Tilapia Piscidin 4 (TP4) Reprograms M1 Macrophages to M2 Phenotypes in Cell Models of Gardnerella vaginalis-Induced Vaginosis

Gardnerella vaginalis is associated with bacterial vaginosis (BV). The virulence factors produced by G. vaginalis are known to stimulate vaginal mucosal immune response, which is largely driven by activated macrophages. While Tilapia piscidin 4 (TP4), an antimicrobial peptide isolated from Nile tilapia, is known to display a broad range of antibacterial functions, it is unclear whether TP4 can affect macrophage polarization in the context of BV. In this study, we used the culture supernatants from G. vaginalis to stimulate differentiation of THP-1 and RAW264.7 cells to an M1 phenotype. The treatment activated the NF-κB/STAT1 signaling pathway, induced reactive nitrogen and oxygen species, and upregulated inflammatory mediators. We then treated the induced M1 macrophages directly with a non-toxic dose of TP4 or co-cultured the M1 macrophages with TP4-treated vaginal epithelial VK2 cells. The results showed that TP4 could not only decrease pro-inflammatory mediators in the M1 macrophages, but it also enriched markers of M2 macrophages. Further, we found that direct treatment with TP4 switched M1 macrophages toward a resolving M2c phenotype via the MAPK/ERK pathway and IL-10-STAT3 signaling. Conversely, tissue repair M2a macrophages were induced by TP4-treated VK2 cells; TP4 upregulated TSG-6 in VK2 cells, which subsequently activated STAT6 and M2a-related gene expression in the macrophages. In conclusion, our results imply that TP4 may be able to attenuate the virulence of G. vaginalis by inducing resolving M2c and tissue repair M2a macrophage polarizations, suggesting a novel strategy for BV therapy.


INTRODUCTION
Bacterial vaginosis (BV) is one of the most common vaginal infectious diseases among women of reproductive age, and it is difficult to manage clinically (1,2). BV is caused by disturbances in the vaginal microflora; lactobacilli are depleted, allowing the mucosa to become dominated by aerobic and facultative bacteria (3). Among BV-associated organisms, Gardnerella vaginalis is thought to play a key role in the condition (4,5). Secretion of virulence factors is a major mechanism of G. vaginalis toxicity, as these factors heighten the inflammatory response, decrease the vaginal epithelial integrity, and inhibit tissue repair (6)(7)(8)(9). These effects are largely responsible for BV symptoms and create a more permissive environment for the acquisition and spread of sexually transmitted infections, such as HIV (10). At present, antibiotic treatments are the most common therapy for BV (11); however, accumulating reports have challenged this treatment strategy. Antibiotics not only indiscriminately destroy bacteria, but they may also cause adverse effects on the host immune system, such as damage to lymphocyte DNA or reductions in the numbers and functions of macrophages and lymphocytes (12)(13)(14). Therefore, it will be beneficial to find new therapeutic options for BV that can improve the risk-benefit ratio for patients by maintaining healthy vaginal microflora and immune function.
Monocytes and macrophages are normally found in the lamina propria of vaginal mucosa and play crucial roles in the mucosal barrier, including prevention of pathogen invasion and response to a variety of microbial virulence factors (15,16). Once bacteria-derived stimuli are detected, macrophages differentiate into the classical inflammatory (M1) macrophages through the production of the microbicidal reactive nitrogen and oxygen intermediates, and the activation of nuclear factor-kB (NFkB) and signal transducer and activator of transcription (STAT) 1. The NFkB/STAT pathway then transcriptionally activates proinflammatory molecules, such as interleukin (IL)-12, tumor necrosis factor (TNF)-a, and IL-1b (17). During later stages of the inflammatory response, macrophages respond to autocrine or paracrine (i.e., received from epithelial cells) signals that shift the macrophages from the M1 phenotype to alternatively activated resolving (M2c) or tissue repair (M2a) types (18). M2c macrophages frequently arise from MAPK pathwayinduced IL-10-STAT3 signaling and serve to prevent excessive inflammation. On the other hand, differentiation of M2a macrophages can be driven by IL-4/13-induced STAT6 signaling, and these cells function to restore normal tissue structures (19). Recently, TNF-a-stimulated gene 6 (TSG-6), an anti-inflammatory protein detected in epithelium, has also been shown to activate STAT6 signaling in macrophages (20,21). The polarization of macrophages into suitable types upon receipt of certain stimuli is necessary for effective tissue repair in many contexts. Dysregulation and prolonged macrophage response cause persistent inflammation and poor tissue repair seen in many diseases including BV, consequently increasing the risk of other pathogenic infections (22)(23)(24).
Antimicrobial peptides (AMPs) are a class of short, cationic and amphipathic molecules secreted by immune and epithelial cells in response to invasive bacteria and their products (25,26). Tilapia piscidin 4 (TP4) was recently identified in Nile tilapia (Oreochromis niloticus) (27). Owing to its alpha-helical secondary structure, net positive charge and optimum hydrophobicity, TP4 has high activity against a broad spectrum of bacterial pathogens, and it can eradicate microbial biofilms (28,29). In addition to its low hemolytic activity and cytotoxicity in various models, TP4 may be useful as a potential treatment for infectious diseases (30)(31)(32). Besides its antimicrobial activities, TP4 also has immunomodulatory functions that can provide a counterbalance to the production of inflammatory cytokines in bacteria-induced inflammation and enhance epidermal wound healing (31)(32)(33). However, the underlying mechanisms of this immunomodulation are still unclear.
In this study, we aimed to investigate the modulatory effect of TP4 on macrophage polarization induced by G. vaginalis. Using G. vaginalis-free culture supernatants, inflammatory M1 phenotypes were induced in THP-1 and RAW264.7 cells. We found that either treatment of TP4 directly to M1 macrophages or indirectly to vaginal epithelial VK2 cells co-cultured with the M1 macrophages could reduce the inflammatory mediators in the M1 macrophages. Further, we discovered that TP4 promoted the phenotype of resolving M2c macrophage through MAPK/ERK pathwayinduced IL-10-STAT3 signaling. In the co-culture model, TP4 induced VK2 and endocervical cells to secrete TNF-a-stimulated gene 6 (TSG-6), which elicited the phenotypes of tissue repair M2a macrophages through STAT6 activation. These results suggest that TP4 may reduce the virulence of G. vaginalis by dampening the induced inflammation and promoting tissue remodeling via modulation of macrophage polarizations.

