Polymorphisms in Inc Proteins and Differential Expression of inc Genes among Chlamydia trachomatis Strains Correlate with Invasiveness and Tropism of Lymphogranuloma Venereum Isolates

Abstract

Chlamydia trachomatis is a human bacterial pathogen that multiplies only within an intracellular membrane-bound vacuole, the inclusion. C. trachomatis includes ocular and urogenital strains, usually causing infections restricted to epithelial cells of the conjunctiva and genital mucosa, respectively, and lymphogranuloma venereum (LGV) strains, which can infect macrophages and spread into lymph nodes. However, C. trachomatis genomes display >98% identity at the DNA level. In this work, we studied whether C. trachomatis Inc proteins, which have a bilobed hydrophobic domain that may mediate their insertion in the inclusion membrane, could be a factor determining these different types of infection and tropisms. Analyses of polymorphisms and phylogeny of 48 Inc proteins from 51 strains encompassing the three disease groups showed significant amino acid differences that were mainly due to variations between Inc proteins from LGV and ocular or urogenital isolates. Studies of the evolutionary dynamics of inc genes suggested that 10 of them are likely under positive selection and indicated that most nonsilent mutations are LGV specific. Additionally, real-time quantitative PCR analyses in prototype and clinical strains covering the three disease groups identified three inc genes with LGV-specific expression. We determined the transcriptional start sites of these genes and found LGV-specific nucleotides within their promoters. Thus, subtle variations in the amino acids of a subset of Inc proteins and in the expression of inc genes may contribute to the unique tropism and invasiveness of C. trachomatis LGV strains.

INTRODUCTION

The chlamydiae are a large group of obligate intracellular bacteria. They include Chlamydia trachomatis, which causes ocular and genital infections in humans. These infections are the leading cause of preventable blindness in developing countries (62) and the most prevalent cause of bacterially sexually transmitted diseases worldwide (4). C. trachomatis strains include trachoma and LGV (lymphogranuloma venereum) biovars (48). The trachoma biovar comprises ocular and urogenital strains, which cause localized infections of the epithelial surface of the conjunctiva and genital mucosa, respectively; strains of the LGV biovar cause invasive urogenital disease, due to their ability to infect macrophages and spread into lymph nodes. C. trachomatis strains can be further classified into ocular serovars A to C, urogenital serovars D to K, and LGV serovars L1 to L3.

The genomic sequences of different ocular, urogenital, and LGV strains exhibit >98% identity and a high degree of synteny (12, 25, 29, 31, 50, 53, 55, 59, 60). Therefore, the determinants of the different types of infection (invasive or noninvasive) and tissue tropism (eyes, genitals, and lymph nodes) must rely on the few genes present in some strains but not in others and on nucleotide differences which may lead either to proteins with disease group-specific amino acids or to differential gene expression. Some of these determinants were suggested in previous studies: the tryptophan (trpRBA) operon (10, 19, 51) and genes encoding cytotoxin (11), phospholipase (40), polymorphic membrane proteins (Pmps) (24), and Tarp (34).

Chlamydiae are characterized by a developmental cycle involving the interconversion between an infectious form, the elementary body, and a noninfectious form, the reticulate body (1). Throughout development, the bacteria reside and multiply within a membranaceous compartment, known as the inclusion, and manipulate host cells by using a type III secretion (T3S) system to translocate effector proteins into host cells (6, 61). Chlamydial T3S substrates have been found by the identification of an N-terminal secretion signal using Salmonella (26), Shigella (56, 57), and Yersinia (14, 15, 20, 21, 27) as heterologous hosts. These include inclusion membrane (Inc) proteins, characterized by a bilobed hydrophobic motif thought to mediate their insertion into the inclusion membrane (16, 21, 45, 56, 57). Inc proteins from the same chlamydial species are normally unrelated to each other (16, 35), and only a subset of ∼25 Inc proteins is conserved between species (16, 35). The C. trachomatis Inc proteins CT119/IncA, CT115/IncD, CT147, CT229, and CT813 have been shown or suggested to subvert host cell vesicular and nonvesicular transport (17, 18, 46). However, virtually nothing is known about the biological role of most Inc proteins, which also reflects the lack of straightforward methods for genetically manipulating Chlamydiae.

In this work, we used phylogenetic, molecular evolution, and gene expression analyses to determine whether Inc proteins could affect the type of infection and tissue tropism associated with C. trachomatis. Our studies suggest that a subset of Inc proteins might play a role in the unique capacity of C. trachomatis LGV strains to infect macrophages and disseminate into lymph nodes.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

C. trachomatis prototype strains B/Har36, C/TW3, E/Bour, L2/434, and L3/404 (from the American Type Culture Collection [ATCC]) and clinical strains F/CS465-95 and L2b/CS19-08 (from the collection of the Portuguese National Institute of Health) were used as detailed below and propagated in HeLa 229 cells (from the ATCC) using standard techniques (49).

Escherichia coli TOP10 (Invitrogen) was used for construction and purification of the plasmids. Yersinia enterocolitica ΔHOPEMT [MRS40(pIML421) yopH_Δ1–352 yopO_Δ65–558 yopP₂₃ yopE₂₁ yopM₂₃ yopT₁₃₅] (28), which is deficient for the Yersinia T3S effectors YopH, -O, P, -E, -M, and -T but T3S proficient, was used for T3S assays. To construct a T3S-deficient derivative of ΔHOPEMT, we deleted in this strain the complete coding sequence (codons 1 to 354) of the yscU gene, which encodes an essential component of the Y. enterocolitica T3S system (54). This was done by allelic exchange with the mutator plasmid pLY16 (54). The resulting Y. enterocolitica ΔHOPEMT ΔYscU [MRS40(pFA1001)] strain was also used in T3S assays. The yscU(Δ1-354) mutation had been previously shown to be nonpolar (54). E. coli and Y. enterocolitica were routinely grown in liquid or solid Luria-Bertani medium with the appropriate antibiotics and supplements. Plasmids were introduced into E. coli or Y. enterocolitica by electroporation.

Construction of plasmids.

