Identification of α-Mangostin as an Agonist of Human STING
Abstract: A xanthone derivate DMXAA is a potent agonist of mouse STING (stimulator of interferon genes) but cannot activates human STING. We here report that -Mangostin with the xanthone skeleton is an agonist of human STING but activates mouse STING to a lesser extent. Protein and cellular assays indicate that -Mangostin binds to and activates human STING leading to activation of the downstream interferon regulatory factor (IRF) pathway, and production of type I interferons. Furthermore, our studies show that -Mangostin has the potential to repolarize human monocyte-derived M2 macrophages to M1 phenotype. The agonist effect of -Mangostin in the STING pathway might account for its antitumor and antivirus bioactivities.
Introduction
STING (Stimulator of Interferon genes, also referred as ERIS, MITA, TMEM173, MPYS),[1-4] is an endoplasmic reticulum adaptor transmembrane protein that plays a pivotal role in innate immune pathway. STING binds both exogenous and endogenous cyclic dinucleotides (CDNs) via its C-terminal domain and then initiates the downstream TBK1-IRF3 cascade to induce type I interferons.[5] Accumulated evidences have demonstrated that STING is essential for innate immune responses in tumor microenvironment. CD8+ T cell priming which correlates with better prognosis in cancer requires activation of STING pathway in antigen-presenting cells (APCs).[6] Radiation-dependent type I Interferon (IFN) production and ionizing radiation mediated-tumor regression are defective in STING knockout mice but not in those lack of MyD88.[7,8] Moreover, efficacy of checkpoint blockades such as antibodies against CTLA4, PD1/PD-L1 and CD47 highly depends on STING pathway.[9-11] Direct activation of STING by intratumoral injection of agonists effectively eradicates local and distant tumors.[11,12] Combination of STING agonists with PD1/PD- L1 or CTLA4 antibodies further boosted antitumor immunity and enforced control of tumor growth.[10,11,13] Above-mentioned studies highlight that STING is an attractive target in cancer treatment.
Direct STING agonists have been widely used in preclinical and clinical studies for cancer treatment, but most STING agonists are natural cyclic dinucleotides (CDNs) derived from bacterial and human sources,[14-18] which clinical applications are limited by their highly polar character and proteolytic liability.[19] Chemical modifications of CDNs by replacing non-bridging oxygen atoms of phosphodiester linkage with sulfur atoms lead to more favorable analogues,[12,13,19-21] some of them are currently in clinical trials (NCT03172936, NCT03010176 and NCT02675439). However, discovery of more drug-like small molecule STING agonists remains an unmet medical need and highly challenging.
5’6’ dimethylxanthenone-4-acetic acid (DMXAA, also known as ASA404, Figure 1a), a xanthone derivative, has great antitumor effect in mouse models, but ultimately failed in phase III human clinical trial.[22-24] Further mechanistic studies suggest that DMXAA is a potent agonist of mouse STING rather than human STING, which probably account for its clinical failure.[25-27] Another structurally related molecule 10-carboxymethyl-9-acridanone (CMA) is also reported to be a mouse STING-specific agonist.[28] -Mangostin is a dietary xanthone which has been shown to have antitumor and antivirus activities in extensive studies,[29] but the underlying mechanism remains unclear. Given that -Mangostin and DMXAA share the same xanthone skeleton, we hypothesize that -Mangostin might be a potential STING agonist.
In this study, we demonstrated that -Mangostin is an agonist of human STING albeit also weakly activates mouse STING. Further studies show that -Mangostin has the potential to repolarize human monocyte-derived M2 macrophages to M1 phenotype. The agonist effect of -Mangostin on the STING pathway might account for its antitumor and antivirus bioactivities.
Results and Discussion
-Mangostin Potently Induced Type I IFN Production Depending on STING Pathway.We first confirmed that -Mangostin induces type I IFN production in 293T cells transfected with human STING plasmids including both hSTINGH232 and hSTINGR232 in a dose-dependent manner, in which hSTINGH232 is more sensitive compared with hSTINGR232 (Figure 1a-d). In contrast, DMXAA and CMA cannot induce IFN- signal in the same system even under high concentration of 50 μM. On the other hand, DMXAA and CMA potently induces IFN- signal in 293T cells transfected with mouse STING plasmid, while -Mangostin has much weaker activity, suggesting -Mangostin as a potential agonist of human STING rather than mouse STING (Figure 1e). In consistent with previous studies,[16,18] both human and mouse STINGs actively respond to 2’3’-cyclic GMP-AMP dinucleotide (2’3’-cGAMP).
