Antigen presenting cell response to polysaccharide A is characterized by the generation of anti-inflammatory macrophages

Abstract

Polysaccharide A (PSA) is the immunodominant capsular carbohydrate from the gram negative commensal microbe Bacteroides fragilis that has shown remarkable potency in ameliorating many rodent models of inflammatory disease by eliciting downstream suppressive CD4⁺ T cells. PSA is composed of a zwitterionic repeating unit that allows it to be processed by antigen presenting cells (APCs) and presented by MHCII in a glycosylation-dependent manner. While previous work has uncovered much about the interactions between MHCII and PSA, as well as the downstream T cell response, little is known about how PSA affects the phenotype of MHCII⁺ APCs, including macrophages. Here, we utilized an unbiased systems approach consisting of RNAseq transcriptomics, high-throughput flow cytometry, Luminex analysis and targeted validation experiments to characterize the impact of PSA-mediated stimulation of splenic MHCII⁺ cells. The data revealed that PSA potently elicited the upregulation of an alternatively activated M2 macrophage transcriptomic and cell surface signature. Cell-type-specific validation experiments further demonstrated that PSA-exposed bone marrow-derived macrophages (BMDMs) induced cell surface and intracellular markers associated with M2 macrophages compared with conventional peptide ovalbumin (ova)-exposed BMDMs. In contrast to macrophages, we also found that CD11c⁺ dendritic cells (DCs) upregulated the pro-T cell activation costimulatory molecule CD86 following PSA stimulation. Consistent with the divergent BMDM and DC changes, PSA-exposed DCs elicited an antigen-experienced T cell phenotype in co-cultures, whereas macrophages did not. These findings collectively demonstrate that the PSA-induced immune response is characterized by both T cell stimulation via presentation by DCs, and a previously unrecognized anti-inflammatory polarization of macrophages.

Introduction

Commensal microbes have shaped host immune systems throughout evolution by providing essential signals for the establishment and maintenance of immune tolerance and homeostasis. Correspondingly, dysbiosis has been associated with a wide range of diseases, particularly those involving immune overactivation (Hevia et al. 2014; Shamriz et al. 2016). Therefore, characterizing how commensal microbes alter host immunity could identify ways in which the growing epidemic of inflammatory and autoimmune diseases can be addressed (Troy and Kasper 2010; De Luca and Shoenfeld 2019; Sofi et al. 2019; Erturk-Hasdemir et al. 2021).

Bacteroides fragilis is a gram negative anaerobic commensal bacterium that typically resides in the human colon and is famous for its ability to drive peripheral immunomodulation by suppressing several models of inflammatory disease, including abscess formation (Tzianabos et al. 1995), inflammatory bowel disease (Mazmanian et al. 2008), asthma (Johnson et al. 2015a, 2015b; Jones et al. 2019) and experimental autoimmune encephalomyelitis (Ochoa-Repáraz, Mielcarz, Ditrio, et al. 2010; Wang et al. 2014). The dominant antigen of B. fragilis is the capsular polysaccharide PSA, a zwitterionic glycan that recapitulates the same suppressive properties of B. fragilis in mechanisms mediated by CD4⁺ T cells (Tzianabos et al. 1999). PSA is historically significant as the first carbohydrate shown to undergo processing by APCs and presentation by MHCII much like conventional protein antigens (Cobb et al. 2004). Its downstream T cell effects have therefore undergone careful scrutiny as to how the polysaccharide antigens result in functionally suppressive immune activity.

Downstream of presentation by APCs, PSA expands a population of CD4⁺FoxP3⁻CD45Rb^lo T cells of a CD62L⁻CD44⁺ effector memory subset (Rb^loTem cells). Rb^loTem cells were sufficient in attenuating inflammatory disease in naïve recipient mice by robustly stimulating the IL-10 production by FoxP3⁺ regulatory T cells via crosstalk mediated by the combination of IL-2 and IL-4 release (Johnson et al. 2015a, 2015b, 2018; Jones et al. 2019; Zhou et al. 2021). Recently, transcriptomic analyses of bulk CD4⁺ T cells revealed that PSA also stimulates the upregulation of immunoregulatory markers including Tim3, Lag3 and PD1 in responding T cells (Alvarez et al. 2020), suggesting that PSA can induce T cell-mediated suppression through both cell contact-dependent and independent pathways.

