Selected Immunological Mediators and Cervical Microbial Signatures in Women with Chlamydia trachomatis Infection

To our knowledge, this is the first study that investigated the association of C. trachomatis with the cervical levels of lactoferrin and selected inflammatory mediators and their correlation with the different community state types characterizing the female genital ecosystem. C. trachomatis, known as the leading cause of bacterial sexually transmitted diseases, continues to be an important public health problem worldwide for its increasing incidence and the risk of developing severe reproductive sequelae, like pelvic inflammatory disease and infertility. Specifically, C. trachomatis tend to persist in the female genital tract, leading to a chronic inflammatory state characterized by increased production of immune mediators responsible for tissue damage. Therefore, our study may help to broaden the knowledge on the complex interplay between the female genital microbiota and the host immune system in response to C. trachomatis infection.

I n the female genital ecosystem, the protection against infectious agents is due to the epithelial barrier, continuously lubricated by the cervicovaginal fluid (CVF) and populated by numerous microorganisms that exist in a regulated mutual relationship with the host (1,2).
The genital epithelium, in addition to acting as a physical barrier, triggers the production of proinflammatory cytokines, like interleukin 1␣ (IL-1␣), IL-6, and interferons and hence contributes to the host immune response toward pathogens (3). Similarly, the CVF contains several host defense compounds, including lactoferrin, that has recently acquired importance for its extensive biological effects (4). For example, lactoferrin has been demonstrated to possess antibacterial activity by inhibiting the growth and dissemination of pathogens through different mechanisms, like iron sequestration or the inhibition of bacterial adhesion to the host cells, as evidenced in Pseudomonas aeruginosa and adherent Escherichia coli HB101, respectively (4).
The cervicovaginal microbiota in healthy women, mostly populated by Lactobacillus spp., is widely considered an important line of defense against urogenital pathogens (1,5,6). In recent years, culture-independent high-resolution techniques based on the analysis of 16S rRNA gene sequences have contributed to expand our knowledge on the composition of the genital microbiota, leading to its classification into five community state types (CSTs) (6,7). CSTs I, II, III, and V are dominated by Lactobacillus crispatus, Lactobacillus gasseri, Lactobacillus iners, and Lactobacillus jensenii, respectively, whereas CST-IV is characterized by a reduced number of Lactobacillus spp. and an increased diversity of anaerobic bacterial species, including Gardnerella vaginalis, Atopobium vaginae, Prevotella spp., etc. (7). CST-IV is often associated with the condition of dysbiosis, known to increase the risk of acquiring sexually transmitted diseases (STDs), including Chlamydia trachomatis (8,9).
C. trachomatis is the leading cause of bacterial STDs, with more than 130 million new cases per year, according to the most recent World Health Organization's estimates (10), and it is responsible for urethritis, cervicitis, and salpingitis. However, most genital infections in women are asymptomatic and, thus, not treated, potentially leading to severe reproductive sequelae, such as pelvic inflammatory disease (PID), ectopic pregnancy, and obstructive infertility (11,12). In fact, it is estimated that 40 to 60% of cases of PID as well as two-thirds of all cases of tubal factor infertility and one-third of all cases of ectopic pregnancy are most likely due to asymptomatic C. trachomatis infection (12).
C. trachomatis genital infection tends to persist in the host, leading to increased production of genital immune mediators, including IL-6, IL-1␣, and gamma interferon (IFN-␥), responsible for the inflammatory damage underlying the development of severe sequelae (12,13).
In a previous pilot study, we showed that women with C. trachomatis genital infection possessed a heterogeneous microbiota with increased species richness and diversity, dominated by anaerobes (14). Herein, we sought to define the different CSTs associated with C. trachomatis infection and to correlate them with the levels of selected immune mediators to investigate their complex interaction in the female genital ecosystem. To do so, we performed the metagenomic analysis of sequenced 16S rRNA gene amplicons with the Illumina MiSeq sequencer, and we analyzed the host defense mechanisms through the detection of lactoferrin, IL-1␣, IL-6, IFN-␣, IFN-␤, and IFN-␥ in the CVF.

RESULTS
Study population characteristics. One hundred forty-five women of reproductive age were enrolled in the present study, 42 women positive for C. trachomatis and 103 matched healthy controls.
All 145 samples underwent the 16S rRNA amplicon-based microbiome analysis, and 7 samples were excluded from further processing due to a total number of reads of Ͻ100. Population characteristics are summarized in Table 1. A statistically significant association was observed between C. trachomatis infection and high-risk sexual behaviors as well as bacterial vaginosis (P Ͻ 0.05). Filardo et al. Error rate estimation via mock microbial community cosequencing. To assess the error rate of the MiSeq Illumina sequencing, we cosequenced a mock microbial community of known composition that contained eight different bacterial species. The types and relative abundances of the bacteria present in the mock community, as well as the species and relative abundances identified by sequencing, are shown in Table 2. The estimated error rate was 0.0314%.
