Chlamydia trachomatis Polymorphic Membrane Protein D Is a Virulence Factor Involved in Early Host-Cell Interactions

Abstract

Chlamydia trachomatis is an obligate intracellular mucosotropic pathogen of significant medical importance. It is the etiological agent of blinding trachoma and bacterial sexually transmitted diseases, infections that afflict hundreds of millions of people globally. The C. trachomatis polymorphic membrane protein D (PmpD) is a highly conserved autotransporter and the target of broadly cross-reactive neutralizing antibodies; however, its role in host-pathogen interactions is unknown. Here we employed a targeted reverse genetics approach to generate a pmpD null mutant that was used to define the role of PmpD in the pathogenesis of chlamydial infection. We show that pmpD is not an essential chlamydial gene and the pmpD null mutant has no detectable deficiency in cultured murine cells or in a murine mucosal infection model. Notably, however, the pmpD null mutant was significantly attenuated for macaque eyes and cultured human cells. A reduction in pmpD null infection of human endocervical cells was associated with a deficiency in chlamydial attachment to cells. Collectively, our results show that PmpD is a chlamydial virulence factor that functions in early host-cell interactions. This study is the first of its kind using reverse genetics to evaluate the contribution of a C. trachomatis gene to disease pathogenesis.

INTRODUCTION

Chlamydia trachomatis is an obligate intracellular mucosotropic bacterium that is the most common cause of preventable blindness (1, 2) and bacterial sexually transmitted infections worldwide (3–5). The study of these medically important pathogens has been severely limited in the past by the lack of genetic tools. However, newly developed genetic approaches enable us to ask definitive questions about the contribution of specific chlamydial genes to pathogenesis (6–8). Polymorphic membrane protein D (PmpD) is one of nine putative autotransporters (ATs) encoded in the C. trachomatis genome (9). ATs are members of the Gram-negative bacterial type V secretion system and are important virulence factors that function in host-cell interactions and immune evasion (10). PmpD exhibits classical AT processing resulting in a membrane translocator domain that facilitates the presentation of a passenger domain on the chlamydial surface (11). PmpD is highly conserved among all C. trachomatis strains and is the target of broadly neutralizing antibodies (12). Despite its conserved nature, surface localization, and immunological importance, little is known about the function of PmpD in the pathogenesis of C. trachomatis infection.

Here, we made a C. trachomatis pmpD null mutant using a targeted reverse genetic approach (6). pmpD was not essential for C. trachomatis growth, and the pmpD null mutant showed no deficiency in cultured murine cell lines in vitro or in a mouse urogenital infection model. However, the pmpD null mutant was attenuated in cultured human endocervical and conjunctival cells and in a nonhuman primate model of C. trachomatis ocular infection. Our findings show that C. trachomatis PmpD is a virulence factor that functions in early host-cell interactions.

MATERIALS AND METHODS

Chemical mutagenesis, library construction, and mutation screen.

The generation of the low-mutagenized C. trachomatis serovar D library has been previously described (6). Briefly, McCoy cells in 96-well tissue culture plates were infected with 10 inclusion-forming units (IFU) per well. Chlamydiae were harvested at 48 h postinfection and were used to reinfect McCoy cells in 96-well tissue culture plates. Infected cells were harvested and passaged for a third time, and DNA extracted from the third passage was used to PCR amplify pmpD. Amplicons were heat denatured, slowly reannealed, and digested by CEL I endonuclease. Digestion products were visualized by DNA agarose gel electrophoresis. Mutations were sequenced, the two mutants harboring nonsense mutations in pmpD were plaque cloned three times, and their genomes were sequenced (6).

Chlamydia trachomatis strains and cell culture.

Plaque-cloned pmpD mutants and the parental serovar D strain derived from D/UW-3/Cx (6) were propagated in McCoy cells and purified as previously described (13). Immortalized human endocervical epithelial (A2EN), human conjunctival epithelial (HCjE), and primary murine oviduct epithelial (Bm12.4) cells were all cultured as previously described (14–16).

Antibodies and Western blot analysis.

A rabbit peptide antibody to PmpD (anti-nPmpD antibody) (11) and mouse monoclonal antibodies to B-complex major outer membrane protein (MOMP) (L2–I5) and HSP60 (A57–B9) were used in this study. Western blot analysis of PmpD was performed as previously described (11).

