A fluorescent STING ligand sensor for high-throughput screening of compounds that can enhance tumor immunotherapy

Summary

The activation of the stimulator of interferon genes (STING) pathway triggers the release of type I interferons that can potentiate the host immune response against tumors. STING agonism is therefore a promising strategy for the development of cancer immunotherapy; however, sensitive tools and assays for the discovery of STING modulators are currently limited. Here, we develop and characterize a STING ligand sensor, FiSL, to detect STING ligands in vitro. Utilizing FiSL, we identify honokiol, a natural compound derived from Magnolia species, as an orally available STING agonist from a bioactive compound library. Functional studies reveal that honokiol exerts antitumor activity in a STING-dependent manner. Moreover, in STING-humanized mouse tumor models, honokiol enhances the efficacy of anti-PD-(L)1 immunotherapy. Collectively, we have developed FiSL as a tool for high-throughput screening of STING ligands and revealed honokiol as a STING agonist that can be harnessed to treat human cancer.

Graphical abstract

Highlights

• •We develop a fluorescent STING ligand sensor, FiSL
• •Screening with FiSL identifies honokiol as an orally available STING agonist
• •Honokiol enhances antitumor immunity dependent on STING
• •Honokiol enhances the efficacy of anti-PD-(L)1 in a humanized mouse model

Motivation

Owing to its pivotal role in activating innate immunity, stimulator of interferon genes (STING) agonism has the potential to improve the clinical efficacy of cancer immunotherapy. However, existing STING activity assays can be nonspecific and/or have limited quantitative sensitivity, posing challenges for high-throughput screening. To accelerate the discovery of STING agonists, we developed a fluorescent STING ligand sensor, referred to as FiSL, as a tool for high-throughput screening of STING ligands and identified honokiol as a STING agonist that can enhance antitumor immunity.

High-throughput screening of STING agonists remains challenging due to a lack of suitable tools. Sun et al. develop FiSL, a fluorescent STING ligand sensor, and discover honokiol as an orally available agonist for human and mouse STING. Honokiol demonstrates potent antitumor activity and enhances anti-PD-(L)1 efficacy.

Introduction

Stimulator of interferon genes (STING) is a transmembrane dimeric protein that is localized in the endoplasmic reticulum or Golgi apparatus.^1,2,3 STING is activated by the binding of its cytoplasmic ligand-binding domain (LBD) to cyclic dinucleotides (CDNs), for example, 2′3′-cyclic GMP-AMP (2′3'-cGAMP), which is produced by cyclic GMP-AMP synthase (cGAS) in mammalian cells.⁴ The binding of cGAMP to STING induces transformational changes in the LBD of STING and subsequently triggers activation of a downstream signaling cascade involving phosphorylation of TANK-binding kinase 1 (TBK1), nuclear translocation of transcription factor interferon regulatory factor 3 (IRF3), and production of type I interferons (IFN-Is) and other proinflammatory cytokines.^2,5

Cancer immunotherapy based on immune checkpoint blockage has achieved effective and durable responses in a subset of patients with different tumor types. However, most patients receiving immunotherapies do not obtain a clinical benefit because of the inadequate pre-existence of antitumor immunity within the tumor microenvironment (TME).⁶ Owing to its stimulatory property for antitumor immunity, STING agonism has emerged as a promising strategy being pursued for drug development.⁷

Although STING agonism is a high priority being pursued for drug development, the development of methods for high-throughput screening of STING agonists is limited. Several studies detect STING activation by largely relying on measuring the induction of IFN-I as an indirect readout in cells.^8,9,10 However, since IFN-I expression can be triggered by the activation of many pattern recognition receptors, these methods are not specific to detect STING activation. Additionally, the translocation of fluorescent-tagged STING to a perinuclear punctate compartment upon CDN binding is used as a reporter for STING activation.¹¹ While these tools are more direct to detect STING activation, they are limited to a qualitative and binary localization readout. Recent advances have shown that STING activation may be monitored by a genetically encoded fluorescent sensor, which binds and responds to CDNs¹²; however, this tool has rather small dynamic ranges of binder concentration or inappropriate affinity to STING agonists other than CDNs. Thus, the development of tools for high-throughput screening of STING agonists is warranted.

Plant-derived compounds have shown promising utility in the prevention and treatment of various human diseases; for example, honokiol, a natural compound derived from Magnolia species, is one such molecule that is used to effectively treat various types of cancer.¹³ Chemically, honokiol is a polyphenolic compound belonging to the lignan family.¹⁴ However, the mechanism underlying how honokiol exerts its antitumor activity remains unclear.

In this study, we developed a fluorescent STING ligand sensor, referred to as FiSL, based on the human STING LBD (hSTING-LBD) (residues 155–343). Using FiSL as a tool for readout, a potent and orally available agonist of STING, honokiol, was screened from thousands of compounds in a self-house bioactive compound library. Subsequent functional studies revealed that honokiol enhances antitumor immunity in a STING-dependent manner. Moreover, in the syngeneic tumor models constructed by STING-humanized mice, honokiol also exerts potent antitumor activity and markedly enhances the efficacy of anti-PD-(L)1 immunotherapy. In sum, we have developed FiSL as a tool for high-throughput screening of STING ligands and revealed honokiol as an orally available STING agonist that can be harnessed to treat human cancer.

Results

Design and development of STING ligand sensor

Previous studies have shown that the binding of ligands, for example, 2′3′-cGAMP, induces significant conformational changes in STING,² which is, therefore, a promising candidate for sensor design (Figure 1A). Structurally, STING is a transmembrane adaptor protein containing a dimerization region and a 2′3′-cGAMP binding domain at the C terminus.² Our previous study has shown that circularly permuted green fluorescent protein (cpGFP) is highly sensitive to conformational changes in the LBD of the itaconate sensor BioITA.¹⁵ Based on this knowledge, we hypothesized that chimera proteins with a fusion of cpGFP to the hSTING-LBD would provide a platform for the development of cpGFP-based sensors for STING ligands. To this end, we first designed 80 chimeric proteins by connecting cpGFP to the N terminus or C terminus of the hSTING-LBD or inserting cpGFP between amino acid residues located on the random coiled loops of the hSTING-LBD (Figure S1A) and assayed for their fluorescence response to 2′3′-cGAMP (Figure S1B), a potent agonist of STING. Among them, a chimera, referred to as N155, with cpGFP connected to the V155 of hSTING showed a moderate increase in the fluorescence emission when excited at 488 nm upon 2′3′-cGAMP addition (Figure S1B). We then generated a series of variants of this chimera with different linkers inserted between cpGFP and V155 of hSTING (Figure S1C) and found that one variant with a linker consisting of 13 glycine residues displayed a 2.321-fold increase in the fluorescence emission upon 2′3′-cGAMP addition (Figure S1D). This chimera, termed FiSL (Figure 1B), was used for further characterization. In addition, a non-responsive control sensor, designated dFiSL (dead FiSL), was engineered by incorporating the 2′3′-cGAMP binding-deficient mutations of Y240S and T263A (Figure S1E),¹⁶ as validated by the bio-layer interferometry (BLI) assay (Figure S1F). As expected, fluorescence spectra analysis showed that FiSL, but not dFiSL, responded to 2′3′-cGAMP (Figure 1C).

Figure 1.

Design and development of STING ligand sensor

(A) Cartoon model of 2′3′-cGAMP-induced conformational changes of STING.

(B) Schematic drawing of FiSL is shown. Fluorescent protein cpGFP was fused to the N terminus of the dimerized STING-LBD with a linker. The binding of ligands, for example, 2′3'-cGAMP, induces changes in protein conformation and fluorescence. cpGFP, circularly permuted green fluorescent protein; hSTING-LBD, human STING ligand-binding domain (residues 155–343).

