In healthy reproductive age women, the cervicovaginal microbiome is generally dominated by Lactobacillus spp. These beneficial bacteria acidify the cervicovaginal microenvironment via lactic acid production, which contributes to protection against infections by pathogenic and opportunistic microorganisms (O’hanlon et al., 2011). The depletion of Lactobacillus spp. leads to the colonization of the lower female reproductive tract (FRT) by a diverse consortium of facultative and obligate anaerobic bacteria, a disorder is referred to as bacterial vaginosis (BV) (Workowski and Bolan, 2015; Muzny et al., 2020). Importantly, BV is associated with a range of adverse gynecologic and obstetric outcomes including an increased risk of sexually transmitted infections (STI) and preterm birth (PTB). Microbiologically, BV is characterized by the presence of a polymicrobial biofilm covering the surface of cervicovaginal epithelium (Muzny et al., 2019). Mobiluncus mulieris and Eggerthella sp. are two of the many anaerobes that may contribute to the biofilm formation, yet their mechanistic contributions to BV and related adverse gynecologic and obstetric outcomes are still poorly understood (Danielsson et al., 2011; Machado and Cerca, 2015).

Mobiluncus spp. are motile, curved rod-shaped bacteria that are isolated from vaginal secretions from women with BV and are associated with high Nugent scores (a method to diagnose BV) (Sprott et al., 1983; Roberts et al., 1985; Hallén et al., 1987; Teo et al., 1987; Vetere et al., 1987; Moi et al., 1991; Gatti, 2000; Srinivasan and Fredricks, 2008). In addition, M. mulieris has been isolated from extragenital sites, such as breast and umbilical abscesses (Glupczynski et al., 1984). In previous epidemiological studies, M. mulieris have been also linked to PTB (Holst et al., 1994; Hillier et al., 1995; Meis et al., 1995). Genital inflammation has been implicated in PTB and M. mulieris has been hypothesized as a microbial driver in such inflammatory states (Dude et al., 2020). The flagella of M. mulieris has been previously demonstrated to stimulate Toll-like receptor 5 (TLR5) activation, which links to the elevation of key inflammatory markers (IL-6, IL-8, and TNF-α) and PTB (Anahtar et al., 2015; Onderdonk et al., 2016; Dude et al., 2020; Dela Cruz et al., 2021). In addition, M. mulieris has been shown to exert sialidase activity (Culhane et al., 2006). Notably, this bacterial enzyme cleaves sialic acid from highly glycosylated proteins present in the cervical mucus plug and its activity is associated with BV (Smayevsky et al., 2001), PTB and chorioamnionitis (Critchfield et al., 2013; Racicot et al., 2013; Smith-Dupont et al., 2017).

Some bacterial members of the Eggerthellaceae family are highly prevalent in BV, identified in ~85-95% of cases (Fredricks et al., 2005; Fredricks et al., 2007; Srinivasan et al., 2012; Shipitsyna et al., 2013). Interestingly, one member of the Eggerthellaceae family have also been linked to all four of the Amsel criteria (vaginal pH, vaginal odor, vaginal discharge and the presence of clue cells) used to diagnose BV in clinical settings (Srinivasan et al., 2015). Eggerthella spp. [previously classified as Eubacterium (Kageyama et al., 1999)] are non-motile anaerobic coccobacilli that are part of the healthy human gut microbiome (Finegold et al., 1983; Schwiertz et al., 2000). The taxonomy of the Eggerthellaceae family requires further investigation to classify them into their appropriate genus and species. However, Eggerthella spp. can also cause bacteremia and sepsis with high mortality rates (Lau et al., 2004a; Lau et al., 2004b; Thota et al., 2011; Lee et al., 2012). This suggests that in the FRT, Eggerthella spp. might play a role in the pathophysiological processes that manifest as adverse obstetric and gynecologic outcomes.

To determine the individual immunometabolic contributions of Eggerthella sp. and M. mulieris within the cervical microenvironment, we utilized our well-characterized human three-dimensional (3-D) cervical epithelial cell model that recapitulates several physiologically relevant features of in vivo tissue, including TLR expression, microvilli, intercellular junctional complexes and secretory material. We combined this advanced bioreactor-derived 3-D cell culture model with multiplex immunoassay and global untargeted metabolomics approaches to identify key immune mediators and metabolites related to M. mulieris and Eggerthella sp. infections of the lower FRT. We chose the 3-D cervical model since the cervix is a critical area impacted by cervicovaginal microbiota that, when disrupted, can lead to PTB, increased STI acquisition and other gynecological sequalae associated with BV.