Phase microscopy.

McCoy cells seeded onto 12-mm glass coverslips were infected at a multiplicity of infection (MOI) of 0.1 with the wild-type (WT) strain or the pmpD null mutant. At 36 h postinfection, coverslips were mounted in phosphate-buffered saline (PBS), inclusions were evaluated at ×40 magnification using a Nikon Eclipse TS100 microscope, and images were acquired with a Nikon DS-Fi1 camera.

Immunofluorescence microscopy.

McCoy cells seeded onto 12-mm coverslips were infected at an MOI of 0.1 with the WT or the pmpD null mutant. At 36 h postinfection, cells were washed with PBS and fixed in methanol. Coverslips were blocked with 2% normal goat serum in PBS, and inclusions were immunostained with the rabbit anti-PmpD peptide antibody, followed by Alexa Fluor 488 goat anti-rabbit IgG (catalog no. A11034; Life Technologies). Inclusions were also immunostained with the anti-MOMP mouse monoclonal antibody, followed by Alexa Fluor 568 goat anti-mouse IgG (catalog no. A11004; Life Technologies). Nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole; catalog no. D3571; Life Technologies). Coverslips were mounted in Mowiol mounting medium (catalog no. 81381; Sigma-Aldrich) and evaluated on a Carl Zeiss LSM 710 confocal laser scanning microscope.

Transmission electron microscopy.

McCoy cells seeded onto 12-mm Aclar coverslips were infected at an MOI of 0.5 with the WT strain or the pmpD null mutant. At 24 h postinfection, coverslips were washed in PBS and fixed with 2.5% glutaraldehyde in cacodylate buffer (100 mM sodium cacodylate, 50 mM KCl, 2.5 mM CaCl₂) for 30 min at room temperature, followed by overnight incubation at 4°C. Samples were processed and imaged as previously described (17).

Infection and growth analysis in vitro.

McCoy cells (murine fibroblasts) and Bm12.4 (primary murine oviduct epithelial), A2EN (human endocervical epithelial), and HCjE (human conjunctival epithelial) cells were each seeded into 24-well plates at a density of 1.5 × 10⁵ cells per well. At 24 h after seeding, cells were inoculated with the pmpD null mutant or the WT strain in sucrose-phosphate-glutamic acid (SPG) buffer. All cells were inoculated with 1.5 × 10⁴ IFU per well, and plates were centrifuged at 545 × g for 1 h at room temperature. Following infection, the inoculum was replaced with supplemented medium containing cycloheximide. Cells were either methanol fixed at 36 h postinfection for counting of inclusions or harvested in SPG buffer for enumeration of recoverable IFU in McCoy cells.

Murine model of urogenital tract infection.

Eight-week-old female C3H/HeJ mice (The Jackson Laboratory) were treated with 2.5 mg medroxyprogesterone acetate at 10 and 3 days prior to urogenital infection with C. trachomatis. Groups of 10 mice were inoculated intravaginally with doses ranging from 1 × 10² to 1 × 10⁷ IFU per mouse in 5 μl SPG buffer. The vaginal vault was swabbed just prior to inoculation. Chlamydial shedding was monitored weekly by swabbing the vaginal vault and culturing the recovered chlamydiae on McCoy cell monolayers in 48-well plates as previously described (18). The 50% infective dose (ID₅₀) of each strain was calculated using the Reed and Muench method (19, 20). Mice were swabbed until the clearance of infection, indicated by three successive culture-negative weeks.

Murine histopathology.

Female C3H/HeJ mice were infected with 100 times the ID₅₀ or were mock infected with SPG buffer as described above. Five mice from each infection and mock-infected group were euthanized at 1, 2, 4, 5, and 6 weeks postinfection. Genital tracts were excised, fixed in 10% formalin, and embedded into paraffin blocks. Upper genital tract tissue sections, including ovaries, oviducts, and uterine horns, were stained with hematoxylin-eosin (H&E) to visualize inflammatory cell infiltration. Histological scoring of chronic inflammatory cellular infiltrates, including lymphocytes, plasma cells, and macrophages, was performed blinded by a veterinary pathologist using the previously described four-tiered semiquantitative criteria (18, 21). The range of cellular infiltration scoring of 0 to 4 was defined as follows: 0 is none, 1 is minimal, 2 is mild, 3 is moderate, and 4 is severe.