(C) Fluorescence spectra of purified FiSL (left) and dFiSL (right) in the presence of 0 (black) or 200 (red) μM 2′3′-cGAMP. The excitation (Ex; dashed lines) spectrum recorded at an emission (Em) wavelength of 520 nm has a maximum at 488 nm; the Em (solid lines) spectrum recorded at an Ex wavelength of 488 nm has a maximum at 520 nm. Triplicate experiments were performed independently, with similar results.

See also Figure S1.

Characterizing STING ligand sensor

Next, to characterize the properties of FiSL in detail, FiSL and the control sensor dFiSL expressed in E. coli were purified by affinity chromatography (Figure S2A) and gel filtration (Figure S2B). Multiangle light scattering (MALS) analysis showed that FiSL and dFiSL appear as homodimers in solution (Figure S2C). Additionally, we found that 2′3′-cGAMP increased the fluorescence (excitation at 488 nm) of FiSL in a dose-dependent manner (Figure 2A), and the apparent dissociation constant (K_D) of 2′3′-cGAMP toward FiSL is 9.6 ± 1.2 μM (Figure 2B), which is within the range of intracellular 2′3′-cGAMP concentrations reported by previous publications.^12,17 Moreover, 2′3′-cGAMP binding did not affect the fluorescence (excitation at 405 nm) of FiSL and dFiSL (Figure S2D), suggesting that fluorescence from excitation at 405 nm can serve as a loading control for protein levels of FiSL and dFiSL.

Figure 2.

Characterizing FiSL sensor

(A) Fluorescence emission scans (excitation at 488 nm) from FiSL at indicated 2′3′-cGAMP concentrations. The scans using buffer serve as a control. Triplicate experiments were performed independently, with similar results.

(B) Maxima values (F_t) from emission peaks (excitation at 488 nm) of FiSL and dFiSL versus indicated 2′3′-cGAMP concentrations were plotted after normalization to F₀ (without 2′3′-cGAMP). The dissociation constant (K_D) of 2′3′-cGAMP toward FiSL was determined as 9.6 ± 1.2 μM. The data indicated at different 2′3′-cGAMP concentrations are shown as the mean ± SD of three independent experiments.

(C) Flow chart of fluorescence emission scan from FiSL. Fluorescence emission scans were performed before and after the addition of 100 μM 2′3′-cGAMP at a dilution ratio of 1:100, and then the repeated fluorescence emission scans were performed following the 2′3′-cGAMP elution from FiSL by 10 kDa ultrafiltration spin columns. Each time point of fluorescence emission scanning is indicated (top). Fluorescence spectra of FiSL (dashed lines) and dFiSL (solid lines) are shown (bottom). M1–4, measurement 1–4. Triplicate experiments were performed independently, with similar results.

(D) Fluorescence emission scan (excitation at 488 nm) was performed from FiSL exposed to an equal concentration (100 μM) of indicated STING agonists.

(E) Maxima values (F_t) from emission peaks (excitation at 488 nm) of FiSL in the presence of indicated STING agonists were shown after normalization to F₀ (maxima value from emission peak of FiSL only). Data are shown as the mean ± SD of three independent experiments. A.U., arbitrary unit.

See also Figure S2.

We then determine the binding manner between 2′3′-cGAMP and FiSL: fluorescence scanning from FiSL prior to and post 2′3′-cGAMP addition was repeated following 2′3′-cGAMP elution. As expected, following 2′3′-cGAMP elution, an obvious elevation of fluorescence emission was observed from FiSL, but not the control sensor dFiSL, upon repeated addition of 2′3′-cGAMP (Figure 2C), suggesting that 2′3′-cGAMP binds to FiSL in a reversible manner. Furthermore, to test the selectivity of FiSL, we performed fluorescence scanning from FiSL in the presence of other STING agonists. Although FiSL did not respond to DMXAA (Figures 2D and 2E), a mouse-selective STING agonist,¹⁸ a marked elevation of fluorescence emitted from FiSL was observed when FiSL was exposed to an equal concentration (100 μM) of 2′3′-cGAMP and other STING agonists¹⁹ (Figures 2D and 2E). These data suggest that FiSL is capable of detecting a variety of different STING agonists.

It has been known that the performance of sensors may be affected by pH and temperature. Fluorescence scanning from FiSL under different pH values revealed that FiSL and dFiSL exhibit similar responses to pH (Figure S2E), and no alteration of 2′3′-cGAMP-dependent fluorescence responses from FiSL was observed under pH between 6.5 and 8.0 (Figures S2F and S2G). Similarly, FiSL and dFiSL exhibit similar responses to temperatures (Figure S2H), and no alteration of 2′3′-cGAMP-dependent fluorescence responses from FiSL was observed under temperatures between 22°C and 37°C (Figures S2I and S2J). Taken together, these data demonstrate that FiSL can detect the presence of STING agonists under conditions with different pH values and temperatures, which makes it a promising tool for screening STING agonists in vitro.

Discovery of honokiol as a potent agonist of hSTING and mouse STING

Following the characterization of the STING ligand sensor, we screened a self-house bioactive compound library (3,148 compounds at 100 μM) using FiSL and found that honokiol was the top hit besides 2′3′-cGAMP, which serves as a positive control (Figures 3A and 3B; Table S1). We then investigated whether honokiol directly binds to the STING-LBD. Indeed, honokiol binds to the purified hSTING-LBD and the mouse STING-LBD (mSTING-LBD) (Figure S2K) with K_D values of 12.9 ± 0.4 and 17.0 ± 0.5 μM, respectively, as revealed by BLI assay (Figures 3C and 3D). Consistently, the binding between the hSTING-LBD and honokiol was observed when hSTING-LBD crystals were incubated with honokiol (Figure 3E; Table S2). In addition, structural mapping indicated that residues Y163, Y167, and T263 of the hSTING-LBD interact with honokiol (Figure 3E). We then mutated these honokiol-interacted residues to alanine (A) and expressed these hSTING mutants in 293T cells (Figure S3A). As expected, honokiol-induced IFN-β mRNA expression was observed in the cells expressing wild-type (WT) hSTING but not hSTING mutants of Y163A, Y167A, and T263A (Figure S3A). These results suggest that honokiol is a bona fide STING agonist and that the residues Y163, Y167, and T263 of hSTING are required for honokiol-induced STING activation.

Figure 3.

Discovery of honokiol as a potent agonist of human and mouse STING

(A) Maxima values (F_t) from emission peaks (monitored at 520 nm after excitation at 488 nm) of FiSL were measured in the presence of 100 μM compounds and shown after normalization to F₀ (maxima values from emission peaks of FiSL in the absence of compound). A total of 3,148 compounds at a concentration of 100 μM were screened. Among them, honokiol was indicated as one of the top hits.

(B) Chemical structures of honokiol.

(C and D) The interactions of honokiol versus hSTING-LBD (C, left) or mSTING-LBD (D, left) were determined by BLI assay. Dissociation constant (K_D) of honokiol toward hSTING-LBD (C, right) or mSTING-LBD (D, right) is shown. hSTING-LBD, human STING ligand-binding domain; mSTING-LBD, mouse STING ligand-binding domain.

(E) Overall structure of honokiol-bound hSTING-LBD (PDB: 8Z37) is shown. The lines circumscribing the zone of honokiol binding pocket and the key amino acid residues that interact with honokiol are shown.

(F) THP-1 and RAW264.7 cells were stimulated with honokiol (0, 10, and 30 μM) for 2 h. Immunofluorescent analyses were performed with an anti-STING antibody, and representative images are shown (top). DAPI was used to stain the nuclei. Scale bar, 20 μm. The percentage of cells (n = 100) with STING perinuclear aggregation was quantitated, and the data are presented as the mean ± SD of three independent experiments (bottom).