Plasmids were constructed and purified using proofreading Phusion DNA polymerase (Finnzymes), restriction enzymes (MBI Fermentas), T4 DNA ligase (Invitrogen), DreamTaq DNA polymerase (MBI Fermentas), a NucleoSpin gel and PCR cleanup kit (Macherey-Nagel), and a GeneElute plasmid miniprep kit (Sigma), according to the instructions of the manufacturers. In brief, to analyze T3S signals we constructed plasmids harboring hybrid genes encoding the N-terminal 20 amino acids of each C. trachomatis Inc protein or of the Y. enterocolitica T3S chaperone SycT (28) and the mature form of TEM-1 β-lactamase (TEM-1). These hybrids were made by PCR using plasmid pCX340 as the template (13). Each forward primer contained an NdeI site followed by 60 nucleotides encoding the first 20 amino acids of each Inc (or of SycT) protein in frame with a sequence complementary to the 3′ extremity of the transcribed strand of the TEM-1-encoding gene; each reverse primer contained either a HindIII or a XhoI site followed by a sequence complementary to the 3′ extremity of the nontranscribed strand of the TEM-1-encoding gene. Digested PCR products were ligated into pLJM3, a low-copy-number plasmid which enables expression of cloned genes driven by the promoter of the Y. enterocolitica yopE gene (36). A similar strategy was used to construct a plasmid encoding TEM-1 alone, except that the forward primer did not contain an inc gene sequence. For primer design, the DNA sequence of each inc gene in C. trachomatis strain L2/434 (see Table S1 in the supplemental material) was used, except for ct036, ct115 (incD), and ct119 (incA), in which the sequence from strain D/UW3 was used (see Table S1 in the supplemental material); the DNA sequence of sycT was from Y. enterocolitica pYVe227 (accession number AF102990). The sequences of all primers are available upon request. The accuracy of the nucleotide sequence of all the inserts in the constructed plasmids was checked by DNA sequencing.

Y. enterocolitica T3S assays.

These analyses were done as previously described (54). Proteins in bacterial pellets and culture supernatants were analyzed by immunoblotting with mouse monoclonal anti-TEM-1 antibodies (QED Bioscience; 1:500) and with rabbit polyclonal anti-SycO antibodies (1:1,000) (32) and detected with a ChemiDoc XRS+ system (Bio-Rad). The amount of protein in the supernatant relative to the amount of total protein (percentage of secretion) was determined from immunoblot images with Image Lab (Bio-Rad).

DNA sequences.

Regardless of the origin of the C. trachomatis gene, throughout this work we used the nomenclature of the annotated D/UW3 strain (see Table S1 in the supplemental material). The nucleotide sequences of the inc, pmp, and housekeeping genes analyzed were from the available genomes of 51 C. trachomatis strains (12, 25, 29, 31, 50, 53, 55, 59, 60). The strains and corresponding genome accession numbers are listed in Table S1 in the supplemental material. The DNA sequences were retrieved from pairwise alignments obtained by BLAST. All sequences were manually inspected and corrected for accuracy and completeness. We excluded from further analysis a few DNA sequences containing ambiguous nucleotides. Whenever there were distinct annotations in GenBank for the start codon of the same inc gene in C. trachomatis archetype ocular (A/Har13), urogenital (D/UW3), and LGV (L2/434) strains, we used the following: for ct036 and ct119 (incA), in all cases the start codons of each gene as annotated for D/UW3 and A/Har13; for ct115 (incD) and ct192, the start codons of each gene as annotated for A/Har13, D/UW3, or L2/434, in ocular, urogenital or LGV strains, respectively; for ct226, in all cases the start codon as annotated for A/Har13 (see Tables S1 and S3 in the supplemental material).

Sequence alignments and analyses of polymorphisms, phylogeny, and molecular evolution.

Alignments of the amino acid sequences of the Inc, PMP, and housekeeping proteins, deduced from the retrieved nucleotide sequences, were generated using the ClustalW algorithm in MEGA5 software (www.megasoftware.net) (58). All alignments were manually inspected and corrected for artifacts. We excluded all strain-specific pseudogenes (see Table S3 in the supplemental material) from further phylogenomic and evolutionary analyses.

For the analyses of polymorphism, phylogeny, and molecular evolution, various tools present in MEGA5 were used, essentially as previously described (24). Briefly, for analyses of polymorphism, we computed pairwise, overall, within-group (ocular, urogenital, or LGV), and between-group (ocular versus urogenital, ocular versus LGV, or urogenital versus LGV) amino acid p distances. For analyses of phylogeny, trees were generated using the neighbor-joining method (47). The generated phylograms of Inc proteins were inspected for separate branches (segregation) of all ocular, urogenital, and LGV strains. This analysis was supported by comparison of pairwise amino acid p distances between and within disease groups. The phylogeny of an Inc protein was considered to segregate a disease group if the maximum pairwise amino acid p distance within groups was less than any of the pairwise amino acid p distances between groups. For molecular evolution analyses, we used the Kumar method (39) to compute overall means of nonsynonymous (d_N) and synonymous (d_S) substitutions (per nonsynonymous or synonymous site, respectively) and to find genes that may be under positive selection. By using the codon-based Z test of selection in MEGA5, genes were considered under positive selection if they showed a statistically significant value (P < 0.05) to reject the null hypothesis of strict neutrality (d_N = d_S) in favor of both positive selection (d_N > d_S) and lack of neutrality (d_N ≠ d_S). All these analyses were performed by bootstrapping with 1,000 replicates and with the pairwise deletion option selected.

Real-time quantitative PCR.

The expression of inc genes during the developmental cycle of C. trachomatis B/Har36, C/TW3, E/Bour, F/CS465-95, L2/434, L2b/CS19-08, and L3/404 was estimated by determining inc mRNA levels at different times postinfection by real-time quantitative PCR (RT-qPCR). These experiments were essentially done as previously described (7, 42). Briefly, for each strain, six tissue culture flasks with a surface area of 25 cm² containing monolayers of HeLa 229 cells were inoculated at a multiplicity of infection of 1; cells were harvested at 2, 6, 12, 20, 30, and 42 h postinfection by scraping in ice-cold phosphate-buffered saline. The cell suspension was sonicated to disrupt mammalian cells and promote bacterial release, followed by low-speed centrifugation at 4°C. The supernatant was then frozen in liquid nitrogen and stored at −80°C. These samples were used for total RNA purification and generation of cDNA, as previously described (7). Primers (available upon request) were designed for each inc gene using Primer Express (Applied Biosystems), based on identical C. trachomatis sequences between strains. The RT-qPCR assays were done using the ABI 7000 SDS, SYBR green chemistry, and optical plates (Applied Biosystems), as previously described (7, 42). At each time point, raw RT-qPCR data for each inc gene were normalized against the data obtained for the 16S rRNA transcript, as it was previously demonstrated that this is a good endogenous control (7). The final results were based on at least two independent experiments.

Transcription linkage analysis and identification of transcriptional start sites.