Bacterial Culture and Preparation of Bacteria-Free Culture Supernatants
Gardnerella vaginalis strain (ATCC# 14018) was cultured in Tryptic Soy Broth (TSB; BD Biosciences) with 5% defibrinated rabbit blood (Rockland Immunochemicals) under anaerobic conditions with the AnaeroPack system (Mitsubishi Gas Chemical Company) at 37°C for 72 h. The cultures and control medium (TSB broth) were centrifuged three times for 10 min each at 2,500 rpm at 4°C and then sterilized using a 0.22 mm syringe filter (EMD Millipore) to remove any remaining bacterial debris. The samples were stored frozen at -80°C until use. To generate the model of G. vaginalis-free culture supernatants (GV sup)-induced M1 macrophages, PMAdifferentiated THP-1 cells and RAW264.7 cells were incubated with 5 or 10% (v/v) GV sup in RPMI-1640 and DMEM cell growth media, respectively for 24 h. For the subsequent induction of GV sup-induced THP-1 cells and RAW264.7 cells (THP-1/GV and RAW264.7/GV), 10% (v/v) GV sup in the cell growth medium was used. To produce GV sup-induced epithelial inflammation and apoptosis, VK2 cells were treated with 10% (v/v) GV sup in KSFM cell growth medium.

Nitric Oxide (NO) Assay
The production of NO by the cells was assessed by measuring nitrite accumulated in the culture medium through a Griess reaction (35). Briefly, 100 ml of cell culture medium was mixed with 100 ml of the Griess reagent (Sigma) and incubated at room temperature for 15 min before the absorbance was measured at 540 nm in a plate reader (SpectraMax i3 Multi-Mode Microplate Reader). Nitrite levels were calculated from a standard curve with known concentrations of sodium nitrite (Sigma).