While T cell activity downstream of PSA-exposure has been the primary focus in the literature, the phenotypic impact on MHCII⁺ APCs responsible for mediating PSA-presentation remains understudied. Here, we report that PSA stimulated transcriptomic and proteomic changes consistent with antigen presentation in undifferentiated splenic MHCII⁺ cells; however, there was a notable downregulation of genes responsible for macrophage activation. We further confirmed that the genes associated with alternatively activated M2 macrophages were enriched, while the genes associated with classically activated M1 macrophages were not. Indeed, PSA-exposed BMDMs upregulated cell surface and intracellular markers associated with M2 macrophages compared with conventional peptide ovalbumin (ova)-exposed BMDMs, which showed a classically activated M1 macrophage phenotype characterized by CD86 expression. In contrast, CD11c⁺ dendritic cells (DCs) upregulated the costimulatory molecule CD86 in response to both PSA and ova. To determine how this translated to function, we compared the T cell stimulation capabilities between splenic F4/80⁺ macrophages and CD11c⁺ DCs and observed that CD11c⁺ DCs elicited an increased T cell antigen experience phenotype in the presence of PSA, while F4/80⁺ macrophages did not. Altogether, these data suggest a model in which PSA drives both antigen presentation and suppressive T cell activation via MHCII-dependent presentation on DCs, and M2-skewed macrophage activation which collectively compose a potent anti-inflammatory environment characteristic of PSA and B. fragilis immune responses in vivo.

Results

PSA-exposure shows transcriptomic and proteomic changes consistent with antigen presentation

To examine the transcriptomic profile of PSA-experienced professional APCs, murine MHCII⁺ splenocytes were sorted and cultured in vitro with PSA for 24 hours, then harvested for RNAseq analysis. Gene set enrichment analysis (GSEA) comparing antigen processing and presentation of peptide or polysaccharide revealed that the transcriptomic profile of PSA-stimulated MHCII⁺ cells trended toward and was consistent with genes previously identified as enriched following polysaccharide antigen presentation (Figure 1A, Supplementary Figure 1) as previously established by the Gene Ontology (GO) Consortium.

Fig. 1.

PSA-exposure shows transcriptomic and proteomic changes consistent with antigen presentation. (A) GSEA of PSA-stimulated MHCII⁺ cells associated with antigen processing and presentation of peptide or polysaccharide antigen via MHCII. (B) PCA of 255 surface proteins on MHCII⁺ cells after 0, 1 and 7 days of co-culture with CD4⁺ T cells and PSA supplementation. (C) Cell surface expression of I-A/I-E on MHCII⁺ cells after no culture or 7 days of culture with CD4⁺ T cells and PSA supplementation.

In order to determine whether similar changes in MHCII proteins are seen in the context of APC-mediated T cell activation, we co-cultured splenic MHCII⁺ cells with CD4⁺ T cells for 7 days with or without PSA. Flow cytometry of 255 cell surface proteins and unbiased Principle Component Analysis (PCA) validated the mRNA transcript findings in that the MHCII proteins I-E/I-A changed upon exposure to PSA and is a key characteristic of the response (Figure 1B), with the already high baseline expression of MHCII molecules increasing over 3-fold during the course of the culture (Figure 1C). Supporting this notion, CD4⁺ T cells that comprised the other half of the co-culture system showed upregulated cell surface expression of the co-receptor CD4 over the course of 7 days (Alvarez et al. 2020), collectively demonstrating that PSA enhanced the antigen presentation pathway.

Unbiased analyses of RNAseq identify that PSA alters M1- and M2-associated pathways

Unbiased transcriptomic analysis revealed 1648 significantly differentially expressed genes between freshly isolated and unstimulated MHCII⁺ cells and those cultured in the presence of PSA for 24 hours. GO analysis showed that the most significantly enriched terms were those related to immune function, cytokine production, response to bacterium and metabolic or catabolic processes, all collectively pointing to the capacity of PSA to drive an immunologically and metabolically active state in MHCII⁺ cells (Figure 2A). Secondarily, PSA exposure also yielded changes in general cellular and molecular functions consistent with activity associated with immune activation (Supplementary Figure 2A and B). Meanwhile, the downregulated biological pathways mostly related to eye, ear and limb development, and not to more pertinent immune functions (Supplementary Figure 2C).

