Discovering and Differentiating New and Emerging Clonal Populations of Chlamydia trachomatis with a Novel Shotgun Cell Culture Harvest Assay

This assay, coupled with ompA and 16S rRNA sequencing, characterized clonal populations of C. trachomatis.

Increasingly, isolates representing single clones are needed for in vitro and in vivo research, including genomic, murine, and translational studies, to advance our understanding of chlamydial pathogenesis. Although a few studies have described methods for segregating clones of laboratory-adapted C. trachomatis clinical and reference strains (12,15,16), none has clonally purifi ed all 19 C. trachomatis reference strains nor determined optimal methods to clonally segregate clinically mixed samples. Consequently, we modifi ed the plaque-forming assay of Matsumoto et al. (16) to segegrate clones from reference strains and developed a novel cell culture shotgun harvest assay to segregate viable clones from recent clinical samples because typical plaques do not form for most of these samples.
Our culture techniques coupled with outer membrane protein A (ompA ) and 16S rRNA sequencing identifi ed the constituents of mixed infections that represented new and emerging Chlamydiaceae strains and clonal variants in human disease. These results stress the importance of clonal isolation for these types of discoveries. Clonal isolates will also be essential for chlamydial research to ensure reproducibility of experiments among laboratories and to understand the dynamics of in vivo strain-mixing, evolution, and disease pathogenesis.

Plaque Assay for Reference Strains and Clinical Samples
We modifi ed the plaque assay of Matsumoto et al. (16) by using low speed centrifugation at 550 × g and 6-well plates for infections, and 1-dram shell vials (Kimble Chase Inc., Vineland, NJ, USA) for propagation. To ensure detection of mixed infections, 1:3 and 1:1 ratios of IFUs for reference strains E/Bour and D/UW-3, and a 1:1 ratio for clinical strains F and G, were created for inoculation and harvest.
At 48 h, duplicate plates with coverslips were fi xed with methanol and stained with a fl uorescein isothiocyanate (FITC)-conjugated C. trachomatis LPS-MAbs (18). Inclusions on each cover slip were counted to determine IFUs per milliliter per well and effi cacy of infection given the calculated IFUs inoculated for each well.
Plaques were visualized as a central area of cellular debris surrounded by viable infected cells with red staining of cytoplasm at the cell periphery ( Figure 1, panels A and C). Inclusion bodies and nonviable cells remained clear. Any plaque (≈1-2 mm) that was clearly isolated from another plaque, or appeared as a solitary plaque in a well, was se-  C. trachomatis D (7) C. trachomatis E (4) Acute clinical Ja 8 11 Ja (11) -C. trachomatis Ja (11) Acute clinical K 10 11 C. abortus (4) lected. A blunt-ended transfer pipette was used to punch a hole ≈1-2 mm in diameter through the gels over the plaque. The contents were placed into a microcentrifuge tube containing CMGH, sonicated and added to shell vials containing McCoy monolayers for propagation. Centrifugation of shell vials at 2,400 × g for 1 h at 35°C was required to successfully grow each clonally segregated strain. Strains were propagated and purifi ed using gradient ultracentrifugation as previously described (2,(18)(19)(20). The pellet was resuspended in SPG, and stored at -80°C.

Shotgun Harvest, Isolation, and Propagation of Single Clonal Populations for Clinical Strains
Because no visible plaques formed for the clinical strains, except for clinical H, the plates were inspected under 100× and 400× light microscopy. Wells were selected for our shotgun harvest as shown in the diagram (Figure 2). Ten spots per well were numbered where the infections were observed under microscopy. Each spot was harvested (≈2-3 wells × 10 spots per well = 20-30 harvests) using a sterile, blunt-ended transfer pipette.
IAO and FAO were carefully removed, and the wells were stained using FITC-conjugated C. trachomatis LPS-MAb (Virostat). Only harvested areas that corresponded to a confi ned group of infected cells with a clear margin from uninfected cells were selected, sonicated, inoculated, propagated in shell vials and fl asks purifi ed and stored as above. The original clinical samples were also independently propagated in shell vials as described previously (2,(18)(19)(20) for comparison with growth in the plaque assay.

Preparation of Genomic DNA and Sequencing of ompA and 16S rRNA for Each Clone
Purifi ed culture was used for genomic DNA extraction according to High Pure Template Preparation Kit package insert (Roche Diagnostics, Indianapolis, IN, USA). PCR was performed and reagents, thermocycling profi le, and sequencing were used according to previously described protocols (21). Table 2 (22) shows the primers used for PCR and sequencing to identify the strain-type of each clone. Multiple sequences were aligned by using MegAlign software (DNASTAR, Madison, WI, USA) and compared with public sequences (21,23). A variant was defi ned as having >1 nucleotide difference(s) from the sequence of the reference strain for either ompA or 16S rRNA genes.

