Chlamydial clinical isolates show subtle differences in persistence phenotypes and growth in vitro

Urogenital Chlamydia trachomatis infection is the most common sexually transmitted bacterial infection throughout the world. While progress has been made to better understand how type strains develop and respond to environmental stress in vitro, very few studies have examined how clinical isolates behave under similar conditions. Here, we examined the development and persistence phenotypes of several clinical isolates, to determine how similar they are to each other, and the type strain C. trachomatis D/UW-3/Cx. The type strain was shown to produce infectious progeny at a higher magnitude than each of the clinical isolates, in each of the six tested cell lines. All chlamydial strains produced the highest number of infectious progeny at 44 h post-infection in the McCoy B murine fibroblast cell line, yet showed higher levels of infectivity in the MCF-7 human epithelial cell line. The clinical isolates were shown to be more susceptible than the type strain to the effects of penicillin and iron deprivation persistence models in the MCF-7 cell line. While subtle differences between clinical isolates were observed throughout the experiments conducted, no significant differences were identified. This study reinforces the importance of examining clinical isolates when trying to relate in vitro data to clinical outcomes, as well as the importance of considering the adaptations many type strains have to being cultured in vitro.


INTRODUCTION
Chlamydia trachomatis is an obligate intracellular pathogen and is the most common bacterial sexually transmitted infection (STI) worldwide. There are more than 83000 Chlamydia infections recorded in Australia each year [1]. The pathology and sequelae associated with chlamydial diseases are thought to be associated with the infected individual's inflammatory response and potentially influenced by a number of important host factors [2]. In women, chlamydial disease ranges from mild cases of cervicitis, endometritis and salpingitis, to more serious cases of pelvic inflammatory disease (PID), tubal infertility and life-threatening ectopic pregnancy [3][4][5].
Cases of urogenital Chlamydia are treated with either azithromycin or doxycycline. While evidence suggests doxycycline (usually 7 day regimen) is slightly more effective at clearing infection, 1 g single-dose azithromycin is often prescribed for its simple one-off treatment dose [6,7]. Azithromycin is a broad-spectrum macrolide antibiotic with a relatively long half-life (40-68 h) and high lipid solubility, and accumulates within macrophages migrating to the site of infection [8,9]. Despite its high efficacy, instances do occur in which monitored women remain infected after treatment. The reasons for this remain unclear, but potential explanations include: reinfection from an untreated partner, chlamydial gastrointestinal colonization and auto-inoculation of the cervical site or treatment failure [10,11]. Direct macrolide resistance is generally not considered a probable cause due to the antibiotic's mechanism of action, and the scarcity of clinical isolates with validated genotypic or phenotypic resistance to azithromycin [12][13][14][15][16]. Furthermore, mutants created in vitro with single nucleotide polymorphisms (SNPs) in OPEN ACCESS their 23S rRNA and L4 protein-encoding genes show poor biological fitness and viability [17][18][19][20], further supporting that their emergence in the population is likely to be rare.
When combined, in vitro and clinical evidence suggests that C. trachomatis rarely survives treatment, which prpbably reflects both the pathogen and its intracellular niche. One candidate mechanism is chlamydial persistence, which is believed to be a positive and beneficial adaptation unique to the genus Chlamydia [21,22]. The persistence phenotype is characterized by several reversible morphological, transcriptional and metabolic changes [23][24][25][26][27]. The collective profiles of these changes vary depending on the inducer of persistence; however, the morphologically aberrant chlamydial cells observed during persistence, altered inclusion sizes, and the reversible loss of both cultivability and replicative capacity are the universal hallmarks of chlamydial persistence (reviewed previously [28]).
Several environmental conditions and exogenous stimuli have been demonstrated to induce chlamydial persistence in vitro [29]. Penicillin and IFN-γ are two of the most extensively studied of these stimuli, and have both been used to characterize different aspects of the persistence phenotype, as reviewed by Wyrick [28]. Use of the iron-chelating agent deferoxamine mesylate has shown that iron restriction not only induces persistence in C. trachomatis, but also alters the pathogen's signalling pathways that modulate hostcell apoptosis [30,31]. Interestingly, IFN-γ also decreases cellular levels of iron in infected cells by downregulating their transferrin receptor expression [32].
Previous investigations into chlamydial persistence have established that type strains have variations in their degree of responsiveness to certain stimuli, and have altered susceptibilities to antibiotics while in the persistent state [33][34][35][36]. In the present study, we examined the in vitro phenotypes of selected clinical isolates of C. trachomatis isolated during the Australian Chlamydia Treatment Study (ACTS) [11]. Their relative abilities to infect and develop in different cell lines were measured, as were their responses to two widely used and physiologically relevant models of in vitro chlamydial persistence. Their susceptibility to azithromycin during persistence was also investigated, to better understand whether persistence has implications for treatment with frontline antibiotics when analysed on recent clinical isolates.