Quantitative Real-Time PCR (qRT-PCR) Analysis
Total RNA was extracted using TRIzol Reagent (Invitrogen), and 1 mg total RNA from each sample was reverse transcribed into cDNA with RT-PCR Quick Master Mix (Toyobo) according to the manufacturer's instructions. The qRT-PCR was performed using StepOne Plus Real-Time QPCR System (Applied Biosystems) and set up in MicroAmp fast 96 well reaction plate (Applied Biosystems). The reaction mixture was prepared in a final volume of 20 ml per reaction. Briefly, 2 ml of sample cDNA was added to 18 ml reaction mix containing 10 ml of SYBR Green Realtime PCR Master Mix (Toyobo) and 8 ml of nucleasefree water (Sigma) with primers at a final concentration of 0.5 mM. All primer sequences are listed in Supplementary Table 1.
The thermocycling was performed according to the manufacturer's instructions. Experiments were performed in triplicate. The gene expression levels were calculated using the 2 −DCt method first normalizing to GAPDH expression as an endogenous control, and the expression level fold-change was calculated relative to the lowest expression level in the control group.

Cytotoxicity Assay
TP4 peptide (H-FIHHIIGGLFSAGKAIHRLIRRRRR-OH) was synthesized and purified by HPLC (GL Biochem), then diluted in sterile PBS before use. The preparation was sterilized by passing it through a syringe filter (PES membrane, pore size 0.22 mm, Millipore) to remove bacteria. The TP4 cytotoxicity was analyzed by MTS/PMS and lactate dehydrogenase (LDH) release assays. VK2 and 10% (v/v) GV sup-stimulated THP-1 and RAW264.7 cells were seeded on 96-well cell culture plates and treated with the indicated concentration of TP4 for 24 h. The positive control was 0.1% Triton-X 100. After incubation, the media were collected for the LDH release assay, and the wells were filled with 20 ml of MTS/PMS mixture (20:1) reagent (Promega) and 80 ml of cell growth medium, followed by incubation at 37°C for 1 h. Absorbance was recorded at a wavelength of 490 nm using a SpectraMax i3 Multi-Mode Microplate Reader. The collected cell culture supernatants were analyzed using a Cytotoxicity Detection Kit (LDH) (Roche) according to the manufacturer's protocol. Briefly, 100 ml of supernatants were incubated with 100 ml LDH reaction mix for 10 min at room temperature. Then, 50 ml of stop solution was added before incubating the samples for 15 min at room temperature. The absorbance of the enzymatic product at 492 nm was measured using a SpectraMax i3 Multi-Mode Microplate Reader.
In the Annexin-V/PI staining assay, VK2 cells were suspended by trypsinization, washed twice with cold PBS, and incubated in binding buffer (Invitrogen) containing Annexin V-FITC (Invitrogen) and propidium iodide (PI) staining solution (Invitrogen) at room temperature for 15 min. Macrophage Treatments, Cell Co-Culture Model, and Wound−Healing Assay In the direct and indirect TP4 treatment models ( Figure 3 and Supplementary Figure 3), PMA-differentiated THP-1 cells and RAW264.7 cells were stimulated with 10% (v/v) GV sup or control TSB broth in the serum-containing medium without antibiotics for 24 h. Then, the cells were washed and incubated in serum-free medium and treated with TP4 (7.82 µg/ml) or cocultured with VK2 cells using cell culture inserts (pore size 0.4 mm; Corning). The insert-cultured VK2 cells had been pretreated with vehicle (PBS) or 7.82 µg/ml TP4 for 8 h and then were changed to fresh KSFM. After 24 h incubation, THP-1 and RAW264.7 cells were collected, and the macrophage phenotypes were assessed.
To measure the anti-inflammatory and anti-apoptotic abilities of macrophages ( Figure 4), VK2 cells were treated with 10% (v/v) GV sup or control TSB broth in KSFM without antibiotics. In co-culture groups, the VK2 cells were simultaneously co-cultured with THP-1/ GV cells that had been pre-treated with TP4 (7.82 µg/ml) or vehicle (PBS) in complete medium for 24 h, followed by incubation in serum-free medium. After 24 h incubation, the cell culture media and VK2 cells were collected to perform measurements.
To investigate the wound repair ability of macrophages ( Figure 5), a linear scratch was made on 100% confluent VK2 cell monolayers with a sterile pipette. The cells were then maintained in 10% (v/v) GV sup-containing KSFM and simultaneously co-cultured with THP-1/GV cells, which had been stimulated with vehicle (PBS)-treated or TP4-treated VK2 cells. The VK2 cells were subcultivated on an insert and incubated with serum-free medium after treatment. Wound healing was observed under a microscope (Leica) and analyzed by ImageJ software. Wound closure (%) = (A 0 -A n )/A 0 × 100, where A 0 represents the area of initial wound area and A n represents the remaining area of the wound at 24 h. After measurement, the cells were collected, and RNA was extracted.
Detection of TNF-a, IL-1b, and IL-6 by MultiPlex Assay The concentrations of cytokines TNF-a, IL-1b, and IL-6 in the VK2 cell culture supernatants were assessed by a MultiPlex assay (Bio-Rad). According to the manufacturer's instructions, the cell culture supernatants were double centrifuged at 6,500 rpm for 15 min at 4°C to eliminate sediments. The samples were frozen at -80°C until being analyzed using a standard test system on a Bio-Plex 200 analyzer (Bio-Rad). Technical support was provided by the Academia Sinica Inflammation Core Facility, IBMS. Each experiment was performed in triplicate and concentrations were calculated based on a standard curve.

