A commensal symbiotic factor derived from Bacteroides fragilis promotes human CD39⁺Foxp3⁺ T cells and T_reg function

Abbreviations

Introduction

The intestinal microbiota profoundly shapes host immune responses.Citation^1,2 Mice raised under germ free (GF) conditions lack commensal influence during development. As a result, GF animals exhibit disorganized lymphoid tissue and aberrant immune responses compared to specific pathogen free (SPF) mice, which undergo conventional microbial intestinal colonization.Citation³ Such deficits may be corrected by the introduction of commensal species to host mice suggesting a potent modulatory role for the microbiota.Citation⁴ Closer examination of several individual species clearly demonstrates their capacity to provoke divergent immune responses in mature mice. For example, Segmented filamentous bacteria promote inflammatory Th17 responsesCitation⁵ while Bacteroides fragilis (B. fragilis) colonization is associated with potentiating Th1 and T_reg activity.Citation⁴

Of the 8 types of surface capsular polysaccharides expressed by B. fragilis, PSA, a zwitterionic polysaccharide (ZPS), is a known immunomodulator and symbiotic factor, directing host immune responses while promoting maintenance of the organism in vivo. This unique polysaccharide possesses the capacity to elicit T cell responses that are essential to the immune regulatory effects observed in hosts.Citation^4,6 The otherwise Th2-skewed immune system in GF mice is rebalanced to reflect appropriate Th1 responses upon B. fragilis colonization. Furthermore, deficiency in Foxp3⁺ T_regs observed in GF mice is corrected upon exposure to PSA.Citation^7,8 Later studies elaborated the biologically important role these T_regs play in B. fragilis survival in the host. Foxp3 T_regs were shown to be responsible for attenuating host Th17 cells in the gut, which would otherwise limit B. fragilis colonization.Citation⁹ The induction of Foxp3⁺ T_regs likewise is associated with PSA-mediated protection against murine autoimmune pathologies.Citation^8,10 PSA significantly enhances the conversion of CD4⁺ T cells into IL-10-producing Foxp3⁺ T_regs. Furthermore, Foxp3⁺CD4⁺ T_regs in PSA-treated mice demonstrated enhanced functional suppression, increased Foxp3 and IL-10 compared to PBS controls.Citation⁸ Thus the induction of Foxp3⁺ T_regs directly represents a commensal mediated immune response that holds potential benefit for both bacteria and host alike.

To date, the association of PSA exposure and induction of Foxp3 in humans has not been reported. Whether induction of a Foxp3 population in humans is important for the maintenance of B. fragilis in human hosts has not been established. However, promotion of Foxp3 frequency and function by PSA would suggest the capacity of PSA to influence human disease in which T_reg disparities have been observed such as multiple sclerosis (MS). We therefore investigated whether PSA induces Foxp3 in human T cells. In this report we demonstrate that this commensal-derived antigen promotes a CD39⁺Foxp3⁺ population among naïve CD4 T cells while enhancing IL-10 production. Induction of this population required cognate interactions with dendritic cells bearing HLA-DR, CD86, CD40 and PD-L1. PSA also increased the expression of Foxp3, CD39 and HLA-DR in T_regs, and enhanced their suppressive function in vitro.

Results

Dendritic cells are required for PSA-mediated human Foxp3⁺CD4⁺ T cell induction

Studies describing murine T cell responses to PSA in vivo demonstrate the induction of T_regs that protect against 2 distinct models of autoimmunity, experimental colitis (inflammatory bowel disease) and experimental autoimmune encephalomyelitis (multiple sclerosis). To determine whether PSA would promote Foxp3 expression in human T cells, DCs were isolated from whole peripheral blood and co-cultured with autologous naïve CD4⁺CD25⁻ T cells (NCD4) in the presence or absence of PSA. As shown (Fig. 1A), PSA promoted CD4⁺Foxp3⁺ T cells in a DC-dependent manner, as no enhancement of Foxp3 was detected in wells containing NCD4s alone. Other traditional antigen presenting cells derived from the peripheral circulation, including monocytes and B cells, were unable to induce this population (Fig. 1, B and C). Foxp3 induction was only observed in the DC-NCD4 context; by comparison, use of PSA in a mixed population of peripheral blood mononuclear cells (PBMCs), had no effect on Foxp3 expression (Fig. S1).