We searched for disease group-specific nucleotides within the promoter region of three inc genes (ct058, ct192, and ct214) that showed differential gene expression. For this, we determined their transcriptional start sites (TSSs). For ct058 and ct192, it was ambiguous whether their promoters would lie immediately upstream from their predicted start codons (see Fig. 5A). Therefore, we used reverse transcription coupled with PCR (RT-PCR) to determine if ct058 and ct059, or ct192 and ct193, are part of the same transcriptional unit. For this, RNA was isolated from HeLa 229 cells infected for 30 h with C. trachomatis L2/434 using an NZY total RNA kit (NZYTech). cDNA was then generated by using random hexamers and iSCRIPT (Bio-Rad). Primers (see Table S2 in the supplemental material) were designed to generate PCR products containing ∼300 bp upstream and downstream from the predicted start codons. PCR products were obtained using DreamTaq DNA polymerase (MBI Fermentas). As controls for the PCRs, we also used as the template the product of a typical reverse transcription reaction but without iSCRIPT or total DNA isolated using an NZY tissue gDNA kit (NZYTech), from cells that were either infected with strain L2/434 for 42 h or left uninfected.

Fig 5.

Identification of LGV-specific nucleotides in the promoter regions of ct192 and ct214 and within the ct059-ct058 transcript. (A) Genetic organization of ct058, ct192, and ct214 (the nomenclature of D/UW3 is used) depicting the fragments amplified in the transcriptional linkage analysis and the approximate locations of the transcriptional start sites (TSSs) determined by RACE in L2/434. (B) Transcriptional linkage analysis in L2/434: gDNA+, PCR from total DNA isolated from cells infected with L2/434; cDNA+, PCR from cDNA generated with reverse transcriptase (RT) from total RNA isolated from cells infected with L2/434; cDNA−, as for cDNA+ but without RT; gDNA−, PCR from DNA of uninfected cells. (C to E) Schematic view of the nucleotide sequences of the ct059 promoter region, ct059-ct058 intragenic region, and first codons of ct058 (C) and of the promoter regions of ct192 (D) and ct214 (E) in L2/434. The TSSs are labeled +1, and the predicted −10 and −35 σ⁶⁶-like hexamers are underlined. The sequences of the three LGV strains used in RT-qPCR assays (Fig. 5) are 100% identical within the depicted regions (see Fig. S3 and S4 in the supplemental material). The identified LGV-specific nucleotides are indicated with an arrow, and letters below the sequence represent nucleotides present in those positions in the ocular and urogenital strains used in RT-qPCR assays. The full nucleotide sequences of these regions in the strains used in RT-qPCR assays are shown in Fig. S3 and S4.

The identification of the TSSs of ct059 (upstream from ct058 and in the same transcriptional unit; see Fig. 5A), ct192, and ct214 in L2/434 was done by 5′ rapid amplification of cDNA ends (RACE), using a 5′/3′RACE kit (second generation; Roche). We used RNA isolated as described above from HeLa 229 cells infected for 30 h with C. trachomatis L2/434 and the primers listed in Table S2 in the supplemental material. Final PCR amplification of double-stranded cDNA was done with Phusion DNA polymerase (Finnzymes). PCR products were purified after agarose gel electrophoresis, using a High Pure PCR purification kit (Roche), and then subjected to DNA sequencing. To analyze the determined TSSs in the context of the promoter regions of ct059-ct058, ct192, and ct214 in all strains used in the RT-qPCR assays, the corresponding nucleotide sequences were either retrieved from GenBank (E/Bour, L2/434, and L3/404; see Table S1 in the supplemental material) or determined by DNA sequencing (C/TW3, B/Har36, F/CS465-95, and L2b/CS19-08), as previously described (24) and using the primers listed in Table S2. All these manipulations were done according to instructions from the indicated manufacturers.

Nucleotide sequence accession numbers.

The sequences of the promoter regions of ct192 and ct214 and of ct059-ct058 in C/TW3, B/Har36, F/CS465-95, and L2b/CS19-08 determined in this study were submitted to GenBank and are available under accession numbers JX451863 to JX451874.

RESULTS

Identification of T3S signals in C. trachomatis Inc proteins.

We focused on 48 predicted C. trachomatis Inc proteins (i.e., those possessing a bilobed hydrophobic domain) which were singled out and studied by Li et al. (33) (Table 1). Thus far, 23 of these proteins have been detected in the inclusion membrane by immunofluorescence microscopy using specific antibodies (referred to as known Inc proteins) (references 16, 33, and 38 and references therein), but other predicted Inc proteins have not (referred to as putative Inc proteins) (16, 35) (Table 1).

Table 1.

Summary of T3S signals found in known and putative Inc proteins of C. trachomatis analyzed in this work

Inc proteina	T3S signal	Reference
Knownb
CT101	No	This work
CT115/IncD	Yes	56; this work
CT116/IncE	Yes	56
CT117/IncF	Yes	This work
CT118/IncG	Yes	56
CT119/IncA	Yes	56; this work
CT147	NDe
CT222	Yes	This work
CT223	Yes	56
CT225	No	This work
CT226	Yes	16
CT228	Yes	16; this work
CT229	Yes	56
CT232/IncB	Yes	This work
CT233/IncC	Yes	21, 56; this work
CT249	Yes	16; this work
CT288	Yes	56
CT358	Yes	16; this work
CT440	Yes	16; this work
CT442	Yes	56
CT618	Yes	This work
CT813	Yes	This work
CT850	Yes/Nod	16; this work
Putativec
CT005	Yes	This work
CT006	ND
CT036	Yes	This work
CT058	Yes	16; this work
CT134	No	This work
CT135	Yes	This work
CT164	No	This work
CT179	No	This work
CT192	No/Yesd	16; this work
CT195	Yes	16; this work
CT196	Yes	This work
CT214	Yes	This work
CT224	Yes	This work
CT227	Yes	This work
CT300	Yes	This work
CT345	Yes	This work
CT357	Yes	This work
CT365	Yes	This work
CT383	Yes	16; this work
CT449	Yes	This work
CT483	Yes	This work
CT484	No	16; this work
CT565	No	16; this work
CT728	No	This work
CT789	Yes	This work

^aProteins containing a bilobal hydrophobic motif that were analyzed by Li et al. (33), which we selected to study in this work. We did not consider proteins (Cap1 and CopN) which localize to the inclusion membrane but which do not possess the bilobal hydrophobic domain (20, 23). More recent bioinformatics-based analyses identified additional putative Inc proteins (CT018, CT079, CT081, CT244, CT324, CT326, CT556, CT578, CT616, CT618, CT642, CT645, CT788, CT789, CT814.1, CT819, CT837, CT846, and CT873) in C. trachomatis (16, 35), but these proteins were not analyzed in our study.

