Phenylacetic acid, an anti-vaginitis metabolite produced by the vaginal symbiotic bacterium Chryseobacterium gleum

The human microbiome contains genetic information that regulates metabolic processes in response to host health and disease. While acidic vaginal pH is maintained in normal conditions, the pH level increases in infectious vaginitis. We propose that this change in the vaginal environment triggers the biosynthesis of anti-vaginitis metabolites. Gene expression levels of Chryseobacterium gleum, a vaginal symbiotic bacterium, were found to be affected by pH changes. The distinctive difference in the metabolic profiles between two C. gleum cultures incubated under acidic and neutral pH conditions was suggested to be an anti-vaginitis molecule, which was identified as phenylacetic acid (PAA) by spectroscopic data analysis. The antimicrobial activity of PAA was evaluated in vitro, showing greater toxicity toward Gardnerella vaginalis and Candida albicans, two major vaginal pathogens, relative to commensal Lactobacillus spp. The activation of myeloperoxidase, prostaglandin E2, and nuclear factor-κB, and the expression of cyclooxygenase-2 were reduced by an intravaginal administration of PAA in the vaginitis mouse model. In addition, PAA displayed the downregulation of mast cell activation. Therefore, PAA was suggested to be a messenger molecule that mediates interactions between the human microbiome and vaginal health.

www.nature.com/scientificreports/naturally increases during menopause due to hormonal changes and decreased lactobacilli composition, leading to the loss of natural epithelial defenses and, thereby, allowing the invasion of pathogens into the vagina and urinary tract 12,13 .
Clinical signs of vaginal inflammation, such as discharge, itching, and pain, are strongly associated with vaginitis 14 .In particular, vulvovaginal candidiasis is accompanied by pruritus, a prevalent symptom of allergic inflammation 14 .Mast cells are well-known for their crucial roles in itch sensation 15 .The stimulation of mast cells with various stimuli, including irritants and allergens, induces degranulation and the release of eicosanoids and pro-inflammatory cytokines 15 .Vaginal bacteria and/or their products were reported to activate mast cells to secrete mediators 16 .
Members of the genus Chryseobacterium are ubiquitous, but only a few are associated with colonization in humans, including Chryseobacterium gleum (formerly Flavobacterium gleum) [17][18][19] .C. gleum type strain F93 was originally isolated from a human high-vaginal swab 18,19 .The isolation of other C. gleum strains from patients 20 has been reported as extremely rare cases, and their multidrug-resistant nature implies the potential pathogenicity of this species 21 .However, in addition to the low virulence of the genus Chryseobacterium 22 , the scarcity of clinical case reports and the weak biofilm-forming ability of C. gleum 23 indicate that the presence of the species in clinical specimens is likely due to colonization rather than infection.Such unique viability of C. gleum led us to propose the hypothesis that its type strain responds to pathogenic infections by sensing a pH change in the human vaginal environment and produces defensive metabolites to provide competitive advantages.
Our previous study revealed that phenethylamine and N-acetylphenethylamine, discovered from the commensal oral microbe Corynebacterium durum, extended the lifespan of Caenorhabditis elegans by overexpressing SIR-2.1 protein 24 .In the present study, our continued efforts to search for chemical messengers from the human microbiome led to the identification of phenylacetic acid (PAA) from the vaginal symbiotic strain C. gleum F93.We also found that the production level of PAA varied considerably in response to pH changes.Furthermore, PAA displayed antimicrobial and anti-inflammatory activities in a mouse vaginitis model.Here, we describe our biochemical observations in detail and propose that PAA is a messenger molecule produced as a microbial defense mechanism against vaginitis.

