Discovery and Mechanistic Study of a Novel Human STING Agonist

Abstract

Stimulator of interferon genes (STING) is an integral ER membrane protein that can be activated by 2’3’-cGAMP synthesized by cyclic guanosine monophosphate-adenosine monophosphate synthase (cGAS) upon binding of double-stranded DNA and activates interferon (IFN) and inflammatory cytokine response to defend against microorganism infection. Pharmacologic activation of STING has been demonstrated to induce an antiviral state and boost antitumor immunity. We previously reported a cell-based high throughput screening assay that allowed for identification of small molecule cGAS-STING pathway agonists. We report herein a compound, 6-bromo-N-(naphthalen-1-yl)benzo[d][1,3] dioxole-5-carboxamide (BNBC), that induces proinflammatory cytokine response in a human STING-dependent manner. Specifically, we showed that BNBC induced type I and III IFN dominant cytokine responses in primary human fibroblasts and peripheral blood mononuclear cells (PBMCs). BNBC also induced cytokine response in PBMC-derived myeloid dendritic cells and promoted their maturation, suggesting that STING agonist treatment could potentially regulate the activation of CD4+ and CD8+ T lymphocytes. As anticipated, treatment of primary human fibroblast cells with BNBC induced an antiviral state that inhibited the infection of several members of flaviviruses. Taken together, our results indicate that BNBC is a human STING agonist that not only induces innate antiviral immunity against a broad spectrum of viruses but may also stimulate the activation of adaptive immune response, which is important for treatment of chronic viral infections and tumors.

Graphical Abstract

Pattern recognition receptors (PRRs) are host cellular proteins that recognize pathogen associated molecular patterns, such as viral nucleic acids, capsids, bacterial peptidoglycans, lipopolysaccharides or other metabolites¹, and subsequently activate a proinflammatory cytokine response and cell death pathways^2–3. These innate immune responses not only limit the proliferation and spread of microorganisms in the early phase of infection, but also facilitate the induction of a more powerfully adaptive immune response that can ultimately resolve the infections^4–5. In addition, PRR-mediated innate immune response also plays an important role in surveillance of tumor genesis and invasion^6–8. Stimulator of interferon genes (STING) is an endoplasmic reticulum (ER) membrane protein that plays a central role in the activation of innate immune response by multiple cellular DNA sensors^9–10. Particularly, recognition of cytoplasmic double-stranded DNA by cyclic GMP-AMP synthase (cGAS), the primary cytoplasmic DNA sensor^11–13, catalyzes the synthesis of a cyclic dinucleotide c[G(2’,5’)pA(3’,5’)p], or 2’3’-cGAMP for short¹⁴, which binds to STING and induces its dimerization and translocation from the ER membrane to perinuclear vesicles and activates downstream signaling components, such as NFkB and TBK-1/IRF3^15–17, leading to the expression of interferons and other proinflammatory cytokines and chemokines^17–18. STING is expressed in professional innate and adaptive immune cells and plays important roles in innate and adaptive immune response to various pathogens and tumors^{6, 19–21}. It is also well documented that STING responds to mitochondrial damages^22–25, DNA double-strand breaks^26–27, and its over-activation may contribute to the onset of autoimmune diseases such as systemic lupus erythematosus^28–29.

Due to its critical role in host immune responses, pharmacological modulation of STING activity has been considered as a viable immunotherapeutic approach for treatment of pathogen infections, tumors and autoimmune disorders. Indeed, recent studies showed that intra-tumor administration of 2’3’-cGAMP induced profound regression of established tumors in mice and generated substantial systemic immune responses capable of rejecting distant metastases and providing long-lived immunologic memory^30–31. STING agonists had also been demonstrated to potentiate the efficacy of immune checkpoint blockade therapy^32–33 and enhance the immunogenicity of vaccines^34–35. In addition, we and others have demonstrated that STING agonist therapy was able to induce a host immune response to control the infection of influenza A virus³⁶, hepatitis B virus^{18, 37}, herpes simplex virus³⁸ and human immunodeficiency virus³⁹. These studies prove the concept that pharmacologic activation of STING is an attractive immunotherapeutic approach to treat viral infection and cancers. Moreover, recent discovery of a small molecular inhibitor of STING paves the way for treatment of systemic lupus erythematosus and other autoimmune diseases⁴⁰.

