Genistein Targets STING-Driven Antiviral Responses

ABSTRACT Cytoplasmic detection of DNA by cyclic GMP-AMP (cGAMP) synthase (cGAS) is an essential component of antiviral responses. Upon synthesis, cGAMP binds to the stimulator of interferon (IFN) genes (STING) in infected and adjacent cells through intercellular transfer by connexins forming gap-junctions, eliciting a strong IFN-β-driven antiviral response. We demonstrate here that Genistein, a flavonoid compound naturally occurring in soy-based foods, inhibits cGAS-STING antiviral signaling at two levels. First, Genistein pretreatment of cGAMP-producing cells inhibited gap-junction intercellular communication, resulting in reduced STING responses in adjacent cells. In addition, Genistein directly blocked STING activation by the murine agonist DMXAA, by decreasing the interaction of STING with TBK1 and IKKε. As a result, Genistein attenuated STING signaling in human and mouse cells, dampening antiviral activity against Semliki Forest Virus infection. Collectively, our findings identify a previously unrecognized proviral activity of Genistein mediated via its inhibitory effects at two levels of cGAS-STING signaling.

IMPORTANCE Several reports suggest that Genistein exhibits antiviral activities against DNA viruses. Our work uncovers a previously unrecognized proviral effect of Genistein, through inhibition of the cGAS-STING pathway at the level of cGAMP transfer and its sensing by STING. This suggests that the use of Genistein as an antiviral should be taken with caution as it may reduce the protective antiviral effects elicited by host STING activation. KEYWORDS Genistein, STING inhibitor, cGAMP, gap junction F ollowing synthesis by cGAS, cGAMP binding to STING leads to its oligomerisation and interaction with TBK1 and IKK« kinases (1)(2)(3). This results in the phosphorylation of STING and IRF3, promoting a strong antiviral response through IFN-b production (2)(3)(4). In addition to engaging with STING in infected cells, cGAMP rapidly propagates antiviral responses to uninfected cells through direct extracellular secretion (5), via gap junction intercellular communication (GJIC) (6,7) or incorporation in viral particles (8).
Naturally occurring small molecules such as flavonoids and flavonoid-like compounds (e.g., Resveratrol) can modulate GJIC (9). Here we initially posited that pharmacological enhancement of GJIC could be used to increase cGAMP-mediated transactivation of adjacent cells and raise antiviral activity. We initially tested a small panel of flavonoid compounds (Genistein, Apigenin, Quercetin, Resveratrol, Kaempferol, and Epigallocatechin gallate [EGCG]) in a coculture model of GJIC. STING-deficient cGAS-overexpressing human cells (HEK-cGAS low herein), which make constitutive levels of cGAMP, were cultured together with STING competent mouse L929 cells stably expressing an IFN-sensitive responsive element [ISRE] luciferase reporter, referred to as LL171 herein (6,10). Pretreatment of the HEK-cGAS low cells with the flavonoid compounds potentiated ISRE-luciferase expression, except for Genistein which had an inhibitory effect (Fig. 1A). Conversely, Genistein direct treatment of LL171 cells rather induced ISRE-luciferase expression (Fig. 1B). Although not exhibiting any visible toxicity on the HEK-cGAS low cells at the doses used, Genistein treatment reduced cell proliferation above 30 mM (Fig. S1A). To circumvent this, we normalized the number of cGAMP-producing cells after Genistein treatment and confirmed a dose-dependent inhibition of cGAMP transactivation of cocultured LL171 cells by Genistein (Fig. 1C). Similarly, pretreatment of connexin (CX) expressing human MG-63 cells with Genistein (11), that were subsequently transfected with an immunostimulatory DNA (ISD) to activate cGAS (12), significantly decreased ISRE-luciferase expression from cocultured LL171 cells (Fig. 1D). It is noteworthy that intracellular levels of cGAMP were increased by Genistein in HEK cGAS low cells (Fig. 1E), indicating an inhibition of GJIC aligned with the accumulating cGAMP levels observed. Accordingly, GJIC-dependent transfer of the small molecular weight dye Calcein was decreased when the Calcein-loaded HEK donor cells had previously been treated with Genistein (Fig. 1F). Genistein inhibition of Calcein transfer was seen in HEK-STING cells naturally expressing connexin (CX) 43 and 45 (HEK-STING herein) (6) and was comparable to that obtained with carbenoxolone (CBX), a known inhibitor   1F). In agreement with this concept, Genistein treatment of SV40T immortalized mouse embryonic fibroblasts (MEFs) significantly decreased CX43 levels by ;60%, as measured by Western blotting (CX43 was undetectable in HEK cells with this method) ( Fig. 1G and Fig. S1B). Gap junctions are formed by two connexin-forming hemichannels, with one contributed by each adjacent cell to ensure intercellular communication of cGAMP (6,10). Surprisingly, Genistein pretreatment of the MEFs entirely ablated the transactivation from the cocultured HEK-cGAS low cells, measured by the secretion of mouse IP-10 (Fig. 1H). This strong inhibitory effect was unexpected given that pretreatment of donor cGAMP-producing HEK-cGAS low cells only partially decreased transactivation of the recipient cells (Fig. 1C). Critically, pretreatment of the MEFs or LL171 cells significantly blunted IP-10 production or ISRE-luciferase expression upon ISD transfection ( Fig. 1I and J) suggesting a direct inhibition of STING sensing independent of CX inhibition.