Isolation and identification of PAA
Based on the hypothesis that C. gleum produces messenger molecules for vaginal protection by sensing a pH increase, high-performance liquid chromatography (HPLC) data were compared between two acetone extracts of the microbe cultured at pH 5.5 and 7.3.Although a healthy human vaginal pH is 4.5 or less 10,11 , our small-scale pH screening results (data not shown) indicated that the lowest pH value for C. gleum was 5.5.This discrepancy could be attributed to the inherent differences between human vaginal fluid and the culture medium used in this study.Despite the variance in acidic pH levels, the impact of the shift between acidic and neutral pHs was evident in the experimental design, leading to changes in metabolite production.
Each extract was partitioned between water and organic solvent.PAA was observed predominantly in the water-soluble portion partitioned from pH 7.3 culture compared to pH 5.5, while the data obtained from the organic solvent-soluble portion of each extract were almost identical.Interpretation of the mass spectrometry (MS) and nuclear magnetic resonance (NMR) data enabled the identification of PAA, which was isolated by preparative HPLC.The molecular formula C 8 H 8 O 2 was assigned to the compound based on HRESIMS and 13 C NMR data. 1 H NMR shift and integration values for multiplets at δ H 7.24 and a singlet at δ H 3.43 suggested the presence of mono-substituted benzene and an isolated methylene unit, respectively.The two partial structures were connected as a benzyl group by analysis of heteronuclear multiple-bond correlation (HMBC) data.In addition, an HMBC cross-peak of δ H 3.43 to δ C 173.4 indicated a linkage of the methylene group to a carboxyl carbonyl carbon, completing the structure of PAA, as shown in Fig. 1A.These NMR and MS data are consistent with literature reports of PAA [25][26][27] .

Quantitative analysis of microbial PAA productions
The relative abundance of PAA generated by C. gleum cultured in acidic and neutral pHs was calculated to verify that pH was a crucial factor in PAA production.The culture supernatant was directly examined in triplicate at different time points by quantitative LC/MS analysis.The PAA peak areas were normalized by the optical density values of the corresponding cultures.More PAA was detected in the culture at pH 7.3 than at pH 5.5, and the gap between their PAA production levels widened with time until seven days of cultivation (Fig. 1B).

RNA sequencing analysis
Based on the observation that external pH regulates metabolite production in C. gleum, its gene expression profiles at two different pH conditions were assessed by RNA-sequencing (RNA-seq) analysis.As a result, 1158 and 958 genes were differentially expressed by more than three-fold (log 2 ) in pH 7.3 and 5.5 conditions, respectively, relative to the other pH condition (data not shown), indicating that pH is an important environmental regulator for C. gleum.The application of gene ontology (GO) analysis to the identified differentially expressed genes (DEGs) provided 856 annotated genes (Fig. S2, Supplementary Information 2).One of the most upregulated genes at pH 7.3, with high average counts per million reads (CPM), was RS19060, the gene product of which was annotated to phenylacetyl-CoA epoxidase subunit A. RS19070 and RS19080, encoding homologs of the phenylacetate-CoA oxygenase subunits PaaC and PaaJ, respectively, were observed at a nearby locus of RS19060 on the chromosome, suggesting that the genes related to PAA metabolism are clustered in C. gleum.Therefore, in conjunction with the chemical analysis data, the RNA-seq analysis results supported the effects of increasing pH on gene expression related to the biosynthesis and/or degradation of PAA.Although no studies on the

Effect of PAA on bacterial vaginosis and vulvovaginal candidiasis in mice
Since G. vaginalis (GV) and C. albicans (CA) represent pathogens that cause bacterial vaginosis (BV) and vulvovaginal candidiasis (VVC), respectively 10,[29][30][31][32] , we investigated the effects of PAA on GV-and CA-induced vaginitis in mice.Numerous studies have reported that myeloperoxidase (MPO) activity is used as a biomarker for GV-and CA-induced vaginitis, indicating an increased accumulation of polymorphonuclear cells in the vaginal tissue of mice 33 .In this regard, the vaginal inoculation of GV and CA in mice has been described to increase MPO activity and cause the development of BV and VVC, respectively [34][35][36] .A clinical study showed that clotrimazole was effective in the local treatment of mixed vaginal infections with BV and VVC 37 .Accordingly, we used clotrimazole as a positive control.As shown in Fig. 3A,B, GV and CA increased vulnerability to uterine horn infections, accompanied by the presence of uterine edema in these mice compared to the normal group.Interestingly, uterine inflammation and morphological changes induced by GV and CA were partially relieved by high-dose PAA treatment, although these changes did not fully return to normal levels.Moreover, GV-and CA-induced MPO activity was inhibited by the intravaginal administration of PAA at doses of 0.2 and 1 mg/mouse.In particular, PAA displayed greater inhibitory effects on MPO at 1 mg/mouse than clotrimazole at 2 mg/mouse in GV-and CA-treated mice, with levels close to those in the normal condition.Additionally, PAA significantly inhibited the GV-and CA-stimulated prostaglandin E 2 (PGE 2 ) levels in the vaginal tissue of mice (Fig. 3C).GV and CA vaginal infections also induced nuclear factor (NF)-κB p65 activation and cyclooxygenase (COX)-2 expression, both of which were significantly reduced by the PAA treatment (p < 0.05; Fig. 3D) in a dose-dependent manner.These results suggested that PAA attenuated the symptoms of BV and VVC through the NF-κB signaling pathway in mice.

