Molecular typing of Trichomonas vaginalis isolates by actin gene sequence analysis and carriage of T. vaginalis viruses

Background The protozoan parasite Trichomonas vaginalis is the most common non-viral, sexually transmitted pathogen. Although T. vaginalis is highly prevalent among women in Kenya, there is lack of data regarding genetic diversity of isolates currently in circulation in Kenya. Methods Typing was performed on 22 clinical isolates of T. vaginalis collected from women attending the antenatal care clinic at Kilifi County Hospital, Kenya, in 2015. Genotyping followed a previously proposed restriction fragment length polymorphism (RFLP) scheme, which involved in silico cleavage of the amplified actin gene by HindII, MseI and RsaI restriction enzymes. Phylogenetic analysis of all the sequences was performed to confirm the results obtained by RFLP-analysis and to assess the diversity within the RFLP genotypes. Additionally, we determined carriage of the four different types of Trichomonas vaginalis viruses (TVVs) by polymerase chain reaction. Results In silico RFLP-analysis revealed five actin genotypes; 50.0% of the isolates were of actin genotype E, 27.3% of actin genotype N, 13.6% of actin genotype G and 4.5% of actin genotypes I and P. Phylogenetic analysis was in agreement with the RFLP-analysis, with the different actin genotypes clustering together. Prevalence of TVVs was 43.5% (95% confidence interval, CI: 23.2–65.5). TVV1 was the most prevalent, present in 39.1% of the strains and 90% of the T. vaginalis isolates which harbored TVVs had more than one type of TVV. None of the isolates of actin genotype E harbored any TVV. Conclusion The presence of five actin genotypes in our study suggests notable diversity among T. vaginalis isolates occurring among pregnant women in Kilifi, Kenya. Isolates of the most prevalent actin genotype E lacked TVVs. We found no association between T. vaginalis genotype, carriage of TVVs and symptoms. Further studies with higher number of strains should be conducted in order to corroborate these results. Electronic supplementary material The online version of this article (10.1186/s13071-017-2496-7) contains supplementary material, which is available to authorized users.


Background
Trichomonas vaginalis is a flagellated protozoan parasite that infects the human urogenital tract, causing the most common non-viral, sexually transmitted infectious disease worldwide [1]. The prevalence of T. vaginalis among women in sub-Saharan Africa is 11.5% [1]. In about half of the infected women, T. vaginalis causes a malodorous vaginal discharge, vulval irritation and inflammation, and a 'strawberry cervix' characterized by punctate hemorrhagic lesions [2]. Men typically remain asymptomatic, but can suffer from urethral discharge, dysuria, urethritis, epididymitis and prostatitis [2,3]. Trichomonas vaginalis has been associated with adverse pregnancy outcomes, such as preterm birth and premature rupture of membranes [4][5][6], increased shedding and acquisition of the human immunodeficiency virus (HIV) [7,8], hence contributing to the HIV pandemic.
The global prevalence of T. vaginalis and the health sequelae associated with it have necessitated the need to understand its genetic make-up. Trichomonas vaginalis is a complex pathogen, with a genome size of~160 megabases, two-thirds of which is composed of repeats and transposable elements [9]. Some strains of T. vaginalis can harbor up to four types of Trichomonas vaginalis viruses (TVVs) [10]. TVVs are members of the family Totiviridae under the distinct genus Trichomonasvirus [11]. Carriage of TVVs has been suggested to upregulate pro-inflammatory host responses [12] and T. vaginalis immunogenic protein P270 [13] and is also associated with differential qualitative and quantitative expression of cysteine proteinases [14]. Thus, since TVVs induce various phenotypic changes that may impact T. vaginalis virulence [14], determining carriage of TVVs seems to be essential in the characterization of T. vaginalis infection. However, no method has been adapted as a standard clinical diagnostic test for TVVs [15].
Better understanding the diversity of T. vaginalis and geographical distribution of various genotypes may improve our knowledge regarding the epidemiology of this infectious agent and contribute to vaccine development efforts. At present, none of the described techniques have been recognized as the "gold standard" for genotyping of T. vaginalis isolates [16][17][18][19][20][21][22][23].
In this study, we opted to sequence the T. vaginalis actin gene to better understand the genetic diversity of T. vaginalis. Actin is a ubiquitous well conserved structural protein in all eukaryotic cells [24] and has been used to clarify the molecular phylogeny of protists, plants, animals and fungi [25].
The data produced by sequencing are unambiguous, reproducible and portable, thus offering the advantage that public databases can be constructed. In Kenya, screening for T. vaginalis is not routinely done. However, recent studies have indicated that there is high prevalence of T. vaginalis among different groups in Kenya [26][27][28][29]. Despite this, no typing has been carried out. In this study, we genotyped T. vaginalis isolated from pregnant women attending ANC in Kilifi, Kenya.

