Culture of Trichomonas and Tritrichomonas has been approved by the Boston University Institutional Biosafety Committees at Boston University and the University of California-Los Angeles. Mouse infections with Tritrichomonas were performed under ABSL-2 protocols (AN-15352) with the approval of the Boston University Institutional Animal Care and Use Committee (BU IACUC). Use of metronidazole-resistant and metronidazole–sensitive Trichomonas, which were previously axenized from de-identified clinical isolates at the CDC, has been reviewed by the Boston University Institutional Review Board and judged to not be human subjects research. The organotypic EpiVaginal tissue cells, which derive from de-identified hysterectomy specimens, are a catalog item from MatTek Corporation (Ashland, MA).

The genome project G3 strain of Trichomonas, the parental strain (B7RC2) for the ricin-resistant mutants, and the two mutants (4–12 and 2E2) have all been described previously [24,47]. The D1 strain of Tritrichomonas was received from Lynnette Corbeil of UCSD [2]. Metronidazole-sensitive and metronidazole-resistant Trichomonas were axenized (grown without bacteria) from clinical isolates at the CDC [9,48]. Parasites were cultured axenically in TYI-S-33 medium containing 10% bovine serum [49]. Human ectocervical cell line (Ect1 E6/E7), which was a generous gift from Dr. Raina Fichorova of the Brigham and Women’s Hospital, Boston MA, was grown in defined medium, as previously described [24,45]. Organotypic EpiVaginal tissue cells were maintained in DMEM medium containing 10% serum [46].

Logarithmic-phase trophozoites were concentrated by low speed centrifugation and washed in chilled phosphate buffered saline (PBS). Cyanovirin-N-labeled trophozoites were washed three times in chilled PBS and then fixed for 10 min at 4°C in 2% paraformaldehyde in 100 mM phosphate, pH 7.4. Cyanovirin-N and griffithsin (1 μg per 100 μl PBS), as well as the anti-2G12 monoclonal antibody made in plants (0.1 μg per 100 μl PBS) (a generous gift of Kevin Whaley of Mapp Biopharmaceutical, San Diego, CA), were labeled with Alexa Fluor dyes and then incubated with intact trophozoites and washed, as described for cyanovirin-N. Recombinant MBL (from Sigma-Aldrich, USA) were also labeled with Alexa Fluor dyes and incubated with intact trophozoites. Nuclei were stained with 0.1 μg/ml DAPI, and organisms were visualized with a DeltaVision deconvoluting microscope (Applied Precision, Issaquah, WA). Images were taken at 100 X primary and deconvolved using Applied Precision’s softWoRx software.

Measurement of the relative amount of cyanovirin-N to Trichomonas was performed using a BD FACSCalibur flow cytometer (BD Biosciences, Woburn, MA). The specificity of binding of cyanovirin-N to N-glycans was determined in three ways. First, we compared the amount of lectin bound to untreated Trichomonas versus Trichomonas grown for two days in the presence of 1 μg/ml tunicamycin, which inhibits the phosphoglucosyltransferase (encoded by the Alg7 gene) that catalyzes the first step in N-glycan precursor synthesis [26]. Second, trichomonads were electroporated with a morpholino to Alg7 gene (ACTCTTGTTGACATATTTGATTGTT) and then cultured for one day prior to labeling with cyanovirin-N, using methods similar to those for morpholino knockdowns in Giardia [50]. Third, a Saccharomyces mnn1/mmn4 double-knockout was made [44]; yeast were broken and walls were purified; and N-glycans were released from yeast walls with PNGaseF. The binding of cyanovirin-N to Trichomonas +/- yeast N-glycans was then measured using flow cytometry. Similar methods were used to measure binding of recombinant MBL to Trichomonas. Each flow experiment was repeated three times.

Agglutination by host MBL and galectin-1 or by anti-retroviral lectins (cyanovirin-N and griffithsin) was performed by incubating log-phase Trichomonas (12 X 10⁴ parasites in 200 μl of TYM medium without serum and antibiotics) with 2.5 μg of lectin for 60 min at 37°C. In this assay, control Trichomonas incubated without lectin (control group) are flagellated and swimming, so that they are evenly distributed between four equal 50 μl volumes collected from top to bottom of the tubes. In contrast, Trichomonas agglutinated by the lectins precipitate to the bottom of the tube. The experiment was performed in duplicate and repeated twice.

