INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Chlamydia trachomatis causes many human diseases, including ocular trachoma (serotypes A to C); urethritis, epididymitis, cervicitis, and salpingitis (serotypes D to K); and lymphogranuloma venereum (LGV; serotypes L1 to L3) (41-43). Serotypes A to K (trachoma biovar) primarily infect columnar epithelial cells of the mucous membranes, while serotypes L1 to L3 (LGV biovar) also proliferate in lymphatic tissue and cause a more systemic infection. The third biovar of C. trachomatis is the murine biovar, consisting of one strain, MoPn, which causes mouse pneumonitis but does not infect humans. In addition, three serogroups among the human biovars that appear to be independent of biovar have been defined: the B complex (serotypes B, Ba, D, Da, E, L1, L2, and L2a), the C complex (serotypes A, C, H, I, Ia, J, K, and L3), and the intermediate group (serotypes F and G) (9, 42, 44).

Serotype specificity is conferred by the major outer membrane protein (MOMP; the product of the omp1 gene). MOMP constitutes ~60% of the protein mass of the chlamydial outer membrane and has been shown to have porin-like characteristics in vitro (5). MOMP is thought to play a role in the structural integrity of the organism (7, 8, 19) and is surface exposed and glycosylated (2, 10, 25, 39). MOMP contains 7 to 10 cysteines which may form homo- or hetero-oligomers with itself and/or other outer membrane proteins (28, 30).

The amino acid sequence of MOMP exhibits heterogeneity that is mainly localized to four hypervariable segments (VS1 to VS4) (44) which are surface exposed and reactive with human immune sera (45, 46). Monoclonal antibodies directed against MOMP are neutralizing in cell culture and in some animal models (2, 27, 31, 45, 46). Although limited, protective immunity is serovar specific, making MOMP a focus of vaccine development.

The outer membrane proteins of obligate intracellular bacteria play a direct role in the process of adaptation by facilitating interactions between the bacterial cell and its host cell. The surface of the chlamydial elementary body must provide components responsible for (i) protection against the environment outside of the host, (ii) defense against host immune response, (iii) attachment to host cells, and (iv) prevention of phagosome-lysosome fusion. Variability in MOMP sequence is presumably a result of host selection and bacterial adaptation. Thus, MOMP has been implicated in the mechanisms of attachment, infection, and/or pathogenesis because of its variability and its exposed location. Consequently, an evolutionary examination of the omp1 gene and the MOMP may provide insight into the role of MOMP in the processes of infection.

Fitch et al. (16) examined the omp1 gene by sequencing VS regions of 15 serotypes (17 strains) and the complete sequence of 9 serotypes (11 strains). The authors concluded that there was no evidence for coevolution of human strains of C. trachomatis and the human host. However, they did not examine MOMP at the amino acid level, and their study was limited to one or two strains from each serovar. The present study combines our omp1 sequences (40 strains from 11 serovars) with omp1 sequences available in GenBank for an analysis encompassing 69 strains from 17 serovars. Our goal was to examine the patterns of substitution in MOMP at both the nucleic acid level (omp1) and the amino acid level (MOMP) to identify residues that may be correlated with serovar-specific tissue tropism or pathogenic potential by virtue of shared evolutionary divergence.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains. Sequences of the MOMPs derived from strains used in this study were analyzed along with sequences from strains available in GenBank (Table 1). Several were epidemiologically linked by being isolates from individuals in a sexual partnership or isolates from the same individual at different times.

Cell culture. Clinical swab specimens were collected and expanded in cell culture as previously described (24). To identify inclusions, immunofluorescence staining was done by using a genus-reactive monoclonal antibody (23). Expansion of each isolate was done via serial passage on glass coverslips in shell vials by using antibiotic-free medium (24).