Effect of PAA on compound 48/80-stimulated mast cell activation
We examined the effects of PAA on mast cell activation that leads to the allergic symptoms of vaginitis, such as itching and irritation.The treatment of HMC-1 cells with PAA concentrations up to 50 μM for 24 h had no effect on cell viability (Fig. 4A).HMC-1 cells were stimulated with compound 48/80 in the presence or absence of different concentrations of PAA (0, 5, 10, and 50 μM).Then, degranulation and eicosanoid production levels were measured by the β-hexosaminidase assay and immunoassay, respectively.PAA decreased compound 48/80-stimulated degranulation (Fig. 4B) and leukotriene C 4 (LTC 4 ) and prostaglandin D 2 (PGD 2 ) production (Fig. 4C,D) in a concentration-dependent manner.Consistent with the inhibitory effect of PAA on mast cell activation, 50 μM PAA reduced the activation of PLCγ1, Akt, p38, ERK1/2, and IKK, which are positive signaling molecules involved in mast cell activation (Fig. 4E).Therefore, PAA was suggested to act as a down-regulator of mast cell activation.

Discussion
The dysbiosis of vaginal microbiota is closely related to vaginitis 38 , most cases of which are induced by an overgrowth of G. vaginalis and C. albicans 10,[29][30][31][32] .BV and VVC require oral or local treatment with antibiotics and antifungals, respectively, based on pathological findings.However, conventional treatment for genital  infections has faced many challenges, which include increasing microbial resistance, adverse effects, and recurrent infections 39 .Recently, vaginal microbiota transplantation (VMT) in patients with recurrent BV showed the long-term recovery of Lactobacillus spp.dominance, as in healthy vaginas 40 .Therefore, the human microbiome has become an emerging source of new antimicrobial agents that maintain balanced microbial communities.One diagnostic criterion for vaginitis is the pH of vaginal fluid.Vaginal pH greater than 4.5 is frequently observed in patients with BV, acute vaginal Candida infection, and trichomoniasis, whereas healthy women maintain a pH level lower than 4.5 10,11 .In addition to the laboratory-based pH test, a microbiological approach has been recently recommended for the comprehensive diagnosis of vaginitis 41 .Furthermore, the presence of two different types of vaginitis, called mixed vaginitis, causes vaginal symptoms and affects the health of women of all ages worldwide 37,42 .Among the various types of mixed vaginitis, the combination of BV and VVC is the most prevalent clinical presentation 37 .Based on the reports that the human microbiota provides chemical messengers for microbe-host and microbe-microbe interactions 4,5 , we postulated that defensive compounds are produced by the vaginal symbiotic bacterium C. gleum in response to a pathogenic invasion by sensing the pH change.
In the present study, C. gleum collected from a human high-vaginal swab was incubated at pH 5.5 and 7.3 to simulate healthy and vaginitis conditions, respectively.Comparison of the HPLC profiles of the two different cultures revealed distinctive differences, which were attributed to PAA by NMR and MS interpretation.The quantitative analysis of PAA produced by C. gleum indicated that its abundance was significantly higher in pH 7.3 conditions over the course of a week relative to pH 5.5.Although the genomic sequencing of C. gleum has not been completed, its draft genome available in public databases enabled the analysis of RNA-seq data.Different pH conditions changed various gene expressions, including the genes involved in the biosynthesis and/ or degradation of PAA, leading to the accumulation of PAA in pH 7.3 conditions.Also, we found that the genes related to PAA metabolism in C. gleum might be clustered in the genome, suggesting the coregulation of their expression by the signal perception of environmental cues, such as pH changes.The chemical synthesis of PAA was achieved in two steps, allowing the examination of the biological activity of the compound against microbes related to vaginitis.PAA showed antimicrobial activity against the two representative vaginal pathogens, G. vaginalis and C. albicans, with IC 50 values of 12.4 and 18.1 mM, respectively, with weaker effects on L. iners, L. gasseri, and L. crispatus.Interestingly, although most vaginal Lactobacillus spp.are known to be beneficial, the contribution of L. iners to vaginal health has been debated 43 .Our in vitro study showed that the reduced viability of L. iners was not statistically significant.
