The Chlamydiaceae are gram-negative obligate intracellular bacterial pathogens that replicate entirely within membrane bound inclusions that develop from the phagocytic vacuoles in which they enter. By mechanisms not understood in detail, inclusions avoid fusing with endosomes or lysosomes that might destroy the pathogen [1–3]. Instead, the inclusions provide an intracellular niche in which the pathogen can survive and complete its developmental cycle. The initial pathogen-containing endocytic vacuoles co-associate with microtubules, and dynein, and traffic to a perinuclear region, ultimately residing at the microtubule organizing center and in close proximity to the Golgi [4, 5]. Here, they intercept exocytic vesicles of the biosynthetic pathway that are derived from the Golgi and continue their developmental cycle [6–9].

There are two genera in the family of Chlamydiaceae [10]. The genus Chlamydia is comprised of the C. trachomatis species, which contains human pathogens, as well as the mouse and swine strains, C. muridarum and C. suis respectively, while the Chlamydophila genus is more diverse, consisting of six species. Of these, C. pneumoniae, and C. psittaci are relevant in human diseases. C. trachomatis is primarily a pathogen of humans, containing the trachoma biovariant (biovar), and the lymphogranuloma biovar (LGV). The mouse pneumonitis agent (MoPn), frequently used in vaccine work [11, 12], is now classified separately as C. muridarumn [10]. Some serological variants (serovars) of the trachoma biovar primarily infect the urogenital tract (e.g., E, K, and F) and are the major cause of significant venereal disease in the United States [13]. Other, trachoma serovars primarily infect the conjunctiva (e.g., A, B, and C), causing trachoma, the major form of preventable blindness, worldwide [14]. The LGV biovar also causes chronic venereal disease, which differs clinically from the sexually transmitted disease caused by strains of the trachoma biovar. C. pneumoniae, another human pathogen, is now implicated in atherosclerosis and cardiac disease, as well as respiratory infection [15, 16]. C. psittaci primarily infects birds, but can also cause disease in humans.

The mechanism by which chlamydiae enter host cells is not well defined, largely because of conflicting reports that these organisms can enter by both clathrin-dependent, as well as clathrin-independent pathways. Recently, we examined the entry mechanisms of representative members of each of the above Chlamydia species and biovars. Our concern was whether their entry might be mediated by the cholesterol and sphingolipid enriched micro-domains in host cell plasma membranes known as lipid rafts [17]. Application of biochemical criteria and confocal microscopy demonstrated that five of the ten strains we examined indeed enter by a mechanism that is dependent on intact lipid rafts [18, 19]. These strains included C. trachomatis serovars E, F, and K, C. pneumoniae (strain AR39), and C. caviae (Guinea Pig Inclusion Conjunctivitis strain; GPIC). In contrast , C. trachomatis serovars A, B, C, biovar LGV (L2 strain), and MoPn do not enter via lipid rafts.

Caveolae are now considered to be a specialized kind of lipid raft. These 50–100 nm flask-shaped invaginations form in lipid rafts when the caveolae marker protein, caveolin-1, is expressed [reviewed in [20]]. In contrast, lipid rafts are ubiquitous entities that form regardless of caveolin-1 expression. The inclusion membranes of all of the chlamydial strains that entered via lipid rafts, as indicated by biochemical criteria, also appeared to acquire caveolin-1 at entry, and accumulated more of the protein during development in caveolin-1 expressing HeLa cells [18, 19]. In contrast to the strains that entered by a raft-dependent process, none of the other strains examined acquired caveolin-1 at entry.

The fact that entry of some chlamydial strains is dependent on lipid rafts in general, rather than on the more specialized caveolae, was demonstrated using Fischer Rat Thyroid (FRT) cells, which do not express caveolin-1 or form caveolae. All ten strains that we examined were able to enter the caveolin-1 negative FRT cells, but not if those cells were treated with pharmacological agents that specifically disrupt lipid rafts [19].

The Golgi is a likely source of most caveolin-1 on the inclusions of those strains entering via lipid rafts. This is suggested by the fact that caveolin-1 immunostaining of those inclusions actually increases as they enlarge and mature, and is consistent with reports that caveolin-1 first becomes associated with raft domains in the Golgi, and subsequently is transported to the plasma membrane on vesicles derived from the Golgi [21–23]. Moreover, chlamydial inclusions are known to intercept vesicles of the biosynthetic pathway that originate from the Golgi, from which they acquire sphingolipids and perhaps other nutrients as well[6, 7]. However, since inclusions of all strains appear to acquire sphingolipids from the Golgi, whereas only a subset of strains acquires caveolin-1, there may be more than one sorting process and class of vesicles with which Chlamydia can interact.

