Trichomonas vaginalis Transports Virulent Mycoplasma hominis and Transmits the Infection to Human Cells after Metronidazole Treatment: A Potential Role in Bacterial Invasion of Fetal Membranes and Amniotic Fluid

Mycoplasma hominis is considered an opportunistic pathogen able to colonize the lower urogenital tract; in females the infection is associated with severe pregnancy and postpartum complications, including abortion, endometritis, preterm delivery, and low birth weight. Molecular mechanisms of pathogenicity and virulence effectors remain poorly characterized. A number of studies in the last decade have demonstrated that M. hominis can establish an endosymbiotic relationship with Trichomonas vaginalis, a urogenital parasitic protozoon, also associated with adverse pregnancy outcomes. Recently, two bacterial genes (alr and goiB) associated with amniotic cavity invasion and a single gene (goiC) associated with intra-amniotic infections and high risk of preterm delivery have been identified in M. hominis isolated from a group of pregnant patients. In this work we demonstrate that a high number of M. hominis intracellularly associated with T. vaginalis have goiC gene, in association with alr and goiB. In addition, we demonstrate that metronidazole treatment of M. hominis-infected T. vaginalis allows delivering viable intracellular goiC positive M. hominis from antibiotic-killed protozoa and that free M. hominis can infect human cell cultures. Results suggest that molecular diagnostic strategies to identify both pathogens and their virulence genes should be adopted to prevent severe complications during pregnancy.


Introduction
Preterm birth is a major cause of neonatal diseases and accounts for 75% of perinatal mortality [1,2]. Preterm labor and premature rupture of membranes can be initiated by multiple mechanisms, but in most cases a precise cause cannot be established. In the last years, several studies have shown a significant and strong association between preterm birth and intrauterine infections, accounting for at least 25-40% of cases [3,4]. Infections induce a robust inflammatory response that can stimulate uterine contractility and trigger spontaneous preterm labor [5,6]. In most cases microorganisms reach the uterus and placental membranes via ascending route from vagina or through haematogenous spread from different sources [7].
Mycoplasma hominis is one of the microorganisms most commonly associated with preterm labor and it has been isolated in 40% of amniotic fluids showing infection [8].
Recent studies have suggested the occurrence of genetic variations among different M. hominis isolates with regard to their potential to invade the amniotic fluid and membranes. Three genes (alr, goiB, and goiC) have been identified in M. hominis isolated from amniotic fluids and the placenta of women with preterm labor, but not in the reference strain PG21, isolated from human intestine [9]. The gene alr encodes for alanine racemase and is involved in the peptidoglycan synthesis, yet the function of this enzyme in Mycoplasma species is not clear. goiB (gene of interest B) has unknown functions and encodes a protein that aligns with an 2 Journal of Pregnancy Ureaplasma urealyticum hypothetical protein (41% identity and 63% similarity), while goiC (gene of interest C) encodes a 55 kDa polypeptide that appears to be strictly specific for M. hominis. Among the three genes, goiC is significantly associated with amniotic infection and preterm labor and can thus be considered a virulence trait of the M. hominis strains able to infect the amniotic cavity and the placenta [9].
Interestingly, M. hominis can establish a symbiotic relationship with T. vaginalis, a sexually transmitted protozoon associated with adverse pregnancy outcomes [10]. M. hominis has the ability to enter and survive in protozoan cytoplasm, where it multiplies in coordination with the eukaryotic host [11]. It has been demonstrated that the percentage of T. vaginalis infected by Mycoplasma hominis ranges from 5 to 89% regardless of the geographic origin [12][13][14][15]. For a recent review see Fichorova et al. [16]. The M. hominis-T. vaginalis consortium strongly influences the pathobiology of the protozoon and contributes to upregulating the host inflammatory response to the infection [17][18][19].
The protozoon Trichomonas vaginalis has been also linked to preterm birth, but in this case the infection is limited to the vagina, without reaching the uterus and placental membranes [20]. The effective role of T. vaginalis in preterm labor is debated, since the mechanisms involved are still unclear. Data on protective effect of metronidazole treatment on adverse pregnancy outcomes are contradictory. In some studies antibiotic therapy seems to be effective in preventing adverse pregnancy complications, but several papers report a failure to prevent preterm delivery by metronidazole treatment in pregnant women with T. vaginalis infection [21,22].
The first objective of this work was to understand if goiC (in association with alr and goiB) is present in M. hominis able to establish intracellular symbiosis with T. vaginalis clinical isolates, thus representing an additional potential risk factor for adverse maternal outcomes during trichomoniasis. In addition, we set up an in vitro model to assess the transmissibility of virulent (i.e., goiC positive) M. hominis released from metronidazole-treated T. vaginalis to humanderived cells. Results obtained suggest an active role for protozoa not only in transport but also in transmission of bacterial infection to human tissues during pregnancy.