Macaque model of ocular infection.

Adult male and female cynomolgus macaques (Macaca fascicularis) were maintained at the Rocky Mountain Laboratories (RML) and cared for under standard practices implemented by the Rocky Mountain Veterinary Branch. Monkeys were housed separately when being used for experimental studies. All handling procedures were approved by the RML Animal Care and Use Committee, and the research was conducted in full compliance with the guidelines in the Guide for the Care and Use of Laboratory Animals (22). The facilities are fully accredited by the American Association for Accreditation of Laboratory Animal Care. Ocular infections were conducted by direct inoculation of chlamydiae onto the conjunctival surfaces (20 μl per eye) of both eyes (13). Six macaques were infected with 2 × 10⁴ IFU/eye of the pmpD null mutant, and six macaques received 2 × 10⁴ IFU/eye of the WT strain. Blinded clinical evaluation of ocular disease was performed weekly by RML veterinary staff. Disease was scored on the basis of hyperemia and follicle formation on the upper conjunctival surface of both eyes (13, 23). Hyperemia was scored in the following manner: 0, no hyperemia; 1, mild hyperemia; and 2, severe hyperemia. Subepithelial conjunctival follicles were scored as follows: 0, no follicles; 1, 1 to 3 follicles; 2, 4 to 10 follicles; 3, >10 follicles; and 4, follicles too numerous to count. A composite disease score for each animal was calculated by adding the left eye and right eye pathology scores, with the maximum score being 12 (13). Ocular infection was monitored by swabbing the conjunctiva and culturing recoverable IFU on HeLa cell monolayers as previously described (13). Macaques were monitored for shedding and disease for 100 days postinfection.

Host-cell attachment assay in human endocervical cells.

Human endocervical epithelial (A2EN) cells were infected with 10⁷ IFU of the pmpD null mutant or the WT strain in SPG buffer in a six-well plate for 1 h at 4°C by centrifugation at 545 × g. After infection, each inoculum was removed and the titer in McCoy cells was redetermined to quantitate unattached infectious organisms. The experiments were repeated using the same inoculum to serially infect A2EN cell monolayers 2 to 3 times consecutively before redetermining the titer in McCoy cells.

RESULTS

pmpD is a nonessential chlamydial gene.

The recently developed targeted reverse genetic approach for C. trachomatis (6) was used to generate pmpD mutants. We used the C. trachomatis serovar D library generated by low-level ethyl methanesulfonate mutagenesis from our previous work, which is known to consist of isogenic or nearly isogenic mutants (6). Screening the library revealed 28 pmpD mutants, 22 of which had nonsynonymous mutations and 6 of which had synonymous mutations. Two of 28 pmpD mutations were premature stop codons. Mutants with each of these mutations were plaque cloned, and their genomes were sequenced (Table 1). The mutant with a C → T transition mutation at nucleotide position 1618 in pmpD (pmpD^C1618T) was selected for further analysis, as it was nearly isogenic with only a single background mutation (Fig. 1A), while the other mutant (the pmpD^C3280T mutant) was used to confirm all in vitro observations (data not shown). The pmpD^C1618T and the wild-type (WT) serovar D parental clones were evaluated by Western blotting for pmpD expression (Fig. 1B). PmpD-specific antibodies failed to detect any form of PmpD in the pmpD^C1618T clone, including the predicted truncated form produced by the premature stop codon, indicating that pmpD is not a gene essential for in vitro infectivity and growth in McCoy cells. Therefore, the pmpD^C1618T mutant was designated the pmpD null mutant.

TABLE 1.

De novo genome sequencing of pmpD mutants

Mutant	SNP locationa	Gene	Nucleotide change	Amino acid change
pmpDC3280T	40,983	IGb	C → T
	272,558	hyp	C → T	Val → Ile
	512,571	ompB	C → T	Ser → Asn
	953,816	pmpD	C → T	Gln → stop
pmpDC1618T	952,154	pmpD	C → T	Gln → stop
	960,639	tyrP	C → T	Ser → Phe

^aSNP location refers to the genomic location in the C. trachomatis D-LC annotated reference sequence (GenBank accession number NC_017436.1).