DNA isolation from cultured cells. All tips, tubes, and buffers were UV irradiated to reduce contaminating DNA. Infected McCoy cell monolayers were disrupted and transferred to sterile 2-ml tubes. One hundred to 200 µl of the cell suspension was centrifuged for 5 min, the supernatant was removed, and the cell pellet was resuspended in 500 µl of sodium dodecyl sulfate (SDS) lysis buffer (0.5 M Tris-EDTA [pH 7.5], 1% SDS, 0.1 mg of proteinase K per ml). The suspension was incubated at 55°C for 2 h and at 95°C for 10 min. The resulting lysate was extracted with an equal volume of phenol-chloroform. DNA was precipitated in an equal volume of isopropanol, washed one time with 70% ethanol, dried, and resuspended in 50 µl of UV-irradiated, double-distilled water.

Preparation of urine specimens. First-void urine was collected in sterile cups and stored immediately at 4°C until processed. Precipitates were resuspended by warming the urine at 37°C for 15 to 30 min before processing. Two milliliters of urine was microcentrifuged at high speed (12,000 × g) for 10 min. The supernatant was discarded, and 1 ml of transport medium was used to wash the pellet. The cells were centrifuged again, for 5 min at high speed (12,000 × g). The supernatant was discarded, and the pellet was resuspended in 200 µl of SDS lysis buffer. DNA extraction and isolation were carried out as described above.

DNA amplification. Primers specific to the C. trachomatis MOMP were based on published sequences (Table 1). The 5' primer, MOMP-108, (5'-GGC CAT TAA TTG CTA CAG GAC ATC TTG TC-3') is located 108 bp upstream of the serovar A omp1 gene in the 5' noncoding region. The 3' primer, RVS1163, (5'-CGG AAT TGT GCA TTT ACG TGA G-3') is located at bp 1163 in the serovar A omp1 gene (11). For samples which required reamplification, the following nested primers were designed: MOMP87 (5'-TGA ACC AAG CCT TAT GAT CGA CGG A-3') and RVS1059 (5'-GCA ATA CCG CAA GAT TTT CTA GAT TTC ATC-3').

DNA sequencing. Direct sequencing of PCR products was done by using the BRL dsCycle sequencing kit (Life Technologies, Gaithersburg, Md.) with Isoblue [-³²P]ATP (ICN, Costa Mesa, Calif.)-end-labeled sequencing primers. Reaction products were separated on a 6 to 8% polyacrylamide gel and visualized by autoradiography (Kodak BioMax XR). PCR products were also cloned with the TA cloning kit (pCR 2.1; Invitrogen, La Jolla, Calif.). To ensure sequence fidelity and to resolve ambiguities, sequence was obtained from at least two individual clones and compared to sequence obtained directly from PCR products. Therefore, each gene was sequenced in part at least three times. Several PCR products were also sequenced by using the ABI Prism 373 (Applied Biosystems) automated sequencing system employing dye terminators. Sequence obtained in this way was also collected from both strands to ensure sequence fidelity.

Nucleotide and amino acid sequence alignment. Both nucleotide and amino acid sequences were aligned with the assistance of the program ESEE (6). For regions of high variability, the alignment was based primarily on the most conservative amino acid substitutions and secondarily on nucleotide substitutions. The entire omp1 sequence was not obtained for nine isolates for which nested reamplifications were done because of the intragene location of the nested primer pair. These sequences and partial sequences from GenBank were compared to other sequences but were not included in the phylogenetic analysis. Consequently, the full sequence alignment contained 29 strains and, for each strain, 1,165 characters (nucleotides plus alignment gaps) and 388 characters (amino acids plus alignment gaps). The alignment consisted of the entire omp1 gene sequence except for the last 26 bases (and corresponding eight amino acids) which overlap the RVS1163 primer sequence and the last four nucleotides of the gene, not including the stop codon. This region is known for some of the serovars but not all and was excluded from the phylogenetic analysis.

Nucleotide and amino acid comparisons. After sequences were aligned, full-length nucleotide and amino acid sequences were compared manually in a pairwise fashion for differences. Contiguous insertions or deletions (IN/DEL) at the nucleotide and amino acid levels were counted as individual differences, i.e., a deletion of six nucleotides between two pairs was counted as six differences rather than one deletion event, and the resulting two-amino-acid deletion was counted as two deletions rather than as one deletion event. Pairwise comparisons were also made for strains with partial sequences, and differences were noted but not tallied.

