Convergence of the Secretory Pathways for Cholera Toxin and the Filamentous Phage, CTXϕ

See allHide authors and affiliations

Science  14 Apr 2000:
Vol. 288, Issue 5464, pp. 333-335
DOI: 10.1126/science.288.5464.333


Virulence of Vibrio cholerae depends on secretion of cholera toxin (CT), which is encoded within the genome of a filamentous phage, CTXφ. Release of CT is mediated by the extracellular protein secretion (eps) type II secretion system. Here, the outer membrane component of this system, EpsD, was shown to be required for secretion of the phage as well. Thus, EpsD plays a role both in pathogenicity and in horizontal transfer of a key virulence gene. Genomic analysis suggests that additional filamentous phages also exploit chromosome-encoded outer membrane channels.

Secretory systems are essential mediators of bacterial pathogenesis, for they enable release of a diverse array of bacterial virulence factors. Export of these extracellular effectors usually depends on one of four distinct multiprotein secretion systems (type I to type IV) to transport macromolecules through the inner and outer bacterial membranes (1). Some components of these systems probably evolved from common ancestral genes, and additional related proteins mediate pilus assembly and contribute to viral and conjugal transfer of DNA (2, 3). However, very few proteins are known to be integral components of more than one secretion apparatus (4,5). Instead, for example, similar but distinct outer membrane pores have been shown to be essential for type II and type III secretion, type IV pilus assembly, and release of filamentous coliphages (2).

The genome of CTXφ, the filamentous phage of Vibrio cholerae that encodes cholera toxin (CT), resembles those of the paradigmatic filamentous phages of Escherichia coli (e.g., f1) in many respects (6). However, the CTXφ genome lacks a homolog of f1's gene IV, which encodes the outer membrane pore (“secretin”), pIV, through which f1 exits from its host (7). Interactions between pIV, pI (an f1-encoded inner membrane protein thought to regulate channel opening), and phage coat proteins mediate the simultaneous assembly and secretion of f1 virions (8). Very specific physical constraints underlie the pIV/pI interaction; thus, the secretin from the closely related coliphage Ike cannot substitute for f1 pIV (9). Because CTXφ resembles f1 in that it does not lyse its host to gain release and because CTXφ encodes proteins related to f1 phage coat proteins and assembly proteins (6), secretion of CTXφ from V. cholerae was expected to require an outer membrane channel similar to pIV. Consequently, we investigated whetherV. cholerae produces a pIV-like protein from a chromosomal locus that serves as a CTXφ secretin.

A comparison of f1 pIV with the database of the V. cholerae genome (10) revealed that V. choleraeencodes a similar (51%) protein: EpsD, the putative outer membrane pore for the eps (extracellular protein secretion) type II secretion system. The eps apparatus is essential for secretion of CT, protease, and chitinase by V. cholerae(11). Because the CT released by infecting bacteria induces most of the symptoms of cholera, it is clear that the epssystem plays a critical role in cholera pathogenesis (12). However, like other type II systems, it is only known to contribute to protein secretion.

To assess the role of epsD in secretion of CTXφ, we generated Kn-insertion mutations within epsD in classical (O395) and El Tor (Bah-2) biotype strains of V. cholerae (13). The wild-type and mutant strains were then transformed with pCTX-Ap, the replicative form of an ampicillin (Ap)–marked CTXφ. Quantitative transduction assays revealed that theepsD::Kn mutant strains produced 300 times fewer CTX-Apφ virions per cell than the corresponding wild-type strains (Table 1). Comparable levels of pCTX-Ap DNA could be detected in the wild-type and epsD::Knmutant strains (Fig. 1), indicating that the reduction in virion titer was not due to impaired replication of phage DNA. Furthermore, parallel transduction experiments with O395epsD::Kn and O395 as recipient strains yielded comparable numbers of ApR colonies (14), demonstrating that the epsD mutation did not substantially impair assembly of TCP, the type IV pilus that serves as the CTXφ receptor. Thus, the epsD::Kn mutants did not suffer from a global secretory defect. Instead, EpsD appears to be specifically required for CTXφ secretion.

