Diversity of endocervical microbiota associated with genital Chlamydia trachomatis infection and infertility among women visiting obstetrics and gynecology clinics in Malaysia

The cervical microbiota constitutes an important protective barrier against the invasion of pathogenic microorganisms. A disruption of microbiota within the cervical milieu has been suggested to be a driving factor of sexually transmitted infections. These include Chlamydia trachomatis which frequently causes serious reproductive sequelae such as infertility in women. In this study, we profiled the cervical microbial composition of a population of 70 reproductive-age Malaysian women; among which 40 (57.1%) were diagnosed with genital C. trachomatis infection, and 30 (42.8%) without C. trachomatis infection. Our findings showed a distinct compositional difference between the cervical microbiota of C. trachomatis-infected subjects and subjects without C. trachomatis infection. Specifically, significant elevations of mostly strict and facultative anaerobes such as Streptococcus, Megasphaera, Prevotella, and Veillonella in the cervical microbiota of C. trachomatis-positive women were detected. The results from the current study highlights an interaction of C. trachomatis with the environmental microbiome in the endocervical region.


Introduction
Alongside Trichomonas vaginalis, Neisseria gonorrhoeae, and Treponema pallidum, genital Chlamydia trachomatis infection is one of the four most common curable sexually transmitted infections (STIs) in the world, with approximately 131 million cases of new infections being reported in year 2012 [1,2]. The seroprevalence for C. trachomatis infection was high (45.5%) in Malaysia with a large percentage (94.4%) of infection among prostitutes [3]. Chlamydial a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 infection also occurs frequently among women with fertility disorders and gynecological problems in Malaysia (22.7%-51.1%), indicating widespread infection within the country [4,5].
Although a cure for C. trachomatis infection is achievable using appropriate antibiotics for most cases (� 97%) [6,7], a large proportion of asymptomatic cases (50-70%) combined with high rates of reinfection remain the significant challenges to ongoing efforts targeted at preventing bacterial dissemination and reducing damages related to infection [8,9]. C. trachomatis infections of the genital tract in females are characterized by a vast spectrum of genital tract pathologies that include mucopurulent cervicitis, urethritis, and salpingitis, which could further lead to pelvic inflammatory disease (PID), infertility, ectopic pregnancy and cervical cancer [10][11][12]. Further, genital chlamydial infection is also linked to preterm delivery and spontaneous abortion, as well as neonatal conjunctivitis [8,[13][14][15].
An increasing number of studies have highlighted disruption of vaginal microbiome as a predisposing factor for infection by urogenital pathogens. A healthy cervicovaginal microbiota is typically dominated by bacteria of the genus Lactobacillus [16]. Lactobacillus spp. exert their protective role in the female reproductive tract against the invasion of pathogenic microorganisms by maintaining the acidity of the mucosal environment, inhibiting the adhesion of pathogens, and producing bactericidal compounds such as hydrogen peroxide (H 2 O 2 ) [17]. Under in vitro condition, Lactobacillus spp. are potent inhibitors of C. trachomatis largely due to their lactic acid producing capacities [18][19][20]. In case of bacterial vaginosis, a state of microbial imbalance within the vaginal environment where Lactobacillus spp. were replaced by other anaerobic bacteria has been linked to the increase transmission of STIs, including infections caused by C. trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis, human papillomavirus (HPV), and human immunodeficiency virus (HIV) [21][22][23][24][25].
Microbiota studies in recent years have shown that women with genital C. trachomatis infection exhibit increased colonization of anaerobic bacteria in the cervicovaginal region, although a causal relationship cannot be determined due to the use of a cross-sectional rather than a longitudinal study design. This suggests the presence of a crosstalk between microbiome community composition and the risk of chlamydial infection, which may dictate the subsequent course and outcome of female reproductive tract disorders [26][27][28][29]. However, most of the studies were conducted primarily among the European and South African populations, the data obtained are likely different from the microbiota diversity in the Asian cohort. In this present study, we evaluated the alteration of endocervical microbiome in association with C. trachomatis infection among a cohort of women in Malaysia by performing a 16S rRNA metagenomic sequencing analysis.