Although several analogs of cyclic dinucleotide (CDN)-based STING agonists have been tested in tumor models in mice and demonstrated antitumor activity through activation of anticipated immune response³⁴, their medical application has been significantly limited by the low cell membrane permeability⁴¹. Accordingly, extensive efforts have been made to discover non-cyclic dinucleotide small molecule human STING agonists. One approach is to modify the mouse-specific STING agonist 5’,6’-dimethylxanthenone-4-acetic acid (DMXAA), which has not yet resulted in identification of novel derivatives capable of binding human STING⁴². However, α-mangostin, a dietary xanthone with antitumor and antiviral activities^43–44, was demonstrated to bind and activate both mouse and human STING⁴⁵. Another approach is through biochemistry- or cell-based high throughput screening (HTS). Screening of small molecules that can compete the binding of a radio-labeled cyclic dinucleotide STING agonist to the C-terminal domain of human STING identified a aminobenzimidazole compound and its dimer derivatives as potent human STING agonists⁴⁶. Alternatively, taking cell-based HTS approaches, three chemotypes of compounds have been identified to activate a cytokine response in a human STING-dependent manner^47–49. We report herein another novel human STING specific agonist from a cell-based HTS campaign and its pharmacological properties. Our preliminary medicinal chemistry studies indicate that this carboxamide STING agonist is a good candidate for further lead optimization and development toward a therapeutic agent for treatment of chronic viral infections and tumors.

Results

Discovery of a carboxamide compound that activates cGAS-STING pathway

We previously established a HepG2 cell-derived cell line with reconstituted human cGAS-STING pathway and ISG54 promotor-driven luciferase for high throughput screening of human cGAS-STING pathway agonists⁴⁹. Screening of the NCI Diversity Set V collection of 1,594 small molecule compounds identified a compound, 6-bromo-N-(naphthalen-1-yl)benzo[d][1,3]dioxole-5-carboxamide (BNBC) (Fig. 1A), that significantly induced ISG54-driven luciferase expression in HepAD38/cGAS-STING/ISG54 cells. As shown in Fig 1B, treatment of HepAD38/cGAS-STING/ISG54Luc cells with BNBC for 4 h induced luciferase expression in a concentration-dependent manner. On the contrary, BNBC did not induce luciferase expression in HepAD38/cGAS-STINGΔC/ISG54Luc cells that express human STING with truncation of carboxyl-terminal domain (CTD). Because the CTD of STING is essential for the activation of IRF3 and subsequent induction of cytokines, the cGAS-STINGΔC reporter cell line serves as a negative control¹⁸. While significant ISG54 promotor-driven luciferase activity was detected at low micromolar concentration in cells expressing human STING, no cytotoxicity was detected at up to 200 μM concentration. (Fig. 1C). Furthermore, BNBC also concentration-dependently induced ISG54 driven luciferase expression in HepG2/STING/ISG54 reporter cells lacking cGAS (Fig. 1B). These results thus suggest that BNBC is a cGAS/STING pathway activator and most likely targets a cellular component down-stream of cGAS, but at or up-stream of STING.

Figure 1. Carboxamide compound BNBC activates cGAS-STING pathway in a functional human STING-dependent manner.

(A) Chemical structure of BNBC. (B) Activation of ISG54 promoter activities by BNBC were determined in HepAD38/cGAS-STING/ISG54Luc, HepAD38/cGAS-STINGΔC/ISG54Luc cells and HepG2/STING/ISG54Luc cells. Luciferase activity was measured at 4 h post treatment and expressed as fold of induction (mean ± standard deviation, n=3). * indicates p<0.05 compared to mock treated control. (C) Cytotoxicity was determined in HepAD38/cGAS-STING/ISG54Luc and HepAD38/cGAS-STINGΔC/ISG54Luc cells after 4 h treatment and expressed as percent of mock treated control (mean ± standard deviation, n=3).

BNBC is a human STING specific agonist

In order to precisely determine the cellular target of BNBC, we first examined the effects of BNBC on proinflammatory cytokine gene expression in HepG2 cells reconstituted with human or mouse STING, respectively, using mouse STING specific agonist DMXAA as a control. As shown in Fig. 2A and B, consistent with results obtained with ISG54 reporter assays, BNBC concentration-dependently induced IFN-β, IL29 (IFN-λ1) and TNF-α mRNA expression in HepG2/STING cells (Fig. 2A), but not in parental HepG2 cells, HepG2/STINGΔC cells and HepG2-derived cell line expressing mouse STING (HepG2/mSTING) (Fig. 2B). As expected, DMXAA only induced IFN-β mRNA expression in HepG2/mSTING cells, but not in HepG2/STING cells expressing human STING. Because the only difference between HepG2/hSTING and HepG2/mSTING cell lines is the expression of STING from human or mice, the results thus indicate that as DMXAA is a specific agonist of mouse STING, BNBC should be a specific agonist of human STING. In agreement with this notion, while DMXAA treatment only induced the peri-nuclear translocation of mSTING and nuclear translocation of IRF3, BNBC treatment induced the peri-nuclear translocation of HepG2 cells expressing human STING, but not mouse STING (Fig. 2C). Interestingly, although BNBC treatment induced peri-nuclear translocation of both full-length and C-terminally truncated hSTING, it only induced IRF3 phosphorylation and nuclear translocation in HepG2/STING cells, but not in HepG2/STINGΔC cells (Fig. 2D). These results imply that although BNBC might specifically bind to human STING and induce its subcellular translocation in a CTD-independent manner, its activation of IRF3 phosphorylation and nuclear translocation is CTD-dependent, which is consistent with the essential role of STING CTD in recruitment of TBK1 and phosphorylation of IRF3¹⁷. Taken together, our results presented above indicate that BNBC is a human STING specific agonist.