Cells and Culture Conditions.
A total of 34 T. vaginalis strains were isolated in Italy and Mozambique from 1994 to 2017. Protozoa were isolated from vaginal swabs of women with trichomoniasis by inoculation in Diamond's TYM medium (trypticase, yeast extract, maltose) supplemented with 20% fetal bovine serum (FBS), penicillin (300 U.I/ml), and streptomycin (300 mg/ml), in order to eliminate concomitant vaginal flora [23]. Protozoa were then cultured by 1:16 daily passages in Diamond's TYM medium without antibiotics at 37 ∘ C in a 5% carbon dioxide atmosphere for at least three weeks and stored at −80 ∘ until use.
The immortalized human cell line WISH (ECACC catalog code: 88102403) was maintained in RPMI 1640 supplemented with 10% FBS at 37 ∘ C in a 5% carbon dioxide atmosphere, in T-25 flasks. Once cell cultures reached the confluence (twice a week) adherent cellular monolayer was enzymatically dissociated with trypsin, and detached cells were passed, 1/10-1/20, in complete RPMI medium supplemented with 10% fetal bovine serum.
M. hominis isolated from T. vaginalis and the bacterial reference strain PG21 were cultivated in SP4 agar plates and SP4 broth.
Minimal Inhibitory Concentration (MIC) for metronidazole and gentamicin were calculated by incubating T. vaginalis SS-49 and M. hominis isolated from the same protozoan strain, with serial dilutions of drugs in liquid media (Diamond's TYM and SP4 broth, respectively).

Selection of M. hominis-Parasitized T. vaginalis Isolates.
The presence of M. hominis in T. vaginalis in the 34 strains included in this study was assessed by specific PCR. For each strain, DNA was extracted from 10 6 mid-log-phase trichomonad cells as previously described. DNA was then resuspended in TE buffer (10 mM Tris-HCl pH 8, 1 mM EDTA) at a final concentration of 0.1 g/ l and subjected to a Multiplex PCR assay for the detection of T. vaginalis and M. hominis [24].
M. hominis were isolated from each positive T. vaginalis strain in selective media. Briefly, protozoan cultures were centrifuged at 350 x g, and T. vaginalis-free culture supernatants were filtered through a 0.45-m filter membrane and finally inoculated in SP4 agar plates. Plates were incubated at 37 ∘ C until the appearance of detectable fried egg shape colonies on the agar surface. Bacterial DNA was extracted as described.

Identification of Virulence Genes in M. hominis Associated with T. vaginalis.
The presence of genes alr, goiB, and goiC in Mycoplasma hominis isolated from protozoan strains was assessed by specific PCR as described by Allen-Daniels et al. [9]. The intestinal M. hominis reference strain PG21, lacking the alr, goiB, and goiC genes, and the T. vaginalis reference strain G3 that is not parasitized by M. hominis, were used as negative controls. In order to confirm PCR results, all amplicons were sequenced (BMR Genomics, Padova, Italy).

Transmissibility of M. hominis Infection from Metronidazole-Treated T. vaginalis to Human-Derived Cells.
The transmissibility of intracellular M. hominis from Mycoplasmainfected T. vaginalis to human-derived cells was studied. Exponentially growing T. vaginalis SS-49, naturally infected by M. hominis positive for alr, goiB, and goiC genes, were extensively washed in phosphate buffered saline (PBS), in order to eliminate nonadherent extracellular mycoplasma cells. Protozoa were then resuspended in 500 l of RPMI medium supplemented with 10% FBS, added to a T-25 flask of semiconfluent WISH cells at 2:1 protozoa/human cells ratio, and incubated at 37 ∘ C in a 5% CO 2 atmosphere. After 30 minutes of incubation, cells were extensively washed with PBS to remove protozoa and WISH cells were trypsinized The effect of metronidazole on Mycoplasma transmissibility was assessed in a second experiment: WISH cells, infected with T. vaginalis as described above for 30 minutes, were treated by adding to the flask 25 g/ml of metronidazole (ten times the minimal lethal concentration for the strain SS-49, data not shown). After 24 hours of incubation, cells were extensively washed in PBS to eliminate killed protozoa and cellular debris and trypsinized as described.
In order to demonstrate that M. hominis associated with T. vaginalis can invade human cells and survive intracellularly, the same experiment has been carried out adding a further step. Briefly, after 24 hours of incubation of WISH cells in the presence of T. vaginalis SS-49 with metronidazole, gentamicin was added to the cells at concentration of 50 g/ml (four times the minimal lethal concentration for the M. hominis strain, data not shown) for 3 hours to kill extracellular bacteria. Cultures were then extensively washed in PBS and cultured for three additional weeks in complete RPMI 1640 medium, without any drugs. In all experiment the number of M. hominis cells associated with WISH cells was quantified by qPCR according to the protocol described by Ferandon et al. [25].
In all experiments, one aliquot of trypsinized cells was inoculated into BE broth to reisolate Mycoplasma hominis, and a second aliquot was subjected to DNA extraction as previously mentioned. All the experiments were carried out in triplicate.