Study population
Our patient cohort comprised 77 female subjects attending the gynecology outpatient clinics at the University of Malaya Medical Centre (a referral centre for subspecialties including early pregnancy, infertility, oncology and general gynaecology) from year 2010 to 2014. All patients were thoroughly briefed about the purpose of the study and written informed consents were obtained from each subject prior to participation in this study. The information pertinent to visits to gynecology clinic including reasons for referral, menstruation, symptoms of genital and urinary tract infection, obstetric and medical histories were documented prior to sample collection by physicians. Patients attending the gynecology clinics for different purposes, especially those diagnosed with infertility, were randomly sampled. From the cohort collection of 180 patients, those with low DNA concentration or insufficient DNA quantity were excluded. A total of 77 samples comprising of approximately 42 C. trachomatis-infected and 35 non-infected samples were then randomly selected from the total patient cohort. The criteria for inclusion were females at the reproductive age (18 to 40 years old), while the criteria for exclusion were positive urine pregnancy test, recent antibiotic therapy, and genital tuberculosis. The definition of primary infertility used in the present investigation is the inability to achieve conception after a minimum period of one year of regular sexual intercourse without the use of contraception, whereas secondary infertility refers to the failure to conceive after the last child birth. Ethical approval was granted by the Ethics Committee of the University of Malaya Medical Centre Medical Research Ethics Committee (MREC) before commencement of this project (Reference number 908.109).

Specimen processing and diagnosis
Endocervical swabs were collected by the healthcare workers from the endocervix using UTM-RT universal transport media tubes (Copan, Brescia, Italy) and processed as described previously [5]. In brief, the transport tubes containing the endocervical swab was vortex mixed and the homogenate was centrifuged at 10,000× g. The resultant pellet was lysed and separated using phenol chloroform: isoamyl alcohol. DNA was precipitated with 1:10 volume of 3 M sodium acetate and isopropanol. After overnight storage at -20˚C, the DNA pellet was washed and then eluted in 10 mM Tris-HCl, pH 8.5. Detection of genital C. trachomatis infection was performed using a combination technique of nested-PCR and RT-PCR as described previously [5] using primer pairs (Supporting Information S1 Table) specifically targeting the C. trachomatis MOMP and cryptic plasmid genes alongside a separate amplification of human β-globin gene that served as a positive control for successful DNA extraction. All tests were run along with a positive (bacterial DNA) and negative (non-template control) samples.
A second PCR was performed to introduce unique dual indices to the ends of the amplified DNA. The PCR master mix comprised 5 μl of purified PCR products, 5 μl of Nextera XT index primer 1 and 2 (Illumina), 25 μl of 2× HiFi hotstart readymix (KapaBiosystems), and 10 μl of sterile water to a final reaction volume of 50 μl. Samples were amplified using the following protocol: 95˚C for 3 min, 8 cycles of 95˚C for 30 s, 55˚C for 30 s, and 72˚C for 30 s, as well as a final extension at 72˚C for 5 min. DNA libraries were re-purified with Agencourt AMPure XP beads (Beckman Coulter) and DNA quantity was measured with Qubit dsDNA HS assay (Life Technologies, Carlsbad, CA). Libraries were loaded on Agilent 2100 Bioanalyzer (Agilent Technologies, Palo, Alto, CA) to evaluate average fragment size and yield using Agilent high sensitivity DNA kit (Agilent Technologies) in accordance with the manufacturer's protocol. Library qualities were assessed with an Mx3000P qPCR system (Agilent Technologies) using a library quantification kit (KapaBiosystems).
Following initial library quality control on the MiSeq system with the MiSeq V2 reagent kit (Illumina), the libraries were pooled at equimolar concentration with 15% PhiX spike-in. The final library was then subjected to paired-end 2× 250 bp sequencing on a HiSeq 2500 platform using HiSeq rapid SBS kit V2 (Illumina).