Fig. 2.

Unbiased analyses of RNAseq identifie that PSA alters M1- and M2-associated pathways. (A) GO analysis of genes with increased expression following PSA exposure in MHCII⁺ cells. GO terms of biological processes significantly increased were plotted as −log10. (B) Volcano plot of genes after MHCII⁺ cells was cultured with PSA for 24 h highlights the downregulation of genes associated macrophage activation. (C) GSEA of PSA stimulated cells according to genes that are associated with M2 macrophage polarization and (D) enriched in M1 macrophage polarization.

Interestingly, our analyses identified that three of the most significantly downregulated genes are Fos, Fosb and Jun (Figure 2B), all of which are AP-1 family transcription factors implicated in macrophage polarization to classically activated M1 macrophages (Tugal et al. 2013; Fontana et al. 2015). Their uniform downregulation suggests that PSA opposes the polarization of macrophages into classically activated M1 macrophages or even drives the polarization of alternatively activated-M2 macrophages. We therefore sought to characterize whether PSA polarizes macrophages with respect to this well-established dichotomy (Mills et al. 2000; Orecchioni et al. 2019). GSEA revealed that genes upregulated in M2 polarization (compared with M1) were significantly enriched (Figure 2C), while genes downregulated in M1 polarization (compared with M2) were not (Figure 2D).

GSEA further highlighted a number of other pathways associated with M1 and M2 polarization. Some significantly downregulated M1-associated pathways include peroxisome, reactive oxygen species pathway, mTORC1 signaling and PI3K-AKT–mTOR signaling (Figure 3A). Fittingly, several M2-associated pathways were upregulated, including TGFβ signaling and angiogenesis (Figure 3B); however, the observed dichotomy is not perfect as TNFα signaling, associated with M1 function, is enriched following PSA-mediated stimulation (Figure 3C).

Fig. 3.

Gene set enrichment analyses suggest skewing toward M2-associated cellular processes. GSEA showing (A) enrichment of mTORC1 signaling, PI3K-AKT–mTOR signaling, peroxisome pathway and reactive oxygen species pathway, (B) downregulation of TGFβ signaling and angiogenesis and (C) upregulation in TNFα signaling.

Cell surface proteomic analyses show a mosaic response

In order to characterize the MHCII⁺ APCs on a proteomic level, we moved to a co-culture system with T cells at days 1 and 7 to allow the translation of induced gene transcripts to reach steady state after activation. This was done by using a 255-plex flow cytometry screening assay. Over the course of 7 days, PSA-stimulation dramatically shifted the phenotype of MHCII⁺ cells (Figure 4A). Among the top 20 upregulated genes as measured by fold change, the expression of CD86, a co-stimulatory marker, was increased, as was the expression of galectin-9 and podoplanin. Galectin-9 has been reported to promote M2 polarization, while podoplanin is a marker associated with tumor-associated macrophages (Bieniasz-Krzywiec et al. 2019) which are commonly viewed as a M2-like phenotype (Yang and Zhang 2017) and are implicated in diminishing sepsis-related inflammatory injury (Rayes et al. 2017) (Figure 4B). Among the top 20, downregulated genes were CD23 and CD300c/d, with CD300c having previously been described an activation marker on monocyte-derived cells (Simhadri et al. 2013) (Figure 4C). Also downregulated was CD84, which is necessary to transduce LPS-stimulated signaling in macrophages (Sintes et al. 2010) (Figure 4C).

Fig. 4.

Cell surface proteomic analyses show a mosaic response. (A) Cell surface marker expression of PSA-exposed MHCII⁺ cells identified immunomodulatory markers. High-throughput flow cytometry of MHCII⁺ cells co-cultured with CD4⁺ T cells and PSA collected at days 0, 1 and 7. Data shown are M(edian)FI from MHCII⁺ cells. (A) Heatmap of all 255 cell surface markers at D0, D1 and D7. (B) Top 20 increased and (C) decreased on day 7 compared with day 0. (D) M2- and (E) M1-associated cell surface markers. (F) RNA expression of M1- and (G) M2-associated genes in MHCII⁺ cells cultured with PSA.