In agreement with this, we observed a dose-dependent inhibitory activity of Genistein on ISRE-Luciferase in LL171 cells stimulated with the murine STING agonist DMXAA ( Fig. 2A). To better define how Genistein impacts STING downstream signaling upon activation, we assessed its activity on NF-k B-Luciferase and IFN-b-Luciferase levels in HEK-STING cells that make endogenous cGAMP upon overexpression of cGAS-GFP (7,13). Genistein inhibited both branches of STING signaling equivalently (Fig. 2B), indicating that its effect was mediated upstream of IRF3 and NF-k B. Western blot analysis of Genistein pretreated immortalized mouse bone marrow-derived macrophages (iBMDMs) stimulated with DMXAA showed a 40 to 50% decrease of STING phosphorylation, and, to a lesser degree, IKK« and TBK1, which resulted in decreased IRF3 (most visible at 1 h) and p65 phosphorylation (most visible at 2 h) ( Fig. 2C and Fig. S1C). This was concurrent with a significant decrease of TNF-a production by the iBMDMs (Fig. 2D). Critically, Genistein pretreatment also strongly reduced IP-10 production by human monocytic THP-1 cells stimulated with a cGAMP analog (ADU-S100) or a synthetic STING agonist (diABZI) (14) (Fig. 2E)-the latter observation being replicated in MG-63 cells (Fig. 2F). Genistein also reduced IFN-b-Luciferase driven by overexpression of a gain of function STING variant R284S (1, 15) (Fig. 2G).
Mechanistically, Genistein's inhibitory activity is not targeted at the level of STING trafficking as evidenced by the unaltered formation of STING foci in HEK-STING cells following stimulation with diABZI (Fig. S2A). Similarly, Genistein did not decrease IFN-b-Luciferase driven by overexpression of human TBK1 or IKK« (Fig. 2H). Nonetheless, Genistein decreased the interaction of mCitrine tagged STING with phospho-TBK1 and IKK« following stimulation of iBMDMs with DMXAA and pull-down of mCitrine-STING (Fig. 2I). This suggests that Genistein limits STING signaling through the modulation of the STING/TBK1/IKK« interactions, presumably through its interaction with STING since it did not affect signaling driven by overexpression of human TBK1 or IKK« . Accordingly, in silico molecular docking of Genistein with STING identified 3 locations where Genistein was predicted to bind. We obtained 9 binding poses of interaction, and that with the lowest binding energy score, located in proximity to the STING C-terminal tail required for TBK1 recruitment (16), is shown in Fig. 2J and in Fig. S3. This illustrates the potential for a direct interaction of Genistein with STING. Albeit warranting further validation, these results collectively suggest that Genistein impacts STING signaling by direct targeting and preventing its interaction with TBK1/IKK« . It is also noteworthy that beyond its effect on STING signaling, Genistein also decreased IFN-b-Luciferase upon TLR3 activation, presumably due to its inhibitory effect on NF-k B activity (17) (Fig. S2B).
Finally, to determine whether the inhibitory effect of Genistein on STING signaling sensitized cells to viral infections, we tested whether it was able to block the antiviral activity of the STING agonist DMXXA. While DMXAA priming of LL171 cells promoted a strong antiviral effect against infection by the RNA Semliki Forest Virus (SFV), pretreatment with Genistein prior to DMXAA reduced this antiviral effect (Fig. 2K and Fig. S4A, S4B). This proviral effect of Genistein was comparable to that obtained with anti-IFNAR1 antibody treatment. Accordingly, the combination of Genistein with the anti-IFNAR1 antibody did not result in a robust increase of viral replication (Fig. S4B). This supports the predominant proviral effects of Genistein seen in this system as relying on type-I IFN inhibition. However, we note that Genistein modestly increased the proviral effect of the anti-IFNAR1 antibody, presumably due to its inhibitory impact on NF-k B activity (17) (Fig. S4B). Our findings collectively establish the proviral activity of Genistein through its inhibition the cGAS-cGAMP-STING pathway at two levels. On one hand, Genistein can block the transfer of cGAMP to adjacent cells, through the reduction of GJIC. This diminishes the amplification of antiviral responses mediated by cGAMP intercellular transfer (6,7). Mechanistically, our data in MEFs suggest that this effect is driven by a decrease in CX43 mediated by Genistein. However, cGAMP can also be transferred between cells through hemichannels formed by CX26, CX31, CX32, CX40 and CX62 (6). Therefore, it remains to be determined whether hemichannels beyond those involving CX43 are also targeted by Genistein. On the other hand, Genistein has been shown to directly inhibit STING activation by its agonists (18), which we posit results from a direct interaction of Genistein with STING, thereby impeding its interaction with TBK1 and IKK« . However, further studies will be required to ascertain whether this activity of Genistein also modulates STING-driven autophagy and apoptosis. We also showed that Genistein has a small effect on TLR3-driven antiviral responses suggesting it may have broad immunosuppressive effects on IFN-b production beyond STING. Importantly, Genistein has been shown to have protective effects in several animal models of diseases that have now been linked to STING signaling (19), including acute pancreatitis (20), DMBA-induced cancer (21), nonalcoholic fatty liver and alcoholic liver diseases (22,23).
Several studies have previously reported an antiviral effect of Genistein against DNA viruses such as HSV-1, cytomegalovirus, or Herpes B Virus, leading to the proposition that its topical administration could be a promising antiviral treatment against these viruses (24). Our findings suggest that its inhibitory effects on type-I IFN responses could confound other antiviral activities in the context of DNA viruses that naturally engage this pathway, prompting caution in its therapeutic development as an antiviral.
Data availability. Additional numerical data used to generate published graphs that support the findings of this study are available on request from the corresponding author.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, PDF file, 0.1 MB.