For ectocervical cell assays, parental strain and ricin-resistant Trichomonas mutants were labeled with 10 μM CellTracker Blue CMAC (Invitrogen) and incubated with no lectin or with cyanovirin-N or 1 μg/ml griffithsin (each 1 μg per 100 μl PBS). Parasites (~100,000 per 500 μl PBS) +/- lectins were allowed to adhere to ~300,000 ectocervical cells grown to confluence on a coverslip in a 24 well plate [24]. After 30 min at 37°C coverslips were washed x 4 in PBS to remove nonadherent parasites, fixed in 4% paraformaldehyde, and mounted on slides. Fluorescenated Trichomonas present in fifteen 20X fields were counted using Scion Image for Windows. Each treatment was performed in triplicate, and each experiment was performed at least three independent times.

For organotypic EpiVaginal tissue cell assays, logarithmic phase parental strain and ricin-resistant Trichomonas mutants were labeled with 10 μM CellTracker green (CMFDA) for 5 min at RT (Invitrogen). CMFDA is concentrated in cytoplasmic vesicles of Trichomonas. Some Trichomonas trophozoites were also labeled with cyanovirin-N conjugated to Alexa Fluor 594 (1 μg per 100 μl PBS). Trichomonads were washed, resuspended in culture medium without serum, and incubated with EpiVaginal tissue cells for 60 min at 37°C under anaerobic conditions. Three washes in PBS were used to remove unbound parasites from EpiVaginal tissue cells, which were fixed in 2% paraformaldehyde, mounted on a two-chambered cover glass system (Thermo Fisher Scientific, Waltham MA), and observed with the deconvoluting microscope.

Six weeks old female BALB/c mice were purchased from Taconic (Hudson, NY). Five mice per treatment were infected vaginally with 500 parasites in 10 μl medium +/- 5 μg cyanovirin-N, griffithsin, 2G12 monoclonal antibody, or galectin-1 [2]. Alternatively, Tritrichomonas was infected vaginally with N-glycans released from Saccharomyces mnn1/mmn4 double-knockout. After two days 100 μl PBS was used to wash out Tritrichomonas parasites, which were incubated overnight in TYM medium to release organisms from viscous vaginal secretions. Parasites were fixed and counted. Each experiment was repeated at least three times.

Our hypothesis here is that host MBL, which binds to unmodified N-glycans of HIV, will also bind to unmodified N-glycans on Trichomonas (Fig 1A) [33–36]. In support of this idea, we found that recombinant MBL binds to the surface of Trichomonas (Fig 1B). The non-homogeneous binding of MBL (and of anti-retroviral lectins in Figs 2A, 3A, 3B and 4B) is likely subsequent to the ridged and ruffled surface of trichomonads labeled in solution. The specificity of the binding of MBL to unmodified Trichomonas N-glycans was shown by decreased binding of MBL to Trichomonas pre-treated with an Alg7 morpholino, which decreases expression of the enzyme that is the first step in N-glycan precursor synthesis (Fig 1C) [26]. Trichomonads, which are flagellated and swim as individuals in solution, are agglutinated by MBL and sink to the bottom of a test tube (Fig 2B and 2C). These aggregates resemble “swarms” of Trichomonas formed in response to stress [51]. Similarly, recombinant galectin-1, which binds to the surface of Trichomonas [22], also causes flagellated trichomonads to self-aggregate and precipitate.

Because Trichomonas weakly infects mice while Tritrichomonas makes a robust infection that ascends into the uterus, Tritrichomonas was used to explore the possible role for parasite N-glycans in pathogenesis [2,3]. Tritrichomonas was directly applied to the mouse vagina +/- N-glycans from a Saccharomyces mnn1/mnn4 double knockout, which makes N-glycans similar to those of Trichomonas and HIV [44]. These yeast N-glycans, which have been used to raise antibodies targeted to the N-glycans present on gp120 of HIV, decreased recovery of Tritrichomonas by 40% (P < 0.001) after two days in the mouse vaginal model (Fig 1D). Because we did not perform a histological examination of the vagina nor are we detecting luciferase expressing parasites by bioluminescence, we cannot rule out that some adherent parasites remain after the vaginal wash. Alternatively, some Tritrichomonas may have migrated to the uterus and were not counted.