(G) Wild-type (WT) or STING/Sting-knockout (KO) THP-1 and RAW264.7 cells were stimulated with indicated concentrations (0, 10, and 30 μM) of honokiol for 2 h. The phosphorylation levels of STING, TBK1, and IRF-3 were detected by immunoblotting. The experiments were repeated three times independently, with similar results.

(H) WT or STING/Sting-KO THP-1 and RAW264.7 cells were stimulated with indicated concentrations (0, 10, and 30 μM) of honokiol for 6 h. The relative mRNA level of indicated cytokines (ISG15/Isg15, CXCL-10/Cxcl-10, and IFN-β/Ifn-β) were determined by quantitative PCR (qPCR). The data were normalized to ACTB/Actb mRNA levels and presented as the mean ± SD of three independent experiments.

See also Figure S3 and Tables S1 and S2.

Next, to determine the effect of honokiol on STING activation in macrophages, we treated the THP-1 and RAW264.7 cells, which are human and murine macrophages, respectively, with honokiol. As expected, honokiol induced perinuclear aggregation of STING in THP-1 and RAW264.7 cells (Figure 3F). Moreover, honokiol elevated the phosphorylation levels of STING, TBK1, and IRF3 (Figure 3G) and induced IRF3 target gene expression (Figure 3H) in WT but not STING/Sting-knockout (KO), THP-1, and RAW264.7 cells. Notably, honokiol also elevated phosphorylation levels of STING, TBK1, and IRF3 in CGAS/Cgas-KO THP-1 and RAW264.7 cells (Figure S3B). In addition, isothermal titration calorimetry (ITC) analysis demonstrated that the purified hSTING-LBD lost its binding to 2′3′-cGAMP in the presence of honokiol (Figure S3C), suggesting that honokiol competes the binding of 2′3′-cGAMP to STING. In summary, these data demonstrate that honokiol acts as a STING agonist to activate innate immunity in human and mouse macrophages.

To test the impact of honokiol on 2′3′-cGAMP production upon cGAS activation, we pretreated THP-1 and RAW264.7 cells with honokiol and then transfected these cells with poly(deoxyadenylic-deoxythymidylic) acid (poly(dA:dT)), synthetic DNA molecules that trigger cGAS activation.²⁰ Liquid chromatography-mass spectrometry (LC-MS) analysis demonstrated that honokiol pretreatment did not alter the intracellular 2′3′-cGAMP levels in THP-1 and RAW264.7 cells transfected with poly(dA:dT) (Figure S3D). These results suggest that honokiol does not affect cGAS-mediated 2′3′-cGAMP production in macrophages.

Furthermore, to compare the potency of honokiol and 2′3′-cGAMP for STING activation, we treated THP-1 and RAW264.7 cells with equal concentrations of honokiol or 2′3′-cGAMP and found that honokiol induced lower expression levels of IFN-β/Ifn-β mRNA in these cells compared to 2′3′-cGAMP (Figure S3E). These data support that honokiol is a STING activator with weaker potency than 2′3′-cGAMP.

Honokiol enhances antitumor immunity in a STING-dependent manner

It has been reported that pharmacological activation of STING is a promising immunotherapeutic strategy for cancer.²¹ We next evaluated the in vivo antitumor activity of honokiol. Due to its lipophilic property, which facilitates quick absorption across the gastrointestinal tract,²² honokiol was administered by the oral route in murine syngeneic tumor models. Pharmacokinetic studies demonstrated that honokiol peak exposure was achieved in plasma (Figure S4A) and tumor (Figure S4B) at ⁓1 and ⁓2 h, respectively, post-PO (per os) administration. Then, to test the effect of honokiol on antitumor immunity, B16F10 melanoma cells and MC38 colon cancer cells were subcutaneously inoculated into Sting^+/+ or Sting^−/− mice to generate syngeneic tumor models, followed by PO administration of honokiol (25 or 50 mg kg⁻¹) or vehicle. Sting^−/− mice were validated by genotyping (Figures S4C and S4D) and survived to adulthood without manifesting detectable physical abnormality. Marked inhibition of tumor growth (Figure 4A) and prolonged survival (Figure 4B) were observed in B16F10 and MC38 tumor-bearing Sting^+/+ mice treated with honokiol. Flow cytometry analysis demonstrated that tumor tissues derived from Sting^+/+ mice treated with honokiol possessed more CD8⁺ T cells (Figures 4C and S5A) and exhibited higher proportions of CD8⁺ T cells expressing granzyme B (GZMB) or IFN-γ (Figures 4D and S5B). Consistently, more infiltrated CD8⁺ T cells were observed in the tumor tissues derived from Sting^+/+ mice treated with honokiol, as visualized by immunofluorescent staining (Figure 4E). However, the abovementioned effects resulting from honokiol treatment were not observed in Sting^−/− mice (Figures 4A–4E, S5A, and S5B), although Sting^−/− mice displayed accelerated tumor growth (Figure 4A), poor survival (Figure 4B), less tumor-infiltrated CD8⁺ T cells (Figures 4C and S5A), lower proportions of CD8⁺ T cells expressing GZMB or IFN-γ (Figures 4D and S5B), and less CD8⁺ T cells in the tumor tissues (Figure 4E) compared to the Sting^+/+ mice. These results suggest that honokiol enhances antitumor immunity in a STING-dependent manner.

Figure 4.

Honokiol enhances antitumor immunity in a STING-dependent manner

B16F10 or MC38 cells (1 × 10⁵ cells/mouse) were subcutaneously injected into Sting^+/+ or Sting^−/− C57BL/6J mice (A–D). Seven days after tumor cell inoculation (tumor volume = ∼100 mm³), honokiol (25 or 50 mg kg⁻¹) or vehicle (PBS) was delivered to the mice via PO. Mice (n = 4) were killed and tumor tissues were harvested for analyses on day 19 post-tumor cell inoculation (C–E). Triangles below the horizontal axis indicate dosing days. Data represent the mean ± SEM (C and D and E, right). p values were determined by the one-way ANOVA (A, C, and D and E, right) or two-tailed log rank test (B). ∗p < 0.05 and #p < 0.01; n.s., not significant.

(A) Tumor volume was measured at indicated time points following honokiol treatment (n = 8).

(B) Survival of the mice (n = 8) was monitored.

(C) CD8⁺ T cell infiltration into B16F10 or MC38 tumors was quantified by flow cytometry.

(D) Proportion of tumor-infiltrated CD8⁺ T cells expressing GZMB or IFN-γ was quantified by flow cytometry.

(E) CD8⁺ T cells in the TME were visualized by immunofluorescence staining with CD8 antibody. Representative images are shown (left). DAPI was used to stain the nuclei. Scale bar, 50 μm. Relative CD8⁺ cell numbers were determined by means of the CD8⁺ cell number quantified for ten microscopic fields of each tumor sample.

See also Figures S4 and S5.

Honokiol enhances antitumor immunity in a STING-humanized mouse model

It has been known that the hSTING and mSTING signaling pathways are not completely identical.²³ To test the effect of honokiol on antitumor immunity in the context of the hSTING signaling pathway, B16F10 or MC38 cells were subcutaneously inoculated into STING-humanized mice followed by PO administration of honokiol (25 or 50 mg kg⁻¹) or vehicle. In line with the abovementioned results, honokiol inhibited tumor growth (Figure 5A) and prolonged survival (Figure 5B) of STING-humanized mice bearing B16F10 or MC38 tumors. Flow cytometry analysis revealed that tumor tissues derived from STING-humanized mice treated with honokiol possessed more CD8⁺ T cells (Figures 5C and S5C) and exhibited higher proportions of CD8⁺ T cells expressing GZMB or IFN-γ (Figures 5D and S5D). Moreover, immunofluorescent staining revealed that honokiol increased the number of CD8⁺ T cells infiltrating the tumor tissues (Figure 5E). Taken together, these data demonstrate that honokiol enhances antitumor immunity in syngeneic tumor models constructed with STING-humanized mice, supporting that honokiol could be further developed as a drug for immunotherapy of human cancer.