Phylogenetic Construction of ompA Nucleotide and Amino Acid Sequence Alignments
Nucleotide and amino acid alignments and phylogenetic analyses of the 19 reference strains and clonal variants were performed by using MEGA 3.1 (Center for Evolutionary Functional Genomics, Tempe, AZ, USA) as described (21,23). Briefl y, neighbor-joining trees were calculated using the Kimura 2-parameter model that assumes that nucleotide frequencies and rates of substitution do not vary among sites. For amino acids, neighbor-joining trees were calculated using the gamma distance model that considers the dissimilarity of substitution rates among sites. We used bootstrap analysis (1,000 replicates) to determine confi dence intervals for each branch. Figure 1, panel A, shows typical plaque formation for F/IC-Cal3. Higher inocula resulted in plaques that fused and, therefore, were not suitable for harvest. These fi ndings are similar to those of others who have used plaque-or focus-forming assays for clonal segregation of laboratoryadapted chlamydial strains (16,24). Table 1 shows the day p.i. that plaques were visualized and the number of isolated clones and nucleotide polymorphisms with respect to reference strain sequences. All reference strains formed mature plaques ≈1-2 mm in diameter. Experimentally mixed infections of D/UW-3 and E/Bour resulted in 13 D, 9 E and 3 D/E clones, and 9 D and 0 E clones. The 3 D/E clones were identifi ed as mixed based on electropherograms where 2 peaks were observed in a single nucleotide position that corresponded to D and E sequences for ≈20 nucleotide positions. These 3 mixed infections were further plaque-purifi ed as above and yielded single clonal populations of D or E, which validated our plaque assay for isolating clonal populations.

Detection of Clonal Populations of C. trachomatis Clinical Strains by Shotgun Harvest
The clinical strains Ja, K, F, and G showed no plaques, while persistent strain K showed signs of <0.5-mm plaques at 10 days p.i., and strain H showed typical plaques at 7 days p.i. A longer growth period up to 20 days p.i. did not result in distinct plaque formation for Ja, K, F, or G. Notably, the size of the inclusion body was much smaller for clinical strains than for reference strains. Figure  1, panel E shows typical large inclusion bodies formed by reference F/IC-Cal3 compared with tiny and occasional medium-sized inclusion bodies at day 10 for persistent clinical strain F (Figure 1, panel F). Similar results were observed for persistent clinical strains G and H when compared with respective reference strains. In contrast, acute clinical strains Ja and K had inclusion bodies that were intermediate in size (data not shown). When original clinical samples were propagated in shell vials, inclusions remained small and the rate of growth was similar as for the plaque assay.
Mixed clinical G and F strains yielded 12 G clones (92.31%), 1 F clone (7.69%), and no mixed clones based on sequencing. Figure 3, panel A, represents 1 well after 10 random areas were harvested since no plaque was visible ( Figure 3, panel B). Figure 3, panel C, represents a microscope photo where chlamydial inclusions are diffi cult to visualize due to their small size. Figure 3, panel D, is a fl uorescent image of Figure 3, panel C, displaying small and medium-sized inclusion bodies.
The C. abortus-specifi c primers (Table 2) were used to amplify another seed stock of A/SA-1, C. abortus (from our plaque assay), C. trachomatis strain A/ Har-13, and C. caviae and C. muridanum. For the C. abortus-specifi c PCR amplifi cation, only A/SA-1 and C. abortus samples were positive, while the rest were negative.

Phylogenetic Analyses of ompA C. trachomatis Reference and Clonal Populations
Phylogenetics of ompA nucleotide and amino acid sequence alignments were performed to evaluate divergence of 5 clonal variants of Ba/Apache-2 and F/IC-Cal3. The trees showed the clustering of the 5 clonal variants with their respective parental strains (Figure 4, panels A and B).