We then examined the expression levels of M2-associated cell surface markers and noted strong upregulation following 7 days of culture (Figure 4D). Interestingly, M1-associated cell surface markers were also increased; however, several of these markers, such as CD80 and CD86, describe antigen presentation which is an established property of PSA in MHCII⁺ cells, and the increase in TLR4 can be explained by virtue of PSA being a bacterial component (Figure 4E). The trends align with transcriptomic data of curated M1 markers (Figure 4F) and M2 markers (Figure 4G); however, one noticeably unchanged M1-associated transcript is nos1, which is consistent with the GSEA plot indicating that reactive oxygen species were remarkably downregulated in MHCII⁺ cells upon PSA exposure (Figure 3A). These data collectively support the notion that PSA induces antigen presentation but steers the immune response away from inflammation and toward a suppressed or tolerogenic state.

PSA upregulates factors that recruit T cells and induce Tregs, Th1 and Th17

PSA has a long published history of stimulating T cells in a classical MHCII-dependent fashion (Cobb et al. 2004; Cobb and Kasper 2005). We therefore sought to identify what T cell factors MHCII⁺ cells produce in the presence of PSA. Our 24-h stimulation RNAseq dataset revealed that PSA upregulated MHCII⁺ cell production of the cytokines necessary to differentiate naïve T cells to Th1 (IL-12), Th17 (IL-6 and TGF-β) and Tregs (TGF-β) (Figure 5A). In contrast, the cytokines responsible for differentiating Th2 (IL-4) and Tfh (IL-21) were not upregulated (Figure 5A). We also identified that PSA leads to changes in the expression of the transcripts of STAT molecules (Figure 5B). Of particular interest, STAT6 has been reported to modulate M2 macrophage polarization (Yu et al. 2019).

Fig. 5.

PSA upregulates factors that recruit T cells and induce Tregs, Th1 and Th17. (A) Heatmaps of RNAseq RPKM values of MHCII⁺ cells cultured with PSA showing CD4⁺ skewing cytokines, (B) STAT signaling molecules, (C) chemokines and (D) cytokines. (E) Profile of significantly different soluble factors secreted by PSA-exposed MHCII⁺ cells as determined by Mouse Cytokine/Chemokine 32-plex Luminex analysis after 24 hours of culture.

The expression of cytokines and chemokines, as well as their receptors, was also examined from MHCII⁺ cells cultured alone for 24 h with and without PSA (Figure 5C and D, Supplementary Figure 3A and B), with particular interest in the upregulation of chemokines that attract T cells, including CCL5, CCL16 and CXCL10 (Griffith et al. 2014). Luminex analysis of 31-soluble factors showed that while PSA-stimulation elicited significant changes in some factors, the concentrations of all factors were uniformly low and trended toward T cell chemokines (Figure 5E, Supplementary Figure 3C), which collectively support the ability of PSA to enhance antigen presentation and communication of MHCII⁺ cells with T cells.

PSA elicits an M2-like phenotype in BMDMs compared with conventional peptide antigen

In order to validate the effect of PSA on macrophage polarization, we cultured freshly differentiated bone marrow-derived macrophages (BMDMs) with or without PSA or a conventional peptide antigen ovalbumin (ova) for 24 h. Characterization of macrophage polarization by flow cytometry revealed that the surface expression of CD86, a costimulatory molecule and classical M1-associated marker, significantly increased when BMDMs were cultured with ova (Figure 6A), but not nearly as much when compared with that of the well-established M1-polarizing agents LPS and IFNγ (Orecchioni et al. 2019). In contrast and despite CD86 being enhanced in both transcriptomic and proteomic analysis of PSA-experienced splenic MHCII⁺ cells, PSA stimulation failed to increase the expression of CD86 on BMDMs (Figure 6A). Instead, PSA stimulation increased the expression of the M2-associated marker PDL2 (Loke and Allison 2003), a PD1 ligand, nearly as much as the well-established M2-polarizing cytokine IL-4 (Figure 6B) (Orecchioni et al. 2019). Neither PSA nor ova stimulation increased the expression of the other M2 markers CD206 (mannose receptor) and CD301 (a member of the C-type lectin family) above baseline (Figure 6C and D), although expression of the intracellular M2-associated proteins Arg1, Chil3, Vegfα and Relmα, by qPCR further supported the notion that PSA-exposure yields a M2-like phenotype in BMDMs (Figure 6E). The contrasting ova and PSA-driven M1 and M2-like phenotypes, respectively, illustrated a distinct inversion of macrophage phenotypes between a conventional peptide antigen and PSA.

