Since the early 1970s, several groups, including most prominently that of Matsumoto in Japan, have published a series of electron micropraphs depicting projections at the surface of chlamydiae (reviewed in Matsumoto, 1988, in Microbiology of Chlamydia. Barron, A.L. (ed.). Boca Raton, Florida: CRC Press, pp. 21–45). The projections and outer membrane ‘rosettes’ through which they protrude are present both in the replicating form, the reticulate body (RB), and in the late differentiated infectious form, the elementary body (EB). They have been observed in at least three of the four Chlamydia species, including C. pneumoniae (Miyashita et al., 1993, J Med Microbiol38: 418–425), and one report documents their presence in vivo (Soloff et al., 1982, J Comp Pathol92: 547–558).
One of the most intriguing features of the chlamydial projections is their non-uniform distribution at the chlamydial surface; approximately 20–40 hexagonally arrayed projections are typically observed in a well-defined patch covering up to half of the surface (Fig. 1A). In electron micrographs of thin sections, the projections appear to originate from the cytoplasmic membrane, span the periplasmic space and extend out of the outer membrane, forming needle-like fimbrial extensions with estimated diameters of 10–13 nm (Fig. 1B).
During chlamydial growth, RBs are often seen blanketing the inner face of the parasitophorous ‘inclusion’ membrane, whereas differentiated EBs and intermediate forms typically reside in the lumen of the inclusion. In several provocative micrographs, Matsumoto observed RB projections in direct contact with or even piercing through the juxtaposed parasitophorous membrane (Fig. 1C). In these instances, the projection patch appears to correspond roughly to the area of contact between the chlamydial outer envelope and the parasitophorous membrane.
On the basis of these observations, several possible functions have been proposed for the projections. Matsumoto initially suggested that the projections might promote exchanges between the infecting chlamydiae and the cytosol of the infected cell, but he did not speculate on the nature of the exchanges. In the currently most widely supported hypothesis, the projections function as channels that facilitate the uptake of nutrients from the host cytosol by chlamydiae, a idea christened by Stephens as the ‘soup-through-a-straw’ hypothesis (Stephens, 1993, Infect Agents Dis1: 279–293).
We wish to propose an alternative function for the chlamydial surface projections, based on our recent discovery of type III (contact-dependent) secretion genes in Chlamydia (Hsia et al., 1997, Mol Microbiol25: 351–359) — a finding recently extended by the ongoing Chlamydia trachomatis Genome Project (Stephens et al., 1998, http://chlamydia-www.berkeley.edu:4231/). We postulate that the envelope-spanning components of the chlamydial surface projections and associated fimbrial extensions represent the type III secretion machinery and associated virulence effector proteins respectively. A corollary of this hypothesis is that the projections are induced and/or activated upon contact between RBs and the juxtaposed parasitophorous membrane, from early after EB to RB differentiation until RB detachment from the vacuole membrane, late in the life cycle. As such, the projections might play an essential role in the development of the parasitic relationship between intracellular chlamydiae and host eukaryotic cells. This hypothesis is globally consistent with the ultrastructural observations by Matsumoto and with related structural and functional properties of type III secretion in other systems.
First, the fundamental dependence of type III secretion activity on contact with eukaryotic membranes offers a plausible explanation for the presence of the projections in a restricted patch, i.e. the imprint of the area of contact, at the chlamydial surface (Fig. 1A). The mean number of projections in C. psittaci strain meningopneumonitis was shown to peak at about 45 in the early RB (10 h post-infection), decreasing to about 20 in the late RB (20 h post-infection) and levelling to about 18 in the EB (48 h post-infection) (Matsumoto, 1982, J Bacteriol150: 358–364). This decrease in number might result from a gradual reduction in the area of contact between the chlamydial cell surface and the parasitophorous membrane as RBs are progressively being ‘squeezed out’, and implies a parallel gradual loss of type III secretion activity during development.
Second, the observed anchoring of the projections in both the inner and the outer membranes (Fig. 1B) is consistent with a channel designed for the secretion of proteins directly from the cytoplasm to the outside, without a periplasmic intermediate. Moreover, the observed piercing of the parasitophorous membrane by the projections (Fig. 1C) is consistent with a role in the translocation of proteins directly from the chlamydial cytoplasm into the host cell cytosol. Putative chlamydial type III secretion genes such as homologues of yscU, lcrD of Yersinia might indeed encode components of a projection anchor in the cytoplasmic membrane. The chlamydial yscC homologue, like its Yersinia counterpart, might encode outer-membrane ring-like structures (Koster et al., 1997, Mol Microbiol26: 789–797), i.e. the outer-membrane rosettes, through which the projections extend.
Last, similar surface structures, some of which are known to be induced upon contact with eukaryotic cells, have been observed in several bacterial pathogens harbouring type III secretion systems. Low calcium-induced Yersinia, which harbours the closest relative of the chlamydial type III secretion system by sequence, secretes long filaments formed upon polymerisation of Yops (Michiels et al., 1990, Infect Immun58: 2840–2849). In Salmonella, contact-induced surface structures, termed invasomes, are thought to contain type III secreted effector proteins that directly promote entry into eukaryotic cells (Ginocchio et al., 1994, Cell76: 717–724).Shigella ipaB and ipaD mutants or Shigella grown in the presence of Congo red actively secrete effector proteins that assemble into macromolecular structures via the Mxi-Spa type III secretion pathway (Parsot et al., 1995, Mol Microbiol16: 291–300). Perhaps the best example of a type III secretion-associated surface structure occurs in the plant pathogen Pseudomonas syringae, in which related conserved type III secretion genes are required for the production of surface Hrp pili that are required for virulence (Roine et al., 1997, Proc Natl Acad Sci USA94: 3459–3464). Finally, the transmembrane segment of a prototype chlamydial projection offers some structural resemblance to the basal body of the bacterial flagellum, itself a surface appendage, that is dependent on type III secretion for export and assembly.
As chlamydiae are obligate intracellular parasites whose extracellular forms, the EBs, are essentially inert, we have proposed that, in contrast with other bacterial pathogens, type III secretion Chlamydia style, must occur from within, i.e. while the bacteria are metabolically active and intracellular (Hsia et al., 1997, Mol Microbiol25: 351–359). Type III secretion-associated surface appendages, such as invasomes of Salmonella, appear transiently. Shortly after initial contact is made, i.e. once bacteria are committed to a pathogenic outcome, these surface structures resorb such that infecting bacteria are ultimately devoid of them. In contrast, chlamydial projections and/or rosettes are found on RBs that are detached from the parasitophorous membrane and on late-differentiated RBs and EBs. It is plausible that RB detachment from the parasitophorous membrane is a signal for differentiation, as recently proposed (Hackstadt et al., 1997, Trends Microbiol5: 288–293). As RB to EB differentiation involves global oxidation of outer-membrane cysteine-containing proteins, projections and rosettes might become ‘fixed’ by disulphide bonding as soon as RBs are detached from the parasitophorous membrane and remain as permanent vestiges at the EB surface. An intriguing possibility, then, is that type III secretion Chlamydia style may not only serve the intracellular replicating RB, but may also play a role in EB infectivity, i.e. during the early steps of chlamydial pathogenesis.
We are extremely grateful to Professor Akira Matsumoto for his keen eye and for kindly providing the micrographs presented here.