The genetic basis of plasmid tropism between Chlamydia trachomatis and Chlamydia muridarum.

The development of genetic transformation technology for Chlamydia trachomatis using its endogenous plasmid has recently been described. Chlamydia muridarum cannot be transformed by the C. trachomatis plasmid, indicating a barrier between chlamydial species. To determine which regions of the plasmid conferred the species specificity, we used the novel approach of transforming wild-type C. muridarum carrying the endogenous plasmid pNigg and forced recombination with the C. trachomatis vector pGFP::SW2 which carries the complete C. trachomatis plasmid (pSW2). Penicillin and chloramphenicol-resistant transformants expressing the green fluorescent protein were selected. Recovery of plasmids from these transformants showed they were recombinants. The differences between the pSW2 and pNigg allowed identification of the recombination breakpoints and showed that pGFP::SW2 had exchanged a ~ 1 kbp region with pNigg covering CDS 2. The recombinant plasmid (pSW2NiggCDS2) is maintained under antibiotic selection when transformed into plasmid-cured C. muridarum. The ability to select for recombinants in C. muridarum shows that the barrier is not at transformation, but at the level of plasmid replication or maintenance. Our studies show that CDS 2, together with adjoining sequences, is the main determinant of plasmid tropism.

The newly developed technique of genetic transformation for Chlamydia is set to significantly change the experimental approach to understanding more about this important pathogen. This manuscript adds critical new information in relation to the barriers to genetic transformation and shows, quite unexpectedly, that the cross-species barriers are actually replication-mediated tropisms, rather than transformation per se.

There are nine recognised species of Chlamydia each with distinctive biological properties such as a specific tissue tropism and disease pathology. Some new candidate species have recently been added to this list (Sachse et al., 2014). However, all the chlamydial species share the property of intracellular parasitism and grow within a modified cytoplasmic organelle (known as an inclusion) (Stephens et al., 2009). Chlamydia have a unique, biphasic developmental cycle alternating between an infectious elementary body (EB) and a replicating metabolically active form, the reticulate body (RB) (Ward, 1983). Chlamydia trachomatis is the leading worldwide infectious cause of blindness (trachoma), and genital chlamydial infection is the commonest diagnosed bacterial sexually transmitted infection (STI) in the Western world (Thylefors et al., 1995; Burstein & Zenilman, 1999). Infectious model systems have been set up with various animals (including nonhuman primates) using C. trachomatis to mimic human infections, but none of these accurately reflect natural human disease (Miyairi et al., 2010). A separate species Chlamydia muridarum causes respiratory tract infection in rodents and is the most well-studied, homologous (pathogen/host) small animal infection system. Thus, there is a great deal of interest in translating findings from the C. muridarum systems to understanding human genital and eye disease caused by C. trachomatis.

Almost all strains of C. trachomatis carry an endogenous 7500 bp plasmid; only four viable naturally occurring plasmid-free clinical isolates have been described; thus, these are exceedingly rare (Peterson et al., 1990; Farencena et al., 1997; Stothard et al., 1998; Wang et al., 2013a). Studies from plasmid-cured (where the plasmid has been physically removed by chemical agents) and naturally occurring plasmid-free chlamydia have shown that the presence of the plasmid is associated with the ability to accumulate glycogen, TLR2 activation and infectivity (O'Connell & Nicks, 2006; O'Connell et al., 2011; Russell et al., 2011). In vivo studies using a murine model and naturally occurring plasmid-deficient human genital C. trachomatis have shown that the plasmid is a virulence factor (Sigar et al., 2014). Preliminary experiments, with limited numbers of subjects, have indicated that a plasmid-cured human trachoma isolate of C. trachomatis is avirulent in the monkey eye and that this plasmid-cured isolate can elicit protective immune responses against the wild-type, plasmid-bearing strain (Kari et al., 2011).

Taken together, these studies indicate that the plasmid is a key determinant of virulence in both C. trachomatis and C. muridarum, and thus, there is renewed interest in understanding both its biochemical function and biological role. Eight major coding sequences (CDS) (> 100 bases) have been assigned to the chlamydial plasmid (Thomas et al., 1997). Antibodies specific to the predicted protein products encoded by each coding sequence (CDS) have been used to investigate the expression profile of these genes (Li et al., 2008). All are expressed during the developmental cycle, and the protein ‘pgp3’ encoded by CDS 5 is secreted beyond the inclusion and into the cell cytoplasm.