Preparation of Conditioned Medium
To prepare the conditioned medium (CM), VK2 cells were seeded at more than 90% confluence and treated with TP4 (7.82 µg/ml) or vehicle (PBS) for 24 h. The next day, cells were washed with PBS, changed to fresh cell culture medium, and incubated for another day. The culture supernatants were then collected and centrifuged at 1,000 rpm for 10 min to remove any remaining cells or debris. For experiments that required neutralization of TNF-a-stimulated gene 6 (TSG-6) in the medium, 10 mg/ml of TSG-6 antibody (sc-398307; Santa Cruz Biotechnology) or control IgG (I9388; Sigma) was incubated with CM at room temperature for 1 h (36).

Statistical Analysis
All data are expressed as mean ± SD. Statistical analyses were performed with GraphPad Prism Software. Multiple comparisons were performed by one-way or two-way analysis of variance (ANOVA). Significant intergroup differences were subsequently tested by post hoc Bonferroni analysis. P values of < 0.05 were considered statistically significant in all cases.

Cytotoxicity of TP4 on Induced M1 Macrophages and Vaginal Epithelial Cell Lines
Mucosal macrophage plasticity and polarization are not only controlled by autocrine mechanisms but also by microenvironmental factors, especially those secreted from epithelial cells (18,38). To study the effects of TP4 treatment on macrophage polarization, we first established the non-toxic doses of TP4 on M1 macrophages induced from THP-1 and RAW264.7 cells (THP-1/GV and RAW264.7/GV) and the vaginal epithelial VK2 cells. MTS/PMS and lactate dehydrogenase (LDH) release assays were performed after treating cells with TP4 for 24 h. The MTS/ PMS assay showed that TP4 treatment only slightly affected viability at a concentration of 15.63 µg/ml in THP-1/GV and VK2 cells (Figures 2A, B), and this dose had no cytotoxicity in RAW264.7/ GV cells (Supplementary Figure 2A). In addition, TP4 treatment did not cause LDH release in THP-1/GV cells ( Figure 2C), and it induced LDH release only at concentrations of at least 15.63 µg/ml in VK2 and RAW264.7/GV cells ( Figure 2D and Supplementary Figure 2B). According to these results, we chose 7.82 mg/ml TP4 as the maximum concentration for the following experiments.

TP4 Reprogramming of THP-1/GV and RAW264.7/GV Cells to M2 Phenotypes
Since GV sup strongly induced the M1 phenotype, we then used GV sup-induced macrophages to determine whether TP4 could affect the macrophage status. As shown in Figure 3A and Supplementary Figure 3A Crosstalk between macrophages and epithelial cells may also affect macrophage status (38). Therefore, we used cell culture inserts to co-culture GV sup-induced macrophages with vaginal epithelial (VK2) cells, testing if TP4 treatment of vaginal epithelial cells might also affect macrophage status. As shown in Figure 3A Figure 3H). Together, these results indicated that TP4 can dampen the inflammatory phenotypes of GV sup-induced macrophages and can reprogram them either to the M2c type through direct effects or to the M2a type through effects on vaginal epithelial cells.