Figure 1. -Mangostin induces production of type I interferon. (a) The chemical structure of DMXAA, CMA and -Mangostin. (b-e) HEK293T cells were transiently transfected with empty vector (b), plasmids encoding hSTINGH232 (c)，hSTINGR232(d) or mSTING (e) together with IFNβ-Luciferase reporter. After 24 h, cells were transfected with 2’3’-cGAMP (2 μg/ml) or stimulated with CMA (50 μg/ml), DMXAA (50 μM) and -Mangostin. Luciferase activity was measured 24h after stimulation. Error bars represent s.d. of independent experiments (n=3). * P < 0.05, ** P < 0.01, *** P < 0.001 (Student’s t-test). To validate the induction of type I IFN by -Mangostin in a native system, we employed Real-Time PCR (qPCR) to examine mRNA level of IFN- and downstream target genes in THP1 cells upon treatment of -Mangostin. -Mangostin enhanced the mRNA level of IFN- in a dose-dependent manner (Figure 2a), and the induction of interferon-stimulating genes including ISG15 and CXCL10 by α-Mangostin showed a similar trend (Figure 2b,c). Interestingly, the mRNA level of IFN- by 2’3’-cGAMP is lower than that induced by α-Mangostin, but the induced mRNA level of ISG15 and CXCL10 is significantly higher for 2’3’-cGAMP. This discrepancy can be explained by the different acting rate for 2’3’- cGAMP and -Mangostin. 2’3’-cGAMP generally acts as a fast- acting STING agonist that induces rapid and robust type I interferon signal in a time-dependent manner that peaks at about 2 hours and then decreases by a negative feedback regulation.[30] On the other hand, the signal induced by -Mangostin peaks at about 6 hours upon treatment and lasts a longer time than that of 2’3’-cGAMP suggesting -Mangostin acting as a slow-acting agonist like CMA.[28] To further verify the STING specificity of -Mangostin, we examined the mRNA level of IFN-, ISG15 and CXCL10 of STING-knockout THP1 cells treated by -Mangostin, 2’3’-cGAMP was used as the positive control (Figure 2). Both -Mangostin and 2’3’-cGAMP did not induce the expression of IFN-β, ISG15 and CXCL10 in the STING-knockout THP1 cells, highlighting that α- Mangostin specifically targets STING to induce type I interferons and its downstream target genes. Figure 2. -Mangostin induces type I interferon production in a STING- dependent manner. THP1 cells and STING-/- THP1 cells were stimulated with 8 μg/ml 2’3’-cGAMP or increased amounts of -Mangostin for 9 h. mRNA level of Ifnb (a), Isg15 (b), and Cxcl10 (c) were measured by Real-Time PCR and normalized by GAPDH expression. Error bars represent s.d. of independent experiments (n=3). * P < 0.05, ** P < 0.01, *** P < 0.001 (Student’s t-test). Previous study has demonstrated that stimulation of cytosolic DNA promotes STING translocated from endoplasmic reticulum (ER) to Golgi.[1,31] To verify if -Mangostin induces type I IFN through STING pathway, we first examined the translocation of hSTING upon the treatment of -Mangostin. 293T cells were transfected with GFP and hSTINGH232-GFP plasmid, and then exposed to DMSO or -Mangostin for one hour. In the DMSO control group, GFP is diffused in the cytoplasm, while hSTING- GFP is localized in a more condensed pattern. Upon the treatment of -Mangostin, hSTINGH232-GFP but not GFP forms the perinuclear punctate structures, resembling the puncta formation induced by 2’3’-cGAMP (Figure 3a), in agreement with the translocation of the STING from the ER to the perinuclear puncta upon activation with CDN.[12,31,32] In support of the activation of STING by -Mangostin, we further found that TANK-binding kinase1 and interferon regulatory factor 3 (IRF3), the downstream targets of STING and the main transcriptional activators of type I α-Mangostin Activated STING-TBK1-IRF3 Pathway. Figure 3. α-Mangostin activates STING-TBK1-IRF3 pathway. (a) HEK293T cells were transfected with GFP or hSTINGH232-GFP for 36h. Then cells were treated with 4 μg/ml 2’3’-cGAMP (transfected by lipofectamine 2000) or 25 μM α-Mangostin for another 1 hour, cells were stained by Hochest 33342 and imaged by fluorescence microscopy. (b, c) THP1 cells and STING-/- THP1 cells were treated with 2’3’-cGAMP (8 μg/ml), CMA (50 μg/ml) and increasing amounts of α-Mangostin for 4 hours in (b) or with 25 μM α-Mangostin for indicated time in (c). Whole cell lysates were prepared and the amount of Phospho-TBK1, total TBK1, Phospho- IRF3, total IRF3, STING and GAPDH were assessed by western-blot. IFNs, were also activated in human THP1 cells by α-Mangostin, reflected