A serine protease inhibitor suppresses autoimmune neuroinflammation by activating the STING/IFN-β axis in macrophages

Multiple sclerosis (MS) is a demyelinating autoimmune disease of the central nervous system (CNS). We have shown that oral administration of Bowman–Birk inhibitor (BBI), a soybean-derived serine protease inhibitor, suppresses disease in experimental autoimmune encephalomyelitis (EAE),¹ a model of MS. We show here that the suppression is dependent on stimulator of interferon genes (STING) and the production of interferon-β (IFN-β) by F4/80⁺ macrophages. Furthermore, we show that the absence of type I IFN receptor-α (IFNAR1) in myeloid cells precludes EAE suppression by BBI, demonstrating that IFN-β signaling in these cells is relevant for the beneficial effect of BBI. BBI also induces IFN-β production by human macrophages and monocytes in a STING-dependent manner, suggesting that BBI could have a therapeutic effect in MS similar to the one in EAE.

It has been shown that BBI induces IFN-β production by a human macrophage cell line,² which prompted us to investigate whether BBI has a similar effect in mice. Indeed, BBI induced substantial IFN-β (but not IFN-α or any other IFN) production by peritoneal and bone marrow-derived macrophages in vivo. We immunized C57BL/6 mice with MOG_35–55 peptide for EAE induction and treated them orally with BBI or phosphate-buffered saline (PBS). RNA-sequencing analysis of splenocytes from these mice identified five pathways enriched after BBI treatment; among them were cellular response to IFN-β, several genes related to the antiviral response, such as Ifnar1, Irf3, and Tbk1, and the immunoregulatory genes Foxp3, Il27a, and Il10 (Fig. 1a).

Fig. 1.

BBI suppresses EAE by inducing IFN-β via a STING-dependent pathway. a C57BL/6 mice (n = 5/group) were immunized with MOG_35–55 for EAE induction and 7 days later treated once with 3 mg of BBI or PBS by oral gavage; splenocytes from these mice were harvested at 1, 3, 5, 7, and 24 h after the treatment, and their RNA was analyzed by RNA-seq. Heat map showing the expression levels (Log 2) of the top 15 genes in the IFN-β pathway in BBI-treated EAE mice. b, c IFN-β(YFP)-reporter mice were immunized for EAE induction and received BBI at the peak of disease. Mice were sacrificed 14 h later, and CD45⁺ cells from the CNS were analyzed by flow cytometry. b Ten leukocyte populations were identified: microglia (CD11b⁺ CD45^low); monocytes (CD11b⁺ CD11c⁺ CD45⁺ Ly6chigh Ly6g⁻); macrophages (CD11b⁺ CD45⁺ Ly6c^low Ly6g⁻ F4/80^high); classical DCs type I (CD11b⁺ CD11c⁺ CD45⁺ Ly6c^med MHCII^high CD26⁺); classical DCs type II (CD11b⁺ CD11c⁺ CD45⁺ Ly6c^med MHCII^high CD172α⁺); plasmacytoid DCs (pDCs; Lin⁻ PDCA1^high); B cells (CD45⁺ MHCII^high CD19⁺); CD8 T cells (CD11b⁻ CD45⁺ CD3⁺ CD8α⁺); and CD4 T cells (CD11b⁻ CD45⁺ CD3⁺ CD4⁺). c Cell type distribution and percentage of IFN-β⁺(YFP⁺) cells among the cells described in b. d IFNAR1 cKO (LysM^CreIFNAR1^fl/fl) and Ctrl (IFNAR1^fll/fl) mice (n = 5/group) were immunized for EAE induction and treated daily with BBI or PBS. e CNS F4/80⁺ macrophages were isolated from IFNAR1 cKO and Ctrl EAE mice described in d at 21 days p.i. and their gene expression analyzed by qPCR. Values are normalized to those measured in PBS-treated Ctrl mice and shown as Log 2. f WT and STING^−/− mice (n = 5 per group) were immunized for EAE induction and treated with BBI or PBS. g Mice described in f were sacrificed at day 24 p.i., and gene expression in their CNS F4/80⁺ macrophages was analyzed by qPCR. Values are normalized to values from PBS-treated WT mice and shown as Log 2. h Human monocyte-derived macrophages were cultured with BBI (100 μg/mL) or PBS for 16 h. STING inhibitor (10 nm/mL) was added 1 h prior to BBI treatment. qPCR analysis of Ifnb1 and Il10 expression. i Heat map of cytokine levels in the supernatants from h. Values are shown as relative expression to that in macrophages treated with PBS. j IFN-β concentrations from the culture supernatants described in h determined by ELISA. Data in d, f were analyzed by two-way ANOVA with Bonferroni correction, **p < 0.001; ***p < 0.0005. Bar graphs depict the mean ± standard error of the mean (SEM). Two-tailed unpaired Student’s t test was used for data analyses in h, j, ***p < 0.0001