Figure 1. Dendritic Cells mediate PSA induction of Human CD39⁺Foxp3⁺ CD4⁺ T cells. PSA-mediated induction of CD39⁺Foxp3⁺ T cells was observed in the presence of DCs but not other APCs. 3 × 10⁴ NCD4s were cultured in the presence or absence of 25 µg/ml PSA or Sp1 and 100 U/ml IL-2, alongside 5 × 10³ of one of 3 primary autologous APC populations: CD19⁺ B cells, CD14⁺ monocytes, or total blood dendritic cells. (A) Fold induction of Foxp3⁺ frequency alone or in the presence of autologous DCs, n = 4. (B) Representative histograms showing frequency of Foxp3⁺ cells (red line) as an overlay against isotype control (shaded blue), panels were pre-gated on CD4⁺ cells. (C) Replicates of CD4⁺Foxp3⁺ frequency, n = 3. (D) Representative FACS plots showing frequency of CD25 and CD39 cells, pre-gated on Foxp3⁺ cells. Positive cells (red line) are overlaid against respective isotype controls (shaded blue). (E) PSA dose response of CD4⁺CD39⁺Foxp3⁺ T cells, n = 1 (Bars represent means of 4 culture wells per condition, error bars represent standard error of the mean). (F) Individual repeats of CD39⁺Foxp3⁺ T cells, n = 4–14 per condition. P values were calculated using 2-tailed Student's t-test (A, C) or one-way ANOVA using Dunnett's multiple comparison test (E, F).

Display full size

PSA induction of CD39⁺Foxp3⁺ T cells requires engagement of HLA-DR and costimulatory molecules

Zwitterionic polysaccharides are unique in their capacity to generate αβCD4⁺ T cell responses via presentation on MHC-II.Citation¹⁴ Following endocytosis, PSA is processed prior to expression on the surface of APCs for presentation.Citation¹⁵ Both Sp1 and PSA were shown to interact with MHC-II as demonstrated by immunoprecipitation and confocal fluorescence.Citation¹⁶ Furthermore CD11c⁺ DCs from PSA-treated mice expressed enhanced levels of MHC-II as well as CD86 in a dose dependent manner.Citation^4,10 We examined the capacity of PSA to modulate surface expression of several APC/Immunologic synapse proteins including HLA-DR, CD80, CD86, CD40, and PD-L1 by geometric mean fluorescence intensity (GMFI) approximately 16 hours post culture. PSA significantly upregulated HLA-DR (P = 0.009), CD86 (P = 0.02), and PD-L1 (P = 0.02) while also increasing surface expression of CD80 and CD40 (Fig. 2A).

Figure 2. PSA induction of CD39⁺Foxp3⁺ T cells requires engagement of HLA-DR and costimulatory molecules. Blocking HLA-DR or costimulatory molecules limited PSA-mediated induction of CD39⁺Foxp3⁺ T cells and cytokine production. (A) 5 × 10⁴ DCs were cultured with 25 µg/ml PSA for approximately 16 hours before being stained for surface expression of the above markers, n = 3–6. DCs were incubated with 10 µg/ml of blocking antibody specific to CD86, PD-L1, HLA-DR or CD40L for 30 min prior to co-culture with 3 × 10⁴ autologous NCD4s in the presence or absence of 25 µg/ml PSA and 100 U/ml IL-2. Supernatants were assessed by ELISA for IL-10 or IFNγ. (B) CD4⁺CD39⁺Foxp3⁺ T cell frequency, n = 6–10 per condition (C) IL-10 and IFNγ production, n = 4–10 per condition. Error bars reflect standard error of the mean representing independent experiments, each using cells from different individuals. P values were calculated using 2-tailed Student's t-test (A) or one-way ANOVA using Dunnett's multiple comparison test (B, C).