by the enhanced phosphorylation of both TBK1 and IRF3 in a dose and time-dependent manner, while no such effect was observed in STING-knockout THP1 cells. In contrast, the mouse STING agonist CMA cannot induce the activation of TBK1 and IRF3 in both the wild type and STING-knockout THP1 cells (Figure 3b, c). -Mangostin Binds and Stabilizes the hSTING CTD. To test if -Mangostin activates STING-TBK1-IRF3 pathway through STING, we applied differential scanning fluorimetry (DSF) to examine the direct binding of -Mangostin to human STING CTD (C-terminal domain,139-378). STING agonists generally enhance the thermal stability of STING as reflected by the increased melting temperature of STING upon binding with agonists like 2’3’-cGAMP.[12] Using this assay, we found that both 2’3’-cGAMP and -Mangostin, but not CMA, increased the melting temperature of human STING CTD in a dose-dependent manner (Figure 4a-f), suggesting that α-Mangostin, like 2’3’- cGAMP, directly binds to and stabilizes human STING CTD. To further cross-validate the direct binding of -Mangostin with hSTING, we measured the binding affinity of -Mangostin with hSTING by microscale thermophoresis (MST). MST results confirmed that 2’3’-cGAMP and -Mangostin bind with hSTING while the negative control CMA does not bind with hSTING (Figure 4g-i). These results are consistent with the thermal shift data and also in agreement with the cellular effects as shown in Figure 1. Figure 4. -Mangostin binds and stabilizes the hSTING CTD. (a-f) The interaction of hSTINGH232 CTD with 2’3’-cGAMP (a), -Mangostin (b) and CMA (c) was analyzed by differential scanning fluorimetry (DSF). Purified human STINGH232 CTD was tested with different concentrations of 2’3’-cGAMP, α-Mangostin and CMA. The melting temperature shifts in the presence of 2’3’-cGAMP (d), -Mangostin (e) and CMA (f) were shown. Error bars represent s.d. of independent experiments (n=3). (g-i) The interaction of hSTINGH232 CTD with 2’3’-cGAMP (g), -Mangostin (h) and CMA (i) was quantified using microscale thermophoresis. 2'3'-cGAMP and -Mangostin displayed the equilibrium dissociation constant (Kd) of 42 ± 15 μM and 137 ± 56 μM, respectively. Points are the average and standard deviation from triplicate measurements. -Mangostin Induces Repolarization of M2 Macrophages towards M1 Phenotype. Tumor-associated macrophages (TAMs) are a key component of tumor-infiltrating immune cells and play an important role in growth, progression and metastasis of tumors.[33] Fully polarized M1 and M2 macrophages are the two extremes of a continuum of functional states of TAMs that have anti-tumor and pro-tumor functions, respectively. Many factors determine the conversion between M1 and M2 macrophages. To examine the impact of - Mangostin on macrophage polarization, we evaluated surface and signaling markers for the polarization of M1 and M2 macrophages derived from THP1 cells. Human THP1 monocytes were firstly differentiated into M0 macrophages by a 24 h incubation in the presence of phorbol 12-myristate 13-acetate (PMA). M0 macrophages were then induced to M1-like macrophages by stimulation with LPS/IFN, and induced to M2-like macrophages by stimulation with IL4/IL13, respectively (Figure 5a). Upon treatment of α-Mangostin, macrophage M1 polarization was characterized by classical M1 markers: CD80, TNF and CXCL10. Meanwhile, macrophage M2 polarization was characterized by classical M2 markers: CD206, CCL22, and fibronectin (Figure 5b-e and Figure 6a-d). -Mangostin significantly increased the M1 markers and decreased the M2 markers on M2 macrophages in a dose-dependent manner. This result is consistent with the previous observation that STING agonists DMXAA and 2’3’-cGAMP re-educate M2 macrophages towards an M1 pro-inflammatory phenotype,[34] further confirming that -Mangostin acts as a STING agonist like DMXAA and 2’3’- cGAMP. Activation of STING by -Mangostin activates both IRF3 and NFB pathways,[35] which accounts for the induced expression of CXCL10 and TNF for both M1 and M2 macrophages. To further characterize the M1-like phenotype of repolarized M2 macrophages induced by -Mangostin, we evaluated phagocytosis of U937 leukemia cells by repolarized M2 macrophages. M2 macrophages induced by IL4/IL13 were preincubated with -Mangostin for 24 h, washed, and mixed with U937 cells in the absence of -Mangostin for another 6 h. Figure 5. -Mangostin re-educates M2 macrophages towards M1-like phenotype. (a) Scheme