Discussion
Plaque-and focus-forming assays have been developed to isolate individual clonal populations of reference strains of C. trachomatis (12,15,16) and Chlamydophila pneumoniae (24). The fi rst methods for C. trachomatis used L (15) and McCoy cells (16). More recently, fl ow cytometry has been used to segregate cells infected with C. trachomatis, C. caviae, and Chlamydia suis (12). However, these techniques have focused on laboratory-adapted clinical and reference strains and have not used nonpropagated or nonlaboratory-adapted clinical samples.
The novel shotgun harvest assay that we developed was successful in segregating clonal populations of C. trachomatis strains and variants that were devoid of plaqueforming characteristics. Consequently, our method is an important advance in reliably detecting and purifying clonal isolates from clinical samples. We also modifi ed the plaque protocol of Matsumoto et al. (16), which allowed us to use lower concentrations of reference strains, ensuring widely separated or single plaques. Most important, our methods showed sample collections that contained mixtures of new and emerging strains and variants based on ompA and 16S rRNA sequences.
The most remarkable mixed infection was for reference strain A/SA-1 in which C. abortus was identifi ed. C. abortus was not a likely contaminant because A/SA-1 was an original isolate. PCR of another seed stock of A/SA-1 was positive for C. abortus, and C. abortus had not previously been propagated in our laboratory. The original sample was obtained from the conjunctiva of a trachoma patient in Saudi Arabia in 1957. This fi nding was unexpected because C. abortus has not been described among trachoma-endemic populations. Although C. abortus may be responsible for zoonoses in pregnant women, it resides in a unique niche, the placenta, compared with C. trachomatis (26). Thus, an explanation for our fi ndings is that C. abortus is now capable of crossing species or niche barriers. Indeed, we recently identifi ed mixed conjunctival infections with C. trachomatis, Chlamydophila psittaci, and/or C. pneumoniae in 35% of infected persons residing in a trachoma-endemic region of Nepal (27). The fi ndings were statistically  unlikely to have occurred by chance. Additionally, infection with C. pneumoniae or C. psittaci was signifi cantly associated with trachomatous infl ammation, a precursor for scarring. With mounting evidence for widespread interstrain recombination among intracellular bacteria such as Chlamydiaceae (8,10,(21)(22)(23)28), the A/SA1 coinfection with C. abortus along with those described above are likely the tip of the iceberg in terms of the prevalence of mixed Chlamydiaceae infections and the possibility for recombination that may result in diverged tissue tropism (21,23). We are currently examining samples from other trachoma-endemic populations for coinfection with C. abortus and other Chlamydiaceae species.
Reference strain Ba/Apache-2 also comprised clonal populations of 3 previously unrecognized ompA genotypes, Ba 1 , Ba 2 and Ba 3 , that were distinct from publicly available Ba ompA sequences (6,7,29). The C662T mutation among our clones encoded a nonsynonymous P221L substitution in a constant region (CR) between variable segments (VSs) II and VSIII of MOMP, which contains 5 CRs and 4 VSs. This change from a proline, an imino amino acid with unique "kink," to a nonpolar leucine on CRIII might disrupt the mid-portion ß-strand transmembrane of MOMP (30)(31)(32). Furthermore, the E225K in Ba 2 occurs in VSIII where the subspecies-specifi c epitope for LGV and A-K strains (32) is located, likely changing polarity of the epitope from a negative to a positive charge. These mutations, then, may lead to adaptive structural and/or functional changes for MOMP.
The presence of mutations in Ba 1 , Ba 2 , and Ba 3 suggests that these have occurred under immune selection in vivo, because growing reference strains in vitro has not shown detectable mutations (3,14), although in theory this could occur. On the basis of phylogenetic reconstructions (Figure 4), the clonal variants likely represent natural diversity arising from the respective parental strain. Also, the ability of Ba strains to either mutate specifi c protein regions or recombine may facilitate their invasion of other mucosal sites. Urogenital Ba infections do occur, and we have previously described a Ba/D recombinant that was isolated from the genital tract (8).
Notably, most of the ompA mutations were located within CRs and encoded for nonsynonymous substitutions, the majority of which encoded for nonconservative amino acids with altered properties. For instance, ompA genotype F-III contains a nonconservative G90E substitution. G90E is located in CRII next to VSI, which may decrease membrane hydrophobicity and disrupt the >0.5 nonpolar or hydrophobic index requirement for the MOMP spanning region (32). In 2 separate studies, we identifi ed F-III variants as statistically signifi cantly associated with pelvic infl ammatory disease (PID) (5,33). The F-III mutation may explain, in part, the association with PID. However, additional studies will be required to delineate these associations. In our experimentally mixed infections, recovery of separate clones of D/UW-3 and E/Bour, and of clinical G and F validated each assay ( Table 1). The greater number of clones for D/UW-3 (52%) than for E/Bour (36%), and for G (92.31%) than for F (7.69%) might indicate different growth rates and timelines for plaque formation and characteristics of each strain (15,16). It is also possible that 1 strain produces byproducts of growth that are inhibitory for coinfecting strains. Nevertheless, these data emphasize the importance of selecting multiple wells of low inocula for plaque or shotgun harvests to identify all strains that are present. Additionally, mixed infections may occur where some strains cause plaque formation and others do not, which stresses the importance of the shotgun harvest even when the morphologic features of plaque are present.
In the present study, we analyzed clones by sequencing ompA, the plasmid, and 16S rRNA to enhance strain categorization. The plasmid was evaluated because its absence has been reported to correlate with reduced or no plaque formation (34). However, all of our clones contained the plasmid, which is consistent with other studies (35)(36)(37). The lack of classic plaque formation for clinical isolates likely stems from their slow growth and lack of adaptation to conventional cell culture. This was borne out by their slow growth in shell vials and fl asks, experiments which were performed separately from the plaque assay. Clinical strains may exit the cell without cellular disruption, facilitating subsequent rounds of infection and lack of plaque formation. Beatty recently showed,that EBs could be released without lysis and also be retained by host cells (38). However, our clinical H formed plaques similar in morphology to reference strains. The presence of a complete toxin gene, as in H/UW-4 and J/UW-36 (39), may have contributed to clinical H plaque formation. H/UW-4 has been shown to produce more cytotoxicity than D/UW-3, which contains a partial toxin gene, and C. muridanum, which contains a full-length gene (40). Although all 19 reference strains formed classic plaques morphology, some have no toxin (LGV strains) or a partial gene, which suggests that plaque formation refl ects adaptation to culture that has occurred over decades instead of the effects of the toxin. Further experiments will be required to determine the genetic factors involved in plaque formation.