Figure 5.

Honokiol enhances antitumor immunity in a STING-humanized mouse model

B16F10 or MC38 cells (1 × 10⁵ cells/mouse) were subcutaneously injected into STING-humanized mice (A–D). Seven days after tumor cell inoculation (tumor volume = ∼100 mm³), honokiol (25 or 50 mg kg⁻¹) or vehicle (PBS) was delivered to the mice via PO. Mice (n = 4) were killed and tumor tissues were harvested for analyses on day 23 post-tumor cell inoculation (C–E). Triangles below the horizontal axis indicate dosing days. Data represent the mean ± SEM (C and D and E, right). p values were determined by the one-way ANOVA (A, C, and D and E, right) or two-tailed log rank test (B). ∗p < 0.05 and #p < 0.01.

(A) Tumor volume was measured at indicated time points following honokiol treatment (n = 8).

(B) Survival of the mice (n = 8) was monitored.

(C) CD8⁺ T cell infiltration into B16F10 or MC38 tumors was quantified by flow cytometry.

(D) Proportion of tumor-infiltrated CD8⁺ T cells expressing GZMB or IFN-γ was quantified by flow cytometry.

(E) CD8⁺ T cells in the TME were visualized by immunofluorescence staining with CD8 antibody. Representative images are shown (left). DAPI was used to stain the nuclei. Scale bar, 50 μm. Relative CD8⁺ cell numbers were determined by means of the CD8⁺ cell number quantified for ten microscopic fields of each tumor sample.

See also Figures S5 and S6.

We then evaluated the potential toxicity of honokiol. No effect on body weight was observed in B16F10 or MC38 tumor-bearing mice treated with honokiol (50 mg kg⁻¹) (Figure S6A). Moreover, no abnormality was detected in the routine blood tests (Figure S6B) and blood biochemical examinations (Figure S6C) of these mice. Enzyme-linked immunosorbent assay (ELISA) demonstrated that honokiol (50 mg kg⁻¹) markedly elevated the levels of IFN-β (Figure S6D), interleukin-6 (IL-6) (Figure S6E), and tumor necrosis factor alpha (TNF-α) (Figure S6F) in tumor and serum, with peak levels at ⁓2 h and a return to baseline within ∼24 h (Figures S6D–S6F). These data suggest that honokiol can be well tolerated by tumor-bearing mice. In addition, we found that depletion of CD8⁺ T cells greatly reduced the effect of honokiol on antitumor immunity in the MC38 syngeneic tumor model constructed with STING-humanized mice (Figures S6G and S6H), suggesting that CD8⁺ T cells exert an important role in honokiol-mediated antitumor immunity.

Honokiol enhances the efficacy of anti-PD-(L)1 cancer immunotherapy

Given that honokiol is capable of enhancing antitumor immunity, we next investigated the effect of honokiol treatment on cancer responses to T cell-based immunotherapy. To this end, B16F10 tumor-bearing STING-humanized mice were treated with anti-PD-L1 antibody without or with honokiol (50 mg kg⁻¹). We found that honokiol and anti-PD-L1 antibody inhibited the growth of B16F10 tumors in STING-humanized mice (Figure 6A) and extended the survival time of these mice (Figure 6B). Of note, the combination treatment of honokiol and anti-PD-L1 antibody achieved the most significant antitumor effect (Figure 6A) and the longest survival time of these mice (Figure 6B). In another syngeneic tumor model, similar tumor inhibition and prolonged survival time were observed when the STING-humanized mice bearing E0771 tumors were treated with honokiol and/or anti-PD-1 antibody (Figures 6C and 6D). Importantly, the most effective tumor inhibition and the longest survival time were achieved by the combination treatment of honokiol and anti-PD-1 antibody in this tumor model (Figures 6C and 6D). Taken together, these results suggest that honokiol is a small molecular compound that enhances the efficacy of T cell-based cancer immunotherapy.

Figure 6.

Honokiol enhances the efficacy of anti-PD-(L)1 immunotherapy

B16F10 (A and B) or E0771 (C and D) cells (1 × 10⁵ cells/mouse) were subcutaneously injected into STING-humanized mice. Tumor-bearing mice (n = 8) were treated with PD-(L)1 antibody (4 mg kg⁻¹) or immunoglobulin (Ig)G control without or with honokiol (50 mg kg⁻¹). Honokiol was dosed PO, and antibodies were dosed intraperitoneally (i.p.). Black and green triangles below the horizontal axis indicate dosing days for PD-(L)1 antibody and honokiol, respectively. p values were determined by the one-way ANOVA (A and C) or two-tailed log rank test (B and D). ∗p < 0.05 and #p < 0.01. (A and C) Tumor volume was measured at indicated time points following indicated treatments (n = 8). (B and D) Survival of the mice (n = 8) was monitored.

Discussion

In this study, we have developed and characterized FiSL as a genetically encoded fluorescent STING ligand sensor. FiSL has the capability to detect a variety of STING ligands, providing a useful tool for screening STING activators in vitro. Based on FiSL, an orally available, non-nucleotide-based STING agonist, honokiol, was screened from thousands of compounds in a self-house natural product library. In syngeneic tumor models, honokiol alone induced tumor regression by enhancing antitumor immunity in a STING-dependent manner. In mouse tumor models that are poorly responsive to anti-PD-(L)1 immunotherapy, the combination of honokiol and anti-PD-(L)1 antibody was superior to monotherapy in inhibiting tumor growth and prolonging survival. Moreover, honokiol is amenable to oral administration, a desirable delivery route with high convenience and low cost. These data strongly support the idea that pharmacological activation of STING by honokiol can increase the sensitivity of cancer immunotherapy via induction of host adaptive antitumor immunity; thus, honokiol may be further developed as an orally available drug for cancer immunotherapy.

It is noteworthy that honokiol, as an agonist for both hSTING and mSTING, enhances antitumor immunity in syngeneic tumor models constructed by STING-humanized mice. These data strongly support that honokiol is also capable of enhancing antitumor immunity in human cancer. Furthermore, our results demonstrate that honokiol treatment significantly increases the number of tumor-infiltrated CD8⁺ T cells and the positive rates of IFN-γ and GZMB among these cells. In addition, experiments with CD8⁺ T cell-depleted mice show that CD8⁺ T cell depletion dramatically impaired honokiol-mediated antitumor immunity. Taken together, these observations suggest that honokiol enhances antitumor immunity through activation of the cytotoxic effector function of CD8⁺ T cells in the TME.

Honokiol, a plant bioactive compound that mainly exists in Magnolia species,²⁴ is one of the key chemical components in some traditional Eastern herbal medicines.²⁵ Previous studies have shown a biosafety of honokiol up to 250 mg kg⁻¹.²⁶ Due to its pharmacological safety, honokiol alone or in combination with other chemotherapeutic drugs can prevent and treat cancer.^13,27 In line with the previous studies, we observed that honokiol exerts potent antitumor activity and enhances the efficacy of anti-PD-(L)1 immunotherapy in the syngeneic tumor models constructed by STING-humanized mice. In addition, our study revealed that honokiol is a bona fide agonist for both hSTING and mSTING and that Sting KO abrogated the antitumor activity of honokiol. Collectively, these data suggest that honokiol is a natural compound exerting antitumor activity depending on its ability to activate STING.