Display full size

We investigated whether inhibition of the upregulated synaptic components in response to PSA impacted induction of human CD39⁺Foxp3⁺ T cells by pre-incubating DCs with blocking antibodies either to HLA-DR, CD86, CD40L or PD-L1. Blockade of any of these components significantly attenuated PSA-mediated induction of CD39⁺Foxp3⁺ T cell frequency. Substantial reduction followed blockade of CD86 (P < 0.05), PD-L1 (P < 0.01) or CD40L (P < 0.01) while near complete ablation occurred when HLA-DR (P < 0.001) was inhibited (Fig. 2B). In vitro culture with PSA promotes T cell secretion of IFNγ and IL-10 in mice and humans.Citation^4,17 We determined that compared to media controls PSA enhanced both cytokines in cultures containing PSA-induced CD4⁺CD39⁺Foxp3⁺ T cells (Fig. 2C). No IL-17 was detected in supernatants (not shown). High inter-individual variation in IL-10 production was observed, whereas the levels of IFNγ detected were more consistent. We examined whether the same synaptic components assessed previously were central in mediating PSA-dependent cytokine production. Blockade of each component substantially reduced the average amount of IL-10, with anti-CD86 being least effective. There was significant attenuation of PSA-mediated IFNγ production following antibody blockade of HLA-DR (P < 0.05). In contrast, blocking CD86, or CD40L yielded little to no impact while inhibition of PD-L1 resulted in a further diminished, albeit non-significant, reduction of IFNγ (Fig. 2C).

PSA enhances circulating CD39⁺ T_reg frequency

T_regs expressing high levels of Foxp3 may be directly isolated from circulating human blood.Citation¹⁸ In addition to constitutive Foxp3, these circulating T_regs possess high levels of CD25 and may express low amounts of CD127 (Fig. 3A). Circulating Foxp3⁺ T_regs represent a diverse array of subpopulations with varying suppressive potential. Numerical deficiencies or impaired ability to control pathological inflammation in particular subpopulations of these regulatory cells have been associated with autoimmune diseases, including human MS. Individuals with relapsing-remitting MS (RRMS) possess a reduced number of circulating CD39⁺Foxp3⁺ T_regs compared to healthy individuals.Citation¹⁹ In addition to being numerically deficient, these CD39⁺ T_regs are deficient in their capacity to suppress the production of IL-17 by inflammatory cells when compared to healthy individuals. We next evaluated whether CD39 expression was up-regulated on human circulating T_regs. T_regs cultured alongside DCs in the presence of PSA significantly increased expression of CD39 as measured by GMFI. PSA also enhanced the frequency of CD39⁺ circulating T_regs (Fig. 3B). PSA did not significantly impact circulating T_reg production of IL-10 or IFNγ (Fig. S2).

Figure 3. PSA increases the frequency of CD39-expressing T_regs and T_reg function. PSA exposure significantly enhanced the frequency of CD39 and HLA-DR-expressing circulating T_regs and resulted in greater capacity to suppress monocyte secretion of TNFα. 1 × 10⁴ T_regs were cultured in the presence or absence of PSA and 100 U/ml IL-2, alongside 5 × 10³ autologous DCs. (A) Representative histograms of different T_reg-associated markers. Positive cells (red line) are overlaid against respective isotype controls (shaded blue) (B) CD39 MFI, n = 11 (left), representative CD39⁺Foxp3⁺FACS panel pre-gated on CD4⁺ cells (center), replicates of CD39⁺Foxp3⁺T_reg frequency, n = 12 (right). (C) HLA-DR MFI, n = 7 (left), Representative CD39⁺HLA-DR⁺ T_reg FACS panel pre-gated on CD4⁺Foxp3⁺ cells (center), replicates of CD39⁺ HLA-DR⁺ T_reg frequency, n = 7 (right). (D) Foxp3 MFI of CD4⁺ T_regs, n = 12 (E) Suppression of LPS-induced TNFα production, n = 3. Error bars reflect standard error of the mean representing independent experiments, each using cells from different individuals. P values were calculated using 2-tailed Student's t-test (B–D) or one-way ANOVA using Dunnett's multiple comparison test (E).