The caveolin gene family also contains caveolin-2 [24] which has 38% sequence identity with caveolin-1 and 58% similarity. Caveolin-1 and -2 are co-expressed in most cells, in which they form hetero-oligomeric complexes. These complexes form in the endoplasmic reticulum (ER), becoming associated with lipid rafts in the Golgi, from which they are transported to the plasma membrane. All of the putative functions of caveolae are thus far associated with caveolin-1, and the function of caveolin-2 is essentially unknown. Importantly, unlike caveolin-1, caveolin-2 can not induce caveolae to form in lipid rafts. An experimental finding that is particularly relevant to the current study is that caveolin-2 can not exit the Golgi in caveolin-1 negative FRT cells. However, it can exit when recombinant caveolin-1 is expressed [25, 26]. Thus, formation of hetero-oligomeric complexes of caveolin-1 and -2 facilitates transport of caveolin-2 out of the Golgi and its sorting to the plasma membrane.

In the present study, we ask whether caveolin-2, like caveolin-1, might be associated with inclusions formed by particular chlamydial strains. Unexpectedly, we found that in HeLa cells, caveolin-2 was associated with inclusions of strains that did not enter via lipid rafts and that did not acquire caveolin-1 either at entry or at later times. These strains included C. trachomatis serovars A, B, and C. In the same cells, inclusions of C. trachomatis serovars E, F, and K, C. pneumoniae (A39), and C. caviae (GPIC) were marked by both of the caveolin proteins. Each of these eight strains acquired caveolin-2 even in caveolin-1 negative FRT cells. In addition, the ability of each of these strains to acquire caveolin-2 in the FRT cells was dependent on pathogen gene expression. Thus, chlamydiae appear to activate or induce an additional pathway for the transport of caveolin-2 that is independent of caveolin-1. Inclusions of two strains, LGV (L2) and MoPn, were not marked by either caveolin-1 or -2.

Experimental results presented here demonstrate that for a number of chlamydial strains and species, the intracellular inclusions indeed acquire caveolin-2 during the developmental cycle. These strains include C. trachomatis serovars A, B, C, E, F, and K, C. pneumoniae (A39), and GPIC (Table 1). C. trachomatis serovars E, F, and K, and the C. pneumoniae and C. caviae species of Chlamydophila previously were shown to enter host cells via lipid rafts. Their vacuoles/inclusions were demonstrated to acquire host cell caveolin-1 at entry and to accumulate it during later stages of infection [18, 19].

Consequently, in host cells that express both caveolin proteins, inclusions of these strains display caveolin-2, as well as caveolin-1. However, in this same cell type, inclusions of serovars A, B, and C, are marked only by caveolin-2, a result implying that for chlamydial inclusions this acquisition can occur independently of caveolin-1. This implication is somewhat surprising since there is experimental evidence demonstrating that caveolin-2 usually can not exit from the Golgi and traffic to the plasma membrane in the absence of caveolin-1[25, 26]. However, if recombinant caveolin-1 DNA is expressed by caveolin-1 negative cells, then caveolin-2 can transit from the Golgi and be delivered to the plasma membrane, as a hetero-oligomeric complex with caveolin-1[25, 26, 30, 31].

To confirm that transport of caveolin-2 to chlamydial inclusions indeed might occur independently of caveolin-1, we asked whether chlamydial inclusions might similarly acquire caveolin-2 in FRT cells, a cell line that does not express caveolin-1 [27–29]. In our earlier report [19], we confirmed that FRT cells in fact do not express caveolin-1. Also, we demonstrated that whereas for some chlamydial strains, entry into those cells was dependent on intact lipid rafts (Table 1), in no case was entry dependent on caveolin-1 and caveolae. We show here that in the caveolin-1 negative FRT cells, inclusions of each of the above chlamydial strains in fact acquired caveolin-2. Moreover, since FRT cells do not contain caveolin-2 at the plasma membrane, pathogen-associated caveolin-2 could not have been acquired at entry. Rather, it had to derive from an intracellular source.

Chlamydial inclusions thus acquired caveolin-2 despite the fact that in the absence of caveolin-1 this protein usually does not traffic from the Golgi. We therefore suggested that the pathogen might influence the host cell and actually induce a transport and sorting pathway that normally may not actively function in the uninfected cell. Consistent with this suggestion is the fact that for all strains and species tested in FRT cells, if the pathogens were inactivated by prior EB treatment with UV irradiation, none of the EB containing vesicles acquired caveolin-2. In contrast, in HeLa cells, there were low levels of caveolin-2 associated with vesicles containing UV-treated EBs. The source of these low levels of caveolin-2 on vesicles/inclusions in HeLa cells is not yet clear. Perhaps the EBs are acquiring minimal amounts of caveolin-2 from the membranes of these cells upon entry. This suggestion is supported by the fact that the vesicles of these UV irradiated EBs contain both caveolin-1 and -2 (data not shown); suggesting that the proteins might be co-localizing in hetero-oligomeric complexes.