T. vaginalis Isolates Can Be Parasitized by goiC Positive M. hominis Strains.
We demonstrated by PCR that 29 out 34 T. vaginalis isolates included in this study were stably parasitized by M. hominis. M. hominis were isolated in SP4 medium from all T. vaginalis positive strains, and the presence of goiC, arl, and goiB virulence genes associated with adverse pregnancy outcomes and preterm delivery was assessed by PCR. Among 29 M. hominis tested the goiC gene is present in 17 strains (58.2%), while goiB and alr are detected, respectively, in 11 (37.93%) and 28 (96.55%) isolates. The goiC gene is strongly associated with alr (16/17) and only partially with goiB (6/17). Only 6 out 29 (20.7%) M. hominis have all three genes. M. hominis reference strain PG21 and the T. vaginalis reference strain G3, that is M. hominis free, do not possess alr, goiB, and goiC genes. Results are shown in Figure 1.

T. vaginalis Parasitized by Virulent M. hominis Can
Deliver Bacteria Able to Infect Human Cells. We hypothesize that the massive release of virulent intracellular M. hominis from metronidazole-killed T. vaginalis following antiprotozoan therapy could mediate the infection of amniotic fluid and membranes. In order to demonstrate this hypothesis, we set up in vitro experiments coculturing a human-derived cell line WISH and T. vaginalis infected with M. hominis positive for all virulence genes (alr, goiB, and goiC). First experiment was carried out coincubating protozoa and WISH cells for 30 minutes in absence of metronidazole. Results obtained by qPCR show that M. hominis delivered from Mycoplasmaparasitized T. vaginalis live cells can infect less than 0.2% of WISH cells after 30 minutes of incubation.
On the contrary, M. hominis were massively delivered from metronidazole-killed protozoa after 24 hours of incubation and viable bacteria were able to infect human cells with a multiplicity of infection (MOI) corresponding to 1.2 M. hominis/3 WISH cells. These data suggest that free bacteria quickly adhere to human target cells once delivered from killed protozoa. In all experiments, infecting M. hominis were reisolated from mammalian cells by using mycoplasmaspecific media. A further experiment was assessed adding gentamicin in medium cultures (gentamicin protection assay), in order to establish if M. hominis released by T. vaginalis treated with metronidazole can also invade and chronically survive into human cells. qPCR demonstrate that M. hominis can survive intracellularly for three weeks in WISH cells exposed to gentamicin, with a MOI corresponding to 1 bacterium/4 human cells. Results were also confirmed by isolation of bacteria from gentamicin treated human cells in mycoplasma-specific media.
Results obtained suggest that M. hominis released by T. vaginalis after treatment with metronidazole can lead to chronic infection in fetal membranes and amniotic fluid, causing adverse pregnancy outcomes, and can also explain the pharmacological failure in preventing adverse pregnancy complications by use of metronidazole to treat subacute trichomoniasis.