More than 50% of healthy women had CST-I, and a minority of them had CST-II and CST-III microbiota. In contrast, a higher number of C. trachomatis-positive women had CST-III microbiota than healthy controls, even though it did not reach statistical significance (P ϭ 0.054). More interestingly, the highest proportion of C. trachomatispositive women had a CST-IV microbiota, mostly driven by G. vaginalis and Prevotella spp. (P Ͻ 0.00001). Overall, C. trachomatis infection was significantly associated with a  combination of CST-III and CST-IV microbiota (74% versus 26% in healthy controls; P Ͻ 0.000001) ( Table 3 and Fig. S2).
The presence of genital symptoms in women with C. trachomatis infection was not related to a specific CST (Table 4).
Increased microbial species diversity and richness in C. trachomatis-positive women. Initially, we characterized the diversity and richness of the microbial species populating the cervical microbiota in either C. trachomatis-positive or healthy women, determining the alpha diversity via Shannon's diversity and Shannon-Weaver's evenness indexes that consider both the abundance and evenness of the bacterial species. Both indexes showed that C. trachomatis infection was significantly associated with increased heterogeneity of the bacterial species ( Fig. 2A and B; P Ͻ 0.0001).
Then, we investigated the diversity in microbial communities between C. trachomatis-positive and healthy women, calculating the beta diversity via the weighted and unweighted UniFrac analyses, the first based on sequence distances in the phylogenetic tree and on their relative abundances, whereas the latter was based solely on the sequence distances. Both analyses showed a statistically significant separation of the cervical microbiota from the two groups ( Fig. 2C and D; P ϭ 0.001).
Specific taxa associated with C. trachomatis infection. To identify a biomarker for C. trachomatis infection in our metagenomic data, we utilized two different approaches, namely, the linear discriminant analysis (LDA) coupled with effect size measurement (LEfSe) (15) and the analysis of composition of microbiomes (ANCOM) (16).
The relative abundances of the key phylotypes successfully differentiated in both analyses (G. vaginalis, P. amnii, P. buccalis, P. timonensis, A. christensenii, and V. guangxiensis) were further compared between the two groups, and all of them were significantly associated with C. trachomatis infection (Fig. 4).
A distinct microbial network associated with C. trachomatis infection. To identify correlations between taxonomic units within the microbial communities of C. trachomatis-positive women and healthy controls, we performed the sparse correlations for compositional data (SparCC) analysis, retaining only correlations with a |r| of Ͼ0.35, as previously described (17).
Overall, a higher number of correlations and, thus, a more complex bacterial network was found in C. trachomatis-positive women than in healthy controls ( Fig. 5 and Fig. S3).  Specifically, the cervical microbiota of women with C. trachomatis infection was characterized by positive correlations within bacteria associated with dysbiosis, like Prevotella spp., Porphyromonas spp., and G. vaginalis, whereas strong negative correlations were observed between these species and Lactobacillus spp., including L. crispatus, L. iners, and L. jensenii.
Increased lactoferrin and proinflammatory cytokine levels in C. trachomatispositive women. We found significantly increased levels of lactoferrin in CVF samples from women infected by C. trachomatis compared to healthy controls (Fig. 6); however, no association was found with a specific CST.
Concerning the proinflammatory cytokines, the levels of IL-1␣ and IL-6 were significantly higher in the CVF samples from C. trachomatis-positive women than healthy controls. Similarly, significantly increased levels of IFN-␣ and IFN-␤ were observed in Chlamydia-infected women, whereas no statistically significant difference was found for IFN-␥ between the two groups. Overall, no CST in C. trachomatis-positive women was associated with high or low levels of IL-1␣, IL-6, IFN-␤, and IFN-␥. However, significantly lower levels of IFN-␣ were observed in C. trachomatis-infected women with CST-I compared to those with CST-III and CST-IV microbiota (Fig. 6).
No significant association was observed between the levels of lactoferrin or proinflammatory cytokines and genital symptoms in C. trachomatis-positive women (Table 6).

DISCUSSION
This is the first study investigating the association of C. trachomatis with the cervical levels of lactoferrin and proinflammatory cytokines and their correlation with the different CSTs characterizing the female genital ecosystem.