Figure 1

Wild-type and epsD::Knmutants contain comparable levels of pCTX-Ap DNA. Southern blots of Eco RI–digested plasmid DNA, which was isolated from parallel pairs of logarithmic-phase cultures, were probed with a CTXφ-specific DNA probe to detect pCTX-Ap. Lane 1, O395/pCTX-Ap; lane 2, O395epsD::Kn/pCTX-Ap; lane 3, Bah-2/pCTX-Ap; and lane 4, Bah-2 epsD::Kn/pCTX-Ap.

Table 1

CTX-Apφ and CT secretion by wild-type andepsD::Kn strains of V. cholerae. Data are representative of assays performed at least three times. NA, not applicable.

View this table:

We also characterized protein secretion by theepsD::Kn mutants. As expected, GM-1 enzyme-linked immunosorbent assays (ELISAs) revealed that O395epsD::Kn secreted less than 10% of the toxin it produced (Table 1), substantially less than the 95% secreted by wild-type O395, but similar to the amount secreted by othereps mutant strains (11). CT secretion by derivatives of the nontoxigenic strain Bah-2 (15) could not be assessed; however, a zymogram of culture supernatants revealed that Bah-2 epsD::Kn/pCTX-Ap secreted less hemagglutinin/protease, the major V. cholerae secreted protease (16), than did its wild-type counterpart (Fig. 2). Thus, EpsD plays its predicted role in protein secretion as well as mediating export of CTXφ.

Figure 2

Bah-2 epsD::Kn secretes a reduced amount of hemagglutinin/protease. Supernatants from early stationary phase cultures were subjected to SDS–polyacrylamide gel electrophoresis on a 10% polyacrylamide gel containing 0.1% gelatin. The gel was washed two times for 45 min in 2.5% Triton X-100 and for 30 min (room temperature) and 2 hours (37°) in tris-buffered saline + 5 mM CaCl2 and then stained with Coomassie Blue. Lane 1, strain 3083; lane 2, HAP-1, a 3083 derivative containing an inactivated form of the gene encoding hemagglutinin/protease (hap::Kn) (16); lane 3, Bah-2/pCTX-Ap; lane 4, Bah-2 epsD::Kn/pCTX-Ap; and lane 5, Bah-2epsD::Kn/pCTX-Ap/pGZ119:EpsD. Numbers at left indicate molecular weight markers.

To confirm that the secretory defects of theepsD::Kn strains were linked to theepsD mutations, we assessed the impact of plasmid-encoded EpsD on secretion by the mutant strains (17). TheepsD plasmid complemented the epsD mutation: It enabled the El Tor and the classical mutant strains to secrete wild-type levels of virions and partially restored secretion of proteins (Table 1 and Fig. 2). In contrast, introduction of the control vector lacking epsD into theepsD::Kn strains did not influence CTX-Apφ or protein secretion.

Although secretion of CTXφ through alternate membrane channels may account for the residual CTXφ produced by ourepsD::Kn mutants, it is more likely that these rare transducing particles were detected because theepsD::Kn mutation does not completely inactivateepsD. The KnR cassette was inserted only 30 amino acids from the COOH-terminus of EpsD (13); thus, a mutant protein might still form minimally functional outer membrane channels. Several earlier attempts to generate epsD alleles with different disruptions were unsuccessful (14), and recent work indicates that EpsD deficiency greatly impairs growth of V. cholerae(18). Consequently, we believe that ourepsD::Kn mutant strains are not entirely devoid of functional EpsD and that their residual EpsD activity enables bacterial survival, some secretion of eps protein substrates (Fig. 2), and a very low level of CTXφ secretion.