^bKnown Inc proteins contain a bilobal hydrophobic motif and have been localized to the inclusion membrane by immunofluorescence microscopy using specific antibodies (references 16, 33, and 38 and references therein).

^cPutative Inc proteins contain a bilobal hydrophobic motif but have not yet been localized to the inclusion membrane.

^dConflicting data between our observations and previous analyses obtained by using Shigella flexneri as a heterologous bacterial host.

^eND, not determined.

As Inc proteins are believed to be transported into the inclusion membrane by a T3S mechanism (16, 21, 56, 57), we used Y. enterocolitica as a heterologous bacterium to identify a possible N-terminal T3S signal in putative Inc proteins of C. trachomatis by comparison to known Inc proteins. We sought to obtain additional indirect evidence (besides the bilobed hydrophobic motif) that putative Inc proteins localize to the inclusion membrane. We analyzed secretion of hybrid proteins comprising the first 20 amino acids of each putative or known Inc and TEM-1 by T3S-proficient (ΔHOPEMT) or T3S-deficient (ΔHOPEMT ΔYscU) Y. enterocolitica (Fig. 1). The different Y. enterocolitica strains were incubated under T3S-inducing conditions (54), followed by fractionation of the bacterial cultures into culture supernatants and bacterial pellets and subsequent immunoblotting analyses of the proteins in the two fractions (examples in Fig. 1A and B). In total, we analyzed T3S signals in 24 putative Inc proteins (18 had not been previously analyzed for T3S) and in 15 known Inc proteins (7 had not been previously analyzed for T3S) (Fig. 1C and Table 1). The expression levels of the TEM-1 hybrid of putative Inc CT006 were extremely low, which hampered the analysis of a T3S signal in this protein (data not shown). These experiments led to the identification of a T3S signal in 18 putative Inc proteins and in 12 known Inc proteins (Fig. 1C and Table 1). This revealed 5 known and 14 putative Inc proteins as novel C. trachomatis T3S substrates (Fig. 1C and Table 1). However, we did not detect a clear T3S signal in three known Inc proteins (CT101, CT225, and CT850) (Fig. 1 and Table 1). Their T3S signals could extend beyond the first 20 amino acids or might not be recognized by the Y. enterocolitica T3S system; alternatively, they may be transported into the inclusion membrane by a distinct mechanism. Regardless of the exact explanation, this implies that the lack of a detectable T3S signal could not be taken as a definitive indication that a putative Inc does not localize to the inclusion membrane. Furthermore, the percentages of putative and known Inc proteins analyzed that displayed an N-terminal region recognized by the Y. enterocolitica T3S machinery were nearly identical [75% (18 of 24) and 80% (12 of 15) for putative and known Inc proteins, respectively]. Overall, these analyses indicated that most of the putative Inc proteins analyzed are T3S substrates.

Fig 1.

Type III secretion (T3S) signals in C. trachomatis Inc proteins. Y. enterocolitica T3S-proficient (ΔHOPEMT) and T3S-defective (ΔHOPEMT ΔYscU) bacteria were used to analyze secretion of hybrid proteins comprising the first 20 amino acids of Inc proteins or of Y. enterocolitica SycT, fused to the mature form of TEM-1 β-lactamase (TEM-1). (A and B) Immunoblots show the result of representative assays in which proteins in culture supernatants (S, secreted proteins) and in bacterial pellets (P, nonsecreted proteins) from ∼5 × 10⁷ bacteria were loaded per lane. SycT and SycO are strictly cytosolic Yersinia T3S chaperones (28, 32). SycT₂₀-TEM-1 was a negative control for the T3S assays. Immunodetection of SycO ensured that the presence of TEM-1 hybrid proteins in culture supernatants was not a result of bacterial lysis or contamination. (C) The percentage of secretion of each TEM-1 hybrid was calculated by densitometry as the ratio between the amount of secreted and total protein. The threshold to decide whether a protein was secreted was set to 5% (dashed line), based on the percentage of secretion of SycT₂₀-TEM-1. Data are means ± SEM from at least three independent experiments. Please note that in this work we did not analyze all described putative Inc proteins (see Table 1) and that T3S signals were not analyzed for all known Inc proteins.

Differences in the amino acid sequences of Inc proteins among C. trachomatis strains correlate with the type of infection and with tissue tropism.

The nucleotide sequences of the genes encoding the selected 48 known and putative Inc proteins were retrieved from 51 fully sequenced C. trachomatis genomes (7 ocular, 23 urogenital, and 21 LGV strains) (see Table S1 in the supplemental material). In an initial analysis, 7 inc genes proved to be pseudogenes in different C. trachomatis strains (ct058, ct101, ct135, ct192, ct227, ct228, and ct300) (see Table S3 in the supplemental material). ct358 was described as a pseudogene in L2/434 and L2/UCH-1 (59), and its nucleotide sequence is 100% identical in all LGV strains. However, analysis of its nucleotide sequence suggests that ct358 may encode a functional protein that is 7 amino acids shorter at its C terminus than CT358 (178 amino acid residues) in ocular and urogenital strains (see Table S3). In the seven inc genes that we identified as pseudogenes, the full-length gene is disrupted by a mutation that leads to a significantly truncated protein. The only disease group-specific correlation was observed with ct300, which is a pseudogene in all LGV strains analyzed. Therefore, the encoded protein is expendable for LGV infections. In addition, ct058 and ct101 revealed to be pseudogenes in almost all analyzed ocular and urogenital strains, respectively (see Table S3). ct135, ct192, ct227, and ct228, are only pseudogenes in a few strains, with no obvious correlation with ocular, urogenital, or LGV disease groups (see Table S3). Furthermore, 12 inc genes showed small deletion and insertion events (see Table S3).

To understand if the amino acid sequences of Inc proteins vary among strains, we determined the overall mean genetic distance (amino acid p distance) for each Inc protein among all 51 C. trachomatis strains (discarding strain-specific pseudogenes) (Fig. 2A). As a reference, we also analyzed the 9 Pmps and 9 housekeeping proteins of C. trachomatis previously shown to be polymorphic (43) (Fig. 2A). The Pmps should localize to the bacterial outer membrane and the housekeeping proteins within the bacterial cell. The average p distance was 0.017 (standard error of the mean [SEM], 0.002) for Inc proteins, 0.020 (SEM, 0.007) for Pmps, and 0.013 (SEM, 0.003) for housekeeping proteins. Based on this, and considering the average p distance for Inc proteins as a cutoff value, we defined 19 (40% of the total) Inc proteins as polymorphic (p distance ≥ 0.017) (Fig. 2A). As comparison, 4 Pmps (44%) and 2 housekeeping proteins (22%) displayed a p distance of ≥0.017. This showed that the overall degree of polymorphism in Inc proteins among C. trachomatis strains is similar to that of Pmps and is higher than that of known polymorphic housekeeping proteins.