Sequencing data analysis
Raw reads generated from Illumina paired-end sequencing were processed and quality filtered by using Mothur version 1.4.0 [30]. Briefly, 46,503,524 paired-end sequences were joined into contigs using Make.contigs command. Sequences were filtered by using the following criteria: minimum 20 bp overlapped, maximum 6 bp homopolymer and no ambiguous nucleotide. Approximately 50% of the sequences was removed based on the filter and a total of 23,029,897 sequences were used for subsequent procedures.
Clustering and assignment of operational taxonomic unit (OTUs) was performed using Recreated SILVA SEED Database Release 132 (https://mothur.org/wiki/Silva_reference_files). Chimeric sequences were identified and removed using VSEARCH within the Mothur pipeline. The final dataset consisted of 669 OTUs from 10141568 sequences (min = 40476; max = 211854), with mean length of 464 bp. Sequences for OTUs were subjected to a search in NCBI's database using the BLASTn algorithm to identify taxa at the species level (http://blast. ncbi.nlm.nih.gov/Blast.cgi). For ease of comparison, the data was rarefied to equal depth of 40476 sequences per sample. Alpha diversity metrices including Shannon Diversity Index, Simpson Diversity Index and Pielou's Evenness were calculated and projected in bar graphs using MicrobiomeSeq package. In addition, bar charts were constructed using phyloseq package [31] to display the proportional differences in genus and phylum across groups.
The beta-diversity (overall differences in bacterial composition across infection i.e. noninfected versus C. trachomatis-infected; and fertility i.e. fertile versus infertile/miscarriage status) was evaluated using Canonical analysis of principal coordinates (CAP) and Permutational Multivariate Analysis of Variance (PERMANOVA). Further, taxa showing significant differences in abundance between infection status i.e. chlamydial-infected and non-infected patients were identified using negative log binomial model implemented in DESeq2 R package [32]. Correction of multiple correction was conducted using Benjamini-Hochberg (BH) procedure implemented in DESeq2. Differentially expressed phylum genus and OTUs were selected by using BH adjusted P-value cut off of 0.01.

Patients' demographics
The study cohort consisted of 77 voluntarily participating women of reproductive age  presenting to the Obstetrics and Gynecology clinic at the University of Malaya Medical Center from the year 2010 to 2014. After quality filtering and chimera removal, 7 samples which generated fewer than 40,000 16S rRNA gene amplicon reads were omitted from further analyses. The final population comprised 70 subjects with ages ranging from 20 to 44 (mean age: 31.4; IQR: [27][28][29][30][31][32][33][34][35]. Of the 70 subjects enrolled, 50 were married, 3 were divorced, and the remaining 17 women had not disclosed their marital status. The study cohort comprised 74.3% (52/70) subjects from the Malay ethnic group, Chinese and Indians each contributing 11.4% (8/70), while 2.4% (2/70) were of other ethnic backgrounds, which describes the demographic diversity of the multiethnic communities in Malaysia (Table 1). Among all, 51.4% (34/70) subjects were grouped as infertile including 20 diagnosed with primary or secondary infertility and 14 patients with miscarriage experience. Besides, 22.9% (16/70) of the subjects reported menstrual irregularity.

The relationship of C. trachomatis infection with infertility
A total of 40 out of 70 participants were infected with genital C. trachomatis based on the diagnostic test result. No significant correlation between chlamydial infection was detected with demographics such as age, marital status, as well as ethnicity (Table 1). When the subjects were grouped by infertility status, 88% (30/34) of the subjects from the infertile group was infected by C. trachomatis, as opposed to only 28% (10/36) in the fertile group. Indeed, a significant association between genital chlamydial infection and infertility was established (OR:19.50, 95% CI:5.459-69.66, P < 0.0001 ��� ). It is noteworthy that high prevalence of C. trachomatis infection was detected among the patients suffering from primary or secondary infertility (OR:103.50, 95% CI:5.722-1871, P < 0.0001 ��� ) and miscarriage (OR:6.50, 95% CI:1.652-25.58, P < 0.0089 �� ), which were 100% (20/20) and 71% (10/14), respectively. However, the prevalence of C. trachomatis infection was not associated with menstrual health.

Association between C. trachomatis infection and the endocervical microbial community composition
The phylum and genus based taxonomic distributions are provided in Fig 1 and compared using negative log binomial model. Among all the patients, the dominant phyla were Firmicutes, followed by Bacteroidetes and Tenericutes (Fig 1A). When the comparison was made at phylum level, an enriched level of Chlamydiae was observed in the subjects diagnosed with (+) genital C. trachomatis infection; while those without (-) genital C. trachomatis infection was found to harbor higher levels of Tenericutes and Proteobacteria. At genus level, higher level of Delftia, Streptococcus, Pseudomonas, Cloacibacterium, Prevotella, Veillonella, Megasphaera, Ureaplasma, and Ralstonia were obvious among the subjects with C. trachomatis infection (+) compared to the group without chlamydial infection (Fig 1A). When we restricted our analysis to infertile women, elevated level of phyla Bacteroidetes was detected the C. trachomatis-infected samples (+). A lower prevalence of genus Megasphaera was detected among the subjects with C. trachomatis infection (+) in comparison to the non-infected group (Fig 1B).
Sequence for each OTU was blasted using NCBI database to identify the microorganisms at species level. The species which matched the highest sequence homology with the input are as shown in Table 2. The abundant species in the C. trachomatis infected group were Lactobacillus gallinarum or Lactobacillus crispatus (OTU2), Streptococcus agalactiae (OTU3), Prevotella colorans (OTU15), Pseudomonas mendocina or Psudomonas oleovorans (OTU33), Veillonella ratti or Veillonella seminalis (OTU39), and Delftia lacustris or Delftia tsuruhatensis (OTU70). Whereas the abundant species in the non-chlamydial infected group were recognized as Aerococcus christensenii (OTU13), Stenotrophomonas maltophilia (OTU32). Some OTUs, for example OTU10 and OTU29, were unable to be identified at species level because of limited information in the database.