Study design, population and specimens processing
From July through to September 2015, we conducted a cross-sectional study at the antenatal care clinic of Kilifi County Hospital, Kenya. The main aim of that study was to describe the prevalence and predictors of curable sexually transmitted infections (STIs) among pregnant women attending to the antenatal care clinic [29]. Women were eligible if they met the following criteria: age 18-45 years, gestation ≥ 14 weeks, resident of the Kilifi Health and Demographic Surveillance area, willingness to undergo free STI and bacterial vaginosis screening procedures, and able and willing to give written informed consent [29]. The current study presents a secondary objective of the above-mentioned study, namely to perform typing of T. vaginalis clinical isolates from this study population.
A total of 349 pregnant women were included in the study. A nurse collected vaginal secretions from the vaginal introitus using a sterile cotton swab. The vaginal swab was inoculated at the clinic in the upper-chamber of an InPouch system (BioMed Diagnostics, White City, Oregon, USA). The inoculated InPouch was transferred to the laboratory within 15 min for direct microscopy of the upper chamber, after which it was merged with the lower chamber and incubated at 37 ± 1°C under aerobic conditions. Daily microscopic observation of the InPouch system was performed and media with motile trichomonads within 5 days of culture were considered positive for T. vaginalis. Two ml of the contents of each InPouch system positive for T. vaginalis were transferred into a 2.0 ml Eppendorf tube and stored at -80°C until shipment to the Laboratory of Bacteriology Research (Ghent University, Belgium) using shipping boxes filled with dry ice (-78.5°C). The samples were stored at -80°C until molecular analysis was performed.
Women found to be positive for T. vaginalis using direct microscopy, i.e. microscopy of the upper chamber, were treated on the same day, while women who were negative for T. vaginalis using direct microscopy but positive on culture were contacted to return to the clinic for treatment immediately after the culture turned positive. Partners of women treated for T. vaginalis infection received presumptive treatment.

Nucleic acid extraction
Frozen cultures were first thawed at room temperature for 30 min, after which the tubes were vortexed for 30 s to ensure that a homogeneous mix was achieved before starting the nucleic acid extraction. Total nucleic acid extraction was performed using the NucliSENS® easyMAG® (bioMérieux, Marcy l'Etoile, France) according to the manufacturer's instructions (generic protocol 2.0.1).

Gel electrophoresis
PCR amplification products were visualized under UV light after electrophoresis on 1% agarose gels in Tris-acetate-EDTA buffer pH 8.5 (30 min at 10 V/cm) and staining with ethidium bromide (0.5 mg⁄ l; Sigma, Bornem, Belgium). The size of the amplified products was assessed by comparison with a commercial weight marker, Smart Ladder (Eurogentec, Liege, Belgium).

Sequencing
Amplicons (20 μl) were sent for sequencing to GATC Biotech (Constance, Germany), using the Sanger sequencing technique. Sequencing was performed bi-directionally using the same primers used in PCR amplification of the actin gene.