Phylogenetic analysis. Residues corresponding to the RVS1163 primer and all regions of ambiguous alignment (gap IN/DEL positions) were excluded from phylogenetic analysis. Only the coding region was considered. Consequently, 1,128 of 1,165 nucleotides and 376 of 388 amino acids were compared. All phylogenetic reconstructions were carried out by using the Phylip package version 3.5c (14). Once the sequences were aligned, the corrected nucleotide divergence of them was calculated by using DNADIST, incorporating the Kimura two-parameter correction for multiple substitutions at a single site. The branching pattern of taxa was estimated from the distance matrix by neighbor-joining (34) by using the program NEIGHBOR. A cladistic reconstruction of the phylogenetic relationships among taxa was also performed by using the parsimony program DNAPARS. Phylogenetic reconstructions based on the inferred amino acid sequence of MOMP were also performed, by using PROTDIST, with the Kimura formula, and PROTPARS, with threshold parsimony. Confidence levels for the branching pattern were estimated by a bootstrap resampling of the data. Bootstrap values for the trees were obtained from a consensus tree based on 1,000 randomly generated data sets by using SEQBOOT with jumbled sequence addition. Analyses were performed on an IRIX Indigo (Silicon Graphics) UNIX workstation. The mouse pneumonitis strain, MoPn, was used as the outgroup in all analyses to place the root joining the human strains of C. trachomatis.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Sequence comparisons. The nucleotide and amino acid sequence comparisons for full-length sequences are summarized in Table 2. All sequence alignments are available upon request (to D.R.S.). Several strains had identical sequences (Table 1).

Phylogenetic analysis of the omp1 gene. Figure 1 shows the distance neighbor-joining tree based on the nucleotide sequences of the omp1 gene with consensus bootstrap values indicated at the nodes to which they apply. Bootstrap values lower than 50 are considered insignificant and are not shown. The most ambiguous regions of the tree are also those regions which join closely related taxa, namely, C-complex strains (Fig. 1). Both parsimony and neighbor-joining analyses produced similar trees, with some discrepancy in branching order of C-complex strains. Both analyses produced trees which segregated the serovars of C. trachomatis into three clades consistent with serocomplexes. Parsimony analysis of the omp1 produced eight equally parsimonious trees (data not shown). Five of these trees differed in the branching order of D isolates. The other two trees differed in the branching order of strains within the C complex.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   The distance neighbor-joining tree based on the C. trachomatis omp1 nucleotide sequence from 29 strains. The tree was rooted by using the C. trachomatis MoPn strain. Branch lengths are proportional to the amount of sequence divergence between taxa in the tree, as illustrated by the bar. Bootstrap values, based on 1,000 bootstrap replicates, are given at the nodes to which they apply. Bootstrap values below 50 are considered insignificant and are not shown.

Phylogenetic analysis of the MOMP. Figure 2 presents the consensus neighbor-joining tree based on the amino acid sequences of the MOMP with bootstrap values at the nodes to which they apply. Parsimony analysis of the protein data produced 42 equally parsimonious trees, all of which indicated an early divergence of serovars F and G, prior to the divergence of the B- and C-complex strains. This branching pattern also agreed with the consensus neighbor-joining tree. However, the neighbor-joining analysis of the MOMP data set (Fig. 2) produced a tree where the B- and C-complex strains diverged first, prior to the divergence of the B-complex strains, similar to the nucleotide tree.

View larger version (25K): [in this window] [in a new window]   FIG. 2.   The distance neighbor-joining tree based on the C. trachomatis MOMP amino acid sequence for 29 isolates. The tree was rooted by using the C. trachomatis MoPn strain. Branch lengths are proportional to the amount of amino acid sequence divergence between taxa in the tree, as illustrated by the bar. Bootstrap values, based on 1,000 bootstrap replicates, are given at the nodes to which they apply. Bootstrap values below 50 are considered insignificant and are not shown.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Lampe et al. (26) recently showed that a MOMP variant could escape neutralization by both serovar-specific monoclonal antibodies and human immune sera. If this adaptive process occurs, the patterns of substitution in MOMP among serovars should be reflective of the biological relationships each serovar has with host cells. However, a detailed examination of amino acid and nucleotide substitutions in sequences of MOMP and omp1, respectively, shows no evolutionary pattern with respect to the biological relationships among human strains of C. trachomatis.