Taxonomic diversity profiles of the endocervical microbiota in the presence of C. trachomatis infection
Alpha diversity measures including Pielou's evenness alongside Simpson's and Shannon's indices were examined (Fig 3A). There is a general trend of higher alpha diversity indices in non-C. trachomatis infected than those in the C. trachomatis-infected subjects. However, the difference did not achieve statistical significance at alpha = 0.05 (Fig 3A). Beta diversity between the groups was further assessed by CAP ( Fig 3B) and PERMANOVA analysis ( Table 3). In CAP, separation between non-infected and infected groups was observed along axis CAP1. Indeed, a significance in bacterial compositional difference between groups with different infection status was apparent (pseudo-F = 1.4744, P (perm) = 0.039). However, statistical significance was not achieved between groups with different fertility status (pseudo-F = 1.384, P(perm) = 0.079).

Discussion
In this study, most of the patients diagnosed with C. trachomatis (88%) experienced primary or secondary infertility (100%, P<0.0001) as well as miscarriage (71%, P = 0.0089). A prior epidemiological study has indicated that women with genital C. trachomatis infection are 30% more likely to develop PID, ectopic pregnancy, as well as tubal factor infertility (TFI) than those without, whereas infections with C. trachomatis elevate the risk of PID by an additional 20% [33]. Our results are consistent with this finding, whereby a significant positive correlation between C. trachomatis and infertility was detected. Given the high prevalence of C. trachomatis infection and the severity of disease that ensues from the infection, a more effective screening and treatment strategy is warranted to curb the dissemination of C. trachomatis in the population. Notably, our study cohort comprised three major ethnic groups, namely the Malays, Chinese, Indians, as well as ethnic minority participants of other nationalities. However, due to the small sample size, the differences between ethnicities were not compared.
The vaginal flora, which constitutes 9% of the human microbiome, have been shown to play a critical role in the maintenance of overall health condition and disease state of the female reproductive tract [34]. Many studies have demonstrated that women with bacterial vaginosis are at higher risk of other sexually transmitted pathogens that includes C. trachomatis, signifying the importance of microbiota homeostasis and its role as a determinant in disease outcome [21,23,24,35]. While the precise etiologic agents in bacterial vaginosis remain unclear, the prevailing view is that the disease is polymicrobial in nature caused by overgrowth of a consortium of anaerobic bacteria such as Gardnerella vaginalis, as well as microbes belonging to the genera Streptococcus, Veillonella, Prevotella, Peptostreptococcus and Bacteroides [36][37][38]. Although alpha-diversity analyses did not reveal categorical difference in cervical microbiota diversity across the sample groups, further evaluation of the cervical bacterial community using beta-diversity measure showed that there was significant difference in the taxonomic