Fig. 6.

PSA elicits an M2-like phenotype in BMDMs compared with conventional peptide antigen. BMDMs were cultured with or without PSA, ova peptide, M1- (LPS + IFNγ) or M2- (IL-4) polarizing factors for 24 hours, then analyzed by flow cytometry. After gating F4/80⁺ cells, the histograms and statistical analyses for (A) CD86, (B) PDL2, (C) CD301 and (D) CD206 are shown. (E) Relative expression levels of M2-associated genes of BMDMs cultured with PSA or ova peptide for 24 hours.

The effect of PSA differs by MHCII+ cell type

Our finding that BMDMs do not upregulate the M1 marker CD86 and instead upregulate the M2-marker PDL2 was consistent with the broad unbiased transcriptomic and proteomic analyses on macrophage polarization; however, it did not explain why PSA upregulated CD86 specifically in the RNAseq and flow cytometric analyses of splenic MHCII⁺ cells (Figure 4E and F). We therefore purified F4/80⁺ and F4/80⁻ cells from freshly harvested spleens, then further sorted MHCII⁺ cells from the F4/80⁻ and cultured them with or without ova or PSA. After 24 h, we measured CD86 expression by flow cytometry among F4/80⁺ macrophages, CD19⁺ B cells and CD11c⁺ DCs. We found that splenic F4/80⁺ macrophages responded to ova and PSA in a way extremely reminiscent of BMDMs, where ova enhanced the expression of CD86 and PSA did not (Figure 7A). CD19⁺ B cells responded in much the same way as macrophages (Figure 7B), yet ova and PSA enhanced the expression of CD86 on CD11c⁺ DCs (Figure 7C). These data explain why CD86 is increased in the RNAseq and LegendScreen flow data of MHCII⁺ cells stimulated with PSA (Figure 4E and G), but not among isolated macrophages (Figures 6 and 7).

Fig. 7.

The effect of PSA differs by MHCII⁺ cell type. F4/80⁺ and F4/80⁻ cells from freshly harvested spleens were purified, and MHCII⁺ cells were further sorted from the F4/80⁻ fraction and cultured with or without ova or PSA. After 24 hours, flow cytometric analysis of CD86 expression was determined for (A) F4/80⁺ macrophages, (B) CD19⁺ B cells and (C) CD11c⁺ DCs. (D) Splenic F4/80⁺ macrophages and CD11c⁺ DCs were purified and co-cultured with splenic CD4⁺ T cells with or without PSA as indicated. CD44 MFI (and percentage change of CD44 MFI from T cells cultured without APCs and PSA) or (E) frequency of CD62⁻CD44⁺ T cells (Tem) of CD4⁺ cells were analyzed by flow cytometry. (E) BMDMs were co-cultured with purified splenic CD4⁺ T cells with or without PSA and the frequency of Tem cells was quantified by flow cytometry.

To examine the function of splenic F4/80⁺ macrophages and CD11c⁺ DCs downstream of PSA exposure, we co-cultured purified macrophages and DCs with splenic CD4⁺ T cells with or without PSA. After 4 d, we analyzed the expression of the activation marker CD44 on T cells and observed that PSA potently elicited upregulation in CD44 on T cells only after co-culture with DCs and not with macrophages (Figure 7D). This trend remained the same when multiplexing with CD62L, where CD11c⁺ DCs elicited a significant increase in CD62L⁻CD44⁺ T effector memory cells (Tem), whereas co-culture with F4/80⁺ macrophages did not (Figure 7E). Lastly, we repeated this experiment by co-culturing BMDMs with T cells and did not observe an increase in the Tem profile of T cells (Figure 7F), which is in strong contrast to previous work illustrating the potent antigen presentation and downstream T cell activation elicited by DCs.

Altogether, these data illustrate a differential response to PSA which may simultaneously drive suppressive T cell activation by DCs and M2-skewing of macrophages to collectively inhibit inflammation in vivo.