BV is characterized by colonization of the cervicovaginal epithelium by a diverse community of anaerobic bacteria. M. mulieris and bacterial species from the family Eggerthellaceae are relatively understudied bacteria compared to many of the other BV-associated microorganisms. In this study we aimed to examine how two vaginal isolates: Eggerthella sp. strain MVA1 and M. mulieris strain UPII-28I, influence the immunometabolic landscape in the context of the lower FRT, as well as the potential pathophysiological contributions of these species to BV.

In recent years, there has been a reclassification of members belonging to the Eggerthellaceae family leading to questions related to the contributions of specific family members to the BV state. Although genomic information is limited for Eggerthella sp. strain MVA1, there is a sequence read SRX655730 available in the NCBI Sequence Read Archive. The 16S rRNA gene was sequenced from isolate Eggerthella sp. MVA1 (JX103988) and shares 99% sequence identity with Eggerthella lenta, however the species was not assigned by the BEI repository. There is very little information about E. lenta in the FRT and the vaginal microbiome since it is predominantly a gut microbe. E. lenta is found at low abundance in the FRT and may be transferred to the vagina from the gastrointestinal tract (Priputnevich et al., 2021). Unfortunately, until the comparative genomic analysis is performed, we cannot classify a species for the strain Eggerthella sp. MVA1. The family Eggerthellaceae also contains Coriobacteriales bacterium DNF00809, previously classified as Eggerthella sp. type 1 by Srinivasan et al. (Srinivasan et al., 2016) before culture and whole genome sequencing of the species. Previous studies showed that this bacterium was present in 85-95% of women with BV compared to those without BV (Fredricks et al., 2005; Fredricks et al., 2007; Srinivasan et al., 2012; Shipitsyna et al., 2013). These previous studies have also shown that E. lenta is much less prevalent in the FRT than Coriobacteriales bacterium DNF00809, which is no longer classified as an Eggerthella species. Although beyond the scope of this study, an in-depth analysis of taxonomic classification of vaginal bacteria belonging to the family Eggerthellaceae should be further investigated and clear taxonomic nomenclature should be referenced for this family to reflect the complexity of its lineage and putative role in BV and the cervicovaginal environment. Our study provides data related to the immunometabolic contributions with one of these understudied vaginal strains from the family Eggerthelleacae.

It is important to investigate BV-associated bacteria in the cervical epithelium since disruption of the microbiota at this mucosal site can lead to PTB, increased STI acquisition and other gynecological sequalae (Brunham and Paavonen, 2020). BV can also lead to ascension of pathogenic bacteria to the upper FRT, therefore promoting endometritis and pelvic inflammatory disease (Eckert et al., 2002). To determine the individual immunometabolic contributions of Eggerthella sp. and M. mulieris within the cervical microenvironment, we utilized a bioreactor-derived 3-D cervical epithelial cell model (Barrila et al., 2010; Hjelm et al., 2010; Gardner and Herbst-Kralovetz, 2016) in combination with multiplex immunoassays and global untargeted metabolomic approaches to identify key metabolites and immune mediators, respectively, related to M. mulieris and Eggerthella sp. infections.

Organotypic 3-D human cervical epithelial cell models recapitulate many features of parental tissue that are not observed in monolayer cell culture models. Our advanced 3-D cell culture model exhibits physiologically relevant features, such as TLR expression, microvilli, intercellular junctional complexes and secretory material that could influence how BV-associated bacteria can colonize the model in a way similar to in vivo tissues (Gardner et al., 2020; Łaniewski and Herbst-Kralovetz, 2021; Salliss et al., 2021) These features provide a more accurate representation of the in vivo state which is more amenable to translational research efforts and studying host-microbe interactions; however, each model system has its strengths, weaknesses and utility (Hjelm et al., 2010; Radtke and Herbst-Kralovetz, 2012; Radtke et al., 2012; Herbst-Kralovetz et al., 2016). The close relation of 3-D human cervical epithelial cells to cervical tissue allows us to investigate how these pathogens can cause pathophysiological changes to the cervicovaginal microenvironment and epithelia. Using SEM, we demonstrated colonization of the 3-D cervical epithelial cell model by Eggerthella sp. and M. mulieris ( Figure 1 ). We evaluated the cytotoxicity of each bacterium and determined that neither Eggerthella sp. nor M. mulieris infections induced significant cytotoxicity in cervical cells.