Display full size

Discussion

The capacity of gut derived commensal bacterial antigen to induce immune regulation is an important paradigm in understanding immune homeostasis. Homeostasis is achieved in the gut microbiome by the wide and diverse range of microorganisms that either effect inflammation or conversely regulate the host. This binary system of induction and regulation is essential to maintain immune balance. When the microbiome is distorted, the dysbiosis can lead to a disease state either by an excessive inflammatory response or poorly regulated immune control. In certain autoimmune conditions, including human MS, current evidence indicates that dysfunctional T_regs may be where homeostasis is lost leading to active disease. We have observed that a specific symbiotic factor and capsular polysaccharide antigen derived from the human commensal B. fragilis has the capacity to induce a population of phenotypically specific and biologically active regulatory T cells.

In this work, we demonstrate that in vitro culture of PSA with NCD4s in the presence of primary peripheral blood DCs leads to the generation of Foxp3⁺ T cells. These Foxp3⁺ cells bear high levels of CD25 and low amounts of CD127, matching fundamental surface phenotypic definitions for T_regs. The role of Foxp3 in T_reg generation and function has been revised in recent years. Multiple studies reveal that expression of Foxp3 alone, such as that mediated by strong T cell receptor activation in combination with TGF-β, does not guarantee suppressive function.Citation^22–26 Contemporary Foxp3⁺ T_reg definitions describe a heterogeneous and plastic continuum of Foxp3⁺ CD4 T cells ranging in regulatory potential. A combination of appropriate transcriptional, epigenetic and environmental factors in vivo determines the acquisition, maintenance and enrichment of Foxp3 facilitating regulatory phenotype.Citation²⁷ Although these studies are limited by the in vitro context of our experimental system we do show for the first time that an isolated commensal antigen is capable of inducing Foxp3 expression in human NCD4s in vitro without polyclonal anti-CD3 signaling or supplementation by exogenous agents such as TGF-β, and rapamycin.

CD39 along with CD73 are part of a sensitive purigenic signaling network by which extracellular nucleotides are detected and degraded in the pericellular environment thereby influencing immune responses.Citation²⁸ Aside from direct contributions to suppression via generation of adenosine, CD39 may function to prolong T_reg longevity by resistance to apoptosis and facilitate entry to inflamed tissue sites.Citation²⁹ We show that CD39 is up-regulated by PSA on human Foxp3⁺ T cells. CD39 up-regulation was also observed after in vitro stimulation along with exposure to rapamyacin endowed suppressive function to human Foxp3⁺ T cells.Citation¹² Interestingly, culture with Sp1 was unable to induce this population, stressing a unique functional role for PSA in mediating human T cell responses. Perhaps the endogenous presence of Bacteroides fragilis within the microbiota of donors used in this study and it's reported role in maturing host immune systems may contribute to the observed in vitro response to exogenous stimulation with PSA. Although CD39 possesses direct regulatory functions (in conjunction with CD73) and may be highly expressed on Foxp3⁺ T_regs, human lymphocytes of both regulatory and inflammatory disposition are capable of expressing CD39.Citation^30,31 CD39⁺ Th17 cells have been reported in a number of studies, showing the enrichment of these cells within inflamed synovial and intestinal tissue of individuals with rheumatoid arthritis and inflammatory bowel disease.Citation^32,33 IL-17 secretion by CD39⁺Foxp3⁻ cells was consistently reported in these instances. It is therefore noteworthy that PSA cultures, despite containing CD39⁺Foxp3⁻ CD4 T cells, did not produce detectable levels of IL-17.