^bIG, intergenic region.

FIG 1.

pmpD^C1618T is a null mutant. (A) Summary of the principal forms of WT PmpD and sequencing information showing the location of the pmpD^C1618T stop codon. (B) Western blot analysis of whole EB proteins with a PmpD-specific antibody showed strong reactivity with the WT and no reactivity with pmpD^C1618T. Anti-HSP60 antibody was used as a loading control.

The pmpD^C1618T mutant is a null mutant with atypical morphological and ultrastructural phenotypes.

Microscopy was performed to determine if the absence of the PmpD protein resulted in morphological changes. Phase microscopy of live McCoy cells infected with the pmpD null mutant revealed large atypical forms throughout developing inclusions (Fig. 2A). Confocal microscopy also revealed an atypical organism distribution, including areas that appeared to lack any chlamydial organisms (Fig. 2B). The WT clone exhibited typical inclusions and a typical organism distribution. Transmission electron microscopy was performed to evaluate ultrastructural differences between the pmpD null mutant and WT clones (Fig. 2C). Interestingly, pmpD null mutant reticulate bodies (RBs) showed a reduced association with the inclusion membrane. At 24 h postinfection, only 6.3% of the pmpD null mutant RBs were associated with the inclusion membrane, whereas 37.6% of WT RBs with intact PmpD were associated with the inclusion membrane. The associations at other time points were also significantly different (see Table S1 in the supplemental material). Transmission electron microscopy performed on infected HeLa cells showed a similarly reduced RB-inclusion membrane association. This observation suggests that PmpD may influence the RB-inclusion membrane interactions and the absence of PmpD destabilizes that close physical association.

FIG 2.

The pmpD null mutant exhibits unique morphological and ultrastructural phenotypes in vitro. (A) Phase microscopy of McCoy cells at 36 h postinfection revealed an atypical organism distribution associated with the pmpD null mutant but not the WT strain. (B) Confocal immunofluorescence microscopy at 36 h postinfection showed no immunostaining of the pmpD null mutant with a PmpD-specific antibody. Immunostaining of the MOMP and DAPI-stained DNA confirmed the atypical organism distribution in pmpD null mutant inclusions. (C) Transmission electron microscopy at 24 h postinfection showed a reduced association of pmpD null mutant RBs with the inclusion membrane compared to that seen in the WT strain (see Table S1 in the supplemental material).

In vitro and in vivo murine models reveal no detectable role for C. trachomatis PmpD.

The ability of the pmpD null mutant to infect and propagate in two murine cell lines was evaluated in vitro. McCoy cells (mouse fibroblasts) and Bm12.4 cells (primary murine oviduct epithelial cells) (14) were infected with the pmpD null mutant or WT strain (Fig. 3A). Immunostaining of inclusions at 36 h postinfection revealed no (McCoy cells) or very little (Bm12.4 cells, 1.22-fold) difference in the ability of the strains to infect the murine cell lines. Infectious progeny were also enumerated at 36 h postinfection, but no difference in the number of recoverable IFU was detected between the strains (Fig. 3B). The ID₅₀, infectious burden, and duration of shedding were evaluated for the pmpD null mutant and the WT strain in the C3H/HeJ murine model of urogenital tract infection (18). Surprisingly, the ID₅₀s for the pmpD null mutant and the WT strain were almost identical, 10^3.025 and 10^3.1742, respectively (see Table S2 in the supplemental material). The lowest infectious dose that resulted in 100% infection was 10⁵ IFU for both strains. The groups infected with 10⁵ IFU were evaluated weekly for infectious burden and duration of shedding (Fig. 3C). There was no difference in the chlamydial burden or duration of shedding between the strains. Histopathological evaluation was conducted at five time points postinfection at which an upper genital tract pathology for C. trachomatis is known to be shown (18). No statistically significant differences in the clinical disease caused by the pmpD null mutant and the WT were detected in upper genital tract tissues (Fig. 3D and E).

FIG 3.