Recently, we developed a plasmid-based gene transfer system for C. trachomatis (Wang et al., 2011). This has proved useful in preliminary studies aimed at defining the biological function of several plasmid coding sequences (CDSs) and their protein products. Studies using natural mutants and employing this technology to make simple gene deletion and/or gene inactivations have shown that several plasmid genes are essential and likely have a role in plasmid maintenance. A focus of work on potential virulence factors has been the protein pgp4 (encoded by CDS 6), which is proposed as a ‘transcriptional regulator’, and the proteins encoded by CDS 7 (pgp5) and CDS 5 (pgp3) genes are dispensable for growth in vitro (Gong et al., 2013; Song et al., 2013; Wang et al., 2013a). Transformation studies have now been extended to C. muridarum (Song et al., 2014). Chlamydia muridarum cannot be transformed by a C. trachomatis plasmid and vice versa, and this has been cited as an example of ‘transformation tropism’ (Song et al., 2014). The genetic basis for this plasmid-mediated, apparent tropism is unclear. To investigate the barriers to plasmid transformation between these chlamydial species, we cured C. muridarum (strain Nigg) of its plasmid using novobiocin as described previously (O'Connell & Nicks, 2006). The plasmid-free C. muridarum, designated C. muridarum Nigg P- (Plasmid minus), was purified by three rounds of plaque purification and confirmed to be plasmid free by PCR (data not shown). Repeated attempts to transform C. muridarum Nigg P- as the recipient host with the C. trachomatis/Escherichia coli plasmid shuttle vector pGFP::SW2 were unsuccessful in our hands, consistent with previous observations (Song et al., 2014).

To attempt to generate recombinants of pGFP::SW2 in vivo which can replicate in C. muridarum, we transformed wild-type, plasmid-bearing C. muridarum Nigg P+ with the C. trachomatis plasmid vector pGFP::SW2 (Wang et al., 2011). This encodes the GFP, bla and cat genes allowing expression of the green fluorescent protein and conferring resistance to penicillin or chloramphenicol, respectively. The transformation protocol was as previously described (Wang et al., 2011) except prolonged passage under penicillin (10 units mL−1) selection was used. After more than 2 weeks under penicillin selection (four passages), resistant inclusions emerged, but only a small portion of inclusions expressed green fluorescence. At this point, chloramphenicol (0.4 μg mL−1) was applied for selection as recently described (Xu et al., 2013), and after four rounds of chloramphenicol selection, almost all inclusions fluoresced green. Green fluorescent, penicillin and chloramphenicol-resistant C. muridarum were expanded by multiple passaging and a stock of bacteria produced. Whole DNA extracted from these C. muridarum transformants was used to transform E. coli to rescue ampicillin-resistant plasmids from the transformed C. muridarum. A total of 52 colonies expressing the green fluorescent protein were selected for plasmid DNA extraction. Fifty of the clones displayed the same Bgl II and/or Sal I restriction endonuclease digestion patterns. One of these was selected for sequence analysis and named pSW2NiggCDS2 (Fig.1a).

Figure 1

The structure of the recombinant plasmid pSW2NiggCDS2. (a) The whole plasmid map. (b) The CDS 2 region and flanking features. (c) The alignment of the sequence (in the region shown in b) with plasmids pL2, pSW2 and pNigg (GenBank accession numbers in brackets). The sequence from pNigg is represented by the green line, and the areas where recombination occurred are indicated by red boxes. Pas, antisense promoter; PCDS2, CDS2 promoter; 44 bp, the characteristic 44 bp tandem repeat in pSW2. The start codon for CDS 2 and the stop codons of CDS 1 and CDS 2 are underlined in all the sequences. The promoters (Pas and PCDS2) are underlined only in the pL2 sequence. The −35, −10 and tsp features are highlighted in BOLD in pL2 and pSW2 (Ricci et al., 1995).

Plasmid pSW2NiggCDS2 is 11 536 bp in size. Sequencing showed that pSW2NiggCDS2 carries a minimum of 1055 bp originating from the plasmid pNigg replacing the equivalent 1058 bp from pGFP::SW2 (it is not possible to map the exact recombinational breakpoint as the sequences are identical for short stretches, see Fig.1c). Thus, for the productive replication of plasmid pGFP::SW2 in C. muridarum, the C. trachomatis progenitor plasmid's entire CDS 2, with some short flanking regions was replaced by the orthologous region from the C. muridarum plasmid pNigg. The replacement pNigg sequence started from the antisense promoter (Pas) and ended at ∼ 20 bp before the unique 2 × 44 bp repeat in pSW2, and included the complete coding sequence for CDS 2. The recombinational breakpoints are located within the 26 bp identical sequences covering the CDS 1 stop codon and the 27 bp (or 30 bp) identical sequences covering the start of the first characteristic 44 bp tandem repeat, which is immediately upstream of CDS 3 (Fig.1b and c). To confirm whether the other clones were identical, we chose a further five clones and sequenced the CDS 2 region and found they were all exactly the same as pSW2NiggCDS2.

Our data build on recent work (Song et al., 2014) who reported that C. trachomatis plasmids could not be used to transform C. muridarum. Our results are consistent with these findings in which no transformants were observed when we attempted to transform C. muridarum Nigg P- with plasmid pGFP::SW2. For experimental rigour, it was necessary to show that recombinant C. trachomatis plasmid pGFP::SW2 with the replacement of the C. muridarum CDS 2 region (i.e. pSW2NiggCDS2) replicates in a C. muridarum background. Thus, we transformed plasmid-free C. muridarum Nigg P- with pSW2NiggCDS2. The transformants were rapidly recovered (within a week) under 10 units mL−1 of penicillin selection, and green fluorescent inclusions were visible at passage 3. Transformants were propagated for ten passages with penicillin selection. From this stock, transformants were grown with or without penicillin for a further five passages, and there were always some ‘nonfluorescent’ inclusions under both conditions. Figure2c and d show these transformants grown without penicillin, some of inclusions display the characteristic empty-centre or bull's eye phenotype under phase contrast microscopy (compare morphology with the parental host in Fig.2b), these inclusions do not fluoresce green. Figure2e and f show the transformants under constant penicillin selection; aberrant inclusions characteristic of sensitivity to penicillin through retardation of the developmental cycle occur (arrowed in Fig.2e); these inclusions do not fluoresce green. These results indicate that the loss of the transforming plasmid occurred in the presence or absence of selection with penicillin.