The failure of plasmid-encoded EpsD to fully restore secretion of CT by O395 epsD::Kn may reflect an altered level of EpsD relative to the 11 other proteins produced from the eps gene cluster. In previous studies of the additional eps genes, comparable incomplete complementation has been observed (11). However, plasmid-encoded EpsD restored CTXφ titers to wild-type levels, suggesting that eps genes other than epsD are not required for phage secretion. To test this hypothesis, we assayed phage production by a previously described epsE mutant, O395epsE::Kn (11). O395epsE::Kn/pCTX-Ap yielded virion titers exceeding those produced by wild-type O395/pCTX-Ap (6.8 × 105 A 600 −1 ml−1 versus 1.2 × 105 A 600 −1ml−1, where A 600 is absorbance at 600 nm), despite secreting only 9% of synthesized toxin. Similarly, a Bah-2 derivative that produced a dominant negative EpsE (19), which blocks protein secretion, still efficiently secreted virions. This strain, Bah-2/pCTX-Kn/pMS43, yielded slightly more virions (4.6 × 107 A 600 −1 ml−1) than did Bah-2/pCTX-Kn/pMS42 (4.2 × 107 A 600 −1 ml−1), which overexpressed wild-type EpsE (20). Thus, EpsE, and by inference a functional eps secretory structure, is not necessary for efficient CTXφ production. Furthermore, the failure ofepsD::Kn strains to secrete CTXφ virions is not an indirect result of the secretory and membrane changes observed ineps mutant strains.

We propose the following model for CTX phage secretion. CTXφ's homolog of f1 pI, Zot, a presumed inner membrane protein (21), interacts with a multimer of the outer membrane protein EpsD and thereby induces opening of this outer membrane channel, through which the phage is released. Additional interactions between Zot, EpsD, and phage coat proteins are also likely; interactions between CTX phage-encoded proteins and proteins of theeps apparatus other than EpsD are probably not required. It is not known whether a single EpsD multimer can interact simultaneously with components of both secretory pathways or whether the phage and protein secretory processes compete for access to the outer membrane channel.

Phage exploitation of a host secretin has not been demonstrated previously; however, analyses of the genomes of additional filamentous phages suggest that reliance upon a chromosome-encoded secretin may be a common strategy for phage secretion. Within the GenBank database, there are at least five filamentous phages other than CTXφ—fs1, Vf12, Vf33, Cf1c, and Pf1—that do not appear to encode a pIV homolog and thus may rely upon a chromosomal protein instead. All these phages infect bacterial species that contain type II secretion systems. In contrast, the filamentous coliphages that encode a phage-specific secretin infect a host that generally does not produce a secretory apparatus (22). Thus, coliphages may have been constrained during their evolution to rely upon a phage-encoded secretin. Alternatively, phage-encoded secretins may grant access to a broader range of host species or confer some other evolutionary advantage.

It is somewhat surprising that EpsD can mediate both CTXφ and CT secretion. Most secretins are unable to function within heterologous systems, even systems composed of very similar proteins with highly related substrates (9, 23). Furthermore, the two secretory processes to which EpsD contributes are markedly different. Phage export releases a cytoplasmic DNA molecule coated with inner membrane–derived coat proteins, whereas type II secretion systems export only free, periplasmic proteins. Nonetheless, in V. cholerae, these two disparate classes of secretion substrates both appear to pass through an outer membrane pore composed of EpsD. The convergence of phage and protein secretion pathways may be a clue that structurally similar periplasmic complexes are assembled during each process. Indirect evidence in support of this hypothesis has already been provided by the findings that both pathways bear similarities, at either the sequence or structural level, to type IV pilus assembly (2). Our finding that a filamentous phage and a type II secretion apparatus use the same secretin provides additional evidence that the two export systems have a common evolutionary origin and suggests that they may still maintain mechanistic similarities.

  • * To whom correspondence should be addressed. E-mail: mwaldor{at}


View Abstract

Navigate This Article