There is a lack of correlation between disease phenotype and the omp1 and MOMP phylogenies. Serovars L1, L2, and L3 are responsible for LGV, a disease that is distinct from the urogenital and ocular infections associated with serovars A to K. Yet, there is no pattern in the omp1 or MOMP phylogenies that groups the LGV serovars distinctly from the urogenital serovars. Likewise, serovars A, B, Ba, and C are most commonly associated with ocular trachoma. While serovars A and C branch within the C-complex clade, they have no specific association with each other. In the nucleotide tree (Fig. 1), they do appear to be distinct from the rest of the C complex, as they are among the first to diverge and have many unique substitutions. However, the substitutions do not appear to be unique to the trachoma biovar since they are not shared by serovars A and C.

If the omp1 and MOMP phylogenies were reflective of disease, then serovars B and Ba, which are associated with both trachoma and urogenital disease, should be intermediate between the trachoma serovars and the urogenital serovars. However, this is not the case. An examination of ocular and urogenital isolates of serovars Ba and C by Frost et al. (17) showed eight nucleotide and two amino acid differences in conserved segment 1 of MOMP (CS1) between sequences of urogenital Ba isolates (named serovar Bb) and those of ocular Ba isolates. These Bb isolates differed from the prototype D/B120 and H/UW4 in the CS1 by only one base, possibly indicating "urogenital-specific" substitutions in CS1. We have also identified this genotype in the present study, for isolate B/IU-1226. It differs from serovar D and H genotypes by one base in CS1, but it is identical to the serovar Ba and B genotypes throughout the rest of the gene. Therefore, if urogenital-specific substitutions are present, they must be located in CS1 and not in a variable region. A more comprehensive study of serovar B ocular and urogenital genotypes would be necessary in order to determine whether a consistent difference in genotype occurs between urogenital and ocular isolates of serovar B.

There is lack of concordance between host cell niche (tissue tropism) and omp1 genotype. Serovars A to K (trachoma biovar) infect epithelial cells, while serovars L1 to L3 (LGV biovar) infect epithelial and lymphatic tissue. Therefore, their intracellular niches are different. However, this difference is not reflected in the phylogeny of the omp1 and MOMP. LGV, trachoma, and urogenital serovars are intermixed in both the omp1 and MOMP phylogenies. This would suggest that there are no unique amino acid residues in the MOMP that have evolved to recognize and/or bind to tissue-specific receptors. What leads an LGV, a trachoma, or a urogenital strain to infect a certain subset of cells cannot be identified in the MOMP. Batteiger et al. (3) and Allen et al. (1) have identified differences in the omp2- and omp3-encoded membrane proteins between LGV and trachoma strains. Therefore, these proteins could be responsible for some of the biological differences between urogenital/trachoma and the LGV strains.

Our data reveal a lack of correlation between success (as measured by prevalence in the population) and genotype. Serovars D, E, and F are the most common serovars found in cases of urogenital infection; therefore, they are the most successful. However, if MOMP residues are responsible for success and consequent fecundity, it is not reflected in the omp1 and MOMP phylogenies. While variants have been described for serovars D and F, few are described for serovar E, and the substitutions are frequently synonymous. Herein, we sequenced 19 isolates of serovar E and found only 3 to be variant, containing one nonsynonymous substitution, and the three variants were linked epidemiologically.