Figure 2. BNBC is a human STING specific agonist.

(A) HepG2/STING cells were treated with indicated concentrations of BNBC for 6 h. The cytokine mRNAs were measured by qRT-PCR and expressed as fold of induction (mean ± standard deviation, n=4). (B) HepG2, HepG2/STINGΔC and HepG2/mSTING cells were treated with indicated concentrations of BNBC or DMXAA for 6 h. Induction of IFN-β mRNAs were determined by qRT-PCR and expressed as fold of induction (mean ± standard deviation, n=4). * indicates p<0.05 compared to mock treated control. (C) HepG2/STING, HepG2/STINGΔC and HepG2/mSTING were mock treated or treated with 100 μM of BNBC or 50 μM of DMXAA for 2 h. The expression and subcellular localization of STING and IRF3 were determined by immunofluorescent staining (green). Cell nuclei were stained with DAPI (blue). (D) HepG2/STING and HepG2/STINGΔC cells were treated with 100 μM BNBC for 2 h. Total and phosphorylated IRF3 were determined by Western blot assay. β-actin served as a loading control.

BNBC induces a proinflammatory cytokine response and establishment of antiviral state in human foreskin fibroblast cells (THF) in a STING-dependent manner.

THF is a cell line derived from primary human diploid foreskin fibroblasts (HFF)⁵⁰ and expresses physiologically relevant levels of STING^48–49. As observed in HepG2-derived cell line ectopically expressing functional human STING, BNBC concentration-dependently induced IFN-β, IL-28A and IL-6 mRNA expression as well as secretion of IFN-β into culture media in THF cells (Fig. 3A and B) in the absence of cytotoxicity (Fig. 3C). Also consistent with its role as a STING agonist, knocking out the expression of STING, but not TRIF or IPS-1 (MAVS), in THF cells completely abolished BNBC-induced IFN-β mRNA expression (Fig. 3D).

Figure 3. BNBC activation of inflammatory cytokine response in THF cells is STING-dependent.

(A) THF cells were treated with indicated concentrations of BNBC for 6 h, followed by quantification of cytokine mRNAs by qRT-PCR assays. The levels of cytokine mRNA were expressed as fold of induction (mean ± standard deviation, n=4). (B) THF cells were treated with indicated concentrations of BNBC or cGAMP for 18 h, followed by detection of IFN-β in culture medium with ELISA assay (mean ± standard deviation, n=4). (C) Cytotoxicity was determined in THF cells and expressed as percent of mock treated control (mean ± standard deviation, n=3). (D) Parental THF cells as well as THF cells with knockout of STING, IPS-1 or TRIF were treated with 50 μM of BNBC for 6 h. IFN-β mRNA was detected by qRT-PCR and expressed as fold of induction (mean ± standard deviation, n=4). * indicates p<0.05 relative to mock treated control. (E and F) RNAseq analysis was performed using RNA samples extracted from THF cells treated with 50 μM of BNBC for indicated time periods. Venn diagram (E) illustration of numbers of up- or downregulated genes with fold change greater than 2 and FDR <20%, relative to mock treated control. The heat map (F) is expressed as the log2 fold difference.

To determine the BNBC-induced gene expression profile, RNAseq analysis was performed with RNA extracted from THF cells harvested at 4, 6 and 8 h post BNBC treatment. As illustrated in Fig. 3E and F, BNBC altered the expression of 132, 211 and 310 genes at 4, 6 and 8 h of treatment, respectively. Ingenuity canonical pathway analysis indicated that many components of PRR and interferon signaling pathways, including downstream factors of IRF3 and IRF7, STAT1 and STAT3, were up-regulated (Table 1). In addition, several top regulator effect networks being altered after BNBC treatment are related to the activation of antigen presenting cells, myeloid cells and leukocytes. The RNAseq analyses thus revealed that BNBC treatment induced a typical innate proinflammatory cytokine response gene expression profile consistent with the activation of STING pathway.