of the THP1 macrophages differentiation. (b-e) M2 macrophages were treated with increased amounts of -Mangostin for 24h. Protein expression of M1 macrophage marker CD80 (b) and M2 macrophage marker CD206 (c) was assessed by flow cytometry. CD11b+CD80+ macrophages (d) and CD11b+CD206+ macrophages (e) were calculated from FACS results and shown in column. Error bars represent s.d. of independent experiments (n=3). * P < 0.05, ** P < 0.01, *** P < 0.001 (Student’s t-test). Figure 6. -Mangostin significantly decreases the M2 markers (a,b) and increases the M1 markers (c,d). mRNA level of M1 and M2 macrophage related genes was studied by Real-Time PCR and normalized by GAPDH expression. Error bars represent s.d. of independent experiments (n=3). * P < 0.05,** P < 0.01, *** P < 0.001 (Student’s t-test). Phagocytosis was then assessed by cell counting. Pretreatment of α-Mangostin significantly enhances phagocytic capacity of M2 macrophage, which is comparable to that of M1 macrophage (Figure 7a, b), highlighting the M1-like phenotype of repolarized M2 macrophages. Conclusions -Mangostin has a broad spectrum of biological activities including anti-tumor, anti-virus and immunostimulatory effects.[36] Specifically, we are intrigued by the mechanism underlying its pleiotropic properties. In this study, we have demonstrated that α- Mangostin binds and activates human STING, leading to the activation of its downstream TBK1-IRF3 pathway and the production of type I interferon. We further demonstrated that α- Mangostin is capable to re-educate M2 macrophages towards an M1 pro-inflammatory phenotype. Our data suggest that the agonist effect of -Mangostin in the STING pathway may account for its pleiotropic biological activities. M2 macrophages are a major component of tumor-associated macrophages that actively promote tumor initiation, growth, and development. Re-education of M2 macrophages towards M1 phenotype by -Mangostin would form the foundation to develop human STING agonists with more drug-like properties in cancer treatment. Figure 7. -Mangostin enhances phagocytic activity of M2 macrophages. (a) Differentiated M2 macrophages were treated with -Mangostin (0, 5, 7.5, 10 and 12.5 μM) for 24h, -Mangostin was then removed by washing with PBS, and polarized macrophages were mixed with U937 cells in fresh RPMI1640 medium without FBS. After 6h, images were carried out using microscope. (b) Evaluation of macrophages phagocytosis by counting cells. Error bars represent s.d. of independent experiments (n=3). * P < 0.05 (Student’s t-test). Macrophage Phagocytosis Assay. Differentiated M2 cells were treated with -Mangostin (0, 5, 7.5, 10, 12.5 μM) for 24h and then washed with PBS twice. Next, these macrophages (4ⅹ104 cells per test) were mixed with U937 cells (2ⅹ105 cells per test) in fresh RPMI1640 medium without FBS. 6 hours later, images were carried out using Olympus IX73 microscope and cell number was calculated using hemocytometer. Differential Scanning Fluorimetry. Human STINGH232 (140-379) was expressed and purified as described above. Each 20 μl reaction (buffer: 20mM Tris-HCl, 150mM NaCl, 10% DMSO, pH 7.5), with or without 3.125 μM/6.25 μM/12.5 μM/25 μM of 2’3’-cGAMP/-Mangostin/CMA, contained 1 mg/ml of STING and a dilution of 1:500 of SYPRO Orange dye (Sigma). Thermal Shift Assay was performed on CFX 96TM Real-Time System (Biorad) with its SYBR mode using filters Ex/Em: 450–490/560–580 nm. Samples were subjected to a temperature gradient from 15 to 85 °C, with an increment of 0.5 °C and incubation steps of 15 s. Each sample was measured in triplet and the Tm values were determined by fitting the melting curves to a Boltzmann sigmoidal equation using Origin 8 (Origin Lab Corp). Microscale thermophoresis assay (MST). The affinity of the purified hSTINGH232 protein with small molecules (2’3’-cGAMP, CMA and - Mangostin) were calculated in triplicate using Monolith NT. 115. hSTINGH232 protein samples were labelled with NT-647-NHS kit in HEPES buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, and 0.05% Tween 20).The reaction solutions were incubated for 30 minutes under room temperature. Labeled samples were used at a final concentration of 100-200nM in HEPES buffer. Excitation power was optimized by varying power to obtain fluorescence intensities above 200 counts.Vadimezan Other MST settings were: medium MST-power (40 %), laser on-time (20 s), laser off-time (1 s),temperature control (25 ℃ ).