The requirement of an APC for Foxp3 induction among human NCD4s is in line with well-defined mechanisms underlying processing and presentation of PSA to activate T cells. While other APCs may facilitate ZPS-mediated human T cell activation, DCs play a prominent role during PSA mediated protection from inflammatory disease in vivo as well as in vitro T cell responses to PSA. Fluorescently labeled PSA was shown to associate with DCs in the mesenteric lymph nodes of mice when orally delivered.Citation⁴ Further, DCs are essential mediators of PSA's downstream therapeutic effects. In response to oral treatment with PSA, CD103⁺ DCs were enriched in the cervical lymph nodes of EAE-afflicted mice mediating induction of Foxp3⁺ T_regs.Citation¹⁰ A recent study showed that plasmacytoid dendritic cells (pDCs) were the primary APCs responsible for promoting CD4⁺ T cell-produced IL-10 and protection from experimental colitis.Citation³⁴ Our results are in agreement with this DC-centric paradigm, as PSA-mediated Foxp3 acquisition in human NCD4s required DCs whereas other APCs isolated from PBMCs were insufficient to promote this induction. Interestingly, we note that human primary pDCs isolated from the blood were unable to induce Foxp3 (not shown) suggesting that myeloid DCs may be primarily involved in PSA-mediated Foxp3 expression in humans consistent with our previously reported observations in mice. Finally, previous work that employed a culture system that utilized T-cell depleted PBMCs as APCs found that PSA did not induce Foxp3 among human CD4⁺ T cells.Citation¹⁷ We observed a similar readout when PSA was cultured with total PBMCs (data not shown). This lack of T_reg activation potential may be due to the paucity of circulating DCs in the whole peripheral blood culture (less than 1% of total PBMCs).

Zwitterionic polysaccharides significantly up-regulate immunologic synapse components MHC-II, CD86, CD40, and PD-L1 on DCs as well as on other APCs. Further, HLA-DR, CD86, and CD40 are pivotal to downstream ZPS-mediated effects on murine and human T cells.Citation^{4,10,16,34,35} We utilized blocking antibodies to assess the role of CD86, PD-L1, HLA-DR and CD40L in the induction of human CD39⁺Foxp3⁺ T cells and cytokine production. Alongside CD39⁺Foxp3⁺ induction, both IL-10 and IFNγ were significantly enhanced in PSA-containing cultures. Disrupting PSA signal 1 by blocking HLA-DR consistently prevented both induction of Foxp3⁺ T cells and cytokine production.Citation^4,16 By comparison, blockade of CD86 significantly reduced PSA-mediated induction by approximately 50%. This reduction of Foxp3⁺ T cell frequency by anti-CD86 parallels the central role of CD86 during human T cell proliferation mediated by the ZPS Sp1.Citation³⁵ We observed that CD86 blockade also gave rise to a substantial (50%) reduction of the average amount of IL-10 produced. This was consistent with IL-10 inhibition observed when pDCs from CD86 knockout mice were used in vitro during a previous study.Citation³⁴ However, given the lack of statistical significance, CD86, while necessary for activation and induction of Foxp3 by PSA, may play a less prominent role for PSA-mediated cytokine production by human cells in vitro.

PD-L1 is typically known as a negative regulator of T cell responses; however, this costimulatory molecule also plays an important role in generation, maintenance and function of murine inducible Foxp3⁺ T_regs.Citation^36,37 We observed significant reductions in CD39⁺Foxp3⁺ T cell frequency, and marked decrease in IFNγ and IL-10 levels when DCs were incubated with anti-PD-L1. Likewise, the CD40-CD40L axis appears necessary for human in vitro responses to PSA as CD40L inhibition substantially disrupted both Foxp3 and cytokine induction to a similar degree.