Table 1.

Analysis of gene alterations using Ingenuity analysis

Ingenuity canonical pathway	p value	Overlap
Interferon signaling	4.8E-13	27.8% (10/36
Activation of IRF by cytosolic PRR	2.0-E10	15.9% (10/63)
Role of PRR in recognition of bacteria and viruses	3.9E-08	8.0% (11/137)
Upstream regulator	p value	Predicted activation
IRF7	2.5E-57	Activated
STAT1	6.1-E46	Activated
IRF3	5.7E-43	Activated
STAT3	8.4E-36	Activated
NKX2–3	4.6E-34	Inhibited
Top regulator effect networks	Consistency score	Diseases & functions
CNOT7, EF1, IFI16, IRF1,3,5,7, NFATC2, NKX2–3…	76.42	Activation of APCs
CNOT7, EBF1, ESR1, HIF1A, IFI16, IRF1,3,4,5…	48.91	Accumulation of myeloid cells
HIF1A, IFI16, IRF5,7, KLF2, STAT4…	28.74	Activation of leukocytes

In agreement with its induced gene expression profile that indicates the induction of antiviral state, BNBC treatment efficiently protected Dengue virus (DENV), Yellow fever virus (YFV) and Zika virus (ZIKV) infection of THF cells. As shown in Fig. 4, pre-treatment of THF cells with BNBC for 8 h concentration-dependently reduced the intracellular DENV, YFV and ZIKV RNA levels by up to 90, 150 and 800-fold, respectively. Similarly, pre-treatment of THF cells with another small molecule human STING agonist, G10⁴⁸, reduced virial RNAs in similar potencies against the three flaviviruses.

Figure 4. Pretreatment of THF cells with BNBC inhibited the replication of Dengue, Yellow fever, and Zika viruses.

THF cells were treated with indicated concentrations of BNBC or control compound G10 for 8 h. Cells were then infected with DENV (A), YFV (B), or ZIKV (C) at MOI of 0.1 for 1 h and cultured for additional 48 h. Viral RNAs were determined by qRT-PCR, and expressed as values relative to mock infected control (mean ±standard derivations, n=3). * indicates p<0.05 compared to mock treated control.

BNBC induced inflammatory cytokine response in human peripheral blood mononuclear cells (PBMCs) and promoted PBMC-derived dendritic cell maturation

STING is expressed in professional innate and adaptive immune cells and plays important roles in innate and adaptive immune response to various pathogens and tumors^{6, 51}. In order to investigate the effects of BNBC on the function of professional immune cells, we first examined BNBC-induced cytokine response in human PBMCs and demonstrated that BNBC efficiently induced expression of type I and type III IFN, but to a lesser extent, TNF-α (Fig. 5A). Interestingly, while cGAMP significantly induced the expression of IFN-β, IL-28A and IL-29 in PBMCs derived from all three donors (Fig. 5B), BNBC and G10 treatment only significantly enhanced type I and III IFN mRNA expression in PBMCs derived from 2 out of 3 donors (Fig. 5B). It remains to be further investigated whether the unresponsiveness of PBMC-derived from donor 1 is associated with STING gene variation that may impede the binding and activation by BNBC and G10 or other physiological factors that affects the metabolism of the non-cyclic dinucleotide STING agonists.

Figure 5. Activity of BNBC in human PBMCs.

(A) PBMCs freshly isolated from human blood were treated with indicated concentration of BNBC for 6 h and followed by detection of cytokine mRNAs by qRT-PCR. The levels of cytokine mRNA were expressed as fold of induction (mean ± standard deviation, n=4). (B) PBMCs freshly isolated from human blood of additional three healthy donors were treated with 15 μM of 2’3’-cGAMP, 60 μM of BNBC or G10 for 6 h. The mRNAs of IFN-β, IL-28A, IL-29 and TNFα were detected by qRT-PCR and expressed as fold of induction (mean ± standard deviation, n=3). * indicates p<0.05 compared to mock treated control.

Dendritic cells are professional antigen presentation cells that function at the interface of innate and adaptive immune responses and play a critical role in immune control of virus infection. It has been shown that activation of pattern recognition receptors (PRRs) by pathogen-associated molecular patterns (PAMPs) in dendritic cells initiates essential innate immune signaling cascades resulting in maturation of dendritic cells and up-regulation of their antigen presenting machinery as well as costimulatory molecules such as CD80. As shown in Fig. 6A, treatment of PBMC-derived dendritic cells with BNBC (100 μM) or LPS (10 μg/ml) for 6 h efficiently induced the expression of IFN-β and IL-29 mRNAs and to a lesser extent, TNF-α and IL-6 mRNAs. Moreover, treatment of PBMC-derived dendritic cells with BNBC (50 μM) or LPS (10 μg/ml) for 48 h significantly induced expression of CD80 (Fig. 6B and C). These results suggest that activation of STING is indeed able to induce the maturation of dendritic cells.