Cell culture and cultivation of Chlamydia
McCoy B, MCF-7, CACO-2, HeLa, SiHa and ARPE-19 cell lines (details in Table 1) were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10 % (v/v) fetal calf serum (FCS; Sigma), 4 mM analyl-glutamine (Sigma), 100 μg ml −1 streptomycin (Life Technologies) and 50 μg ml −1 gentamicin (Life Technologies). A range of cell types available to the study were selected to profile any possible differences in the clinical isolates. All cell lines were incubated at 37 °C in a humid environment containing 5 % CO 2 . All cell lines were regularly confirmed to be free of Mycoplasma contamination using either an in-house PCR, or the Mycoalert Plus Mycoplasma Detection Kit assay (Lonza). All six clinical isolates used were obtained from the ACTS. This was a cohort study of women diagnosed with and treated for genital Chlamydia [11] to examine factors associated with repeat chlamydia infection. In one case, two isolates from one participant after a repeat positive event were included in the study in the event that they prove to have some detectable difference in persistence phenotypes to those analysed here. Isolates were collected from infected women using swabs stored in a 2 ml cryovial tube containing a sucrose-phosphate glutamate (SPG) buffer, at −80 °C, and couriered on dry ice. A unique code has been generated to identify each isolate purely for the purposes of this paper.
Isolates were cultured from the original cervical swabs from women using a series of culture steps to attain sufficient culture for these experiments (six to 12 passages depending on the isolate). Stocks were stored and propagated from SPG buffer which consisted of 5 mM glutamic acid, 10 mM sodium phosphate and 250 mM sucrose balanced to pH 7.4. The five clinical isolates used in this study are detailed in Table 2.

Infectivity and determination of viable progeny
McCoy B, MCF-7, CACO-2, HeLa, SiHa and ARPE-19 cells were cultured in 96-well plates, in triplicate for each condition and experimental analysis conducted. Cells were All C. trachomatis genotype determinations utilized a 5 µl aliquot of PBS swab homogenate elution, utilizing a series of quantitative PCR (qPCR) amplification assays targeting the ompA gene of C. trachomatis as described previously [37].

Penicillin and iron deprivation persistence models
Persistence models were conducted in MCF-7 cells. In total, 2.5×10 4

Data analysis and graphing
Raw data were compiled using Microsoft Excel 2010 before being transferred into GraphPad Prism version 8.0.0 for Windows for statistical analysis and graphing, with each value and any applicable statistical testing described in the respective figure legends.

RESULTS
Clinical isolates differed in infectivity and growth more profoundly than the differences caused by the cell lines Six cell lines were selected, to compare how susceptible they were to infection by four of the clinical isolates, and the type strain D/UW-3/Cx (Fig. 1) (Fig. 1).

Clinical isolates show similar susceptibilities to conditions of iron deprivation at lower doses than type strain D/UW-3/Cx
To more closely examine if persistence phenotypes were impacted by ompA genotypes, and the known outcomes in participants in the ACTS trial, an additional isolate was included in these persistence experiments to have a close genotype to the isolate from a participant who experienced repeat infection. Five clinical isolates and the type strain were cultivated in MCF-7 cells, treated with 100, 200 (D/UW-3/Cx only) and 400 µM bipyridal (Bpdl)±100 µM iron as FeCl 3 , with infectious progeny determined at 96 h PI. As seen in Fig. 2a, the type strain was able to recover from up to 100 µM Bpdl (7.1×10 4 IFU ml −1 before recovery) with the supplementation of FeCl 3 into culture (3.1×10 8 IFU ml −1 after recovery). The five clinical isolates showed lower recovery, even from the lowest dose of Bpdl, as seen in Fig. 2. For example (Fig. 2c), at the 100 µM dose of Bpdl with and without recovery, 600(13) yielded 3.1×10 3 and 7.0×10 2 IFU ml −1 (near the limit of detection for this assay) respectively. All five isolates produced infectious progeny in a manner that was dose-dependent, with slightly lower levels of infectious progeny present in cultures treated with 400 µM than those treated with 100 µM.