No detectable role for C. trachomatis PmpD in murine infection models. (A) Enumeration of inclusions in murine fibroblasts (McCoy cells) and primary murine oviduct epithelial (Bm12.4) cells revealed no (McCoy cells) or very little difference (Bm12.4 cells, 1.22-fold) in the ability of the pmpD null mutant to infect cells in vitro compared to that of the WT. (B) Recoverable IFU harvested from McCoy and Bm12.4 cells at 36 h postinfection also showed no differences in the ability of the pmpD null mutant to grow in these cell lines compared to that of the WT. (C) The recoverable numbers of IFU and the duration of chlamydial shedding following urogenital infection of C3H/HeJ mice were similar for the pmpD null mutant (n = 10) and the WT strain (n = 10). No statistically significant difference was found at any time point (two-tailed t test). (D) H&E staining of C3H/HeJ upper genital tract tissues (ovaries, oviducts, and uterine horns) infected with the pmpD null mutant or the WT strain showed no histopathological differences up to 6 weeks postinfection. Average clinical disease scores with standard deviations are shown (n = 5). (E) H&E staining of oviducts (OD) is shown for pmpD null mutant-, WT-, and mock-infected animals.

The pmpD null mutant is significantly attenuated in nonhuman primates.

Despite the lack of detectable infection and growth differences in murine model experiments, the conserved and immunogenic nature of C. trachomatis PmpD argues that PmpD has an important role in human chlamydial infections. Since C. trachomatis PmpD has only a 72% sequence similarity to its ortholog in Chlamydia muridarum, the natural chlamydial pathogen of mice, we decided to test the C. trachomatis pmpD null mutant in a nonhuman primate model, a model that mimics human infections much more closely than any murine system. The cynomolgus macaque model of ocular infection was used to evaluate infection and clinical disease in vivo. Six animals were infected with the pmpD null mutant, six animals were infected with the WT, and the animals were evaluated at weekly intervals postinfection. Surprisingly, and contrary to the results of the mouse experiment, the chlamydial burden was significantly reduced in monkeys infected with the pmpD null mutant compared to that in monkeys infected with the WT strain during the first 2 weeks of infection (Fig. 4A). The average total chlamydial burden during the first 2 weeks of infection for the pmpD null mutant-infected animals was 558 IFU, whereas it was 7,282 IFU for WT-infected animals, resulting in a 13-fold reduction in burden. The peak shedding difference was 25-fold at 1 week postinfection (see Table S3 in the supplemental material). Interestingly, there were no significant differences at any time point after week 2 postinfection. The duration of infections was also similar for pmpD null mutant and WT clones. Nevertheless, the average total infectious burden during the entire study for the pmpD null mutant-infected animals was 795 IFU (range, 32 to 1,477 IFU), whereas it was 7,782 IFU (range, 2,686 to 18,263 IFU) for the WT group (P = 0.002), resulting in an approximately 10-fold reduction (see Table S3 in the supplemental material). A similar trend in pathological inflammation was observed. Clinical pathology based on hyperemia and follicle formation was significantly reduced in pmpD null mutant-infected animals at 2 and 3 weeks postinfection compared to that in WT-infected animals (Fig. 4B). Similar to the infectious burden, the pathologies at later time points were not significantly different. We resequenced the pmpD gene of organisms isolated from the mutant-infected animals at 2, 4, and 6 weeks postinfection to test for revertants; none were detected.

FIG 4.

The pmpD null mutant is attenuated in a nonhuman primate infection model. Six cynomolgus macaques were ocularly infected with the pmpD null mutant and six were infected with the WT strain to evaluate infection, the duration of shedding, and clinical pathology. (A) Ocular infection was monitored by swabbing the conjunctiva and culturing recoverable IFU on HeLa cell monolayers. Averages and standard deviations are shown on a log₁₀ scale. Detailed chlamydial shedding data are shown in Table S3 in the supplemental material. (B) Disease was scored on the basis of hyperemia and follicle formation on the upper conjunctival surfaces (score 0 = no disease, score 12 = maximum disease). Averages and standard deviations are shown. Macaques were monitored for shedding and disease for 100 days postinfection. *, statistically significant difference (P < 0.05, two-tailed Mann-Whitney U test, n = 6).

The pmpD null mutant is deficient in host-cell attachment during early host-cell interactions.