Fig 2.

Polymorphisms in C. trachomatis Inc proteins. Polymorphic membrane proteins (Pmps) and housekeeping proteins (HKs) were analyzed as references. (A) Overall mean genetic distance (polymorphisms) based on the p distance between all possible pairs of amino acid (aa) sequences of Inc proteins, Pmps, and HKs among C. trachomatis strains. Proteins marked with an asterisk have a p distance that is equal to or higher than the average value for Inc proteins (0.017 [dashed line]). (B) Average mean genetic distance, based on the p distance between all possible pairs of amino acid sequences within (ocular [OC], urogenital [UROG], or LGV) or between (OC-UROG, OC-LGV, or UROG-LGV) C. trachomatis disease groups. (C) Venn diagrams showing the phylogenetic segregation of C. trachomatis disease groups based on neighbor-joining trees of Inc proteins, Pmps, or HKs and on pairwise p distances between and within disease groups for all possible pairs of Inc protein, PMP, or HK sequences from the C. trachomatis strains analyzed. All these analyses were performed by bootstrapping with 1,000 replicates. Error bars represent SEM.

To understand if the amino acid differences between Inc proteins were related to the type of infection and with tissue tropism, we determined the average p distances of Inc proteins within and between the three groups of strains (ocular, urogenital, and LGV) (Fig. 2B). This showed that the differences were largely due to variations between Inc proteins from LGV strains and ocular (average p distance = 0.031; SEM, 0.004) or urogenital strains (average p distance = 0.029; SEM, 0.004) (Fig. 2B). These average p distances were significantly higher (in all cases, P < 0.0001; two-tailed t test) than those between Inc proteins from ocular and urogenital strains (average p distance = 0.011; SEM, 0.002) or within Inc proteins from the same disease groups (average p distance < 0.006) (Fig. 2B). Similar observations were made for Pmps except that, in contrast to Inc proteins, the average p distance between Pmps from ocular and urogenital strains was not significantly different than that between Pmps from LGV strains and ocular or urogenital strains (in both cases, P > 0.05; two-tailed t test) (Fig. 2B). Also in contrast to the Inc proteins, the variation in the amino acid sequences of housekeeping proteins between each of the three groups was clearly not different (in all cases, P > 0.05; two-tailed t test) (Fig. 2B). Furthermore, the amino acid sequences of housekeeping proteins varied nearly as much between or within groups (Fig. 2B). The main exception was the LGV group, within which the housekeeping proteins, like Inc proteins and Pmps, proved to be extremely highly conserved (average p distance < 0.002 for the three cases) (Fig. 2B). Thus, the separation (p distance) between LGV strains and ocular or urogenital strains is much more marked for Inc proteins than for Pmps or housekeeping proteins.

We then made and analyzed phylogenetic reconstructions based in the amino acid sequence of Inc proteins. The phylograms of five Inc proteins (10%) showed tropism, i.e., segregation of the three disease groups, and those of 38 Inc proteins (84%) evidenced segregation of at least one disease group (Fig. 2C; also, see Table S4 in the supplemental material). The phylograms of a total of 35 Inc proteins (73%) showed segregation of LGV strains, while only 12 (25%) and 8 (17%) displayed clustering of ocular and urogenital strains, respectively (Fig. 2C; also, see Table S4). This scenario was mirrored by the phylograms of Pmps but was in contrast to the phylograms of housekeeping proteins, in which segregation by disease group was less often seen (Fig. 2C; also, see Table S4).

In summary, there are significant differences in the amino acid sequences of Inc proteins among C. trachomatis strains, and they correlate with the type of infection and with tissue tropism. In particular, differences are mostly between Inc proteins from LGV strains and Inc proteins from ocular or urogenital strains.

inc genes of C. trachomatis have distinct evolutionary dynamics, and several inc genes are likely under positive selection.

To understand the underlying evolutionary pressures that drive amino acid changes in Inc proteins, we analyzed the molecular evolution of inc genes. We first determined overall d_N/d_S values for inc genes by comparing them to the 9 pmp genes and the 9 selected housekeeping genes of C. trachomatis. We found that 24 inc genes (50%) had d_N/d_S values of >1 and that in four inc genes, all substitutions were nonsynonymous (Fig. 3A; also, see Table S5 in the supplemental material). In contrast, only two pmp genes (22%) d_N/d_S values of >1, and all housekeeping genes had d_N/d_S values of <1 (Fig. 3A; also, see Table S5).

Fig 3.

Evolutionary dynamics of inc genes. Genes encoding polymorphic membrane proteins (pmp genes) or housekeeping (HK) proteins were used as reference. (A) Ratio of nonsynonymous (d_N) to synonymous (d_S) substitutions among C. trachomatis strains. The dashed line indicates neutrality (d_N/d_S = 1). The arrows specify genes likely under positive selection, according to the codon-based Z test of selection (see Materials and Methods). (B) Distribution of d_N versus d_S values for the 10 inc genes likely under positive selection, comparing the impact of artificially discarding ocular, urogenital, or LGV strains relative to the analysis with all C. trachomatis strains. The straight line in each graph indicates neutrality (d_N/d_S = 1). In all cases, inc genes likely under positive selection (codon-based Z test of selection) are depicted as black circles, while inc genes for which statistical support of likely positive selection can no longer be detected after discarding a particular group of strains are depicted as white circles. All these analyses were performed by bootstrapping with 1,000 replicates. For sake of clarity, SEMs for values in both panels are presented only in Table S5 in the supplemental material.

Analyses of the overall d_N/d_S values using the codon-based Z test of selection (see Materials and Methods) yielded statistically significant values for 10 inc genes (ct116 [incE] ct118 [incG], ct119 [incA], ct222, ct223, ct228, ct229, ct249, ct288, and ct813) but only for one pmp gene and for no housekeeping gene (Fig. 3A; also, see Table S5 in the supplemental material). This indicated that these 10 inc genes are likely under positive selection. We then aimed to understand which group of strains might cause the detection of this possible evolutionary trend. For this, we assessed the impact of artificially removing ocular, urogenital, or LGV strains from the analyses (Fig. 3B). This showed that discarding LGV strains caused major alterations on the d_N and d_S values and confined the ability to detect Z-test-based likely positive selection to only one inc gene, whereas discarding ocular or urogenital strains had less pronounced effects (Fig. 3B). In addition, within each of the 10 proteins encoded by inc genes likely under positive selection, we found 113 amino acid residues that are disease group specific (Table 2). Among these residues, 104 (92%) were in Inc proteins from LGV strains, 77 (74%) of which localized in regions of the proteins predicted to be on the cytoplasmic side of the inclusion membrane (Table 2).