Figure 6. BNBC activated type I and III interferon dominated inflammatory cytokine response in PBMC-derived dendritic cells, and promoted dendritic cell maturation.

(A) Human PBMC-derived dendritic cells were treated with 10 μg/ml LPS or 100 μM of BNBC for 6 h. Cytokine mRNAs were detected by qRT-PCR and expressed as fold of induction relative to mock treated control (mean ± standard deviation, n=3). (B and C) Human PBMC-derived dendritic cells were treated with 10 μg/ml LPS or 50 μM of BNBC for 48 h. CD80 expression was gated from CD11c+ cells (B). Fluorescent intensity of CD80 from 4 independent experiments using PBMCs from 2 donors were expressed as mean ± standard deviation (MFI) (C). * indicates p< 0.05 compared to mock treated control.

Preliminary structure-activity-relationship studies identified active pharmacophores of BNBC

BNBC is an amide with two fused rings on each side of the acid and amine with a MW of 370 and calculated logP of 4.0. The necessary pharmacophores of BNBC were determined through evaluating the analogs with structural alterations around B ring (57093Z, 57057Z), C ring (57071Z, 57076, and 59011Z), or both B and C rings (59016Z) (Table 2). BNBC analogs were prepared in house and tested for their activities to induce ISG54 promoter-driven luciferase expression in HepG2/hSTING cells and ranked by the Minimum Effective Concentration to induce 5-fold luciferase activity (MinEC_5X) relative to the mock-treated controls. As shown in Table 2, introduction of a more electronegative nitrogen atom to either B ring (57057Z) or D ring (57076) will make the resulting products more polar and thus more likely improve water solubility. However, both compounds lost activities to induce the ISG54 promotor-driven luciferase expression. Similar inactivation was also observed when a polar and water solubilizing group, morpholine, was added to the 2-position of the C ring (59011Z). These results suggested that the binding site of BNBC might be highly hydrophobic. In contrast, two analogs of BNBC with either opened A ring (57093Z) or D ring (57071Z) maintained low MinEC_5X values of 2.5 and 2.6 μM, respectively, similar to that of BNBC (2.2 μM). However, compound 59016Z, with both opened A and D rings, has a much higher MinEC_5X of 50 μM, suggesting a reduced activity in activating STING. Interestingly, the compound 57093Z has significantly increased maximum fold of induction of ISG54 promoter activity indicating that it might be a more potent STING activator. These results suggested that either A ring or D ring can be opened and substituted with alkyl, halogen, or other groups to maintain or improve the STING activation activity and thus established a base for further chemical modification.

Table 2.

Structure-activity relationship study of BNBC

Cmpd*	BNBC	57093Z	57057Z	57071Z	57076	59011Z	59016Z
Structure
1MinEC5X (μM)	2.2	2.5	> 200	2.6	> 200	> 200	50
1Maximum induction (fold)	20.5	28.0	-	17.3	-	-	7.8
2CC50 (μM)	> 200	> 200	> 200	> 200	> 200	> 200	> 200

^*compound.

¹Luciferase activity was determined in HepG2/STING/ISG54Luc cells and expressed as Minimum Effective Concentration to induce 5-fold luciferase activity or maximum fold of luciferase induction.

²Cytotoxicity was determined in HepG2/STING/ISG54Luc cells by MTT assay.

To evaluate the pharmacological properties of BNBC and its active analogs, 57071Z and 57093Z, an in vitro ADME profiling studies were performed. As summarized in Table 3, BNBC has good caco-2 permeability and acceptable plasma protein bindings, which predict a favorable oral absorption rate if the molecule can be made in solution. A panel of 9 cytochrome P450 (Cyp) enzymes (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, CYP2E1 and 3A4 with two different substrates) were analyzed in 10 reactions. While BNBC and 57071Z each inhibits 5 out of 10 Cyps/reactions, including relatively more common enzymes CYP3A4 (determined with Midazolam and Testosterone) and/or CYP2D6, 57093Z, a more potent BNBC analog with an opened A ring, exhibits an improved Cyp inhibition profile and inhibits 2 out of 10 Cyp enzymes, CYP1A2 and CYP2C9, which are not considered to be as critical as CYP3A4 and 2D6. One major issue of BNBC and its analogs is the poor aqueous solubility in both pH 4 and pH 7.4 at sub- or low-micromolar. Further chemical modification is necessary to improve the solubility, since an extremely low aqueous solubility is hard to overcome even through proper formulation and dosing as a suspension. As shown in Table 3, BNBC and its two analogs apparently have very rapid metabolic rate in both human and mouse liver microsomes, predicting a low metabolic stability in vivo and raising concern for insufficient overall drug exposure. However, since high systematic exposure is the major concern for STING agonist to cause systematic inflammatory side effect, the metabolic property of BNBC predicts a low systematic exposure after oral dosing and favors intrahepatic STING activation, which may reduce systematic toxicity and is thus beneficial for treatment of chronic HBV infection.