Genotyping of T. vaginalis
Trichomonas vaginalis actin sequences were edited using Chromas Lite 2.01 (http://technelysium.com.au/wp/). Furthermore, we identified the T. vaginalis actin genotypes amongst our isolates by means of in silico RFLP-analysis Webcutter version 2.0; http://rna.lundberg.gu.se/cutter2/ with the three restriction enzymes (HindII, MseI and RsaI) used by Crucitti et al. [20]. The in silico analysis was done on our clinical isolates and on retrieved sequences representing the genotypes that have been proposed. To identify if there were additional T. vaginalis actin genotypes not captured by the previous proposed scheme, in silico RFLP was performed on actin sequences retrieved from the GenBank whose actin genotype is yet to be documented. The sequences were of at least 1100 bp in length (Additional file 1: Table S1).
To compare results obtained in the RFLP analysis, and to determine the genetic variation among the identified genotypes, we performed phylogenetic analysis of the sequences. Sequences obtained in our study were aligned using MEGA software and compared with analogous sequences representative of known T. vaginalis actin genotypes identified in from our isolates.
Evolutionary distances were calculated by Kimura's two-parameter model (Kimura, 1980) and a phylogenetic tree was generated using the Maximum Likelihood method using the MEGA software (version 7.0) [30]. Finally, confidence levels were estimated using bootstrap resampling on 1000 randomly selected pseudoreplicates.

Statistical analysis
The socio-demographic and clinical data of the participants were entered in the REDCapTM electronic data capture tool, version 6.5.0 (Vanderbilt University, Nashville, Tennessee). Prevalence of T. vaginalis was expressed as a percentage with exact 95% binomial confidence intervals (CIs). Univariable logistic regression was used to determine associations of T. vaginalis presence with socio-demographic data, hygienic data, sexual behavior, and vaginal signs & symptoms. Variables significant at P ≤ 0.1 in bivariable analysis were entered into a multivariable model to identify independent associations. Crude odds ratios (COR) and adjusted odds ratios (AOR) were calculated. These statistical analyses were done using STATA, version 13.1 (Stata-Corp, College Station, Texas).

Characteristics of the study population
The median age and gestation age including the interquartile range (IQR) for the participants was 27 (22)(23)(24)(25)(26)(27)(28)(29)(30)(31) years and 25 (20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) weeks, respectively. The majority of the participants were married (93.6%) and Christian (72.3%). Approximately a quarter of the participants had received secondary school education and above. The median age and IQR of sexual debut was 18.6 (16)(17)(18)(19)(20) years, although 17.4% of the participants were not sure or did not respond to this question. Seventy-three percent of the participants had given birth before and the number of children ranged between 1 and 10. Sixtyfive percent of the participants had previously experienced signs/symptoms associated with reproductive tract infection and about one third had ever received syndromic treatment for genital signs or symptoms of infection. The socio-demographic characteristics of the participants are summarized in Table 1.

Prevalence, clinical signs and symptoms
A total of 23/349 (6.6%, 95% CI: 4.2-9.7) women were culture-positive for T. vaginalis, and 34.8% of these cases were positive by direct microscopy prior to further incubation. Based on symptoms routinely used in syndromic management of STIs (i.e. genital ulcer, lower abdominal pain or abnormal vaginal discharge), 43.5% of the 23 women with T. vaginalis infection were asymptomatic. The most commonly self-reported symptoms amongst all participants included vaginal discharge

Predictors of T. vaginalis infection
Univariable analysis indicated that T. vaginalis infection was more common among participants who were traditionalist or reported having no religion compared to participants who were Christians or Muslims, used water to clean themselves after visiting the toilet compared to those who used tissue paper or other solid materials, reported having ≥ 2 lifetime sexual partners, reported dyspareunia, had genital warts, and/or had a genital ulcer (Table 1). In multivariable analysis, the only independent predictor associated with T. vaginalis was having a genital ulcer (AOR = 7.6, 95% CI: 1.4-42.3).