The anti-retroviral lectins cyanovirin-N and griffithsin and the anti-carbohydrate 2G12 monoclonal antibody bind to the surface of Trichomonas and Tritrichomonas (Figs 2A, 3A, 3B and 4B) [37–40]. The specificity of cyanovirin-N for Trichomonas N-glycans was shown in two ways [30]. First, treatment of Trichomonas with an Alg7 morpholino or with tunicamycin, both of which target the first enzyme in N-glycan precursor synthesis, decreases binding of cyanovirin-N to the parasites, as measured by flow cytometry (Fig 3C) [26,50]. Second, binding of cyanovirin-N is inhibited by co-incubation with N-glycans released from a Saccharomyces mnn1/mmn4 double-knockout (Fig 3D) [44]. As a control for the adherence experiments (see below), we showed that binding of cyanovirin-N is slightly greater to ricin-resistant mutants (2E2 and 4–12) than to the parent strain (B7RC2) (Fig 3E) [24].

We tested the binding of cyanovirin-N to axenized clinical isolates of Trichomonas from the CDC, which are either sensitive to metronidazole (two isolates) or resistant to metronidazole (five isolates) [9,48]. All but one of the clinical isolates binds cyanovirin-N better than the parent strain (B7RC2) for the ricin-resistant mutants (Fig 3F). The two clinical isolates that bind cyanovirin-N best include one isolate that is metronidazole-sensitive and one that is metronidazole-resistant. These results suggest N-glycan synthesis and metronidazole activation are unrelated, and so anti-retroviral lectins might be used as an additional, local treatment for metronidazole-resistant Trichomonas or for pregnant women, who cannot take tinidazole.

Like MBL and galectin-1, anti-retroviral reagents (cyanovirin-N, griffithsin, and the 2G12 monoclonal antibody) each cause flagellated trichomonads to self-aggregate and precipitate (Fig 2A and 2B). Previous studies have shown that ricin-resistant mutants of Trichomonas adhere less well than the parental strain to ectocervical cells in vitro and that galectin-1 increases adherence [22,24]. Here we found that cyanovirin-N nearly doubles the adherence of ricin-resistant mutants to ectocervical cells, while griffithsin has a more modest effect on adherence (Fig 4A). In contrast, neither of the anti-retroviral lectins have much effect on adherence of the parent strain (B7RC2), which adheres well to ectocervical cell monolayers in the absence of lectins.

Close examination of Trichomonas adhering to organotypic EpiVaginal tissue cells, which form a multilayered squamous epithelium that closely resembles that of the vagina [46], showed three findings. First, cyanovirin-N increases adherence to EpiVaginal cells of ricin-resistant mutants, as was shown with ectocervical cell monolayers (data not shown). Second, cyanovirin-N binds well to EpiVaginal cells, which have unprocessed N-glycans on their surface in addition to complex N-glycans that do not bind cyanovirin-N (Fig 4C). Third, when trichomonads adhere to the EpiVaginal cells and transform from a flagellated to an ameboid form, they shed most of the cyanovirin-N that is present on swimming parasites (Fig 4B and 4C). Although cyanovirin-N agglutinates flagellated parasites and causes them to sink to the bottom of a tube or flask (Fig 2B and 2C), cyanovirin-N and the other lectins that bind to Trichomonas do not cause them to convert to an ameboid form, which only occurs when trichomonads adhere to host cells or to a derivatized surface [13–18,22,24,25].

Topical application of cyanovirin-N or griffithsin at the time of infection with Tritrichomonas reduced the recovery of parasites from the vagina of mice after two days by 63% or 70%, respectively (Fig 5). Although the 2G12 monoclonal antibody had little effect on the recovery of Tritrichomonas from the mouse vagina, galectin-1 reduced recovery by 51% (Fig 5). These results, which have the same caveats as experiments with the yeast N-glycans with regards to missing adherent parasites or those that have migrated to the uterus, suggest anti-retroviral lectins have a modest potential for treating or preventing human infections with Trichomonas.