Inflammation is a driver of many disease processes including BV where it has been previously associated with PTB (Meis et al., 1995; Goldenberg et al., 2000; Romero et al., 2007). Our immune mediator analysis revealed that both M. mulieris and Eggerthella sp. induced a proinflammatory response in 3-D cervical epithelial cell models. Eggerthella sp. significantly increased IL-1α while M. mulieris significantly elevated IL-6, IL-8, MCP-1 and TNFα. M. mulieris has been linked to the significant elevation of IL-6, IL-8 and TNFα in vitro (Anahtar et al., 2015; Dude et al., 2020). Flagella, such as those expressed by M. mulieris, have been linked to activation of TLR5 (Hayashi et al., 2001; Dela Cruz et al., 2021) which leads to the stimulation of the NF-κB signaling pathway and, in consequence, upregulation of IL-6, IL-8 and TNFα (Nasu and Narahara, 2010). Intriguingly, upregulation of IL-6 and IL-8 as well as the NF-κB signaling pathway have been connected to PTB (Romero et al., 2014). Clinical studies showed that IL-6, IL-8 and TNFα are some of the most common cytokines associated with PTB (Murtha et al., 1998; Coleman et al., 2001; Romero et al., 2014; Ville and Rozenberg, 2018). However, there is no clear diagnostic marker for PTB and proinflammatory cytokine and chemokine profiles differing among women who deliver pre-term (Wei et al., 2010; Fettweis et al., 2019). In addition, elevation of proinflammatory markers including IL-6 and IL-8 are associated with HIV infection risk (Mlisana et al., 2012; Rodriguez Garcia et al., 2015). Genital inflammation can lead to an increased risk of STI acquisition, with the infected epithelia being damaged, allowing the pathogens that cause STIs access to deeper tissues (Passmore et al., 2016). Inflammation of cervical and vaginal tissues induces recruitment of immune cells to the lower FRT which facilitates the spread of HIV and other STIs (Masson et al., 2014; Anahtar et al., 2015; Masson et al., 2015; Fichorova et al., 2020).

Oxidative stress has been closely linked to inflammation (Reuter et al., 2010). Oxidative stress becomes damaging when there are a disproportional amount of reactive oxygen species (ROS), that overwhelm the antioxidants capacity of glutathione (Schafer and Buettner, 2001). High levels of ROS can induce cellular damage and promote many inflammatory states including cancer (Perwez Hussain and Harris, 2007; Reuter et al., 2010; Burton and Jauniaux, 2011) by recruiting inflammatory markers such as cytokines and chemokines and stimulating NF-κB signaling (Reuter et al., 2010). In this study, we found that M. mulieris significantly altered multiple metabolites associated with oxidative stress, including cysteinylglycine, cysteinylglycine disulfide and 2-hydroxyglutarate, whilst Eggerthella sp. significantly altered 2-hydroxybutyrate and cysteinylglycine disulfide. Depletion of cysteinylglycine and cysteinylglycine disulfide, two intermediates in the glutathione synthesis pathway, could either indicate increased glutathione biosynthesis or signify an increase in the levels of ROS. Disruption of redox balance due to the elevated levels of ROS has been linked to activation of cell signaling pathways including those responsible for the regulation of inflammatory cytokines and PTB (Schieber and Chandel, 2014; Moore et al., 2018). Notably, increase of ROS can lead to lipid peroxidation which can free lipids from cell membranes (Burton and Jauniaux, 2011; Moore et al., 2018). The lipid peroxidation and membrane damage by ROS might be a mechanistic link behind the increased concentrations of lipids following infections of 3-D cervical models with Eggerthella sp. and M. mulieris.

Through our global untargeted metabolomics analyses, we found that Eggerthella sp. significantly altered twice as many lipids as M. mulieris. Specifically, sphingolipids were significantly elevated exclusively by Eggerthella sp. We also found similar significantly altered glycerolipids and sphingolipids as those reported by Salliss et al., 2021 that were elevated following infection with another BV-related microorganism: Megasphaera micronuciformis (Salliss et al., 2021). Sphingolipids are components of eukaryotic cell membranes and have been related to proinflammatory signaling pathways and apoptosis (Kolter and Sandhoff, 2006; Hannun and Obeid, 2008; Hannun and Obeid, 2018). Our results demonstrate that M. mulieris induced higher abundance of sphingolipids and most glycerolipids relative to Eggerthella sp. and PBS mock-infected controls, although the levels did not reach significance. We hypothesize that this may be related to the observation that M. mulieris has been shown to increase membrane permeability in cervical epithelial cells grown on transwells inserts (Dude et al., 2020) potentially by freeing these membrane-associated lipids. Both bacteria induced extracellular accumulation of multiple lipids related to epithelial barrier function (Bittman, 2013; Jernigan et al., 2015). Considering these results, we hypothesize that Eggerthella sp. and M. mulieris may play a role in increasing membrane permeability, although significant cytotoxicity was not observed in our experiments. Unexpectedly, compared to other lipids long-chain fatty acids (LCFAs) were significantly depleted by M. mulieris infection. It is possible that the depletion of LCFAs could result from an ability of M. mulieris to catabolize these liberated LCFAs a as an energy source. Unfortunately, the genomic sequence of M. mulieris is not fully annotated, therefore it is unclear if this bacterial species synthesizes all proteins necessary to facilitate the catabolism of the LCFAs.