In addition to the reported antimicrobial activity 26,27 , PAA has been described to play a messenger role in various ecosystems.The bacterium Phaeobacter gallaeciensis produces PAA to mediate bacteria-algae interactions, in which PAA converts to roseobacticides, antibiotic compounds that protect an environmentally valuable marine microalga 44 .PAA was also proposed to be involved in the interaction between fungal pathogens and rhizobacterial communities.The invasion of plant roots with the fungus Rhizoctonia solani induced a shift in the bacterial composition of the rhizosphere and secondary metabolite production, including PAA, as a part of the defensive mechanisms 45 .
Vaginitis symptoms, including vaginal discharge, itching, and burning, are closely associated with inflammatory responses 46 .Diclofenac is a nonsteroidal anti-inflammatory drug (NSAID) that contains a PAA moiety as a pharmacophore 47 .The mechanism of action of NSAIDs involves the inhibition of COX, an enzyme that oxidizes arachidonic acid to prostanoids.Currently, a new formulation that combines clotrimazole with diclofenac (ProF-001) is under phase 3 clinical trials for recurrent VVC 48 .In addition, the topical application of ibuprofen, another representative NSAID with a PAA structure, was reported to reduce the symptoms of vaginitis in clinics 49 .A recent study revealed that patients with chronic idiopathic vaginitis responded well to mast cell-targeted treatment 50 , indicating that mast cells are involved in the pathophysiology of vaginitis.In the present study, PAA was able to reduce compound 48/80-stimulated degranulation and LTC 4 and PGD 2 production in mast cells, as well as the production of PGE 2 and expression of COX-2 in mouse vaginal tissues infected with BV and VVC.Therefore, the anti-vaginitis activity of PAA was suggested to be mediated by the downregulation of mast cell activation and PGE 2 /COX-2 expression.
In summary, the vaginal symbiotic bacterium C. gleum was found to produce PAA, which was significantly enhanced by a pH increase similar to the physiological condition of vaginal infections.Notably, we demonstrated that the intravaginal administration of PAA was effective in reducing GV-and CA-induced vaginal symptoms and MPO activity in vivo.PAA was also able to inhibit GV-and CA-induced NF-κB and PGE 2 activation and COX-2 expression, indicating that its anti-vaginitis effect was partly due to the modulation of immune responses in the vagina.Moreover, PAA attenuated compound 48/80-stimulated mast cell activation determined by degranulation and eicosanoid production.Our findings suggest the potential role of PAA in treating BV and VVC, as well as chronic vaginitis caused by abnormal mast cell activation 50 .These results support our hypothesis that PAA production might be involved in the mechanism that protects the vaginal microenvironment from pathogenic invasion as a part of microbe-host interactions.However, its mechanistic actions remain to be elucidated.

Extraction
The vaginal bacterial strain C. gleum F93 was cultured in two groups, both in BHI medium for four weeks at 37 °C.The pH of one group (4 × 1 L) was adjusted to 5.5 by adding HCl, while the other group (4 × 1 L) was incubated without HCl (pH 7.3).Adsorption of the organic materials in the culture was accomplished by adding sterilized XAD-7-HP resin (20 g/L), followed by shaking the mixture at 200 rpm for 2 h.The resin collected by filtration through cheesecloth was extracted with acetone to afford 11.0 g (pH 5.5) and 21.1 g (pH 7.3) of brown residue.