Foxp3⁺ T_reg dysfunction is a significant contributor to autoimmunity. Mutations in human Foxp3 gene yield lethal multi-organ systemic autoimmunity in humans with a similar phenotype replicated during Foxp3 deletion in mice.Citation³⁸ Additionally, numeric and functional defects in Foxp3⁺ T_reg subsets, including those expressing CD39, are also described in chronic autoimmune conditions such as MS.Citation^19,29 PSA significantly increased the expression of CD39, on the surface of Foxp3⁺ T_regs in addition to enhancing the frequency of these CD39⁺Foxp3⁺ T_regs in vitro. Interestingly, 3 months of treatment with sphingosine-1-phosphate receptor antagonist FTY720 similarly enhanced CD39 GMFI as well as frequencies of CD39⁺Foxp3⁺ T_regs in patients, suggesting the therapeutic value of these cells.Citation³⁹ Further support for the importance of these cells in MS stems from the positive correlation of increased CD39⁺ T_reg frequency and disease remission.Citation⁴⁰ Finally, these results parallel our findings in mice showing enhanced CD39⁺Foxp3⁺ T_regs in the CNS of PSA treated animals during EAE. In addition to reduced presence, attenuated suppressive function among T_regs has also been reported. Foxp3 protein levels correlate with the suppressive function of Foxp3⁺ T_regs.Citation⁴¹ That Foxp3 levels are reduced in MS patients corresponds with additional reports of diminished suppressive capacity among Foxp3⁺ T_regs in MS.Citation^41–44 Exposing circulating T_regs to PSA-activated DCs increased Foxp3 protein expression among these cells. We also observed a significant increase in the frequency of HLA-DR⁺ T_regs, which co-expressed CD39. HLA-DR⁺ T_regs possess reduced suppressive activity in individuals with RRMS pointing to a potential role in regulating autoimmune disease.Citation⁴⁵ Both low frequencies of HLA-DR⁺ T_regs, as well as low HLA-DR expression among these cells correlate with preterm labor and allograft rejection emphasizing the role of these cells in maintaining tolerance.Citation⁴⁶ Lastly, we directly demonstrated that PSA promoted enhanced function by showing a greater reduction in the average amount of LPS-induced monocyte TNFα by cultures containing PSA-exposed T_regs compared to media T_reg controls.

The intestinal microbiota can shape the development of murine host immune systems and impact a number of experimental autoimmune conditions. Indeed, preliminary studies profiling taxonomic shifts within the intestinal microbiota report differences between healthy individuals and those with relapsing MS or inflammatory bowel disease. Interestingly, Faecalibacterium prausnitzii, a noted producer of Foxp3⁺ T_reg-inducing short-chain fatty acid butyrate,Citation^47,48 was decreased in abundance for both autoimmune conditions.Citation^49,50 This demonstration suggests that underlying shifts in the microbiota could relate to Foxp3⁺ T_reg disparities implicated in human autoimmune disease. An increase in Foxp3⁺ T_reg frequency and function by B. fragilis PSA in EAE raises the prospect of using microbiota-based interventions to understand and possibly treat those human conditions associated with autoimmune-related Foxp3⁺ T_regs deficiencies. Our pre-clinical observations in murine EAE and now human blood-derived cells demonstrate consistency in the capacity for PSA to promote Foxp3⁺ T_reg frequency and function. Oral treatment with PSA derived from the human commensal B. fragilis may represent a novel shift in the paradigm for treating autoimmune disease such as human MS. To this end we have made preliminary observations that show that PSA can substantially enhance in vitro production of IL-10 by PBMCs from those with relapsing MS to a greater degree than healthy individuals. Further studies using human PBMCs focused on the capacity of PSA to induce both enhanced T_reg functional activity as well as regulatory cytokine production in those with MS are currently underway.

Materials and Methods

Disclosure of potential conflicts of interests

No potential conflicts of interest were disclosed.

Author Contributions

KMT conducted the research, performed the experiments and wrote the paper; YW, JOR, and SBH contributed to experiment design, data and provided critical review. AP, MC, and CK provided technical assistance. DLK graciously provided purified zwitterionic polysaccharides and comments on the research. LHK supervised the research and reviewed the manuscript.

Funding

Funding sources: The National Multiple Sclerosis Society (RG 4662A2/1) and National Institutes of Health (1R 41AI110170–01).

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.