The main findings of our study follow: (i) correlation of C. trachomatis infection with a cervical microbiota categorized CST-IV (predominance of anaerobic bacteria) (P ϭ 0.000002); (ii) identification of a specific bacterial network, characterized by G. vaginalis, P. amnii, P. buccalis, P. timonensis, A. christensenii, and V. guangxiensis, as a potential biomarker of C. trachomatis infection (P Ͻ 0.05); (iii) association of C. trachomatis infection with significantly increased production of lactoferrin, IL-6, IL-1␣, IFN-␣, and IFN-␤ (P Ͻ 0.01); (iv) detection of very low levels of IFN-␥ in C. trachomatis-positive women.  Concerning the cervical microbiota, our findings add to the notion that the presence of C. trachomatis is associated with a varied microbiota mostly dominated by anaerobes, including G. vaginalis and Prevotella spp., as evidenced by the prevalence of CST-IV in our patients. Interestingly, the combination of two statistical methods for biomarker discovery (LEfSe and ANCOM) revealed the association between C. trachomatis and G. vaginalis, the most predominant species, as well as P. amnii, P. buccalis, P. timonensis, A. christensenii, and V. guangxiensis. More importantly, through the statis-  (17), we observed a strong cooccurrence within the above-mentioned bacterial species, suggesting the compelling hypothesis of a specific bacterial network as a biomarker for C. trachomatis infection; in fact, such a bacterial network is an intriguing hypothesis that may better describe the peculiar genital microenvironment associated with C. trachomatis.
Alongside the resident microflora, the female genital tract possesses other defense systems that protect against infectious threats. Among them, lactoferrin, synthesized and released in the CVF by mucosal epithelial cells and neutrophils, has recently been shown to modulate the host immune response and to exert antimicrobial activity against different pathogens (4).
Here we observed a higher production of lactoferrin in C. trachomatis-positive women than in healthy controls. This is not surprising, since increased levels of lactoferrin in the genital secretions of women with other STDs have been reported (18,19). Moreover, in previous studies, we demonstrated the protective role of lactoferrin against C. trachomatis infection by inhibiting its adhesion and invasion to cervical epithelial cells (20,21).
A more interesting finding of our study shows that cervical lactoferrin levels were   In addition to lactoferrin, the host genital defense systems include a plethora of immune cells that release a wide array of cytokines in the extracellular space once a pathogen is detected (3).
Our findings show the ability of C. trachomatis to induce a distinct inflammatory state in the cervical microenvironment through the activation of Th1 and Th2 responses, as evidenced by the high levels of IL-1␣, IL-6, IFN-␣, and IFN-␤ determined in the CVF samples from chlamydia-infected women. Surprisingly, the levels of IFN-␥ in the CVF samples from our patients were very low, like those observed in healthy women.
IL-1␣ is a Th2-type cytokine that contributes to the amplification of the inflammatory process following C. trachomatis infection (21)(22)(23), eliciting the production of other proinflammatory cytokines, like IL-6, and recruiting innate immune cells (23,24). Simultaneously, tissue-resident macrophages and dendritic cells release type I interferons, namely, IFN-␣ and IFN-␤ (25). These cytokines, once considered only antiviral agents, mediate numerous antibacterial and immunoregulatory effects, including the regulation of IL-12 expression, that, in turn, induces the activation of natural killer cells as well as the development of Th1 CD4 ϩ T cells (25). Subsequently, the immunoregulatory activity of type I interferons leads to the production of IFN-␥, a critical cytokine of the innate and adaptive immune defenses against obligate intracellular pathogens like C. trachomatis (26,27). Indeed, IFN-␥ is considered an antichlamydial agent for its ability to reduce tryptophan availability through the upregulation of the enzyme indoleamine 2,3-dioxygenase that metabolizes tryptophan into kynurenine, thus inhibiting chlamydia intracellular replication (28). Interestingly, evidence in the literature showed that the high kynurenine-to-tryptophan ratio as well as high levels of IFN-␣ and IFN-␤ decreased IFN-␥ production by altering T-helper cell activity or negatively affecting IL-12 production and, hence, downregulating the immune response to infectious agents (25,29).
In this regard, in our patients, the observation of low levels of IFN-␥, as well as high levels of IFN-␣ and IFN-␤, leads to the hypothesis that increased production of type I interferons may be a potential mechanism underlying IFN-␥ downregulation. However, we cannot exclude the possibility that an increase in the kynurenine-to-tryptophan ratio may also be responsible for decreased IFN-␥ levels (29), hinting at complex regulatory pathways in response to C. trachomatis infection.
It should be noted that the low levels of cervical IFN-␥ observed in our patients may also favor the development of chlamydial persistence, which is well-known to elicit a chronic inflammatory state, worsening, de facto, the clinical course of the infection and leading to severe reproductive sequelae (12,13).
Interestingly, among all the cytokines observed in our patients, only IFN-␣ was influenced by the composition of the cervical microbiota in C. trachomatis-positive women, as evidenced by lower levels of this cytokine in women with CST-I than in women with CST-III or CST-IV microbiota (P Ͻ 0.05). This suggests that a cervical microbiota dominated by L. crispatus (CST-I) might possess anti-inflammatory properties, in accordance with other studies that observed increased inflammation in the absence of L. crispatus (30,31). Nevertheless, the determination of other proinflammatory cytokines (IL-8, IL-12, tumor necrosis factor alpha [TNF-␣]) will be required to fully characterize the host immune response to C. trachomatis in association with the different CSTs.