Discussion

Bacteroides species comprise ~25% of the organisms in the human colon, making it one of the most common anaerobes in the gut (Wexler 2007). Outside of the gut, B. fragilis has been implicated in clinical pathogenesis and infection, with its dominant antigen PSA being the first carbohydrate antigen shown to be presented on MHCII and having a decades-long history of inducing T cell-dependent formation of abscesses (Shapiro et al. 1986; Gibson et al. 1998) and adhesions (Onderdonk et al. 1978). More recently, however, the focus on B. fragilis and PSA has turned to their robust suppressive ability in multiple rodent models of inflammatory and autoimmune diseases (Ochoa-Repáraz, Mielcarz, Wang, et al. 2010; Ochoa-Repáraz, Mielcarz, Ditrio, et al. 2010; Wang et al. 2014). In fact, PSA-experienced T cells originating from the spleen were capable of suppressing house dust mite-induced asthma (Johnson et al. 2015a), demonstrating the profound systemic effect that gut commensals and their byproducts have on the host immune response.

The suppressive mechanisms of PSA have been ascribed primarily to CD4⁺ T cells, where exposure leads to the expansion of regulatory T cells (Tregs) as well as conventional CD4⁺ T cells with high expression of the checkpoint inhibitors Lag3, Tim3 and especially PD1 (Alvarez et al. 2020). In addition, PSA has been shown to expand CD4⁺Rb^lo T cells of an effector memory subset that robustly induces IL-10 release by FoxP3⁺ Tregs (Jones et al. 2019). Further studies indicate that the IL-2 and IL-4 secreted by these PSA-exposed T cells were necessary to trigger synergistic and robust amounts of IL-10 release by Tregs, and could themselves be used therapeutically to treat mouse models of asthma and multiple sclerosis (Zhou et al. 2021). While the suppressive functionality of PSA-exposed T cells and their associated mechanisms have been well characterized, the APC half of the story has been largely unexplored in terms of immune suppression.

In this study, we used unbiased and targeted validation approaches to comprehensively characterize the MHCII⁺ APC response to PSA in a murine system. Based on transcriptomics analysis, we found that PSA-responding MHCII⁺ cells were enriched in the genes associated with M2 macrophages (Figure 2C) and depleted of genes associated with M1 macrophage activation (Figure 2D). Culture of BMDMs in the presence of PSA responded in an M2-like manner compared with the conventional peptide, ovalbumin (Figure 6A–E), suggesting that PSA has a significant and direct role in driving anti-inflammatory macrophage differentiation, potentially assisting in the polarization of the downstream DC-mediated T cell response in the early phases of PSA exposure in vivo.

M2-like alternatively activated macrophages are critical for preventing and maintaining overly exuberant immune responses, particularly in tissue locations that are constantly exposed to environmental antigens such as the GI-tract and lungs, where they antagonize or regulate inflammatory responses (Gordon and Pluddemann 2017). The suppressive ability of M2-like tumor-associated macrophages has also been co-opted by solid tumors, where it maintains a mechanism by which malignant cells adopt the ability to evade immune destruction (Yang and Zhang 2017). Whether in support of health or disease, our data show that on the transcriptomic level, PSA-stimulated MHCII⁺ cells produce the chemokines associated with T cell migration, as well as the TGF-β to induce the differentiation of inducible FoxP3⁺ Tregs (Figure 5A). Given that the suppressive action of PSA-stimulated CD4⁺ T cells is also dependent upon resident FoxP3⁺ Tregs (Jones et al. 2019), these findings suggest a mechanism by which PSA can promote immune suppression beyond T cell activation (Round and Mazmanian 2010).

Our unbiased–omics analyses also pointed to the upregulation of molecular components involved in antigen processing and presentation, which appear at odds with the induction of an M2 macrophage phenotype. Upon looking at MHCII⁺ cells individually, we found that CD11c⁺ DCs were the only surveyed cell type in which upregulation of surface expression of CD86 was noted in response to PSA (Figure 7C). This suggests a delineation of function among the various types of MHCII⁺ cells in response to PSA, whereby the enhanced CD86-expression of CD11c⁺ DCs supports antigen presentation and T cell activation, while the M2-like polarization of macrophages could support a more tolerogenic environment and thereafter promote anti-inflammatory skewing of DC-mediated T cell activation such that global suppression and homeostasis is achieved in both the innate and adaptive arms of the immune response.