Previous studies showed that adherence of Trichomonas to ectocervical cells is mediated in part by host galectin-1, which is bivalent and binds to LacNAc on parasite LG and to LacNAc on host cells (Fig 4D) [22–24]. The present experiments show that galectin-1 agglutinates flagellated Trichomonas, so they are less able to swim and divide (Fig 2B and 2C). In contrast, galectin-1 decreases recovery of Tritrichomonas in the mouse vaginal model (Fig 5). This discrepancy, which is also the case for the anti-retroviral lectins, suggests the importance of the in vivo model, even if it is with the bovine rather than the human parasite and even if there is a possibility that all parasites have not been obtained in the washes [2,3].

MBL, another important player in the host innate immune response, binds to unmodified N-glycans on the surface of Trichomonas and causes parasites to self-aggregate (Figs 1A–1C, 2B and 2C) [33–36]. Whether complement activation by MBL plays a role in Trichomonas infections was not examined. Recovery of Tritrichomonas in the mouse vaginal model is decreased in the presence of N-glycans from the Saccharomyces mnn1/mmn4 double-knockout (Fig 1D) [44]. This result suggests roles for N-glycans of the host and/or parasite in colonization of trichomonads in vivo.

The results here, as well as recent experiments that demonstrate roles for parasite exosomes, bacterial vaginosis, and host cytokines [7,8,13–15,22–24,52], suggest that mechanisms of pathogenesis by Trichomonas are multiple and complex. Similarly, how anti-retroviral lectins reduce recovery of Tritrichomonas in vivo (next section) is likely more complicated than how these same reagents block invasion of HIV invasion into host cells [37–40,53].

The logic for testing anti-retroviral lectins versus Trichomonas included the following. Spectrometric studies show that many Trichomonas surface and secreted proteins have no known function and are unique to the parasite, so it is difficult to choose vaccine candidates (our unpublished data and [29]). The majority of unique Trichomonas proteins are absent from the predicted proteins of Tritrichomonas (our unpublished data), so that it is necessary to overexpress the Trichomonas vaccine candidate in transformed Tritrichomonas and then use these parasites in the mouse model [2,3,54]. In contrast, nearly all Trichomonas surface and secreted proteins have N-glycan sites. Like HIV and HSV, Trichomonas is sexually transmitted, so an anti-retroviral reagent present in the vagina might be used to prevent all three infections [1,6,37–40]. The big advantage of using the anti-retroviral reagents versus Trichomonas is that GMP production, pharmacokinetic studies, and toxicity tests, all of which are likely prohibitively expensive for a new anti-trichomonad reagent, will have already been performed in an effort to prevent heterosexual transmission of HIV [39,40]. Finally, the use of a topical anti-retroviral reagent along with systemic treatment with tinidazole or metronidazole to treat drug-resistant Trichomonas is similar to the idea of combining anti-retroviral lectins with conventional anti-retroviral drugs to target HIV [55].

Similar to host galectin-1 [16], anti-retroviral lectins and 2G12 antibody increase adherence of trichomonads to ectocervical cells in vitro (Fig 4A). This effect is most likely by cross-linking unprocessed N-glycans on the surface of parasite and host (Figs 3B–3F and 4E). Agglutination of flagellated parasites by the anti-retroviral lectins may also contribute to their effect on adherence in vitro and colonization in vivo (Fig 2A–2C). While anti-retroviral lectins have been shown to decrease infectivity of numerous viruses in animal models, this is the first time that these lectins have been tested versus a parasite in a mouse model (Fig 5) [41–43]. The modest reduction in Tritrichomonas recovered in the presence of anti-retroviral lectin (with many caveats about the counting) is much less than the total elimination of parasites that can be accomplished with metronidazole or tinidazole [9]. Indeed the partial reduction in parasite recovery is more similar to that seen by vaccination against Trichomonas, Entamoeba, or Giardia [10,56,57]. An alternative conclusion from the studies here is that the use of anti-retroviral reagents to prevent heterosexual transmission of HIV and HSV is unlikely to have a large negative effect on Trichomonas infections. Because Trichomonas does not infect easily infect mice, anti-retroviral lectins might be tested in an excellent but expensive pigtail macaque model of Trichomonas infection +/- HIV [58–60].