To investigate if the reduced infectivity and virulence in nonhuman primates correlates with a detectable in vitro phenotype in human cells, we tested a human endocervical cell line, A2EN (15), and a human conjunctival cell line, HCjE (16); both of these are epithelial cell lines. A2EN and HCjE cells were infected with the pmpD null mutant or the WT strain using McCoy cells as infection controls. Contrary to the mouse cell line infections, the pmpD null mutant showed an approximately 70% reduction in the ability to infect the human cell lines (Fig. 5A). A similar 70% decrease in the number of recoverable IFU (Fig. 5B) was observed, indicating that the two chlamydial strains have a similar burst size. These data suggested that the absence of PmpD significantly altered early host-cell interactions in human cells, while it had no influence on growth or replication in vitro.

FIG 5.

The pmpD null mutant is deficient in attachment to human cells. (A) Infection of human endocervical (A2EN) and conjunctival (HCjE) cells with the pmpD null mutant yielded significantly fewer inclusions than the number obtained by infection with the WT (P < 0.05, two-tailed Mann-Whitney U test, n = 3). pmpD null mutant inclusions are shown as a percentage of the number for the WT control for each cell line. (B) A decrease in the number of recoverable IFU concomitant with decreased numbers of inclusions was detected in both human cell lines following infection with the pmpD null mutant and the WT strain (P < 0.05, two-tailed Mann-Whitney U test, n = 3). The amount of pmpD null mutant recoverable IFU is shown as a percentage of the number of recoverable IFU for the WT control for each cell line. (C) A2EN cells were consecutively infected with the same inoculum zero, one, two, or three times before the titer of the nonattached EBs in the inoculum in McCoy cells was redetermined. The number of viable EBs remaining in the inoculum after A2EN cell infections, determined by redetermining the titer in McCoy cells, is shown (results are averages and standard deviations). The numbers of unattached pmpD null mutant IFU remaining in the inocula after one, two, and three infections of A2EN cells were statistically significantly higher (P < 0.05, two-tailed Mann-Whitney U test, n = 6).

To further characterize the role of PmpD during early host-cell interactions, we investigated the reason that the pmpD null mutant infections produced reduced numbers of IFU in the human endocervical cells, the primary target cell type during natural C. trachomatis serovar D urogenital infections. A2EN cells were infected with the pmpD null mutant or the WT strain. After infection, the titers of the inocula containing the unattached elementary bodies (EBs) were redetermined in McCoy cells to determine the number of infectious EBs still present. Similar experiments were conducted by serially infecting A2EN cell monolayers before redetermining the titer in McCoy cells. Interestingly, pmpD null mutant inocula contained a high number of viable EBs even after several infections of A2EN cell monolayers, while WT EBs were more than 90% reduced by just one infection of A2EN cells (Fig. 5C). The high number of EBs detected in the pmpD null mutant inocula strongly suggests that the absence of PmpD in C. trachomatis EBs significantly inhibits chlamydial attachment to human endocervical cells and that this deficiency is likely the cause for the reduced pmpD null mutant infectivity in human cells (Fig. 5A).

DISCUSSION

C. trachomatis was historically considered a genetically intractable organism. However, recent advancements in chlamydial genetics now provide new approaches to study gene function (6–8). C. trachomatis PmpD has been implicated as an important chlamydial virulence factor, as it is a conserved surface antigen and target of neutralizing antibodies (12). However, direct evidence supporting this hypothesis is lacking. Here, we used a targeted reverse genetics approach to generate a pmpD null mutant and have studied the infection phenotype of this mutant in vitro and in vivo. We showed that pmpD is not an essential chlamydial gene but the absence of PmpD interferes with the RB-inclusion membrane association in both human and murine cells. The pmpD null mutant had no detectable infection or growth deficiencies in cultured mouse cell lines or in a murine chlamydial infection model. However, in human endocervical and conjunctival cells and in a nonhuman primate infection model, we observed a significant reduction in infectivity and virulence for the C. trachomatis pmpD null mutant. We also showed that the infection deficiency in human cells and nonhuman primates is likely the result of the null mutant's significantly impaired chlamydial attachment to host cells. Taken together, the findings implicate C. trachomatis PmpD as an important virulence factor that is involved in early host-cell interactions.