Overall, PSA appears to possess unique properties that allow it to drive immune inhibition to an extent needed to reduce or prevent inflammatory diseases and maintain immune homeostasis. This study supports decades of research surrounding B. fragilis and its dominant surface antigen, carbohydrate antigen PSA. It also unveils more complexity about the relationship surrounding host and commensal organisms, and how they work together to tune the host immune system.

Materials and methods

Mice

C57Bl/6 J (JAX #000664) mice were acquired from the Jackson Laboratory (Bar Harbor, ME). Mice were housed in a specific pathogen-free facility with a 12-h light/dark cycle. Mice were fed a standard diet (Purina 5010). Enrichment and privacy were provided in mating cages with sterile wood-pulp huts (Bio-Serv, Flemingtton, NJ, USA). Mouse housing and studies were approved by and performed according to the guidelines established by the Institutional Animal Care and Use Committee at CWRU. Sex matched mice between the ages of 12–22 weeks were used for experiments.

PSA purification

PSA was isolated from log-phase ∆44 stain of NCTC9343 B. fragilis expressing only PSA as previously described (Alvarez and Cobb 2019). Polysaccharide purity was determined by SDS-PAGE, BCA assay for protein and absorbance scans for protein and nucleic acid.

BMDM differentiation and culture

To isolate the bone marrow, mouse femurs were dissected and harvested by detaching the bone at the knee joint and guiding the scissors up the femoral shaft, where it was clipped to expose the bone marrow. The femur was placed clipped side down into a nested 0.5-ml microcentrifuge tube with punctured hole at the bottom in a 1.5-ml microcentrifuge tube, and centrifuging at 2000 ×g for 10 min (Amend et al. 2016). Bone marrow was then resuspended in media, counted and plated on petri dishes at a density of 5 million cells per 10 cm² petri dish with RPMI supplemented with 10% fetal bovine serum, L-glutamine and penicillin/streptomycin, as well as 25% L929-conditioned media containing macrophage-differentiating growth factors. Additional media with 25% L929 media was supplemented after 3 days of culture. After 7 days of differentiation after initial plating, plates were washed, and adherent macrophages were scraped and used in experiments.

Cell purification

Primary splenocytes were isolated from mice by reducing spleens to a single cell suspension by passing them through a sterile 100-μM nylon mesh cell strainer (Fisher Scientific, Hampton, NH) using the stopper of a 1-ml syringe. Single cell suspensions were labeled with anti-mouse CD4 (L3T4), MHCII or streptavidin magnetic microbeads (Miltenyi Biotec, San Diego, CA) and underwent positive selection with LS columns according to the manufacturer’s protocols.

PSA-mediated activation

One hundred thousand CD4⁺ T cells and MHCII APCs were cultured in a 1:1 ratio in Advanced RPMI (Gibco/Fisher Scientific, Waltham, MA) supplemented with 5% Australian-produced heat-inactivated fetal bovine serum, 55-μM β-mercaptoethanol, 100 U/ml and 100-μg/ml Penicillin/Streptomycin and 0.2-mM L-glutamine (Gibco/Fisher Scientific, Waltham, MA) at 37°C 5% CO₂ for the indicated duration of time. Cultures were supplemented with 50-μg/ml PSA.

Ova and PSA stimulation

After magnetic bead sorting, cells were cultured in 96-well round bottom plates (Corning, Corning, NY) at a density of 100,000 cells per well for one day. Ova or PSA was added at a concentration of 50 μg/ml at the start of culture (day 0).