Table 2.

Amino acids residues within Inc proteins encoded by genes likely under positive selection that are specific of C. trachomatis disease groups

Protein	Length (aa)a	TM segmentsa,b	Residue(s) in strain typec
			LGV	Urogenital	Ocular
CT116/IncE	132	36–59, 64–87	V37A, V38A, S39C, S64G, I68L, V72I, I84T, D92N	D19G
CT118/IncG	167	33–57, 63–88	A33V, F77C, C79Y, N87S, F135L, G136R, H165R
CT119/IncA	273	35–59, 64–84	I75T, Q207E, V211A, V212A
CT222	129	39–63, 69–93	Y10C, I75-, C89Y, L125I, V126S, Y127V, S128F, N129H
CT223	270	38–61, 67–91	G11R, A65T, K100R, I108L, K127G, N130D, P134-, C153Y, E160D, T166K, H200Y, E204D, N207R, L208M, R231L, V241A, P251L, D261Y, G263-	L225F	A40V, A75V, D142G, V161M
CT228	196	38–59, 65–86	I22T, A68V, A72P, C90Y, V119A, A143V, I144M, V146F
CT229	215	42–65, 71–90	A39S, V99I, G112E, K114E, F121S, Q125R, V126A, H137Y, Q144K, Y152H, E154A, E164K, G173R, S181N, T201A	Q102H, I156V	C157S
CT249	116	51–72, 78–97	H8Y, D24N, V80T, A89I
CT288	563	36–58, 65–88, 242–263, 269–291	A67T, I69V, E102A, S121P, A194T, S198T, N203K, T209I, A258T, I260V, I261V, L318W, S346G, L388V, –474D, A516P, V531L, A532T
CT813	264	41–61, 68–94	N7T, V76I, E107K, K125E, R128Q, T145A, E161K, A163G, E168K, E171K, E172Q, I181V, I236T

^aBased on the protein sequence annotation of C. trachomatis strain D/UW3.

^bPositions of transmembrane (TM) domains in the corresponding Inc proteins, obtained from reference 16, except for CT223, for which we found only the two indicated TM domains.

^cResidues specific to Inc proteins from each disease group are in bold, relative to the amino acid in the same position in the other disease groups, and those in regions predicted to be on the cytoplasmic side of the inclusion membrane are underlined (we considered that the loop region within two TM segments faces the lumen of the inclusion); a dash indicates a deletion or an insertion.

Overall, this suggested that the evolutionary dynamics of inc genes is distinct from that of pmp genes or housekeeping genes and that LGV-specific amino acid residues in a subset of Inc proteins might be involved in the unique ability of C. trachomatis LGV strains to infect macrophages and disseminate to lymph nodes.

Disease group-specific expression of C. trachomatis inc genes.

To analyze if there were differences in the expression of inc genes between C. trachomatis strains that correlate with the type of infection or with tissue tropism, we used RT-qPCR to determine the mRNA levels of the 48 selected inc genes throughout the developmental cycle of C. trachomatis. We aimed to find inc genes showing differences in the highest mRNA levels during the cycle (peak of expression) or in the variation of mRNA levels throughout development (profile of expression) between C. trachomatis strains.

We first infected HeLa 229 cells with ocular (C/TW3), urogenital (E/Bour), or LGV (L2/434) prototype strains. Total RNA was isolated at 2, 6, 12, 20, 30, and 42 h postinfection, which was used to generate cDNA for RT-qPCR assays (complete data are shown in Table S6 in the supplemental material). Generally, the comparison of the peak of expression revealed differences from 30- to 60-fold (depending on the strain) between inc genes (Fig. 4A). However, the averages of the peaks of expression of inc genes in C/TW3, E/Bour, and L2/434 were not significantly different between strains (Fig. 4A). Regarding the profile of expression, essentially as previously described (5, 41, 52), we identified inc genes whose expression was highest at 2 or 6 h postinfection and then either decreased or remained constant throughout the cycle (early-cycle genes), inc genes whose expression was highest only at 12 or 20 h postinfection and then decreased or remained constant at later time points (midcycle genes), and inc genes whose expression was highest only at 30 or 42 h postinfection (late-cycle genes) (Fig. 4B; also, see examples in Fig. S1 in the supplemental material). We also identified four inc genes in E/Bour and five inc genes in L2/434 that were simultaneously early- and late-cycle genes, showing identically high mRNA levels both at 2 and 6 h postinfection and at 30 and 42 h postinfection but lower expression at midcycle (Fig. 4B). This was typically the case of ct214 and ct288 in L2/434 (see Fig. S2 in the supplemental material). Generally, 35 inc genes showed the same profile of expression in the three C. trachomatis strains, and the majority of inc genes showed an early-cycle profile of expression (34 in C/TW3, 27 in E/Bour, and 32 in L2/434) (Fig. 4B). In spite of these common features, we identified 9 inc genes (ct005, ct058, ct192, ct214, ct232 [incB], ct249, ct288, ct440, and ct442) whose peak of expression consistently showed >2-fold differences between strains and/or whose profile of expression displayed differences that could not be explained by distinct growth kinetics of the strains (see Fig. S2 in the supplemental material). With the exception of ct214, all these genes showed consistently higher peaks of expression in L2/434 than in C/TW3 and/or E/Bour; ct442 also showed a higher peak of expression in E/Bour than in C/TW3 (see Fig. S2).

Fig 4.

mRNA levels of inc genes during the developmental cycle of different C. trachomatis strains. The mRNA levels of 48 inc genes (A and B) and of ct058, ct192, and ct214 (C) were analyzed by RT-qPCR throughout the developmental cycle of the indicated prototype (B/Har36, C/TW3, E/Bour, L2/434, and L3/404) and clinical (F/CS465-95 and L2b/CS19-08) strains. (A) Peak of expression (highest mRNA levels during the developmental cycle) of each inc gene. The P values were calculated by two-tailed t tests. (B) Number of inc genes showing the indicated profiles of expression (variation of mRNA levels during the developmental cycle). (C) The expression values (mean ± SEM) resulted from raw RT-qPCR data (10⁵) of each gene normalized to that of the 16S rRNA gene and are from at least two independent experiments. Complete data are shown in Table S6 in the supplemental material.