Figure 2

Images of Chlamydia muridarum infected McCoy cells at 28hpi. (a) Wild-type, plasmid-bearing C. muridarum (Nigg P+); the arrow shows a typical inclusion. (b) Plasmid-cured C. muridarum Nigg strain (Nigg P-, without plasmid); the arrow shows an inclusion with the distinctive plasmid-free bull's eye phenotype (i.e. ‘bright hole’ in the centre). (c–f) Transformants of C. muridarum Nigg P- with plasmid pSW2NiggCDS2, either without penicillin selection (c & d, same field), or under continuous penicillin selection (e & f, same field). When the transformants were grown without penicillin selection, some inclusions displayed the distinctive Nigg P- phenotype (an example is arrowed, (c), and they did not fluoresce green (d, arrowed). When the transformants were grown under penicillin selection, some inclusions displayed the distinctive aberrant inclusion phenotype of penicillin sensitivity (an example is arrowed in (e), and they did not fluoresce green (f, arrowed). (Scale bar = 20 μm)

The ability to select for recombinants in C. muridarum shows conclusively that the barrier is not at the point of transformation, but at the level of plasmid replication or maintenance. Our studies show that CDS 2, together with adjoining sequences (see Fig.1c), are the main determinants of plasmid tropism between C. trachomatis and C. muridarum. The low level of spontaneous loss of the plasmid pSW2NiggCDS2 on passaging suggests that these recombinants are less stable than the wild-type pNigg plasmid in C. muridarum, indicating other factors are also likely to contribute to plasmid retention, although CDS 2 is the main factor. Spontaneous plasmid loss occurs in C. trachomatis during culture (Matsumoto et al., 1998); this phenomenon has not been studied in detail in C. muridarum.

The cross-species barriers are actually replication-mediated tropisms rather than transformation tropism per se. Regardless, their presence has been confirmed by our results. By contrast to the report of ‘transformation tropism’ between C. trachomatis biovars (Song et al., 2014), our shuttle vector pGFP::SW2 (a genital tract based plasmid) can be used to transform LGV biovar isolates. Looking at the sequence of CDS 2 across the C. trachomatis biovars, it is evident that this is the most highly conserved gene of all the plasmid sequences, with only a single nonsynonymous SNP between them (Seth-Smith et al., 2009). Moreover, within the plasmid of C. trachomatis, the protein encoded by CDS 2 is not required for plasmid replication (Gong et al., 2013). Given the very high levels of sequence conservation, it seems unlikely that the CDS 2 region would determine C. trachomatis biovar ‘transformation tropism’. Thus, we need to seek an alternative explanation for the determinants of the tropism for C. trachomatis plasmids across different C. trachomatis biovars. Phylogeny has revealed high levels of diversity as well as the existence of recombinants within biovars, and natural recombinants between LGVs and non-LGV C. trachomatis have also been described (Jeffrey et al., 2013). The newly defined and distinctive clades T1 and T2 also encompass ocular and urogenital isolates (Harris et al., 2012); thus, barriers to plasmid replication between C. trachomatis biovars (Song et al., 2014) may not be biovar-specific, but strain-specific. The explanations for varying abilities to transform plasmids into C. trachomatis biovars may lie within the limits of the transformation technique itself (low transformation frequency and large vector, rendering definitions of ‘failure to transform’ unsatisfactory, especially as these are subjective rather than quantitative measures). Alternatively, the explanations may reside in the intrinsic properties of individual plasmids chosen for the studies. In this respect, our results may be unique for transformation of LGV isolates with the shuttle vector pGFP::SW2 (Wang et al., 2011, 2013a, b). The pSW2 plasmid is from the Swedish new variant of C. trachomatis, and this plasmid has two distinct features: a 377 bp deletion within CDS 1 and duplication at the 5′ end of CDS 3 (Seth-Smith et al., 2009). These features precisely flank the segment of DNA that has recombined from the endogenous C. muridarum plasmid pNigg to form pSW2NiggCDS2. The 377 bp deletion appears to inactivate CDS 1. The 44 bp repeated sequence at the 5′ end of CDS 3 duplicates the transcriptional start point (tsp) for CDS 2 (which is transcribed in the opposite direction) (Ricci et al., 1995; Albrecht et al., 2010). It is thus intriguing to hypothesise that the unique 44 bp duplication in the plasmid from the Swedish new variant is a favourable mutation that may have a significant biological role in conferring greater potential promiscuity to the wild-type plasmid pSW2 amongst C. trachomatis.