The epithelial barrier function and the physiological properties of the mucosal membranes lining the FRT are crucial in protecting the cervix from BV-associated bacteria (Rodriguez Garcia et al., 2015). One of the key pathophysiological changes during BV is disruption of the epithelial barrier function, which allows pathogenic bacteria to access deeper tissues and induce inflammation (Muzny et al., 2019). Sialic acid is the terminal sugar moiety on glycans of cell surface glycoproteins and mucins. The epithelium of the FRT is lined with highly glycosylated mucins which limit adhesion and colonization of pathogenic bacteria during BV (Linden et al., 2008; Barrila et al., 2010; Lewis and Lewis, 2012; Radtke et al., 2012). In addition, sialic acid residues can bind to pathogens and induce host cell signaling to generate an immune response (Macauley et al., 2014; Bhide and Colley, 2017). Significant alterations in the levels of sialic acid following bacterial infections indicates bacteria-mediated sialidase activity. Previous clinical studies have revealed elevated levels of sialidase and sialic acid in the cervicovaginal fluids of women with BV (Briselden et al., 1992; Moncla et al., 2015). Infection of 3-D cervical cell models with both M. mulieris and Eggerthella sp. significantly altered the levels of sialic acid, indicating that both species exert sialidase activity. As M. mulieris decreased the levels of extracellular sialic acid, we hypothesize that this species catabolizes sialic acid residues that are liberated from the cell surfaces (Culhane et al., 2006), similarly to other BV-associated bacterium Gardnerella vaginalis (Lewis et al., 2013). The sialidase activity of M. mulieris could play a role in PTB since the mucus plug created during pregnancy contains multiple mucins which could be degraded by sialidase, therefore allowing pathogenic bacteria to ascend to the uterus (Mcgregor et al., 1994; Lewis and Lewis, 2012; Smith-Dupont et al., 2017; Baker et al., 2018). Consequently, ascension of pathogenic bacteria into the upper FRT during pregnancy can lead to chorioamnionitis and PTB (Galinsky et al., 2013). M. mulieris and elevated levels of IL-8 have been previously associated with amniotic infection and PID (Hillier et al., 1988; Larsson et al., 1989; Hitti et al., 2001).

Sialidase activity has been noted as a potential diagnostic marker for BV (Briselden et al., 1992) along with several cervicovaginal metabolites, some of which are highlighted by Srinivasan et al., (2015). Amongst the metabolites associated with BV, biogenic amines are also considered key players in many aspects of BV pathogenesis (Nelson et al., 2015; Srinivasan et al., 2015; Borgogna et al., 2021). The Amsel criteria and Nugent scores are two methods to diagnose BV (Amsel et al., 1983; Nugent et al., 1991). Putrescine and cadaverine have been linked to decreased in vitro growth of Lactobacillus spp. and high Nugent scores in women with BV (Borgogna et al., 2021). Both putrescine and cadaverine are associated with increased vaginal pH, vaginal amine odor and vaginal discharge that manifest during BV (Srinivasan et al., 2012; Yeoman et al., 2013; Nelson et al., 2015). Through our metabolomics analysis we found that Eggerthella sp. significantly elevated both putrescine and cadaverine in the extracellular milieu, which corresponds with observations from Srinivasan et al. (Srinivasan et al., 2012). In contrast, M. mulieris did not elevate any biogenic amines in our current study. Previously, M. mulieris has been linked to elevated trimethylamine (Spiegel, 1991; Africa et al., 2014), however, this polyamine was not detected in our metabolomics analysis. Phenyllactate was found to be significantly elevated for Eggerthella sp. with the highest fold change (1150-fold) of all the metabolites measured. Although the role of this metabolite in the cervicovaginal microenvironment is still not clear, we have previously observed its accumulation following infections with other vaginal bacteria (Łaniewski and Herbst-Kralovetz, 2021; Salliss et al., 2021). The severity of BV has been associated with increased risk of HIV and other STI acquisition (Allsworth and Peipert, 2011); thus, contribution of Eggerthella spp. and M. mulieris to clinical symptoms of BV mechanistically links these species to poor health outcomes related to BV.