It has been reported that honokiol acts as an anti-inflammatory mediator in the progression of various diseases, such as renal fibrosis,²⁸ acute lung injury,²⁹ lupus nephritis,³⁰ and silicosis.³¹ Our study demonstrates that honokiol functions as a STING agonist that is capable of triggering IFN-I expression and enhancing antitumor immunity in syngeneic mouse tumor models. Based on these results, we propose that honokiol may play different immunomodulatory roles in the context of different pathogenesis conditions.

In conclusion, our work develops FiSL as a fluorescent sensor for high-throughput screening of STING ligands. Utilizing FiSL as a readout tool, honokiol was screened as a potent and orally available agonist for both hSTING and mSTING from thousands of compounds in a self-house bioactive compound library. In vivo experiments demonstrate that PO administration of honokiol is capable of enhancing antitumor immunity in a STING-dependent manner. Thus, honokiol could be further developed as an orally available drug for immunotherapy of human cancer.

Limitations of the study

Based on the STING ligand sensor FiSL, we identified honokiol as a potent STING agonist by screening a self-house bioactive compound library containing 3,148 compounds. Due to the limitation of the compound library pool, STING agonists with better antitumor activity and low toxicity could be missed in our current screening assay; thus, further study for the screening of STING agonists based on a larger pool of compound libraries is warranted. In addition, our studies demonstrate that honokiol exposure can be observed in plasma and tumor post-PO administration. However, to achieve the optimal antitumor efficacy and tolerability profile, detailed in vivo pharmacokinetic and pharmacodynamic properties of this compound remain to be investigated.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Xinjian Li (lixinjian@ibp.ac.cn).

Materials availability

All reagents, including the unique plasmids for the expression of FiSL, dFiSL, and h/mSTING-LBD, antibodies, cell lines, and mouse strains, can be found in the key resources table. Further information and requests for resources and reagents are available from the lead contact upon completion of appropriate material transfer agreements.

Data and code availability

• •The atomic coordinates and structure factors generated in this study have been deposited at RCSB Protein Data Bank (https://www.rcsb.org/structure/8Z37). Uncropped immunoblotting data and Excel datasheets, including values underlying each graph, have been deposited at Mendeley Data (https://data.mendeley.com/datasets/skv4hchsss/1).
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported by the Natural Science Foundation of Beijing, China (grant no. 5252016 to X.L.); the National Key R&D Program of China (grant no. 2024YFA1307400 to X.L.); the Scientific Research Fund for the Doctoral Scholars, Shanxi University of Chinese Medicine (grant no. 2024BK08 to P.S.); the Open Fund of Shanxi Key Laboratory of Innovative Drug for the Treatment of Serious Diseases Basing on the Chronic Inflammation, Shanxi University of Chinese Medicine (grant no. SXInFDL2024-005 to P.S.); and the Fund of the State Key Laboratory of Epigenetic Regulation and Intervention, CAS (grant no. O4CCSGZ3 to X.L.).