Measurement of Tissue Remodeling Abilities in Different Types of TP4-Induced M2 Macrophages
Exposure to G. vaginalis induces inflammatory mediators and inhibits tissue repair in the vaginal epithelium, causing BV pathogenesis (37,39). The actions of M2 macrophages include dampening inflammatory response (M2c type) and promoting tissue repair (M2a type), largely via secreted factors (18,19). Since TP4 reprogrammed the status of inflammatory macrophages, we next evaluated the M2-related functions of TP4-treated macrophages.
First, we investigated the ability of TP4-induced M2c macrophages to prevent GV sup-induced epithelial inflammation and cell death. As shown in Figure 4A, VK2 cells were treated with control TSB broth [treatment 1] or 10% (v/v) GV sup [treatment 2] for 24 h to induce inflammation and cell death. Meanwhile, the GV sup-treated VK2 cells were co-cultured with THP-1/GV cells that had been previously treated with vehicle PBS [treatment 3] or TP4 [treatment 4] on the cell culture inserts. Compared to control broth (lane 1), GV sup (lane 2) significantly induced nitrite production (lane 2; Figure 4B), ROS levels (lane 2; Figure 4C), and secretion of inflammatory mediators (TNF-a, IL-1b, and IL-6) (lane 2; Figure 4D) in VK2 cells. Co-culture of GV sup-treated VK2 cells with TP4-treated THP-1/GV cells (lane 4) significantly decreased GV sup-induced inflammation (Figures 4B-D). Furthermore, Annexin-V/PI staining ( Figure 4E) and the cleaved poly (ADPribose) polymerase (PARP) ( Figure 4F) showed that exposure to GV sup caused apoptosis of VK2 cells, which could be attenuated by co-culture with TP4-treated THP-1/GV cells. These results indicated that TP4-induced M2c macrophages performed resolving functions to reduce inflammation and cell apoptosis.
Second, we investigated if TP4-induced M2a macrophages could promote wound repair activities in GV sup-treated VK2 cells. As shown in Figure 5A Increased matrix metalloproteinase-9 (MMP-9) expression was also detected ( Figure 5D). Moreover, we found that expression of the tight junction component, zonula occludens-1 (ZO-1), was enhanced after 48 h healing time ( Figure 5E). These results indicated that TP4-induced M2a macrophages can not only promote epithelial cell migration but also increase epithelial barrier integrity after GV sup stimulation.

TP4 Treatment Induces IL-10-STAT3 Activation via MAPK/ERK Pathway
In response to inflammatory stimulation, IL-10 is rapidly expressed by macrophages through activation of the MAPK pathway, which results in suppression of NFkB signaling and activation of downstream STAT3 and M2c-related gene transcription (40,41). Since TP4 is known to have a significant effect on MAPK signaling (42), and we found IL-10 was strongly induced upon TP4 treatment in macrophages ( Figure 3G and Supplementary Figure 3G), we hypothesized that TP4 may regulate IL-10 expression through activation of the MAPK pathway. Using qRT-PCR and immunoblotting, we observed that mRNA ( Figure 6A and Supplementary Figure 5A) and protein ( Figure 6B and Supplementary Figure 5B) levels of IL-10 were dose-dependently increased by TP4 treatment, and this observation corresponded to increased STAT3 phosphorylation (p-STAT3) and decreased p-p65 in THP-1/GV and RAW264.7/ GV cells. Although the active forms of ERK (p-ERK) and p38 (p-p38) were increased after TP4 treatment ( Figure 6C and Supplementary Figure 5C), only pre-treatment with an ERK pathway inhibitor, U0126, could suppress the TP4-induced IL-10 expression and p-STAT3 ( Figures 6D-F and Supplementary  Figures 5D-F). Further, inhibition of the ERK pathway also repressed the TP4-induced expression of M2 marker CD206 ( Figure 6G and Supplementary Figure 5G), as well as SPHK1 in THP-1/GV cells ( Figure 6H), and CD163 in RAW264.7/GV cells (Supplementary Figure 5H). These results demonstrated that treatment of TP4 induces IL-10 expression through the MAPK/ ERK pathway, resulting in activation of downstream STAT3 and M2c-related genes. Results are shown as relative percentage to positive control. Data are presented as mean ± SD of three independent experiments (**P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not statistically significant, compared to 0 mg/ml). were measured by qRT-PCR. Data were represented the normalized target gene amount relative to the group of treatment 1. Data are presented as mean ± SD of three independent experiments (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not statistically significant).