Treatment with azithromycin during iron deprivation-induced persistence decreases the number of recoverable infectious progeny among clinical isolates
Each of the five clinical isolates and the type strain were cultured in MCF-7 cells treated with Bpdl, subsequently treated with azithromycin (Az), and then recovered with supplementation of iron. The azithromycin dose used was the MIC that had been determined for each isolate. The MICs were as follows: 600 (1): 0.032 µg ml −1 , 600 (13): 0.125 µg ml −1 , 620: 0.032 µg ml −1 , 628 : 0.064 µg ml −1 ; 649 : 0.064 µg ml −1 , and D/UW-3/Cx: 0.064 µg ml −1 . All strains showed a significant decrease in infectious progeny at 44 h PI when treated with azithromycin, Bpdl or both (Fig. 3, P<0001). Compared to the untreated control at 44 h PI (5.4×10 5 IFU ml −1 ), D/UW-3/ Cx showed an impaired ability to recover from the effects of Bpdl when also treated with azithromycin (2.9×10 4 IFU ml −1 ), even with iron supplementation (5.0×10 2 IFU ml −1 ). Compared to their respective untreated controls at 44 h PI (Fig. 3), all five clinical isolates were found to have only slight differences in their resulting infectious yield when treated with the combinations of azithromycin, Bpdl and FeCl 3 . For example, the untreated control 600 (13) culture produced 2.2×10 4 compared to only 1.5×10 3 IFU ml −1 after treatment and recovery. Analysis of the cultures by confocal microscopy showed morphologies consistent with persistence, as inclusions visible were consistent with persistence after treatment with penicillin; regular development occurred after allowing for recovery from the drug (Fig. 4, representative images from some isolates).

Clinical isolates entered persistence at lower doses of penicillin compared to the type strain
Several doses of penicillin were used to induce persistence in each strain, which was confirmed by measuring viability before and after recovery from the drug. The data shown in Fig. 5   progeny by 110 h PI, all to a similar level, even after treatment with 1 U ml −1 of penicillin. Confocal microscopy examination of the morphology of the cultures confirmed the presence of visible forms consistent with persistent or recovered morphology (Fig. 6). Specifically, at 44 and 110 h PI and in the absence of penicillin, each of the five clinical isolates and the type strain showed typical morphologies consistent with regular development. At the same time point, cultures treated with penicillin showed significantly smaller inclusions with enlarged particles inside, morphologies consistent with  persistence. Imaging of the cultures at 110 h PI (66 h after the removal of penicillin from culture) showed inclusions typical of regular inclusions.

Clinical isolates treated with azithromycin during penicillin persistence showed a dose-dependent decrease in recoverable infectious progeny
To assess whether clinical isolates had altered susceptibilities to azithromycin during persistence, three clinical isolates and D/UW-3/Cx were cultured in MCF-7 cells and treated with both penicillin and azithromycin. Persistence was induced using 0.05 U ml −1 of penicillin for the three clinical isolates and 1.0 U ml −1 for the type strain. At 44 h PI, Fig. 7 shows a complete loss of infectious progeny for the penicillin-treated cultures, which was recoverable (albeit reduced) by 110 h PI. The infectious progeny of the four strains at 44 h PI was also observed to decrease by up to 100-fold when treated with azithromycin alone. In the type strain, this effect was also seen, whereby the recoverable infectious progeny decreased in the presence of azithromycin only, with a complete loss of viability at 44 h PI when both azithromycin and penicillin were added. However, no statistical differences were apparent for any isolates treated with azithromycin during penicillin persistence. It appeared that the recovery of the clinical isolates from persistence was further reduced when treated with azithromycin at the MIC, with two of the isolates [600 (13) and 649] appearing to be more impacted by the combination treatment.