Flow cytometry

Cells were incubated with a Fc receptor-blockade cocktail consisting of hybridoma-conditioned media containing IgG1 (clone KL295; ATCC), IgG2 (clone HB55; ATCC) and IgG3 (clone HKPEG1; ATCC) subtypes, as well as Fixable Viability Stain 510 (BD, Franklin Lakes, NJ) prior to antibody staining. BMDMs were stained with a cocktail of antibodies to F4/80-FITC (BioLegend, San Diego, CA), CD86-PE (BD), CD206-PE-Cy7 (BioLegend), CD301-APC (BioLegend) and PDL2-BV711 (BD). Splenic MHCII⁺ cell flow cytometry, cells were stained with a cocktail of antibodies to F4/80-FITC (BioLegend), CD86-PE (BD), CD19-APC (eBioscience, San Diego, CA), CD11c-APC-Cy7 (BioLegend), CD44-APC (BioLegend), CD62L-PE (BioLegend). Cells were washed in MACS buffer before being analyzed on an Attune Nxt (ThermoFisher, Waltham, MA) with support of the Cytometry & Imaging Microscopy Core Facility of the Case Comprehensive Cancer Center. All FACS data were analyzed using FlowJo (Tree Star, Inc., Ashland, OR).

qPCR

Total RNA was extracted from cell lysates using the RNeasy Plus Micro Kit (QIAGEN, Hilden, Germany). cDNA was synthesized from RNA using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA). Reactions were assembled by combining cDNA with TaqMan Fast Advanced Master Mix (Thermo Scientific) and one of the following TaqMan primers (Thermo Scientific): ARG1 (Mm00475988_m1), Chil3 (Mm00657889_mH), VEGFA (Mm00437306_m1), Retnla (Mm00445109_m1) and B2m (Mm00437762_m1). qPCR reactions were run on the QuantStudio3 Real Time PCR Instrument (Thermo Scientific).

Luminex

Media from the indicated ± PSA culture conditions were snap frozen in liquid nitrogen and shipped to Eve Technologies (Calgary, Ontario, Canada) for mouse 31-plex cytokine/chemokine analysis.

RNAseq and analysis

Cells were harvested from culture and repurified, if from co-cultures, using MHCII⁺ magnetic beads as before to a minimum of 95% purity with two passes through the autoMACS Pro system (Miltenyi). Pelleted cells were snap-frozen in liquid nitrogen and shipped to LC Sciences, LLC (Houston, TX) for RNA extraction, purification and quality check, library creation and high-throughput sequencing (Illumina, San Diego, CA, USA). Differential expression and GO analyses were performed using EdgeR v3.12.1 by LC Sciences, LLC. Genes showing significant differences (FDR > 0.05 and log2CPM > 0) were selected for enrichment analysis using GAGE v2.20.1 by LC Sciences, LLC. We acknowledge our use of the GSEA software (http://www.broad.mit.edu/gsea/) (Subramanian et al. 2005), including the deposited gene sets as noted. All RNAseq data has been deposited in the NCBI Gene Expression Omnibus under the accession number GSE181463.

Data analysis

All experiments were performed at N ≥ 3 and data are represented by mean ± SEM, with ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Data and statistical measurements were generated with GraphPad Prism (v9.0). For comparisons between two groups, Student’s t-test was used; comparisons between multiple groups utilized two-way or three-way ANOVA. Heatmaps were created with gplots in R (v4.0.0). The RColorBrewer package was used to create a smooth color gradient. Volcano plots were generated using ggplot2 and ggrepel. PCA plots were created with the factoextra package.

Funding

Boehringer Ingelheim to B.A.C.; National Institutes of Health (GM115234 and AI154899 to B.A.C., AI089474 to J.Y.Z. and CA043703 to the Case Comprehensive Cancer Center).

Conflict of interest statement

This work was supported by grants from the National Institutes of Health as well as a research contract between BAC and Boehringer Ingelheim Pharmaceuticals. None of the authors have a financial stake in the results herein, nor is any intellectual property associated with any experiment or outcome.

Abbreviations

PSA, polysaccharide A; APC, antigen presenting cell; MHCII, class II major histocompatibility complex; BMDM, bone marrow-derived macrophages; Ova, ovalbumin; Rb^loTem, CD4 + FoxP3-CD45Rblo T cells of a CD62L-CD44+ effector memory subset; M1, classically-activated macrophages; M2, alternatively-activated macrophages; IL, interleukin; STAT, signal transducer and activator of transcription; GSEA, gene set enrichment analysis; DEG, differentially expressed genes; GO, gene ontology; TAM, tumor-associated macrophage; LPS, liposaccharide; TLR4, toll-like receptor 4; Treg, regulatory T cell; Th1, T helper type 1; Th17, T helper type 17; Th2, T helper type 2; Tfh, T follicular help; DC, dendritic cell; MFI, median fluorescence intensity