TP4-Induced TNF-a-Stimulated Gene 6 (TSG-6) Secretion From VK2 Cells Is Required for STAT6 Activation and M2a Macrophage Polarization
Using the co-culture model shown in Figure 3, we found that TP4 treatment of insert-cultured VK2 cells promoted the M2a phenotype. This finding suggested that identifiable determinants of M2a polarization may be produced by VK2 cells after TP4 treatment. To investigate the underlying mechanisms of this effect, we collected the conditioned medium (CM) from TP4treated VK2 cells and used it to treat THP-1/GV and RAW264.7/ GV cells. After treatment, p-p65 was decreased, and compared to p-STAT3, the phosphorylation of STAT6 (p-STAT6) was strongly increased in THP-1/GV ( Figure 7A) and RAW264.7/  Figure 7A). STAT6 plays a pivotal role in M2a macrophage differentiation and can be stimulated by soluble factors, such as IL-4 and TNF-a-stimulated gene 6 (TSG6) (21,43). Indeed, we found that TSG-6 was upregulated by TP4 not only in VK2 ( Figures 7B, D, E) but also in endocervical epithelial (End1) cells in dose-and time-dependent manners (Supplementary Figures 6A, C, D). However, IL-4 was not changed ( Figure 7C and Supplementary Figure 6B), indicating that TP4 may modulate macrophage status through TSG-6 secretion by vaginal epithelial cells. To test this hypothesis, TSG-6-neutralizing antibodies were added to the CM from TP4-treated VK2 cells. We found that the addition of TSG-6-neutralizing antibody reduced the increased STAT6 phosphorylation ( Figure 7F and Supplementary Figure 7B), CD206 upregulation ( Figure 7G and Supplementary Figure 7C), and M2a-related gene expression (ARG-1, YM-1, and RELMa in THP-1/GV cells; ARG-1 and YM-1 in RAW264.7/GV cells) ( Figure 7H and Supplementary Figure 7D). These results confirmed that TP4-induced TSG-6 secretion from epithelial cells activated STAT6 signaling in the macrophages and reprogrammed them to the M2a phenotype.