We next investigated the cellular sources of BBI-induced IFN-β in EAE. We induced EAE in IFN-β-reporter mice that expressed YFP from the IFN-β gene and treated them with BBI at the peak of disease. By analysis of immune cells from the CNS, we observed a 3.3-fold increase (from 4.5% to 15%) in the number of YFP⁺ cells from BBI-treated mice compared to control mice; F4/80^high macrophages (86%) were the predominant producers of IFN-β (Fig. 1b, c). In control mice, type 2 classical dendritic cells (cDCs) (98%) were the only producers of IFN-β (Fig. 1c). These data suggest that BBI acts primarily on a subset of F4/80⁺ macrophages present throughout the body, causing the systemic increase in IFN-β that leads to EAE suppression.

It has been shown that endogenously produced IFN-β acts on myeloid cells to suppress CNS autoimmunity.³ To determine whether this is also the case with BBI-induced IFN-β, we generated IFNAR1 conditional knockout (cKO) mice lacking IFNAR1 in LysM⁺ cells (macrophages and other myeloid populations). IFNAR1 cKO mice were immunized for EAE induction and treated with BBI or PBS daily by gavage. The lack of IFNAR1 signaling in myeloid cells precluded EAE suppression by BBI (Fig. 1d). Furthermore, messenger RNA (mRNA) transcripts associated with IFN-β signaling, such as Mx1, Il10, Stat1, and Stat3, were downregulated in IFNAR1 cKO F4/80⁺ cells after both BBI and PBS treatment; however, we did not find any difference in Sting or Irf3 mRNA levels between BBI-treated control and IFNAR1 cKO mice (Fig. 1e). Furthermore, IFNAR1 cKO F4/80⁺ macrophages exhibited an “M1-like phenotype,” upregulating the expression of several proinflammatory genes, such as Il6, Nos2, Il1b, Tnf, and Il23a, compared to the macrophages from BBI-treated control mice (Fig. 1e). Overall, these data suggest that IFNAR1 signaling in LysM⁺ cells is crucial for EAE suppression by BBI.

Given that BBI induces the STING pathway (Fig. 1a), we tested whether STING is required for EAE suppression by BBI. STING^−/− mice developed typical disease;⁴ however, BBI failed to reduce EAE severity (Fig. 1f). Furthermore, sorted STING^−/− CNS-derived F4/80⁺ macrophages restimulated with BBI failed to upregulate the expression of Tbk1, Irf3, Ifnar1, Ifnb1, or Il10 compared to BBI-treated WT macrophages (Fig. 1g). Overall, these data confirm that STING is required for the activation of immunoregulatory pathways by BBI.

Finally, we investigated whether the effects of BBI could be recapitulated in human macrophages. Human monocyte-derived macrophages from peripheral blood mononuclear cells of healthy subjects treated with BBI produced IFN-β and interleukin-10 (IL-10) (Fig. 1h, j). Moreover, a STING inhibitor precluded the induction of immunoregulatory cytokines by BBI. Indeed, when STING was inhibited, we found a downregulation of IFN-β and IL-10 in BBI-treated macrophages (Fig. 1h, j). Collectively, these data show that BBI in the human system induces immunoregulatory pathways in a STING-dependent manner.

In conclusion, these findings further elucidate the anti-inflammatory mechanisms of BBI and validate its use as a potential oral therapy for MS.