The most surprising result of our study was the complete absence of any detectable deficiency in murine infection models. Although PmpD exhibits 72% sequence similarity between human and mouse chlamydial strains, subtle differences may exist in domains that are critical in the molecular interactions at the host-pathogen interface. Conversely, the cognate PmpD host receptor(s) might exhibit a similar diversity that determines chlamydial interactions. One intriguing hypothesis is that PmpD's interactions with the host cells are species specific and that C. trachomatis PmpD has coevolved with its natural host to optimize its role in chlamydial pathogenesis. Coevolution of human and mouse chlamydial strains and their respective hosts has also been described for genes in the organism's plasticity zone that are thought to function in evasion of host immunity (24). The species specificity of PmpD could be confirmed by generating a C. muridarum pmpD null mutant and evaluating its virulence in the mouse model.

Another unexpected finding was that even in a nonhuman primate model of infection, a model that usually mimics human infection very well, the pmpD null mutant showed only partial attenuation that was limited to early times postinfection, while the bacterial burden and ocular disease were not significantly altered at later time points. pmpD is a member of an expanded and divergent autotransporter family that consists of nine genes (9). This suggests that other Pmp proteins might compensate for the loss of PmpD. C. trachomatis pmp genes exhibit phase-variable expression patterns during the chlamydial growth cycle (25). Although the significance of the phase variation is not known, it could represent a potential compensatory mechanism important to chlamydial pathogenesis. This hypothesis is supported by studies on Rickettsia, an intracellular bacterial pathogen with an expanded autotransporter gene family (26). The deletion of the Rickettsia rickettsii autotransporter outer membrane protein A has been associated with an attenuated phenotype (27). Moreover, other members of this autotransporter gene family are thought to play a role in host tropism that varies between arthropod and mammalian hosts (26). It is possible that the C. trachomatis pmp gene family functions similarly by determining tissue tropism within the human host rather than among disparate mammalian hosts.

The obvious question emerging from our studies is how might PmpD function during chlamydial host interactions? Our current findings support two potential answers to this question. First, our experiments with human endocervical cells suggest that PmpD functions in either attachment or entry. The fact that anti-PmpD antibodies neutralize chlamydial infection in vitro (12, 28) is consistent with this possibility. Furthermore, the surface-exposed PmpD passenger domain contains an integrin binding RGD motif (11) that might function in either attachment or entry. Interestingly, the PmpD passenger domain of the mouse pathogen C. muridarum has this motif replaced by KGD (11), a modification that is known to alter integrin binding specificity (29).

Another possible clue resulting from our work that suggests a specific interaction for PmpD and host-cell cytoplasmic membrane receptors was the failure of pmpD null mutant RBs to intimately interact with the inner surface of the chlamydial inclusion membrane (Fig. 2C; see Table S1 in the supplemental material). The chlamydial inclusion membrane, although modified by chlamydiae (30, 31), is initially derived from host endocytic vesicles following chlamydial entry (32). Thus, the inability of pmpD null mutant RBs to bind the inclusion membrane may be a secondary, infection-independent phenotype that is an indication of an earlier more important PmpD ligand-receptor interaction between EBs and the cytoplasmic membrane that functions in EB attachment or entry. The fact that this intracellular developmental phenotype had no measurable effect on the in vitro growth of the organism in cultured cells argues against an alternative function important to nutrient acquisition or chlamydial protein secretion, but one cannot exclude the possibility that such a host-pathogen relationship might be more important to the pathogenesis of chronic or persistent chlamydial infections. The lack of an in vitro growth defect is also inconsistent with the contact-dependent hypothesis that suggests that the detachment of the RB from the inclusion membrane provides the signal for the late differentiation of RBs into EBs (33, 34).

Further experimentation is needed to fully elucidate the molecular interactions between PmpD and the host cells. Unfortunately, our efforts to complement the null mutant and express PmpD on a shuttle vector in the pmpD null mutant or in serovar L2 were unsuccessful, as no viable transformants were found after numerous attempts, while complementation efforts with identical shuttle vectors carrying similarly sized inserts were successful. Our current efforts are focused on generating null mutants for other members of the pmp family in combination with the pmpD null mutation to investigate their potential redundant functions, generating a C. muridarum pmpD null mutant to confirm the species specificity of PmpD, and introducing nonsynonymous single nucleotide polymorphisms (SNPs) into potentially key parts of the protein, like the RGD motif. In summary, our study suggests that C. trachomatis PmpD is an important chlamydial virulence factor that primarily functions in host-cell attachment and entry.