Author contributions

This study was conceived by X.L.; X.L. and P.S. designed the study; P.S., B.W., C.L., and Z.W. performed the experiments; Y.L. and Y.-B.Q. provided expertise in experimental methodology and data interpretation; and X.L. and P.S. wrote and edited the manuscript with comments from all authors.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-human p-Ser386-IRF3	Beyotime	Cat#AF1594; RRID: AB_2893174
Rabbit monoclonal anti-mouse p-Ser379-IRF3	Cell Signaling Technology	Cat#79945; RRID: AB_2799943
Rabbit monoclonal anti-STING	Cell Signaling Technology	Cat#13647; RRID: AB_2732796
Rabbit monoclonal anti-STING	Cell Signaling Technology	Cat#90947; RRID: N/A
Rabbit monoclonal anti-mouse p-Ser365-STING	Cell Signaling Technology	Cat#72971; RRID: AB_2799831
Rabbit monoclonal anti-human p-Ser366-STING	Cell Signaling Technology	Cat#19781; RRID: AB_2737062
Rabbit monoclonal anti-TBK1	Cell Signaling Technology	Cat#38066S; RRID: AB_2827657
Rabbit monoclonal anti-p-Ser172-TBK1	Cell Signaling Technology	Cat#5483S; RRID: AB_10693472
Rabbit monoclonal anti-IRF3	Cell Signaling Technology	Cat#4302; RRID: AB_1904036
Rabbit monoclonal anti-CD8α	Cell Signaling Technology	Cat#35467; RRID: AB_2799078
Mouse monoclonal anti-β-actin	Affinity Biosciences	Cat#T0022-HRP; RRID: AB_2839417
Rabbit monoclonal anti-mouse cGAS	Cell Signaling Technology	Cat#31659; RRID: AB_2799008
Rabbit monoclonal anti-human cGAS	Cell Signaling Technology	Cat#15102; RRID: AB_2732795
Pacific blue (PB) anti-mouse/human CD45	BioLegend	Cat#157212; RRID: AB_2876534
Phycoerythrin (PE)/Cyanine7 anti-mouse CD3	BioLegend	Cat#100219; RRID: AB_1732057
Allophycocyanin (APC) anti-mouse CD4	BioLegend	Cat#100411; RRID: AB_312696
FITC anti-mouse CD8	BioLegend	Cat#100705; RRID: AB_312744
PE anti-mouse granzyme B	BioLegend	Cat#396405; RRID: AB_2801074
APC anti-mouse IFN-γ	BioLegend	Cat#163514; RRID: AB_3097457
Anti-mouse CD8α	Selleck	Cat#A2102; RRID: AB_3099521
Anti-mouse PD-L1	Selleck	Cat#A2115; RRID: AB_3675704
Anti-mouse PD-1	Selleck	Cat#A2122; RRID: AB_3644244
Rat IgG2b isotype	Selleck	Cat#A2116; RRID: AB_3662740
Horseradish peroxidase (HRP)-conjugated goat anti-mouse	Thermo Fisher Scientific	Cat# G-21040; RRID: AB 2536527
Horseradish peroxidase (HRP)-conjugated goat anti-rabbit	Thermo Fisher Scientific	Cat# G-21234; RRID: AB 2536530
Alexa fluor 488-conjugated goat anti-rabbit	Invitrogen	Cat# A11034; RRID: AB 2536530
Bacterial and virus strains
BL21(DE3) pLysS	Sigma-Aldrich	Cat#C606010
Chemicals, peptides, and recombinant proteins
BamH I	New England Biolabs	Cat#R3136T
Xho I	New England Biolabs	Cat#R0146L
BsmB I	New England Biolabs	Cat#R0739S
NaCl	Sigma-Aldrich	Cat#S9888
Dithiothreitol (DTT)	Sigma-Aldrich	Cat#D9779
phenylmethanesulfonyl fluoride (PMSF)	Sigma-Aldrich	Cat#P7626
IPTG	Sigma-Aldrich	Cat#I6758
Methanol	Sigma-Aldrich	Cat#900641
Chloroform	Sigma-Aldrich	Cat#151823
Glycerol	Sigma-Aldrich	Cat#G5516
DMSO	Sigma-Aldrich	Cat#D5879
DAPI	Sigma-Aldrich	Cat#D8417
Leupeptin	Sigma-Aldrich	Cat#L9783
Bromophenol blue	Sigma-Aldrich	Cat#114391
PVDF	Sigma-Aldrich	Cat#IPVH00010
Bovine Serum Albumin	Sigma-Aldrich	Cat#B2064
NaOH	Sigma-Aldrich	Cat#S5881
Citrate	Sigma-Aldrich	Cat#V900095
EDTA	Sigma-Aldrich	Cat#324503
10-kDa ultrafilter column	Sigma-Aldrich	Cat#PLGC02510
0.45 μm filter	Sigma-Aldrich	Cat#HVLP02500
Skimmed milk	Sigma-Aldrich	Cat#1.15338.0500
β-mercaptoethanol	Santa Cruz	Cat#sc-202966
Puromycin	Santa Cruz	Cat#sc-108071
Polybrene	Santa Cruz	Cat#sc-134220
c-di-GMP	InvivoGen	Cat#tlrl-nacdg
3′3′-cGAMP	Abcam	Cat#ab144865
PMA	Abcam	Cat#ab120297
c-di-AMP	Acmec	Cat#X80635
2′3′-cGAMP	APExBIO	Cat#B8362
MSA-2	Med Chem Express	Cat#HY-136927
SR-717	Med Chem Express	Cat#HY-131454
ABZI	Med Chem Express	Cat#HY-123943
DMXAA	Med Chem Express	Cat#HY-10964
Honokiol	Med Chem Express	Cat#HY-N0003
MCE FDA-approved Drug Library-1724 cmpds	Med Chem Express	Cat#HY-L021
Anti-virus Compound Library-508 cmpds	Med Chem Express	Cat#HY-L001
MetaSci Human Metabolite Library-916 cmpds	MetaSci	Cat#HML 12.99
Collagenase/Dispase	Roche	Cat#11097113001
DNase I	Roche	Cat#10104159001
Corn oil	Beyotime	Cat#ST2308
SDS	Beyotime	Cat#ST2681
Kanamycin	Beyotime	Cat#ST101
Triton X-100	Beyotime	Cat#P0096
Coomassie blue	Beyotime	Cat#ST1119
Paraformaldehyde	Beyotime	Cat#P0099
7-AAD viability staining solution	Selleck	Cat#420404
ProLong Gold antifade reagent	Invitrogen	Cat#P36982
Trizol	Invitrogen	Cat#15596026
Lipofectamine 3000	Invitrogen	Cat#L3000015
Ni-NTA	Smart-Lifesciences	Cat#SA004100
DMEM	Gibco	Cat#C11965500BT
RPMI-1640	Gibco	Cat#C11875500BT
Fetal Bovine Serum	Gibco	Cat#A5669701
Penicillin Streptomycin	Gibco	Cat#15140-122
0.25% Trypsin-EDTA	Gibco	Cat#25200-072
HBSS	Gibco	Cat#14175095
PBS	Gibco	Cat#C10010500BT
Matrigel	BD Biosciences	Cat#356234
Mouse CD16/CD32	BD Biosciences	Cat#553141
Saponin-containing buffer	BD Biosciences	Cat#554722
Optimal cutting temperature (O.C.T.) compound	Tissue-Tek	Cat#4583
Poly (dA:dT)	InvivoGen	Cat#tlrl-patn
Critical commercial assays
HiLoad 16/600 Superdex 200 pg column	GE healthcare	Cat#GE28-9893-35
Superdex 200 Increase 10/300 GL column	GE healthcare	Cat#GE28-9909-44
Dual Luciferase Reporter Gene Assay Kit	Yeasen	Cat#11402ES60
ELISA kits for detecting IFN-β	Jiangsu Meibiao	Cat#MB-6317A
ELISA kits for detecting TNF-α	Jiangsu Meibiao	Cat#MB-2868A
ELISA kits for detecting IL-6	Jiangsu Meibiao	Cat#MB-2899A
Deposited data
Crystal structure of STING LBD in complex with honokiol	RCSB Protein DataBank	https://www.rcsb.org/structure/8Z37
Uncropped immunoblotting data and Excel datasheets for graphs	Mendeley Data	https://data.mendeley.com/datasets/skv4hchsss/1
Experimental models: Cell lines
B16F10 cells	Cell Bank/Stem Cell Bank, SIBCB	Cat#TCM36
MC38 cells	Sigma-Aldrich	Cat#SCC172
THP-1 cells	Cell Bank/Stem Cell Bank, SIBCB	Cat#SCSP-567
E0771 cells	ATCC	Cat#CRL-3461
RAW264.7 cells	ATCC	Cat#TIB-71
293T cells	ATCC	Cat#CRL-3216
Experimental models: Organisms/strains
Sting+/− C57BL/6J mice	GemPharmatech	Cat#T012747
STING-humanized C57BL/6J mice	GemPharmatech	Cat#T049781
Wild-type C57BL/6J mice	GemPharmatech	Cat#N000295
Oligonucleotides
PCR primers and DNA oligonucleotides	This Paper	Table S3
Recombinant DNA
pET28a vector	Sigma-Aldrich	Cat#69864
pET28a-FiSL	This paper	N/A
pET28a-dFiSL	This paper	N/A
pET28a-hSTING-LBD	This paper	N/A
pET28a-mSTING-LBD	This paper	N/A
LentiCRISPRv2 vector	Addgene	Cat#98290
LentiCRISPRv2-non-target gRNA	This paper	N/A
LentiCRISPRv2-STING gRNA	This paper	N/A
LentiCRISPRv2-Sting gRNA	This paper	N/A
pMD2.G plasmid	Addgene	Cat#12259
psPAX2 plasmid	Addgene	Cat#12260
Software and algorithms
GraphPad Prism (Version 8.0.2)	GraphPad Software	http://www.graphpad.com
Rstudio (Version 4.1.3)	Posit Software	https://posit.co/download/rstudio-desktop/
PHENIX (Version 1.18.2–3874)	Phenix Industrial Consortium	https://phenix-online.org/
Pymol (Version 1.3)	Schrödinger	http://www.pymol.org
Imaris (Version 9.0.1)	Oxford Instruments	http://www.imaris.com
FlowJo (Version 10.8.1)	Becton, Dickinson & Company	https://www.flowjo.com/
Other
Liquid chromatography/Quadrupole-TOF mass spectrometer	Agilent 6530	https://www.agilent.com.cn/
ABI Q7 Fast Real Time PCR System	Applied Biosystems	https://www.thermofisher.cn/
BD LSRFortessa	BD Biosciences	https://www.bdbiosciences.com

Experimental model and study participant details

Mouse models

Sting^+/− male mice were crossed with Sting^+/− female mice to generate Sting^−/− mice. Sibling mice (n = 3-8 for each group) with random sex and at the age between 6 to 8 weeks were used for generation of the syngeneic mouse models of melanoma, colon, and breast tumor. All mice (3-5 animals per cage) were bred and housed in a specific pathogen-free environment with a light/dark cycle of 12 h/12 h, temperature of 23 ± 2°C, relative humidity of 50 to 65 %, and ad libitum access to water and food. The use of animals in this study was approved by the Institutional Animal Care and Use Committee of the Institute of Biophysics, Chinese Academy of Sciences (Protocol # SYXK2019030).

To generate syngeneic mouse tumor models, B16F10 cells (1 × 10⁵), MC38 cells (1 × 10⁵), and E0771 cells (1 × 10⁵) resuspended in 50 μL DMEM supplemented with 50% Matrigel (BD Biosciences #356234) were subcutaneously injected into Sting^+/+, Sting^−/− or STING-humanized mice. Seven days after inoculation of tumor cells, mice bearing tumors with similar size were randomly grouped. Honokiol dissolved in 100 μL corn oil or vehicle was administered into tumor-bearing mice (25 or 50 mg kg⁻¹) by oral (PO) route on day 7, 11, and 15 post inoculation of tumor cells. For depletion of CD8⁺ T cells, STING-humanized mice were intraperitoneally (IP) injected with anti-CD8α antibody (30 mg kg⁻¹) or isotype control antibody (IgG2b) 3 days before inoculation of MC38 cells and continued to receive anti-CD8α antibody or IgG2b treatment (10 mg kg⁻¹) once weekly to maintain the depletion of CD8⁺ T cells. For PD-(L)1-based immunotherapy, PD-(L)1 antibodies (4 mg kg⁻¹) or IgG control were intraperitoneally injected into STING-humanized mice bearing tumors in combination without or with orogastric administration of honokiol (50 mg kg⁻¹). Survival status of each animal was recorded daily and the size of each tumor was measured every 2 or 3 days according to the formula volume = length × width² × 0.5. Experiments were terminated if the tumor volume reaches 2000 mm³.