DISCUSSION
The susceptibility and growth permissiveness of different cell lines to infection by C. trachomatis enables examination of the host-pathogen relationship [42][43][44][45]. Such studies frequently find differences between strains in their ability to enter the host cell, and complete their developmental cycle [46]. The infectivity of the clinical isolates in the present study was observed to be highest in the MCF-7 cell line, despite the   long-standing practice throughout the field of using McCoy or HeLa cells for cultivation and isolation of C. trachomatis, especially from clinical samples [47][48][49]. However, infectious yields were highest in the McCoy B cell line for each of the strains tested. The ARPE-19 cell line was observed to have a very low susceptibility to infection by, and permissiveness of growth to, all strains except D/UW-3/Cx, yet was nonetheless able to be infected by each of the strains, even though it is not an epithelial cell type. The inability to detect progeny from this cell line for the clinical isolates may reflect the low yields being below detection of the assay, or that the clinical isolates in these cells are unable to form infectious elementary bodies. Overall, these findings reinforce that there are phenotypic differences between type strains and clinical isolates that probably reflect the adaptation of D/UW-3/Cx to growth in vitro [50].
Although it has been shown that type strains of C. trachomatis have varying levels of susceptibility to the in vitro effects of IFN-γ, fewer studies have examined how different strains respond to penicillin or iron deprivation [51]. In this study, we aimed to determine how clinical isolates respond to such conditions. Both were selected on the basis that penicillin has been used widely as a chlamydial persistence model, so our findings can be interpreted in light of other studies [36,[52][53][54]; iron deprivation is potentially a more clinically relevant model when considering the physiology of the female reproductive tract [55][56][57]. Penicillin is known to induce persistence by interacting with chlamydial penicillin-binding proteins (PBPs), while host cell-derived iron is essential for chlamydial development. Both points combined raise the possibility that genetic variation among infecting strains could result in differing thresholds at which they divert from the regular developmental cycle into persistence.
To assess the susceptibility of each strain to penicillin, they were treated with a dose range in MCF-7 cells. None of the clinical strains showed any notable difference in their levels of susceptibility to penicillin persistence, with all entering a viable but non-cultivable state at 0.05 U ml −1 of the antibiotic. In contrast, the type strain D/UW-3/Cx remained cultivable up to the maximum dose used, which was 1 U ml −1 .
A recent study into the effects of beta-lactam antibiotics on C. trachomatis showed that type strain E/UW-3/Cx entered persistence at 0.02 U ml −1 of benzylpenicillin [36]. Although using different host cells, this suggests there may be different susceptibilities among type strains.
Similarly, there were no large observed differences in the responses of clinical isolates to the effects of iron deprivation. In a recent review, Pokorzynski and colleagues postulated a complex system by which C. trachomatis may be able to both passively and actively acquire ferrous and ferric iron from within the host cytoplasm by modulating the host's own iron trafficking pathways [58]. As with other instances of chlamydial persistence, it is possible that strain differences may in turn result in slight differences in the proteins that conduct these functions.
The effects of azithromycin were tested on a selection of clinical isolates during both active and persistent development. We observed no differences in the susceptibilities of active and persistent infections, in any of the clinical strains tested. Previous findings by the Caldwell group demonstrate that during IFN-γ-mediated persistence, a serovar D type strain of C. trachomatis is significantly more susceptible to azithromycin [35]. However, Wyrick and Knight have shown that a serovar E type strain is less susceptible to the same antibiotic during penicillin-mediated persistence [52]. This reinforces previous findings that different inducers of persistence produce different phenotypes [59], which probably reflects the physiological stress the inducer places upon C. trachomatis.
Collectively these findings show that clinical isolates respond to the effects of penicillin, iron deprivation and azithromycin in a similar but more pronounced way than type strain D/ UW-3/Cx. This is important because these are recent clinical isolates, indicating that persistence may occur more frequently and with lower thresholds in vivo. Here we demonstrated that clinical isolates showed subtle variation in thresholds for persistence that may be more distinct and impactful in the complex in vivo environment. These variations may translate into phenotypes in vivo that could relate to heterogenic survival of antibiotic treatment, or different capacities to survive in different tissue niches (e.g. rectal compared to urogenital) that could be relevant for future investigation of chlamydial variation.

Funding information
This study and ACTS was funded by a National Health and Medical Research Council project grant (APP1023239), awarded to J.H., P.T., W.H. and others.