To analyze whether the differences found in the expression of ct005, ct058, ct192, ct214, ct232 (incB) ct249, ct288, ct440, and ct442 were strain specific or disease group specific, we determined the mRNA levels of these genes during the developmental cycle of additional C. trachomatis strains. For this, we infected HeLa 229 cells with ocular B/Har36 (prototype), urogenital F/CS465-95 (clinical isolate), LGV L2b/CS19-08 (clinical isolate), or LGV L3/404 (prototype) strains. The infected cells were processed for RT-qPCR assays and analyzed as described above (complete data are shown in Table S6 in the supplemental material). We detected disease group-specific differences in gene expression only for ct058, ct192, and ct214 (Fig. 4C): ct058 showed an early-cycle gene profile of expression in which mRNA levels were evident for LGV strains but only vestigial for ocular and urogenital strains; ct192 showed only a clear early-cycle gene profile of expression in LGV strains, and its expression levels were generally higher in LGV strains than in ocular or urogenital strains; ct214 displayed an early- and late-cycle gene profile of expression in LGV strains but a late-cycle gene profile of expression in ocular or urogenital strains. Therefore, we have identified three inc genes (ct058, ct192, and ct214) with differences in expression between C. trachomatis strains that correlate with the type of infection and tissue tropism, in particular with LGV isolates.

Identification of LGV-specific nucleotides in the promoter regions of ct058, ct192, and ct214.

We next attempted to obtain insights into the genetic basis for the disease group-specific expression of ct058, ct192, and ct214 by analyzing the promoter regions of these genes in C. trachomatis L2/434. The gene organization of these inc genes suggested that the promoter region of ct214 should lie between the start codons of ct214 and ct215 or within the first codons of ct215 (Fig. 5A). However, the localization of the promoter regions of ct058 or ct192 was unclear, as these genes could be cotranscribed with ct059 or ct193, respectively (Fig. 5A). Therefore, we used RT-PCR with a cDNA template generated from total RNA of HeLa 229 cells infected with C. trachomatis L2/434 to determine the possible transcriptional linkages between ct058 and ct059 and between ct192 and ct193. This indicated that ct058 is cotranscribed with ct059, which likely encodes a ferredoxin, and transcription of ct192 is unlinked from ct193 (Fig. 5A and B). We previously detected clearly measurable mRNA levels of ct059 in C. trachomatis ocular and urogenital strains (7). We confirmed this using the same biological samples in which the levels of ct058 mRNA were vestigial (data not shown). Therefore, we tentatively propose that expression of ct058 in ocular and urogenital strains might be downregulated by specific 3′-to-5′ posttranscriptional processing of the ct059-ct058 transcript.

To precisely define the promoter regions of ct059-ct058, ct192, and ct214, we determined their TSSs by RACE, using as the template total RNA of cells infected with C. trachomatis L2/434 and primers complementary to the ct059, ct192, or ct214 mRNA (Fig. 5A and C to E). When we used a primer complementary to the ct058 mRNA, we were unable to identify a ct058-exclusive TSS upstream from its start codon in L2/434 (data not shown). The TSS of ct214 matched the one previously identified by deep sequencing in strain L2b/UCH-1 (2), while the TSSs of ct059-ct058 and ct192 have not been identified before.

C. trachomatis encodes three σ factors: σ⁶⁶, the homolog of the E. coli main σ factor (σ⁷⁰); σ²⁸, a minor σ factor; and σ⁵⁴, an alternative σ factor (37). By inspecting the nucleotide sequences immediately upstream from the determined TSSs of ct059-ct058, ct192, and ct214 for σ⁶⁶-, σ⁵⁴-, and σ²⁸-like promoters (37), we identified only σ⁶⁶-like promoters (Fig. 5C to E; also, see Fig. S3 and S4 in the supplemental material). Finally, we analyzed the promoter regions of ct059-ct058, ct192, and ct214 and the ct059-ct058 transcript for LGV-specific nucleotide differences in all strains used in the RT-qPCR assays. We found LGV-specific nucleotides upstream from the predicted −35 region of ct059 (Fig. 5C; also, see Fig. S3). Furthermore, we found seven LGV-specific nucleotides scattered within the coding region of ct059 and between the stop codon of ct059 and the start codon of ct058 (Fig. 5C; also, see Fig. S3). In particular, three of these LGV-specific nucleotide differences were clustered 30 to 35 nucleotides upstream from the start codon of ct058 (Fig. 5C; also, see Fig. S3). However, it is unclear how these LGV-specific differences could explain the vestigial mRNA levels of ct058 in ocular and urogenital strains or have a discriminatory role in the proposed hypothetical degradation of the ct058 transcript in those strains. The scenario was simpler for ct192 and ct214, as we identified discrete LGV-specific nucleotides within the promoter regions of these genes that may explain their disease group-specific expression (Fig. 5D and E; also, see Fig. S4).

DISCUSSION

We found that amino acid differences between Inc proteins and distinct mRNA levels among inc genes throughout the developmental cycle of C. trachomatis strains correlate with the specific invasiveness and tropism of LGV isolates. Thus, we propose the novel hypothesis that a subset of Inc proteins may contribute to the specificity of infection by LGV strains. In fact, the vast majority of amino acid differences between Inc proteins are due to variations between proteins from LGV and trachoma biovars (Fig. 2). This could simply reflect the evolutionary history of C. trachomatis (25). However, most inc genes had d_N/d_S values of >1 among C. trachomatis strains, and according to the Z test of selection, 10 inc genes are likely under positive selection. In contrast, pmp genes or selected housekeeping genes of C. trachomatis mostly had d_N/d_S values of <1 (Fig. 3). Moreover, polymorphisms in C. trachomatis genomes are essentially driven by fixation of silent mutations (8). This suggests that the amino acid differences between Inc proteins should not be explained solely by genetic drift. In addition, almost all disease group-specific amino acids of Inc proteins encoded by genes likely under positive selection were found among proteins of LGV strains, and the majority of these amino acids localized in regions of the proteins predicted to face the host cell cytosol (Table 2). However, it must be clarified that other proteins in addition to Inc proteins are likely involved in the specificity of infection by LGV strains (8, 30). Moreover, the overall determinants of tissue tropism of C. trachomatis should be complex and multifactorial and should certainly also include, e.g., the products of the trpRBA operon (10, 19, 51) or of the cytotoxin gene (11) and Tarp (34).