Mouse genotyping

A small piece of toe tissue (∼10 mg) clipped from mouse was heated at 98°C for 30 min in 100 μL of 50 mM NaOH solution. The samples were neutralized with 10 μL 1 M Tris-HCl (pH [8.0]), and then centrifuged at 13, 200 × g for 10 min. The supernatants were collected and used as templates for PCR amplifications using the primer pairs spanning the genomic locus of Sting. The genotyping was performed following the instructions provided by GemPharmatech. For PCR amplicon 1 amplified by primer pair F1 + R1, PCR product with size of 447 bp represents the mouse genotype of Sting^−/−; For PCR amplicon 2 amplified by primer pair F2 + R2, PCR product with size of 493 bp represents the mouse genotype of Sting^+/+. The sequences of PCR primers used for mouse genotyping were listed in Table S3.

Serum and plasma preparation

For serum preparation, whole blood (⁓500 μL) from orbital veins of tumor-bearing mice was drawn onto blood collection tubes without addition of anticoagulant at different time points post honokiol administration. The whole blood samples were kept at 4°C for 30 min, then centrifuged at 9, 300 × g for 30 min at 4°C. Serum located at upper layer of the blood samples was collected for the downstream experiments. For plasma preparation, whole blood (⁓500 μL) from orbital veins of tumor-bearing mice was drawn onto blood collection tubes treated with anticoagulant (citrate) at different time points post honokiol administration. The samples were immediately centrifuged at 2, 000 × g for 15 min at 4°C, and then the resulting supernatants were collected as plasma.

Cell lines and cell culture conditions

THP-1 cells were maintained in 1640 medium, RAW264.7 cells, 293T cells, E0771 cells, MC38 cells and B16F10 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific #10091148) and 1% penicillin-streptomycin (Thermo Fisher Scientific #15140122) at 37°C with 5% CO₂. THP-1 cells were differentiated by incubation with PMA (100 ng mL⁻¹) for 48 h prior to treatment. All cell lines were authenticated by Short Tandem Repeat (STR) profiling and routinely tested for mycoplasma contamination.

Method details

Confocal microscopic analysis

For immunostaining in cultured cells, macrophages seeded on 8-well chamber slides at a density of 2 × 10⁴ per well were treated with different concentrations of honokiol. The cells were fixed with 4% paraformaldehyde for 10 min followed by permeabilization with 0.1% Triton X-100 for 5 min at room temperature. After blocking with 5% goat serum and 1% BSA for 30 min, the cells were incubated with antibodies recognizing STING (Cell Signaling Technology #90947) at a dilution of 1:200 overnight at 4°C, followed by incubation with an Alexa Fluor 488-conjugated secondary antibody for another 1 h at room temperature. The cells were counterstained with DAPI (1 μg mL⁻¹) for 2 min followed by mounting with ProLong Gold antifade reagent (Thermo Fisher Scientific #P36982).

For immunostaining in tissue samples, mouse tumor tissues embedded in O.C.T. compound (Tissue-Tek # 4583) were snap-frozen in dry ice, sectioned in 5-μm thickness in a cryostat (Leica CM1860 UV) at −20°C, transferred to room temperature microscope slides, and then immediately fixed in 4% paraformaldehyde for 10 min. After blocking with 5% goat serum and 1% BSA for 30 min, the slides were incubated with FITC-conjugate antibodies recognizing CD8α (Cell Signaling Technology #35467) at a dilution of 1:100 overnight at 4°C. The nuclei were counterstained with DAPI (1 μg mL⁻¹) for 2 min followed by mounting with ProLong Gold antifade reagent (Thermo Fisher Scientific #P36982).

Lastly, fluorescence images of the cells were acquired using a LSM700 inverted confocal microscope (Zeiss) equipped with a 40× objective and processed with ZEN Blue software (Version 2012) (Zeiss).

Flow cytometry

To analyze the tumor infiltrating lymphocytes (TILs) or splenocytes, tumor or spleen tissues (⁓200 mg) collected from tumor-bearing mice were minced into small pieces (⁓0.5 mm³) and digested in DMEM (10 mL g⁻¹) containing 10% FBS, 1 mg mL⁻¹ collagenase (Roche #11097113001) and 0.1 mg mL⁻¹ DNase I (Roche #10104159001) for 1 to 2 h(s) at 37°C with gentle agitation. The suspended cells were harvested by centrifugation at 350 × g for 5 min, and then digested with prewarmed 3–5 mL HBSS containing 0.25% trypsin and 0.1% EDTA for 5 min at 37°C. Trypsin digestion was terminated by addition 10 mL DMEM supplemented with 10% FBS. Cells were passed through a 70-μm cell strainer (BD Biosciences), centrifuged at 350 × g for 5 min, and then resuspended in FACS buffer (PBS containing 2% FBS and 2 mM EDTA). CD45⁺ TILs in single-cell suspension were enriched using CD45 microbeads (Miltenyi Biotec #130-097-153), incubated with antibodies against mouse CD16/CD32 (BD Biosciences #553141) for 10 min on ice to block Fc receptors, and then stained for 30 min with one or a combination of various fluorochrome-conjugated monoclonal antibodies. Following staining, cells were washed with FACS buffer and resuspended in 7-AAD viability staining solution (BioLegend #420404). Dead cells and doublets were excluded by forward and side scatter. For intracellular antigens, cells were fixed/permeabilized with saponin-containing buffer (BD Biosciences #554722) for 30 min at 4°C before staining. Stained samples were acquired on a BD LSRFortessa flow cytometer (BD Biosciences). Flow cytometry data were analyzed using FlowJo software (Version 10.8.1).

Immunoblotting analysis

THP-1 cells and RAW 264.7 cells grown in 60-mm dishes with a confluence of 50–80% were washed twice with PBS, and then lysed with 200 μL modified lysis buffer (50 mM Tris-HCl pH [7.5], 0.1% SDS, 1% Triton X-100, 150 mM NaCl, 1 mM dithiothreitol, 0.5 mM EDTA, 100 μM PMSF, 100 μM leupeptin). Total cell lysates were centrifuged at 13, 400 × g for 10 min at 4°C. The supernatants were harvested and total protein concentration was measured using a BCA protein assay kit (Thermo Fisher Scientific #23225). Equal amount of total protein was mixed with 5 × loading buffer (250 mM Tris-HCl pH [6.8], 8% SDS, 40% [v/v] glycerol, 100 mM dithiothreitol, 0.1% bromophenol blue) followed by heating at 98°C for 5 min. The heated samples were separated on 8%, 10%, or 12% polyacrylamide gels and transferred onto a PVDF membrane (Bio-Rad) by a wet transfer. The blotted membranes were blocked in 5% skimmed milk for 1 h at room temperature, incubated with primary antibodies overnight at 4°C, and then HRP-conjugated secondary antibodies for 1 h at room temperature. Immunoblots were visualized by a ChemiScope 6000 Exp instrument (CLinX, China) following incubation of the membranes with the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific).