LC-MS analysis of PAA production levels
BHI liquid media at pH 5.5 and pH 7.3 was produced with and without HCl, respectively.After autoclave sterilization, 198 mL of each medium was inoculated with 2 mL of C. gleum culture (OD 600 = 0.1 ± 0.02).The resulting cultures were shaken at 120 rpm at 37 °C.At each time point, OD 600 , pH, and LC-MS data of the samples collected from each C. gleum culture were recorded.The relative abundance of PAA (t R = 6.23 min) was calculated by LC-MS (2.5 µm, 2.1 × 150 mm, 0.4 mL/min), eluted with 15% MeCN/H 2 O (0.1% formic acid) for 1.5 min, followed by a linear gradient to 100% MeCN/H 2 O (0.1% formic acid) over 8.5 min.The electrospray ionization mass spectrometry (ESIMS) was set to the negative ion mode with a capillary voltage of 3.00 kV, cone voltage

RNA sequencing
Total RNA concentration was calculated by Quant-IT RiboGreen (Invitrogen, #R11490).Samples were run on TapeStation RNA ScreenTape (Agilent) to assess the integrity of the total RNA.Only high-quality RNA preparations, with an RNA integrity number (RIN) greater than 7.0, were used for RNA library construction.The libraries were independently prepared with 1 μg of total RNA for each sample using the Illumina TruSeq Stranded mRNA Sample Prep Kit (Illumina, Inc., San Diego, CA, USA, #RS-122-2101).The libraries were quantified using KAPA Library Quantification Kits for Illumina Sequencing platforms according to the qPCR Quantification Protocol Guide (KAPA), and qualified using TapeStation D1000 ScreenTape.Indexed libraries were then submitted to Illumina HiSeqXten, and paired-end (2 × 151 bp) sequencing was performed by Macrogen Incorporated (Seoul, Republic of Korea).

Statistical analysis of gene expression levels
The gene model of the C. gleum genome was obtained from the National Center for Biotechnology Information (NCBI) public database.The RNA-seq results matched with the C. gleum genomic data were analyzed using InfoBoss and Macrogen pipelines for obtaining gene expression levels.Briefly, the relative abundance of genes was measured in read counts using HTSeq.Statistical analysis was performed to identify DEGs using the abundance estimates for each gene in the samples.The statistical significance of the differential expression data was determined by nbinomWaldTest using DESeq2 and fold change, in which the null hypothesis was that no difference exists among groups.For the DEG set, hierarchical clustering analysis was performed using complete linkage and Euclidean distance as a measure of similarity.Gene enrichment and functional annotation analysis, and pathway analysis for the significant gene list were performed based on blastGO (http:// geneo ntolo gy.org/) and the Kyoto Encyclopedia of Genes and Genomes (KEGG; http:// kegg.jp).

Antimicrobial activity in vitro
All microbes used for antimicrobial activity were incubated at 37 °C in tryptic soy, Penassay, and MRS broth for C. albicans, G. vaginalis, and Lactobacillus spp., respectively.Anaerobic and 5% CO 2 conditions were applied for Lactobacillus spp.and G. vaginalis cultures, respectively.A stock solution of PAA was prepared in 20% polyethylene glycol (PEG) in sterile water.Serial dilutions of the PAA stock solution were made in a liquid medium, and 50 μL was added to each well of 96-well plates.The microbial culture was adjusted to an OD 600 of 0.1 ± 0.02, and 50 μL was transferred to each well containing PAA solution, producing final concentrations of PAA ranging from 0.2 to 30.3 mM (in 0.02% to 2.5% PEG).The plate was sealed with cover film and incubated at 37 °C for 20 h prior to measuring OD 600 values.