Genotyping analysis
The  Table 2). The most prevalent actin genotype was E, representing 50.0% of the isolates. The other genotypes were, in order of descending frequency, N (27.3%), G (13.6%) and I and P (each 4.5%).
Multiple sequence analysis to compare polymorphic sites found on our actin sequences, and those retrieved from the GenBank, revealed a total of 33 single nucleotide differences in the open reading frame of the actin gene. Three of these single nucleotide polymorphisms were exclusively found in actin sequences from our study (Additional file 2). The nucleotide sequences obtained of the actin gene for all the 22 isolates were submitted in GenBank under accession numbers: (MF350322-MF350343). The phylogenetic analysis ( Fig. 1) showed that actin genotype E clustered with a bootstrap value of 99. Lower bootstrap values were observed for actin genotypes N, G, I and P.

Prevalence of T. vaginalis viruses
TVVs were present in 43.5% (95% CI: 23.2-65.5) (10/23) of T. vaginalis isolates. Trichomonas vaginalis virus type 1 (TVV1) was the most prevalent (39.1%), followed by indicates the presence of a restriction cut site TVV2 (26.1%), TVV3 (17.4%) and TVV4 (13.0%). Nine out of 10 T. vaginalis isolates with TVVs harbored more than one type of TVV (Table 3). TVV1 was present in all virus-positive T. vaginalis isolates but one (TV207). All 11 genotype E isolates were virus-negative. Finally, the distribution of symptomatic and asymptomatic cases was not linked to any particular T. vaginalis actin genotype. Similarly, presence or absence of TVVs did not appear to have an influence as to whether a patient was symptomatic or asymptomatic.