Table 3.

ADME profile of BNBC and its analogs

Cmpd*	Aqueous solubility (μM)		Metabolic stability T1/2 (min)		Caco-2			Cyp#	PPB (%)
	PH = 4	PH =7.4	Human	Mouse	Papp (A-B) (10−6, cm/s)	Papp (B-A) (10−6, cm/s)	Efflux Ratio	IC50 < 10 μM	Human	Mouse
BNBC	3.57	2.62	19.96	2.21	14.46	5.98	0.41	5/10	97.34	97.88
57093Z	0.28	0.26	9.50	2.70	ND	ND	ND	2/10	ND	ND
57071Z	4.14	3.94	15.39	1.81	ND	ND	ND	5/10	ND	ND

^*compound.

^#Cyps inhibited by BNBC are 1A2, 2C19, 2D6, 3A4 (Midazolam) and 3A4 (Testosterone); Cyps inhibited by 57093Z are 1A2 and 2C19; Cyps inhibited by 57071Z are 1A2, 2B6, 2C19, 3A4 (Midazolam) and 3A4 (Testosterone);

Discussion

BNBC is a carboxamide compound from NCI Diversity Set V compound library and was identified to induce the expression of interferon and other proinflammatory cytokines in HepG2 cells expressing functional human, but not mouse STING, suggesting that BNBC is a human STING specific agonist (Fig. 2). In agreement with this notion, BNBC induced STING peri-nuclear translocation as well as phosphorylation and nuclear translocation of IRF3 (Fig. 2). Moreover, BNBC induced cytokine response in THF and PBMCs. Its activity to induce cytokine response in THF cells was dependent on the expression of STING, but not MAVS and TRIF, adaptor for RIG-I-like receptors and TLR3/TLR4, respectively (Figs. 3 and 5). Importantly, BNBC was able to induce a gene expression profile that is consistent with STING activation (Fig. 3) and established an antiviral state in treated cells to restrict the infection of three different flaviviruses (Fig. 4). Although our biological data presented herein suggest that STING is likely the target of BNBC, it remains to be determined whether BNBC can directly bind to STING.

Due to its critical role in innate and adaptive immunity, STING is an important therapeutic target for diseases with dysfunctional immune response, particularly, cancers and chronic viral infections. Preclinical and clinical studies with different formulations of CDN-based STING agonists demonstrated promising efficacy to various tumors and adjuvant activity to induce more robust cellular and humoral immune responses. However, due to its poor cell permeability and metabolic stability, the CNDs were administrated via intra-tumor injection, which limited their application, particularly for treatment of viral infections and non-solid tumors. In fact, studies with DMXAA in mice clearly demonstrated that systematic administration of STING agonists is efficacious to tumors and viral infections^52–53. However, all the non-CDN STING agonists discovered through cell-based HTS thus far only specifically activate human, but not mouse STING^48–49. This species specificity impeded the in vivo evaluation of their therapeutic efficacy in mice models. Interestingly, α-mangostin and amidobenzimidazole derivative appear to activate both human and mouse STING, indicating that the discovery and development of human and mouse dual active non-CDN STING agonists is possible^45–46. Further investigation into the mechanism of different agonists to activate STING should provide clues for the development of dual active STING agonists.

Considering the mode of action on STING agonist therapy of tumors, the current experimental evidence indicates that the activation of STING in tumor cells can induce a cytokine response and cell death^{31, 34}, which will enhance cross-presentation of tumor antigens by dendritic cells and activate T cell immune response against tumors. In addition, activation of STING in tumor stromal cells and infiltrated immune cells may also contribute to the induction of efficient anti-tumor immune response. In the case of chronic viral infection, such as chronic hepatitis B, while activation of the cytokine response in either virus infected cells or non-infected cells may inhibit viral replication, restoration of host adaptive antiviral response is essential to control the persistent viral infection and achieve clinical cure^54–56. Therefore, in the course of developing STING agonist, it is important to not only evaluate its activity on cytokine induction but should also assess its ability to enhance adaptive antiviral immunity in relevant cell culture and in vivo models. Our studies reported herein showed that BNBC not only activates an inflammatory cytokine response, but also induces dendritic cell maturation (Fig. 6). With this encouraging observation, we will further investigate immune modulating activity of BNBC and its derivatives (Table 2) in vivo in humanized mice models with human hepatocytes and an adaptive immune system^57–58.