We have done a focused and in-depth study of the variability of Inc proteins and evolution and expression of inc genes among C. trachomatis serovars, encompassing most of the available genomic information. We reveal that the overall degree of variation in the amino acid sequences of Inc proteins among strains is similar to that of a characteristic family of C. trachomatis polymorphic proteins (Pmps). Our results strengthen previous studies that suggested that some C. trachomatis Inc proteins could contribute to tissue tropism (8, 9, 35, 43) and confirm recent data suggesting that many inc genes could be under positive selection (8, 30). However, almost all of these previous studies analyzed a limited number of sequences, as the majority of the 51 genomic sequences used in our work became available only very recently (25).

Our findings also revealed that differential gene expression could be a mechanism contributing to the different invasiveness and tissue tropism of C. trachomatis strains. A previous work have identified differences in expression between pmp genes from reference and clinical strains (42), but disease group-specific differences in expression of C. trachomatis genes have not been noticed before. The inc genes ct058, ct192, and ct214 evidenced LGV-specific gene expression (Fig. 4), and we have further identified LGV-specific nucleotides in the promoter regions of ct192 and ct214 and within the ct059-ct058 transcript (Fig. 5). This was directly analyzed for the strains used in RT-qPCR assays, but the specificity is maintained within all 51 C. trachomatis genomes (see Table S1 in the supplemental material) that we used in our studies (data not shown). As only 3 of 48 inc genes showed LGV-specific gene expression, it is unlikely that this specificity is a common feature among C. trachomatis genes.

For ct214, we tentatively propose that the LGV-specific nucleotides could differentially affect its expression at either the transcriptional or posttranscriptional level. For example, a recent study showed that a Salmonella small noncoding RNA can discriminate mRNA regions that differ by a single nucleotide (44). The picture is more complex for ct058 and ct192. Regarding ct058, in L2/434 this gene is cotranscribed with ct059, and mRNA levels of ct059 were also detected previously in ocular and urogenital strains (7), which we confirmed (data not shown). We speculate that specific processing of the ct059-ct058 transcript selectively reduces the levels of ct058 mRNA in ocular and urogenital strains. Furthermore, ct058 is a pseudogene in many ocular strains (but not in those used in the RT-qPCR assays) (see Table S3 in the supplemental material; also data not shown), and ct192 is a pseudogene at least in the ocular strain B/Jali20 (see Table S3). This does not necessarily have an impact on the interpretation of the data, as for example the LGV-specific pseudogene ct300 showed similar mRNA levels during the developmental cycles of C/TW3, E/Bour, and L2/434 (see Table S6 in the supplemental material). Another issue is that there are different annotations in GenBank of the start codon of ct192 in the archetypal ocular (A/Har13), urogenital (D/UW3), and LGV (L2/434) strains (see Table S3). This alone could explain the differential expression of ct192, owing to dissimilar promoter regions. However, inspection of the nucleotide sequence immediately upstream from the annotated start codons of ct192 in each of these three strains reveals a strong putative ribosome binding site only in L2/434 (see Table S3). Therefore, it is likely that the start codon of ct192 is conserved between C. trachomatis strains and corresponds to the one annotated in L2/434. In this situation, ct192 is a pseudogene in all ocular strains (see Table S3), and the differences in the mRNA levels of ct192 between LGV and ocular or urogenital strains may be tentatively explained by the single-nucleotide differences found within its promoter region (Fig. 5D; also, see Fig. S4 in the supplemental material).

It is not possible to make a rigorous side-by-side comparison between our RT-qPCR data and previous analyses of inc gene expression by RT-PCR (52) and microarrays (5, 41), as the sensitivities of the methods used are quite different. In general, we have confirmed that inc genes display different profiles of expression but that they are mostly early-cycle genes, which supports the idea that many Inc proteins should play a role in modifying the inclusion membrane early during development (6, 61). We also found inc genes displaying an early- and late-cycle profile of expression. A similar profile has been observed in microarray studies (5), and it was suggested to result from “carryover” mRNA of highly expressed late genes from the previous infectious cycle. However, the first time point we analyzed was at 2 h postinfection, when carryover mRNA would have been degraded. Furthermore, at 30 or 42 h postinfection, several late-cycle inc genes showed levels of expression at least comparable to those of early- and late-cycle inc genes (see Table S6 in the supplemental material). This suggests that the expression of some inc genes could be induced early in the cycle, downregulated at midcycle, and induced again at late cycle.

Not all predicted C. trachomatis Inc proteins have been detected on the inclusion membrane. We addressed this by the analysis of T3S signals in 48 C. trachomatis Inc proteins. This led to the identification of 19 novel T3S substrates, including 14 putative Inc proteins (Fig. 1 and Table 1). In particular, and importantly, the putative Inc proteins CT058, CT192, and CT214 showed a T3S signal. Furthermore, all 10 inc genes likely under positive selection encode known Inc proteins. Our data for T3S signals in 14 Inc proteins that were previously analyzed using Shigella flexneri as the heterologous bacterium (16, 56) (Table 1) differed only for CT192 and CT850. This could be related to the distinct heterologous hosts used, or to the analysis of different N-terminal regions of CT192 (due to the different annotations of the start codon of ct192 in GenBank).

Although human macrophages have strong antimicrobial activity against C. trachomatis ocular and urogenital strains, they support the growth of LGV strains (63). We hypothesize that specific amino acids in Inc proteins, or their earlier and/or higher expression, could specifically enable LGV strains to inhibit phagolysosomal fusion in macrophages and/or prevent the formation of reactive oxygen or nitrogen species (22). Unfortunately, little is known about the function of the Inc proteins that we have identified as potentially involved in these processes. CT222 and CT118/IncG colocalize with kinases of the Src family in discrete regions of the inclusion membrane that associate with host cell centrosomes (38), and CT223 could inhibit host cell cytokinesis (3). However, it is unclear how this may relate to our hypothesis. On the other hand, Inc proteins that could manipulate intracellular membrane trafficking, such as those that have SNARE-like motifs (CT119/IncA, CT223, and CT813) (17) or interact with Rab GTPases (CT229) (46), are good candidates to selectively inhibit macrophage phagolysosomal fusion. Not all C. trachomatis Inc proteins tentatively proposed to be specifically involved in inhibiting phagolysosomal fusion have homologues in other chlamydial species, which also avoid this host cell degradation pathway (16, 34). It is possible that each chlamydial species evolved particular Inc proteins, and other virulence proteins, to account for the specificity of each type of cell it infects.