The main strengths of our study are the sequencing error estimation and the selection of a population whose size was estimated a priori to attain a statistical power of at least 80% and a level of significance of 5%. Furthermore, we enrolled only women with well-characterized healthy genital conditions in the control group in this study. In fact, all the participants positive for any genital pathogen other than C. trachomatis were excluded from the study, since an altered cervical microbiota has also been associated with other pathogens. Last, we excluded women with behavioral factors (antibiotic, probiotic, and/or prebiotic treatments) that could affect the composition of the cervical microbiota.
The selection of an appropriate number of participants and the application of stringent inclusion criteria allowed us to greatly diminish the impact of confounding bias related to the selection of the study population and to increase the probability of observing clinically significant differences between the groups, although the casecontrol nature of our study limited the possibility of discovering a causal relationship.
In conclusion, our findings show a distinctive signature of C. trachomatis genital infection, characterized by a specific bacterial network and by increased levels of lactoferrin and proinflammatory cytokines (IL-1␣, IL-6, IFN-␣, and IFN-␤), accompanied by low levels of IFN-␥. This complex picture may have physiopathological relevance, since it might be responsible for the incomplete clearance of C. trachomatis, leading to a persistent infection and, hence, to the development of chronic reproductive sequelae.
In the future, it will be important to monitor the changes in the cervical microbiota as well as in the host immune responses before and after exposure to C. trachomatis to clarify their temporal dynamics.

MATERIALS AND METHODS
Study design and sample collection. From January to July 2017, 42 women of Italian origin positive for C. trachomatis genital infection were selected among the patients attending the STIs Diagnostic Centre, Sant'Anna Hospital, Turin, Italy, and the Department of Gynaecology, Obstetrics and Urology at the University of Rome "Sapienza" for regular check-ups. For the control group, 103 racially and age-matched healthy women were recruited.
Women with sexual activity in the week prior to sampling, recent or current antibiotic and/or hormonal medications (oral or topical), as well as use of probiotics and/or prebiotics, were excluded from the study.
All study participants were also examined for bacterial vaginosis (BV) and for the presence of symptoms. BV was assessed using Amsel criteria and confirmed using Gram stain criteria (Nugent score).
All study participants gave their written informed consent prior to sampling and provided a detailed personal, medical, and gynecological history. This study design and protocol was approved by the Umberto I University Hospital ethical committee (reference number 367/16) and conducted according to the principles expressed in the Declaration of Helsinki.
Sequencing data analysis. To analyze the amplicon sequence data obtained from the MiSeq Illumina sequencer, we followed the standard operating procedure (SOP) developed for the mothur software package (version 1.39.5) (33) by Kozich et al. (34) and illustrated on the mothur website (https://www.mothur.org/wiki/MiSeq_SOP), accessed in January 2018.
The microbial community of each sample was assigned to a specific CST by following the approach described by Gajer et al. (7).
Shannon's diversity and Shannon-Weaver's evenness indexes were used as metrics for alpha diversity rarefaction analysis, whereas unweighted and weighted UniFrac distance matrixes were used as beta diversity measures, as previously described (35).
Differential taxonomic units between groups were identified using the LEfSe and ANCOM analyses as previously described (15,16).
To identify correlations between taxonomic units within microbial communities, we performed the SparCC analysis as described by Friedman and Alm (17).
Statistical analysis. Relative abundances of taxa were expressed as means Ϯ standard error of means (SEM). Nonparametric t test based on Monte Carlo permutations was used for alpha diversity comparisons, and Adonis was used for category comparisons of distance matrixes, all calculated in QIIME 2 (36). The relative abundances of the taxa successfully differentiated in both LEfSe and ANCOM were compared between the groups using unpaired Wilcoxon rank sum test, considering ties among values.
Cytokine and lactoferrin levels were expressed as means Ϯ standard deviations (SD) for at least two replicates. One-way analysis of variance (ANOVA) for multiple comparisons followed by pairwise t test for Filardo et al. equal variances or Games-Howell post hoc test for unequal variances between groups were used for comparison of means. Bonferroni correction was used to correct for multiple hypothesis testing when needed. The chi-squared test was used for assessment of association of frequencies among groups (Fisher's exact test was used when any cell had expected values of Ͻ5). The single or multiple inference significance level was set at 5%. Data availability. Raw sequences were deposited into the NCBI's Sequence Read Archive (SRA) under accession number PRJNA509578.

ACKNOWLEDGMENTS
This study was supported by a grant from "Sapienza" University of Rome, grant C26A15PC5N, to Rosa Sessa. The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.