Conclusion

BNBC represents a new chemotype human STING agonist that can activate innate and adaptive immune response for treatment of viral infections and tumors.

Materials and Methods

Cells and viruses

HepG2 is a human hepatoblastoma cell line obtained from ATCC. HepG2-derived cell lines with reconstituted human cGAS and/or STING, with/without firefly luciferase under the control of an ISG54 promoter (HepAD38/cGAS-STING/ISG54Luc, HepG2/STING/ISG54Luc and HepG2/STING), with reconstituted human STING with 39 amino acid deletion on CTD required for downstream signaling, but not ligand binding (HepAD38/cGAS-STINGΔC/ISG54Luc, HepG2/STINGΔC), and with reconstituted mouse STING (HepG2/mSTING) were described previously^{18, 49}. THF cells are primary human diploid foreskin fibroblasts (HFF) with reconstitution of the catalytic subunit of human telomerase⁵⁰. THF with STING knockout (THF-ΔSTING), IPS-1 knockout (THF-ΔIPS-1) or TRIF knockout (THF-ΔTRIF) were described and characterized previously⁵⁹. Human peripheral blood mononuclear cells (PBMCs) were isolated from the whole blood of four healthy donors (Biological Specialty) using Ficoll-Paque density gradient centrifugation (Miltenyi Biotech). Freshly isolated PBMCs were cultured in RPMI-1640 medium (Corning) containing L-Glutamine and 10% FBS. To isolate PBMC-derived dendritic cells, the suspended cells were removed from freshly isolated PBMCs after overnight culture. The remaining adherent cells were stimulated with human recombination GM-CSF (50 ng/ml, Gibco) and IL-4 (50 ng/ml, Gibco) for 7 days to induce the production of dendritic cells.

Zika virus (PRVABC59 strain) was purchased from ATCC. Yellow fever virus (17D strain) and dengue virus (serotype 2, New Guinea C strain) stocks were produced by electroporation of Huh7.5 cells with in vitro transcribed RNAs from the corresponding cDNA constructs, as described previously^{49, 60–61}.

Chemicals

Diversity Set V compound library containing 1,594 structure diversified small molecule compounds were acquired from the National Cancer Institute (NCI). 6-bromo-N-(naphthalen-1-yl)benzo[d][1,3]dioxole-5-carboxamide (BNBC) and its analogs were synthesized in house with > 95% purity. Synthesis and characterization data of BNBC and its analogs are shown in Supplemental Materials. BNBC and its analogs were dissolved in DMSO at a stock concentration of 50 mM. 2’3’-cGAMP and LPS were purchased from InvivoGen. DMXAA was purchased from AdooQ BioScience. G10, a previously reported human STING agonist, was purchased from Aobious⁴⁸.

ISG54 luciferase reporter assay

HepAD38/cGAS-STING/ISG54Luc cells, HepAD38/cGAS STINGΔC/ISG54Luc cells or HepG2/STING/ISG54Luc cells were seeded in black wall/clear bottom 96-well plates at a density of 4×10⁴/well in 0.1 mL of medium for overnight. HTS and hit identification were as described previously⁴⁹. To determine the activity of BNBC, the cells were treated with 1% DMSO (mock treated control) or indicated concentrations of BNBC (or its analogs) for 4 h. The firefly luciferase activities were measured by using Steady-Glo substrate (Promega) with TopCount (Perkin Elmer)⁴⁹.

Cytotoxicity assay

Cell viability was measured using MTT assay (Sigma).

Analysis of cytokine expression

IFN-β mRNA, TNF-α mRNA, IL-28A, IL-28B, IL-29 mRNA, IL-1β and IL-6 mRNA were measured using total cellular RNA, by quantitative RT-PCR (qRT-PCR) assays^{18, 49}. β-actin mRNA served as internal. Cytokine secreted into culture medium was detected by ELISA assay (RD Systems).

Western blot assay

Cells were lysed with NuPAGE® LDS sample buffer (Thermo Fisher Scientific) supplemented with 2.5% 2-Mercaptoethanol (Sigma) followed by electrophoresis in NuPAGE 4–12% Bis-Tris Gel (Thermo Fischer Scientific) and transfer onto a PVDF membrane (Thermo Fischer Scientific). Primary antibodies against total and phosphorylated IRF3 or β-actin were from Cell Signaling. Secondary antibodies were from LI-COR. Bolts were imaged with LI-COR Odyssey system (LI-COR Biotechnology).

