0.5% phosphotungstic acid pH 6.5 for 3 additional min. The grid was dried on filter paper and visualized in a JEOL 100CX EM with a potential difference of 100 kV.

[35S]-methionine labelling and immunoprecipitation

Strains CJ105 and CJ107 were transformed with pTK43 and grown overnight at 30°C in minimal media supplemented with ampicillin and 0.1% casamino acids (Wolfe et al., 1985). Cultures were then divided into two, diluted 1:100 into fresh minimal medium and incubated at 30°C and 42°C. After 0, 30, 60, 120 and 240 min a 1 ml aliquot was removed and labelled for two minutes with 10 µCi of [35S]-methionine (> 1000 Ci mmol−1, Amersham). Cells were washed once with PBS and then resuspended in 1 ml PBS. After sonication on ice (3 × 10 s pulses with 30 s rest), lysates were cleared by centrifuging at 10 000 g for 30 min at 4°C. The supernatants were recovered and immunoprecipitation and SDS-polyacrylamide gel electrophoresis were carried out as previously described (Wolfe et al., 1983).


We are grateful to Matt Waldor for the generous gift of strain EL30. We thank Louisa Howard for electron microscopy and the following individuals for contributing to the construction of the tcp deletion mutants: Colby Wyatt, Yingzhen Yang, Robin Hulbert and Yiting Liu. Electron microscopy was performed at the Rippel Electron Microscopy facility at Dartmouth College. This work was supported by NIH grant AI25096 to R.T. T.K. was supported by NIH training grant GM08704 and the Rosalind Borison Memorial Fellowship.