Conversely, Dean and Millman (12) found that 16% of their serovar E population was variant, reflected by 11 variant genotypes out of 67 isolates. However, their E/Bour MOMP sequence (E/Bour 1997) also was different from that described by Peterson et al. (E/Bour 1990) (32) by 10 nucleotides corresponding to four amino acids. It also differed from the sequence obtained from our own stock of E/Bour. The source of the differences may be due to the fact that the E/Bour strain has been propagated in culture for many years in different labs and may have accumulated mutations during this time. However, this explanation is unlikely, because it would require a very high rate of mutation in the omp1 gene in the absence of immune selection. Hayes and coworkers (22) found some flux in the prevalence of alleles observed in a trachoma-endemic population over time, but the overall distribution remained constant. Fitch et al. (16) plotted evolutionary distance against year of isolation and found there was no measurable rate of evolution for human C. trachomatis isolates, and their calculations were biased by including only the variable regions. We have investigated the possibility of cell culture-induced mutation by continuously propagating nine clinical isolates for 20 passages in cell culture and then comparing the MOMP sequence of the original isolate to the one obtained after the 20th passage in culture. We found no differences (38a). Therefore, the variation in E/Bour reported by Dean and Millman (12) is not easily explained.

A successful phenotype, as displayed by serovars D, E, and F, may be one in which a less vigorous immune response is elicited by the infecting serovar. In this way, the chlamydial organism could go undetected, untreated, and further disseminated. Few studies on the severity of disease as it relates to specific serovars, have been done, and the results have been contradictory. Dean et al. (11) looked at serovars F and G in lower and upper genital tract infections. They found that F variants produced a more severe disease phenotype than the prototype F genotype does. Batteiger et al. (4) examined 224 cases of urethritis and cervicitis and found a weak correlation between inflammation and serovars F and G but only in men. Overall, they found no association between inflammation and infecting serovar. Van de Laar et al. (40) examined 275 cases and found less obvious disease in men infected with serovars F and G but no differences in women. Therefore, there is no evidence to suggest that production of a more mild disease is a strategy employed by the more prevalent serovars.

Variation within the omp1 is expressed as nucleotide substitutions as well as IN/DEL events. IN/DEL events are consistent with serocomplexes: the B-serocomplex genes are smallest, followed by serovar F and G genes, followed by C-serocomplex genes. Therefore, the variation in the omp1 gene between more closely related serovars occurs by a base substitution mechanism while more distantly related serovars differ by base substitutions and IN/DEL mutations. However, there is no pattern to the IN/DEL events that corresponds to disease or tissue tropism. Therefore, the presence or absence of specific amino acid residues in the corresponding protein sequence does not appear to be responsible for tissue-specific invasiveness, enhanced transmission, or disease phenotype. In addition, there are no disease-specific or tissue-specific amino acid substitutions that are held in common between related serovars. In other words, there are no synapomorphic (derived and shared) characters that are unique and specific to eye isolates or cervix-urethral isolates or LGV isolates. The only synapomorphic mutations are two LGV-specific substitutions in CS1. However, neither mutation leads to an LGV-specific amino acid substitution at the site. Therefore, it is highly unlikely that substitutions have occurred at these two sites in the LGV strain sequences as a result of selective forces that are specific to the LGV phenotype.

There are eight conserved cysteine residues in the MOMP sequence, all or some of which may be involved in disulfide tertiary or quaternary interactions. However, there were exceptions; most notably, the E/Bour 1997 and E variant isolates from San Francisco (12) had a serine residue at position 207, while all other strains including MoPn, had a cysteine. This finding may be relevant in that all of the E variants described in that study, except the E/Bour 1997 strain, are geographically linked and may be epidemiologically linked as well. The lack of the cysteine residue in these variants may lead to a conformational change in the protein that enhances infectivity or transmission of serovar E in the local (San Francisco) population. However, since detailed clinical or epidemiologic information was not provided for these strains (12), the significance of the substitution is unknown.

In conclusion, while MOMP is highly variable in sequence, suggesting the use of an adaptive mode of evolution to escape immune pressures, one of the most successful MOMP serovars, serovar E, does not appear to be highly variable and thus would seem to contradict the use of this tactic by C. trachomatis. In addition, there is a lack of concordance between omp1 and MOMP phylogenies and disease phenotype, tissue tropism, or epidemiological success. These observations suggest that the role of MOMP in infection may be passive and that it may not be the major ligand responsible for directing infection of various human cell types. This does not mean that MOMP cannot induce protective immunity in humans. However, immune responses against MOMP may confer weak and short-lived immunity, perhaps because macromolecules other than MOMP may be fundamental determinants of cell tropism, disease phenotype, and epidemiological fitness.