Immunofluorescent assay

HepG2/STING, HepG2/STINGΔC and HepG2/mSTING cells were treated with either BNBC or DMXAA for 2 h. The cells were fixed with PBS containing 4% paraformaldehyde followed by incubation with 0.1% Triton X-100 for 20 min. Antibodies against IRF3 or STING were from Cell Signaling. Alexa Fluor 488-conjugated secondary antibody was from Invitrogen⁴⁹. Cell nuclei were stained with DAPI.

RNAseq and data analysis

mRNA sequencing (mRNA-Seq) assays were performed in the Genomics Facility of Fox Chase Cancer Center. RNA sample quantity was measured by UV absorbance at 260, 280, and 230 nm with a NanoDrop 1000 spectrophotometer (ThermoScientific). RNA integrity was determined by bioAnalyzer using an RNA Nano chip instrument (Agilent Technologies). 1000 ng total RNAs from each sample were used to make mRNA-seq library using the Truseq stranded mRNA library kit. Specifically, mRNAs were enriched twice via poly-T based RNA purification beads, and fragmentated at 94 °C for 8 min. The first strand cDNA was synthesized by Superscript II and random primers at 42 °C for 15 min, followed by second strand synthesis at 16 °C for 1 h. Adapters with illumine P5, P7 sequences as well as indices were ligated to the cDNA fragment at 30 °C for 10 min. After Ampure bead (BD) purification, a 15-cycle of PCR reaction was used to enrich the fragments. Sample libraries were subsequently pooled and loaded to the Illumina Hiseq2500. Raw sequence reads were aligned to the mouse genome (mm10) using the Tophat algorithm⁶². Cufflinks algorithm⁶³ was implemented to assemble transcripts and estimate their abundance. Cuffdiff⁶⁴ was used to statistically assess expression changes in quantified genes in different conditions. Genes with false discovery rate <0.2 and fold change >2 were considered significant.

YFV, DENV and ZIKV infection and antiviral assays

THF cells were treated with indicated concentrations of STING agonists for 8 h followed by infection with DENV, YFV or ZIKV at a multiplicity of infection (MOI) of 0.1 for 1 h. Two days post infection, the viral RNAs were quantified by qRT-PCR assays using LightCycler 480II (Roche). β-actin mRNA was quantified to normalize the levels of viral RNA. Primers used for amplification of viral RNA were reported previously^{60, 65}.

Flowcytometry analysis of PBMC-derived dendritic cells

PBMC-derived dendritic cells were washed twice with PBS containing 2% FBS and 1 mM EDTA, and stained using anti-CD11c FITC and anti-CD80 PE antibodies (eBioscience) for 30 min at room temperature. Data were acquired by GUAVA (Millipore) and analyzed with FlowJo.

In vitro adsorption, distribution, metabolism and elimination (ADME) profiling of BNBC and its analogs

In vitro ADME profiling studies were performed by Pharmaron. Aqueous solubility in PBS at pH 4.0 and 7.4 was determined at up to 300 μM. Metabolic stability in human and mouse liver microsomes was determined at 0, 15, 30, 45 and 60 min post incubation and expressed as time of 50% reduction compared to 0 min (T1/2). Permeability in human epithelial colorectal adenocarcinoma cells Caco-2 was assessed and expressed as transportation rate in both directions (apical to basolateral (A-B) and basolateral to apical (B-A)) across the cell monolayer, as well as efflux ratio, an indicator of a compound’s active efflux. Plasma protein binding (PPB) in samples of human or mouse origins was determined using equilibrium dialysis method and expressed as percentage of binding. Inhibition of each of the 10 cytochrome P450 (CYP) isozymes (CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1 and CYP3A4) was determined as greater than 50% inhibition at 10 μM concentration.

Statistics

P values of unpaired t test were calculated using GraphPad QuickCalcs.

Abbreviations

BNBC

6-bromo-N-(naphthalen-1-yl)benzo[d][1,3]dioxole-5-carboxamide

cGAS

cyclic guanosine monophosphate-adenosine monophosphate synthase

DMXAA

5,6-Dimethylxanthenone-4-acetic acid

HBV

hepatitis B virus

HFF

human foreskin fibroblast

IFN

interferon

LPS

Lipopolysaccharides

mDCs

Myeloid dendritic cells

PAMPs

pathogen-associated molecular patterns

PBMCs

peripheral blood mononuclear cells

PRRs

pattern recognition receptors

STING

stimulator of interferon genes

TLR

toll-like receptor

TNF-α

tumor necrosis factor-alpha