Discussion
Preterm birth is a major cause of neonatal illness and death, especially in developing countries. Local and systemic microbial infections are important causes of preterm labor and premature rupture of membranes [1,26]. M. hominis and T. vaginalis infections are both associated with adverse pregnancy outcomes. T. vaginalis limits its colonization to the vagina [27,28], and the infection seems to play a role in adverse pregnancy outcomes by inducing a massive local inflammation and the production of proinflammatory cytokines [29,30]. On the contrary, M. hominis can invade the amniotic cavity, thus directly exploiting virulence mechanisms in this microenvironment. Even if mechanisms of pathogenicity and virulence genes involved in adverse pregnancy complications associated M. hominis are not fully characterized, Allen-Daniels and colleagues recently identified two genes (arl and goiB) in bacterial strains isolated in amniotic fluid and placental tissue and a third gene (goiC) that is significantly associated with amniotic fluid invasion and preterm labor risk [9]. Even if the effective function of the three genes is not clear, the authors hypothesize that goiC could contribute to the colonization of the placenta and the amniotic fluid rather than the vagina [9].
M. hominis infection can be mediated in vivo by T. vaginalis, since the two microorganisms are able to establish a symbiotic relationship, and most protozoan isolates are stably infected by the bacterium [10]. In a study conducted in Italy, 78.6% of women with trichomoniasis were affected also by M. hominis [24]. So far, the presence of alr, goiB, and goiC genes has been verified only in M. hominis isolated from patients that had no history of trichomoniasis. We investigated the presence of alr, goiB, and goiC genes in M. hominis strains that live in symbiosis with T. vaginalis. Our data reveal that not only free M. hominis but also those that live in symbiosis with T. vaginalis can possess the three genes associated with amniotic membranes colonization and adverse pregnancy outcomes.
The ability of M. hominis isolates to locate intracellularly has been previously demonstrated by several authors, in different human-derived cell lines and spermatozoa [31][32][33][34]. Hopfe et al. demonstrated that mycoplasmal infection of host cells is mediated by bacterial cytoadhesins [31]. In addition, Henrich et. al characterized several M. hominis genes involved in HeLa cells intracellular infection [35].
We demonstrate that M. hominis released by T. vaginalis are able to infect WISH cells in vitro. However, since T. vaginalis is highly cytopathic and induces a massive destruction of the cell monolayer in less than 2 hours, we had to limit the coincubation to a very short time. To prevent target cell lysis, we added metronidazole to the cells after 30 minutes of coincubation, demonstrating that M. hominis released by T. vaginalis and killed by the drug can efficiently and stably invade human cells. The quantification by qPCR of M. hominis associated (i.e., membrane associated and intracellular bacteria) with human cells after 24 hours of coincubation of metronidazole-treated T. vaginalis and WISH cells reveals that about 40% of mammalian cells are infected by the bacteria.
Results obtained by gentamicin protection assays demonstrate the ability of M. hominis released by metronidazolekilled T. vaginalis to locate intracellularly in mammalian cells. In fact, since gentamicin kills only the extracellular M. hominis, the detection of bacteria in infected WISH after three weeks of cultivation in the presence of gentamicin indicates that human cells can be chronically infected by intracellular bacteria, as previously demonstrated by Hopfe by using HeLa cells [31]. These data confirm the ability of mycoplasmas released by T. vaginalis to infect human host cells and to locate intracellularly, suggesting a role of T. vaginalis infection in transmission of M. hominis. A similar result has been reported by Fichorova et al.: in an elegant paper the authors demonstrated that metronidazole-killed T. vaginalis can deliver intracellular endobiont dsRNA virus (TVV) and that free viruses, even if they are unable to directly infect human cells, can stimulate a massive proinflammatory response: the production of cytotoxic cytokines can lead to severe complications during pregnancy [36].
These results suggest that the role of protozoan infection in adverse pregnancy outcome could not be limited to the induction of vaginal inflammation. In fact, the peculiar T. vaginalis/M. hominis symbiosis represents an additional potential risk factor for adverse maternal outcomes and preterm delivery during trichomoniasis. T. vaginalis can carry M. hominis possessing the alr, goiB, and goiC genes, protecting them intracellularly from the host immune response and antibacterial therapy, thus allowing their multiplication and transmission. The intracellular localization of bacteria in T. vaginalis cells can explain the paradoxical results described by some authors, reporting the failure of metronidazole treatment of subclinical trichomoniasis to prevent preterm delivery in pregnant women [21,22]. In this scenario, the anti-T. vaginalis treatment with drugs selectively effective against trichomoniasis, probably together with cytolysis of protozoa mediated by host immune response, could induce a massive release of M. hominis from killed T. vaginalis, leading to bacterial invasion of placental membranes and amniotic fluid.

Conclusions
The symbiotic consortium between T. vaginalis and M. hominis implies a role in infections during pregnancy: we can hypothesize a primary role for T. vaginalis as "Trojan horse", able to transport virulent bacteria, protecting them not only from local massive host innate and adaptive immune response, but also from antimycoplasma antibiotics unable to cross the protozoan membrane. In consequence, molecular diagnostic strategies to identify both pathogens and their virulence genes might be adopted to prevent severe complications during pregnancy.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
Trung Thu Tran Thi and Valentina Margarita contributed equally to this work.