As with all experiments and biological models there are limitations (Herbst-Kralovetz et al., 2016). The 3-D cell culture model we have used is a robust tool that can provide mechanistic insights into host-microbe interactions in the cervical microenvironment. Our model, as with most human in vitro cell culture models, requires the use of pH-buffered medium; thus, it cannot mimic the acidic pH found in healthy women in vivo without impacting cellular viability However, the BVAB tested in this study thrive in a more neutral pH environment, which is a characteristic of our model (Barrila et al., 2010; Hjelm et al., 2010; Gardner and Herbst-Kralovetz, 2016). We also acknowledge that the bacterial strains used in this study may not represent the other closely related strains or species. As stated previously, the taxonomy of Eggerthella sp. MVA1 is still incomplete; therefore, we cannot generalize our findings to the other members of the Eggerthellaceae family. Mobiluncus mulieris is closely related to Mobiluncus curtsii; however, the two species are unique from each other in terms of physical characteristics and enzymatic activity. M. curtsii is smaller in size, can hydrolyze starch and hippurate and produce citrulline, ornithine and ammonia from arginine whilst M. mulieris cannot (Spiegel and Roberts, 1984). Future studies utilizing additional well-characterized bacterial isolates in mono- or polymicrobial infections are needed to better understand the individual contributions of these BVAB to poor gynecologic and obstetric outcomes.

Overall, we found Eggerthella sp. infections significantly altered multiple metabolites related to BV symptoms. These metabolites included the biogenic amines putrescine and cadaverine, as well as their precursors, and metabolites linked to epithelial barrier function, such as sphingolipids and glycerolipids (Ghosh et al., 1997; Bittman, 2013; Jernigan et al., 2015; Hannun and Obeid, 2018; Harrison et al., 2018; Heaver et al., 2018). M. mulieris infections significantly elevated multiple proinflammatory markers that are linked to PTB in addition to metabolites related to energy metabolism and oxidative stress ( Figure 7 ). This study sheds light into the mechanisms that Eggerthella sp. and M. mulieris may utilize to promote BV. Our data suggests that Eggerthella sp. plays a key role in the production of biogenic amines, which contribute to the elevated vaginal pH and the amine odor, whilst M. mulieris potentially impacts the membrane permeability and induce proinflammatory immune responses. The increased concentration of lipids present in M. mulieris infection could also link into the increased immune response (Hannun and Obeid, 2018; Albeituni and Stiban, 2019; Sukocheva et al., 2020). The link to inflammation and the altered metabolic microenvironment fits into the hypothesis of Muzny et al. (2020) that proposes early colonizers establish biofilm and evade host defense responses whereas secondary colonizers mediate inflammation, an altered metabolic microenvironment and symptoms associated with BV. Based on our data, we propose that M. mulieris is functioning as a secondary colonizer in this hypothetical model of BV. In contrast, Eggerthella sp. while not inflammatory, exhibits metabolic activity consistent with our definition of a secondary colonizer, but may also participate in the early stages of biofilm formation. However, further in vitro studies investigating these microorganisms in polymicrobial settings in conjunction with longitudinal clinical studies are needed to elucidate microbe-microbe interactions and determine the role of these bacteria in the context of BV biofilms.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

MH-K, PŁ, and JM conceived of the experimental design and interpretation of the data. JM conducted experimental infections of the 3-D cervical cell aggregates and the Bio-Plex analyses. Cell supernatants were sent to Metabolon, Inc. for global untargeted metabolomics analysis. RM carried out the cytotoxicity experiments as well as the analysis of the cytotoxicity data, Bio-Plex and metabolomics data. RM was also responsible for writing the first draft of the manuscript, drafting and editing the figures and revising the manuscript. JM obtained SEM images of infected 3-D human cervical cell models and assisted in the statistical analysis. MH-K, JM, and PŁ provided support and advice on writing, figures and tables, and also read and revised the manuscript. MH-K and PŁ provided guidance of the experimental and writing processes. MH-K supervised the research and provided funding acquisition, project administration and resources. All authors read, revised, and approved the final version of the manuscript.

The funding for this study was supplied by the NIH National Cancer Institute (3P30CA023074-39S3) and the Flinn Foundation (2244) to MH-K.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.