Discussion
To our knowledge, this is the first study determining T. vaginalis genotypes and co-occurrence of T. vaginalis viruses in Kenya. We sequenced the actin gene for 22 isolates and identified five types of T. vaginalis by in silico RFLPanalysis of the amplified actin gene. We found notable genetic diversity by full actin gene sequence analysis among T. vaginalis isolates in Kilifi, as well as those retrieved from GenBank. Prevalence of T. vaginalis in this study (6.6% in 349 pregnant women) was high but, however, fell short of the prevalence of T. vaginalis among women aged 15-49 years in the World Health Organization Africa region in 2012, which was estimated to be 11.5% (95% CI: 9.0-14.6) [1]. This is suggestive of lower rates of T. vaginalis among the general population of women in Kilifi, Kenya compared to other African countries [31].
Although T. vaginalis does not traditionally present with genital ulcers [32], multivariable analysis showed that genital ulcers were the only predictor of an infection with T. vaginalis in our study. Ulcers could not be due to syphilis, which was not diagnosed in any of the women with T. vaginalis, but we did not test for the Herpes simplex virus, which is also associated with genital ulcers. The association of genital ulcer with T. vaginalis is not unique to our study as it has been reported amongst female sex workers in China [33].
Nucleotide sequence analysis of actin sequences showed 33 polymorphic sites, three of which caused amino acid substitution. Two of these amino acid substitutions have been previously reported to occur in genotypes G, N, I and P, in which nucleotide 371 substituted alanine for valine, and nucleotide 904 substituted lysine for glutamine [20,34]. A unique polymorphism leading to an amino acid substitution, in which nucleotide 892 substituted threonine for serine, was observed to be exclusively present on a GenBank sequence (accession number XM_001301892). As such, in silico genotyping of isolates provides an opportunity to distinguish closely related isolates based on these polymorphic sites and to further identify such polymorphic sites.
Our phylogenetic analysis confirms RFLP as a good typing method, as the results from this method were in agreement with phylogenetic analysis. Phylogenetic analysis and detection of carriage of TVVs, revealed that none of the isolates of the most prevalent actin genotype E harbored a TVV. Furthermore, phylogenetic analysis indicated that genotype E formed a distinct phylogenetic lineage, suggesting clonal stability of this genotype [35].
The high prevalence (43.5%) of TVVs found in this study is comparable to a prevalence of 55% (95% CI: 38.4-70.7) among Cuban isolates [36]. Two recent studies have reported lower carriage of TVV, 18.7% (95% CI: 11.5-28.0) in the Philippines [37] and 17.3% (95% CI: 7.8-31.4) among Iranian isolates [38], although the latter study only determined the presence of TVV1. However, higher prevalence rates of TVV have been reported as well, 81.9% (95% CI: 71.1-90.0) in South Africa [39], and 75.0% (95% CI: 55.1-89.3) in Baltimore City, Maryland [40]. The presence of TVV, in addition to metronidazole susceptibility, has been found to differ significantly between T. vaginalis isolates genotyped by a panel of 21 microsatellites and six single-copy genes of T. vaginalis, which classifies T. vaginalis into two types: type 1 and type 2 [23]. Type 2 is characteristically free of TVV and resistant to metronidazole [23]. Metronidazole susceptibility in relation to actin genotypes is yet to be determined.
Fifty-seven percent of isolates in our study did not harbor TVV, suggesting that they might be of type 2. Our study might have been biased towards type 2; 63.6% were recovered from patients who were diagnosed by culture, after direct microscopy had been determined to be negative, suggesting that the parasite load in the patient was low. Conrad et al. [23] observed that type 1 parasites are often diagnosed by direct microscopy and suggested that this may be indicative of higher parasite load in type 1 which harbor TVVs. Additional studies, sampling a more diverse population and other regions in Kenya, are needed to confirm the population type and distribution of T. vaginalis in the country.
A total of 22 TVVs were identified in 10 T. vaginalis cultures, with multiple TVVs detected in nine cultures ( Table 2). The higher prevalence of T. vaginalis cultures with either TVV1 or TVV2 than with TVV3 or TVV4 is consistent with previous reports [10,37]. In these publications, concurrent TVV infection, with at least two or 3 TVVs, was recorded in six and three T. vaginalis sample cultures, respectively. Although we identified single actin genotypes in all T. vaginalis cultures, which is indicative for the presence of only 1 T. vaginalis strain per culture, we cannot rule out that the presence of the multiple TVVs may be the result of a mixture of T. vaginalis strains, each infected with a different TVV. Therefore, our data does not necessarily indicate concurrent infection of TVVs in a single TV strain. The lytic cycle of TVVs is yet to be described, and attempts to infect uninfected isolates have been unsuccessful [41]. Therefore, it is plausible that the virus may solely be acquired through vertical transmission, making its presence an important genetic marker [42].
Fifty-six percent of women with T. vaginalis infection in our study were symptomatic and the symptoms occurred independently of the presence of TVV in the protozoon. In vitro studies have shown that TVVs are sensed by the  Sample ID TVV1 TVV2 TVV3 TVV4 Genotype Symptomatic a   TV279 + - human epithelial cells via Toll-like receptor 3, triggering Interferon Regulating Factor − 3, interferon type I and pro-inflammatory cascades previously implicated in preterm birth and HIV-1 susceptibility [12]. While treatment with metronidazole generally eliminates T. vaginalis, this may aggravate T. vaginalis-associated inflammation caused by the release of TVV by stressed or dying parasites [12]. Additionally, TVVs upregulate levels of phenotypically variable immunogen mRNA P270 of T. vaginalis [13], while also playing a role in T. vaginalis protein composition and its growth kinetics [14].

Conclusion
Our study was limited by the small number of isolates, which rendered it difficult to investigate the implication of TVV carriage on clinical signs and symptoms. Despite the low number of T. vaginalis isolates, the presence of all four types of TVVs in our isolates, in addition to the five actin genotypes, demonstrates there is notable genetic diversity of T. vaginalis isolated from pregnant women in Kilifi, Kenya. Isolates of the most prevalent actin genotype E lacked TVVs; further studies with higher number of strains should be conducted in order to corroborate these results. The actin gene should be considered as a potential genetic marker for molecular epidemiology and genotypic traits of T. vaginalis.

Additional files
Additional file 1: Table S1. Genotype, number and position of restriction sites using HindII, MseI and RsaI restriction enzymes for actin sequences retrieved from GenBank. (DOCX 15 kb) Additional file 2: Figure S1. Alignment of the T. vaginalis actin gene nucleotide sequences retrieved from GenBank and those of from the clinical T. vaginalis isolates of the present study. (PDF 1746 kb) Abbreviations RFLP: Restriction fragment length polymorphism; TVV: Trichomonas vaginalis viruses