Fig 2. Significantly abundant taxa within the cervical groups of subjects with (+) and without (-) chlamydial infection.
Analysis were done in all cohorts after accounting for differences in fertility status across the study groups. Red bars represent bacteria genus which were significantly abundant in C. trachomatis-infected group compared to non-chlamydial-infected group; whereas blue bars represent significant groups that are more abundant in nonchlamydial infected group. Numbers in bracket represents OTU.  compositions between the subjects in relation to C. trachomatis infection (PERMANOVA, pseudo-F = 1.4744, P (perm) = 0.039). In addition, we report here that women with genital chlamydial infection harbor significantly more abundant bacteria such as Streptococcus agalactiae, Pseudomonas sp., Veillonellaceae bacterium, Megasphaera sp., Veillonella sp., Prevotella colorans, Veillonella sp., as well as Delftia sp., the majority of these are anaerobes and facultative anaerobes which have been implicated in BV [36][37][38][39]. Other studies have also reported an increase in the colonization of diverse anaerobic bacteria in the cervicovaginal environment of women with C. trachomatis infection [26-28, 40, 41]. It is currently unknown whether Delftia can modulate the host susceptibility to other pathogens. Delftia tsuruhatensis appears to constitute the normal microbiota of the human vulva but there is evidence to suggest that its abundance in amniotic fluid and blood is associated with preterm birth [42,43]. The significant presence of Delftia sp. in our C. trachomatis-positive group, therefore, merits further study to better elucidate its role in the female endocervical microbiota. Differential abundance analysis (DESeq2) further showed that the taxon Lactobacillus crispatus and L. gallinarum was significantly elevated in the microbiota of women infected with C. trachomatis relative to those without C. trachomatis infection. Evidence is emerging that prevalence of particular Lactobacillus provides fewer protection compared to other members of the same genus. L. crispatus has been found to inhibit C. trachomatis infection, replication, as well as attachment to host cells; these effects are likely related to its production of lactic acid that lowers the surrounding pH and inhibits chlamydial growth [18][19][20]. However, it was reported recently that L. crispatus promotes repair of injured vaginal epithelial cells by promoting the secretion of vascular endothelial growth factor A (VEGF), which may help to reduce subsequent acquisition of urogenital tract infections [44]. L. crispatus was reported as one of the major compositions of endocervix microbiota among the asymptomatic C. trachomatisinfected patients in South Africa [28]. Meanwhile, L. gallinarum has previously been found in a high percentage of HPV-infected women in Beijing, China [45]. L. iners (OTU1) (data not shown) was the most abundant taxon in the endocervical milieu of both infected and noninfected group. Such dominance has been previously reported among normal healthy reproductive Thai and Chinese women [46,47]. L. iners has been shown to be negatively associated with unfavorable reproductive health outcomes. Although commonly detected in the vagina of healthy reproductive age women [48], a high prevalence of L. iners has been shown to be a predictive factor for bacterial vaginosis and pre-term delivery [49][50][51][52]. In fact, L. iners-dominated microbiota has been linked to increased predisposition to acquisition of C. trachomatis, HIV, as well as M. genitalium, indicating its double-edged sword role in female reproductive health [26,[53][54][55]. However, we were unable to unequivocally determine the identity of the Lactobacillus in the present study as the sequences of L. gallinarum and L. crispatus had the identical homology in the BLAST results.
Conversely, subjects negative for C. trachomatis displayed significantly greater abundance of Aerococcus christensenii. Certain members of the genus Aerococcus including A. urinae, A. sanguinocola, and A. viridans are rare pathogens involved in urinary tract infections particularly among individuals with comorbidities [56][57][58]. In a similar study, Di Pietro et al. interrogated the microbial constituents of a population of Italian women infected with C. trachomatis Endocervical microbiota in Chlamydia trachomatis infected females and identified A. christensenii as the species closely associated with chlamydial infection [40]. Balle et al. showed that A. christensenii was only weakly related to C. trachomatis infection in a cohort of South African adolescents [28]. A. christensenii has previously been implicated as a cause of chorioamnionitis in a case report [59]. Beyond that, much remains unknown about the role of A. christensenii in the female genital tract. Currently, there is a paucity of information concerning the significance of Stenotrophomonas in the context of female reproductive health. S. maltophilia is considered as a human commensal and is a frequently isolated microorganism in urine specimens. Although rare, S. maltophilia has been reported to be a cause of nosocomial urinary tract infection [60,61].
In conclusion, our study showed significant difference in endocervical microbiota between C. trachomatis-infected and non-infected cohorts among patients visiting gynecology clinics from Malaysia. However, we highlighted that genital C. trachomatis infection was not directly connected to increased cervical bacterial diversity, as no significant differences were observed in data richness and evenness. Further, we showed that infection with C. trachomatis was related to elevated prevalence of mostly strict and facultative anaerobes, however disruption of the sequencing data by PCR artefact should also be alerted [62]. Future studies involving a larger cohort with more stringent criteria will allow us to delineate the endocervical microbial landscape of women living in Malaysia. Nevertheless, the role of the anaerobes in disease progression is currently unknown, hence a further study to evaluate the involvement of taxa such as A. christensenii and S. maltophilia in chlamydial infection is warranted.
Supporting information S1