Murine vaginal infection models
Female C57BL/6 mice (8 weeks old) were purchased from Orient Bio Inc. (Seongnam, Republic of Korea).All mice were housed in a room with standard environmental conditions (12/12 h light-dark cycle, temperature 22 ± 2 °C, and humidity 50-60%) and provided food and water ad libitum.To overcome the limitation of mouse models and closely mimic clinical situations in humans, the mice were pretreated with β-estradiol-3-benzoate before infection to render them immunocompromised.β-Estradiol-3-benzoate (0.125 mg/mouse) was intraperitoneally injected into all mice three days before inoculation with G. vaginalis or C. albicans.The mice were intravaginally injected with G. vaginalis or C. albicans (both at 1 × 10 8 CFU/20 µL phosphate-buffered saline, PBS), except for the normal control group.In our preliminary experiment, the administration of PAA at 1 mg/ mouse did not induce toxicity or inflammation in the vaginal tissue of mice (Fig. S9).Based on these findings, we opted to use the low and high concentrations, namely 0.2 and 1 mg/mouse, for further investigation.PAA (0.2 and 1 mg/mouse) and clotrimazole (2 mg/mouse) were intravaginally administered once a day for 14 days beginning the day after infection.The vaginas were washed with PBS, and the washed and excised vaginas were stored at − 80 °C for MPO activity and immunoblotting analyses.Euthanasia was performed using cervical dislocation.Animal treatment and maintenance for this study were approved by the Institutional Animal Care and Use Committee (IACUC) of Woosuk University (protocol number WS-2023-11).All methods were performed in accordance with the relevant guidelines and regulations.The study is reported in accordance with ARRIVE guidelines (https:// arriv eguid elines.org).

ELISA
To analyze MPO activity and PGE 2 production, commercial ELISA kits were used.Vaginal tissue lysate was prepared by homogenizing the tissue in RIPA lysis buffer (Biosesang, Seoul, Republic of Korea) containing phosphatase and protease inhibitors.The supernatant was added to a reaction mixture containing 1.6 mM tetramethyl benzidine and 0.1 mM hydrogen peroxide, incubated at 37 °C, and the absorbance at 650 nm was measured over time.The MPO activity assay and PGE 2 ELISA were performed according to the manufacturer's instructions.

Western blot analysis
Vaginal tissues were homogenized in RIPA buffer to extract protein.Protein concentrations were measured by the Bradford assay.The protein samples were mixed with 5× sodium dodecyl sulfate (SDS) sample buffer, boiled for 5 min, and separated by 10% SDS-polyacrylamide gel electrophoresis.After electrophoresis, the proteins were transferred to membranes.The membranes were blocked with 5% skimmed milk for 30 min.After washing

Figure 1 .
Figure 1.Chemical and biological synthesis of phenylacetic acid (PAA).(A) Synthetic scheme and chemical structure of PAA.(B) Relative abundance of PAA production by C. gleum cultured at pH 7.3 and 5.5.The MS peak areas of PAA were normalized by the corresponding OD 600 values, and data were collected from three independent cultures at each time point and pH level (*P < 0.05, **P < 0.01).

Figure 3 .
Figure 3.Effect of PAA on Gardnerella vaginalis (GV)-and/or Candida albicans (CA)-induced vaginitis and the expression of inflammatory markers in female mice.(A) Effect on GV-and/or CA-inflamed vagina and uterus.(B) Effect on myeloperoxidase (MPO) activity in vaginal tissues.(C) Production of PGE 2 .(D) Effect on COX-2 expression and NF-κB activation analyzed by Western blotting.β-Actin was used as an internal control.Female mouse vaginas were infected with GV and CA (both at 1 × 10 8 CFU/mouse), except in the normal control group (NOR, normal group treated with vehicle alone).PAA and clotrimazole (CLO) were intravaginally administered once a day for 14 days.On day 15 post-infection, the mice were euthanized.The expression of COX-2 and activation of MPO, PGE 2 , and NF-kB were measured in vaginal tissues.Western blots from three independent results were quantified by densitometry using ImageJ.All values are shown as the mean ± SD of five replicate mice (n = 5).### P < 0.001 vs. normal control group.**P < 0.01 and ***P < 0.001 vs. GV-and CA-treated control group.