DISCUSSION
BV is a vaginal inflammatory disease caused by dysregulation of commensal bacteria. It is difficult to treat and has high rates of recurrence (44). Metronidazole is the current treatment option for BV, as it can kill G. vaginalis (45). However, almost 60% of G. vaginalis isolates taken from patients are resistant to metronidazole (6). In addition, metronidazole exhibits an unwanted suppressive effect on vaginal Lactobacilli (46). Thus, antibiotics seem to be a suboptimal treatment option for BV, due to their non-selective killing effect on vaginal flora. Instead, modulation of inflammatory status and epithelial barrier function in the vagina have been proposed as alternative therapeutic strategies (47,48). Our study demonstrated that TP4 has multiple beneficial effects in the context Data were represented the normalized target gene amount relative to the group of treatment 1. Data are presented as mean ± SD of three independent experiments (**P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not statistically significant).  (Figures 3, 4). In addition, TP4 can promote tissue remodeling processes of the vaginal epithelium ( Figure 5). Both of these effects are beneficial for relieving BV.
inflammation and tissue repair through induction of IL-10 (41) and regulation of tissue remodeling (57), respectively. The third mechanism may be neutralization of soluble pathogen-associated molecular patterns (PAMPs) or toxins released from G. vaginalis. There are several molecules released from G. vaginalis that are associated with BV, including phospholipases, cholesteroldependent cytolysin, and vaginolysin (58,59). Vaginolysin is considered to be a major cause of vaginal epithelial cytotoxicity and induction of inflammatory response (59,60). Notably, TP4 is a cationic antimicrobial peptide (29), and cationic antimicrobial peptides are able to suppress negatively charged PAMP-induced inflammation via electrostatic attraction-mediated neutralization activity (61). However, the ability of TP4 to directly neutralize G. vaginalis-related PAMPs has not been demonstrated.
Maintaining the integrity of the vaginal epithelial barrier is essential for preventing sexually transmitted diseases caused by harmful microorganisms (62). Patients with BV have increased susceptibility to other sexually transmitted disease pathogens, such as HIV (63). This increased susceptibility is due to immune activation (64), recruitment of HIV target cells (65), and damage to vaginal epithelial barriers (66). Our findings suggest that TP4 might improve vaginal barrier integrity via suppression of epithelial cell inflammation and apoptosis ( Figure 4) and promotion of epithelial cell migration ( Figure 5). M2 macrophages suppress apoptosis via various mechanisms, such as inhibition of proapoptotic mechanisms (67) and induction of IL-10 secretion (41). IL-10 has anti-apoptotic effects on various types of cells (68)(69)(70). Therefore, TP4 might reduce the risk of BV patients contracting other sexually transmitted diseases. Furthermore, TP4 might also facilitate the rebuilding of a healthy vaginal microenvironment and reduce the recovery period of BV. The vaginal epithelium is more than a physical barrier, as it plays a critical role in the regulation of vaginal immunity (71). Impairments in vaginal epithelium function are associated with higher inflammatory status driven by M1 macrophage accumulation (72). The M1 macrophage accumulation usually prolongs the inflammatory phase and increases disease severity via cytokine production (73).
TP4 elevates IL-10 expression in macrophages ( Figure 6 and Supplementary Figure 5), which might provide clinical benefits in BV. IL-10 is a key anti-inflammatory cytokine that is essential for ending the inflammatory phase by promoting macrophage polarization toward anti-inflammatory phenotypes (49). Furthermore, elevated expression of IL-10 is correlated with improved probiotic colonization in the vagina (74). In contrast, low levels of IL-10 in BV patients are associated with increased risk of adverse pregnancy outcomes (74).
In addition to G. vaginalis (58,59), several other pathogens have been implicated in BV (75). TP4 possesses demonstrated killing activity toward several vaginosis-related pathogens, including Candida albicans, Staphylococcus aureus, and Escherichia coli (29). However, whether TP4 has a direct killing effect on G. vaginalis remains unclear. In summary (Figure 8), TP4 exhibits several anti-inflammatory properties ( Figure 3). TP4 elevates IL-10 in macrophages, which is followed by M2c macrophage polarization ( Figure 6). It also induces M2a macrophage polarization via induction of TSG-6 in vaginal epithelial cells (Figure 7). Moreover, the virulence of G. vaginalis could potentially be mitigated by TP4-induced M2c and M2a macrophages via suppression of vaginal epithelial cell inflammation and apoptosis ( Figure 4) and induction of vaginal epithelial cell migration ( Figure 5), respectively. Since we successfully identified the immunomodulatory activities of TP4 in the vaginal microenvironment using an in vitro model, further investigations using primary cells and in vivo models are warranted to verify the pharmacological potential of TP4 in BV.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
C-WL and B-CS wrote the paper. J-YC supervised the study and edited the paper. C-WL studied the experiments. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank the Academia Sinica Inflammation Core Facility, IBMS for technical support. The core facility is funded by the Academia Sinica Core Facility and Innovative Instrument Project (AS-CFII-108-118). We thank Dr. Marcus J. Calkins for manuscript editing. This work was supported in part by the Higher Education Sprout Project from the Ministry of Education (MOE-110-S-0023-A) in Taiwan.

Supplementary
(A) Detection of phosphorylated NFkB p65 subunit (p-p65), STAT3 (p-STAT3), and STAT6 (p-STAT6) after 1 h treatment with the conditioned medium (CM) from TP4 (3.91 or 7.82 mg/ml)-treated VK2 cells to RAW264.7/GV cells by immunoblotting. bactin was used as an internal control to show equal protein loading. (B) The CM from vehicle (Veh; PBS)-or TP4-treated VK2 cells were incubated with TSG-6 neutralizing antibody (anti-TSG6) or IgG control for 1 h and then used to treat RAW264.7/GV cells. After incubation, p-STAT6 in RAW264.7/GV cells was detected. GAPDH was used as an internal control to show equal protein loading. (C) Macrophage surface marker (CD206) was detected by flow cytometry after CM treatments. The mean fluorescence intensity (MFI) is shown. The black bar graph indicates isotype controls. (D) ARG1 (left) and YM-1 (right) expression levels were measured after treatments by qRT-PCR. Data were represented the normalized target gene amount relative to the vehicle-treated group (lane 1). Data are presented as mean ± SD of three independent experiments (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not statistically significant).