It is well known that Chlamydia-containing vacuoles and later, the developing inclusions, are able to intercept and fuse with exocytic vesicles of the biosynthetic pathway that originate from the Golgi [6, 7]. These Golgi-derived vesicles provide the inclusions with sphingolipids and perhaps other key metabolites [7, 32]. In addition, the ability of at least some chlamydial strains to intercept these Golgi-derived vesicles is dependent on pathogen viability and therefore likely on gene expression. Considering these points, and the fact that caveolin-1/2 hetero-oligomers traffic to the plasma membrane from the Golgi [25, 26, 30, 31], it is reasonable to suggest that the previously described pathogen induced interception of Golgi-derived vesicles might account for the acquisition of caveolin-2 by inclusions, as seen in the current study. However, although studies by others indicate that all chlamydial strains appear to intercept vesicles of the biosynthetic pathway, not all strains acquire caveolin-2 and this fact provides a strong counter argument. Thus for example, in the current study LGV (L2) inclusions did not acquire caveolin-2 (Table 1), although they have been reported to intercept Golgi-derived vesicles of the biosynthetic pathway [7]. Furthermore, in the current study, acquisition of caveolin-2 by inclusions of some strains was independent of caveolin-1, while among other strains, inclusions acquired both caveolin proteins. Thus there may be more than one pathway by which inclusions acquire caveolin-2. Importantly, acquisition of caveolin-2 by inclusions of any particular strain did not correlate with entry by a raft-mediated pathway. In contrast, acquisition of caveolin-1 at early as well as late stages in pathogen development did correlate with a route of entry involving lipid raft microdomains. Together, these facts are consistent with the conclusion that acquisition of caveoloin-1 and -2 can represent independent and distinct processes. As expected, similar incubations with anti-caveolin-3 demonstrated this component was not present in any of these cells whether infected with Chlamydiae or not.

To date, no chlamydial protein has been identified that might be secreted into the host cell cytosol to influence host intracellular trafficking. However, several chlamydial proteins, termed Incs, have been identified in the inclusion membrane [33–35]. Several of these Inc proteins have cytoplasmic domains, making them potential mediators for interactions with the host that might influence trafficking [36]. One of these proteins, IncG, interacts with host protein, 14-3-3β [37]. This particular interaction can not underlie the trafficking of caveolin-2 to the inclusions since GPIC (C. caviae) and C. pneumoniae (AR39) do not express IncG [37], although, as shown here, they do acquire caveolin-2. Furthermore, the reverse also is true. Thus, as demonstrated above, mouse pneumonitis strain, MoPn and LGV (L2) do not acquire caveolin-2, but they do express IncG [37]. There are, however, other Inc proteins and they may underlie the phenomena we have presented here.

Despite their presence on chlamydial inclusions, it remains unclear what role, if any, caveolin proteins may play in the developmental cycle of Chlamydia. As noted above, all ten of the chlamydial strains and species we examined were able to enter caveolin-1 negative FRT cells [19]. Since these caveolin-1 negative cells do not express caveolin-2 at the plasma membrane [25, 26], it would appear that for any of these strains neither of the caveolin proteins is necessary for entry per se. Although following internalization all ten strains and species remained viable in the FRT cells [19], the only strains able to generate large inclusions in the FRT cells were GPIC which acquires both caveolin-1 and -2 in caveolin-positive cells, and MoPn which acquires neither caveolin protein. Despite these latter two instances, our findings remain consistent with the possibility that caveolin-1 may yet play a post entry role in the development of at least some strains. The fact that successful MoPn development does occur in inclusions that acquire neither caveolin-1 nor -2 [19] (and current study) could mean that neither caveolin-1 nor -2 has a necessary role in the development of MoPn. The ability of GPIC to develop mature inclusions in FRT cells where it acquires only caveolin-2, despite acquiring both caveolin proteins in caveolin-1 positive cells, may imply that for GPIC caveolin-2 rather than caveolin-1 is a key for development. Likewise, inclusions of C. trachomatis serovars A, B, and C, whether developing in HeLa or FRT cells, display only caveolin-2, suggesting a potentially broader chlamydial requirement for the caveolin-2 protein during development.

The series of merged optical sections demonstrating the association of caveolin-2 with the inclusions (Figure 2) implies that caveolin-2 is in apposition to pathogen cells that are located at the inclusion membrane. Nevertheless, the caveolin-2 is not in direct contact with the pathogens. Caveolin is not a transmembrane protein, a fact that can be deduced from the experimental finding that cell-surface biotinylation does not label caveolin proteins [38]. Thus, these membrane proteins are not accessible from the extracellular milieu, and this originally extracellular membrane surface, is topologically equivalent to the lumen of the chlamydial inclusion. Hence, whereas our findings imply a specific association between the pathogens and caveolin-2, this association appears to be indirect. The paucity of information concerning caveolin-2 does not enable us to suggest an identity for the linking molecule. Moreover, how this indirect association between caveolin-2 and the pathogen might relate to the protein's function in the chlamydial developmental cycle, or in its acquisition by the pathogen, is also not yet clear.

Among the Chlamydiae, depending on the serovar, or the species, one or both of the caveolin proteins may play an important role in the developmental cycles. Further studies of the interaction of the Chlamydiae with these enigmatic host cell proteins may well help to clarify caveolin functions in the host cell. Likewise, such studies may indicate factors influencing caveolin expression and trafficking from the Golgi, and elucidate their significance to the developmental cycle of these pathogens. The function of caveolin-2, either in the uninfected cell or in the chlamydial developmental cycle, remains to be elucidated. Nevertheless, this second caveolin protein can now be added to the small number of host proteins that are reported as associated with the inclusions of this obligate intracellular pathogen.