RNA isolation and quantitative PCR

THP-1 cells and RAW 264.7 cells seeded at a density of 5 × 10⁵ cells per well in 6-well plates were treated with honokiol (0–30 μM) for different duration of time. Total RNA of cells was extracted using TRIzol reagent (Invitrogen #15596026) followed by reverse transcription into complementary DNA (cDNA) using a PrimeScript RT reagent Kit (Takara Bio #RR037A) following the manufacturer’s protocol. Quantitative PCR (qPCR) was performed using a PerfectStart Green qPCR Super Mix kit (TransGen Biotech #AQ602) in an ABI Q7 Fast Real Time PCR System (Applied Biosystems). Housekeeping gene ACTB/Actb served as the normalization gene in these studies. ΔCt was obtained by subtracting the Ct of target genes (IFNB/Ifnb, IL6/Il6, CXCL10/Cxcl10 and ISG15/Isg15) from the Ct of ACTB/Actb and the relative expression levels for the target genes were defined as 2^ΔCt. The sequences of primers used for qPCR were listed in Table S3.

Plasmid construction

For construction of the bacterial expression plasmids, DNA fragment encoding human (h) or mouse (m) STING ligand-binding domain (h/mSTING-LBD) (155-343aa and 146-370aa for human and mouse STING, respectively) was synthesized in Genewiz company. DNA fragments encoding FiSLs or dead FiSL (dFiSL) were obtained by fusing the coding sequences of hSTING-LBD and cpGFP (circularly permuted green fluorescent protein) using PCR amplifications with primers containing overlapped sequences. Then, the bacterial expression plasmids were constructed by inserting the DNA fragments encoding h/mSTING-LBD or FiSLs into the BamH I/Xho I-digested pET28a vector (Sigma-Aldrich #69864) containing a N-terminal poly-Histidine (His) tag and TEV protease cleavage site. For construction of the lentiviral plasmids expressing guide RNAs (gRNAs) and hSpCas9, gRNAs targeting STING or Sting were designed using an online tool (https://www.zlab.bio/resources). Two different promising gRNAs were chosen, synthesized, and then subcloned into the BsmB I-digested lentiCRISPRv2 vector (Addgene #98290) containing a puromycin selection marker. The presence of the gRNA in lentiviral vector was examined by Sanger sequencing. The sequences of gRNAs and PCR primers used for molecular cloning are listed in Table S3.

Lentivirus production and gene knockout

To produce lentiviruses, 293FT cells (1 × 10⁶) plated in 60-mm dishes were co-transfected with 3 μg of lentiCRISPRv2-gRNA plasmid, 1.5 μg of pMD2.G plasmid (Addgene #12259), and 1.5 μg of psPAX2 plasmid (Addgene #12260) using Lipofectamine 2000 (Thermo Fisher Scientific). Forty-eight hours after transfection, cultural supernatants containing lentiviruses were collected, centrifuged at 600 × g for 10 min, and then passed through filters with pore size of 0.45 μm. THP-1 cells and RAW 264.7 cells were infected with lentiviruses (MOI = 1) in the presence of 8 μg mL⁻¹ polybrene. Infected cells were maintained in cultural medium supplemented with 1 μg mL⁻¹ puromycin for 7 days. The efficiency of gene knockout (KO) was evaluated by immunoblotting analysis.

Protein expression and purification

Single clone of BL21 (DE3) pLysS cells transformed with pET28a-based plasmids expressing h/mSTING-LBD or FiSLs was inoculated in 5 mL LB medium containing 20 μg mL⁻¹ kanamycin, and then cultured at 37°C overnight with 220 rpm shaking, the bacterial suspensions were transferred to 250 mL LB medium at a ratio of 1:100 followed by induction with 1 mM IPTG for 16 h at 18°C when the cell density of bacteria reaches about 5 × 10⁸ cells mL⁻¹ (optical density of the bacterial suspension sample measured at a wavelength of 600 nm is about 0.6, OD₆₀₀ ≈ 0.6). The cells were harvested, homogenized in the lysis buffer (20 mM Tris, pH [7.4], 400 mM NaCl, 10% glycerol, 5 mM β-ME, and 1 mM PMSF), and then centrifuged at 18, 000 × g for 15 min. The supernatants were collected, loaded onto a Ni-NTA column, washed with washing buffer (20 mM Tris, pH [7.4], 400 mM NaCl, 10% glycerol and 20 mM imidazole), and then eluted with elution buffer (20 mM Tris, pH [7.4], 400 mM NaCl, 10% glycerol and 200 mM imidazole). The eluted target proteins were desalted, incubated with TEV protease (20 μg mL⁻¹) at 16°C overnight followed by reloading into the Ni-NTA column to remove the uncleaved His-tagged proteins and cleaved His tag. Next, the partial purified proteins were desalted, washed twice with gel filtration buffer (10 mM Tris, pH [7.4], 200 mM NaCl), concentrated to about 10 mg mL⁻¹ with a 10-kDa ultrafilter column, and then loaded onto a HiLoad 16/600 Superdex 200 column (GE Healthcare). The peak fractions were collected, separated by SDS-PAGE, followed by visualization with Coomassie blue staining.

Fluorescence spectroscopy

The FiSLs (250 nM) dissolved in buffer (100 mM Tris, 150 mM NaCl) were placed into a white 96-well plate with flat bottom. Excitation and emission spectra of FiSLs were measured using a Synergy H1 spectrofluorometer (BioTek) at the set temperatures in the presence of different concentrations of compounds. Excitation spectra were monitored at 520 nm with excitation wavelength from 400 to 500 nm and emission spectra were measured from wavelength 500 to 600 nm by excitation at 488 or 405 nm. Fluorescence was read every 2 nm with an integration time of 1 s. Unless stated, the fluorescence of FiSLs was monitored in a total volume of 100 μL under the condition of temperature 25°C and pH [7.4].

STING ligand screening

The FiSL (250 nM) dissolved in 50 μL buffer (100 mM Tris, pH [7.4], 150 mM NaCl) was mixed with 50 μL buffer (vehicle) or thousands of compounds at a final concentration of 100 μM in white 96-well plates with flat bottom. After incubation for 1 min, FiSL-derived fluorescence (Ft and F₀ for compound- and vehicle-treated wells, respectively) (excitation at 488 nm and emission at 520 nm) in each well was measured using a Synergy H1 spectrofluorometer (BioTek) at 25°C. The screening results were expressed by ratios of Ft to F₀. The self-house bioactive compound library was constructed by combining MCE FDA-approved Drug Library-1724 cmpds (#HY-L021), anti-virus Compound Library-508 cmpds (#HY-L001) and MetaSci Human Metabolite Library-916 cmpds (#HML 12.99).

Multiangle light scattering (MALS) analysis

Purified FiSL was analyzed by a multiangle light scattering instrument (Agilent Technology) connected to a Superdex 200 Increase 10/300 GL column (GE Healthcare), a DAWN Heleos-II multi-angle light scattering detector (Wyatt Technology) and an Optilab T-rEX refractive index detector (Wyatt Technology). The multiangle light scattering instrument was equilibrated with buffer (100 mM Tris pH [7.4], 150 mM NaCl) at a flow rate of 0.4 mL min⁻¹ for 12 h. The ultraviolet light (280 nm) absorption, light scattering and refractive index were simultaneously monitored for 1 h following the loading of 100 μL protein sample onto the multiangle light scattering system. Molecular mass of FiSL was calculated using ASTRA software (Version 5.3.4) (Wyatt Technology).

Bio-layer interferometry (BLI) assay

BLI assay was performed using a ForteBio Octet system (Sartorius). Briefly, ligand binding signal was detected by a Ni-NTA sensor loaded without or with 40 μg mL⁻¹ His-tagged h/mSTING-LBD using a circulatory detection program as follows: association with serial concentrations (0–80 μM) of honokiol for 60 s, and then dissociation in binding buffer (20 mM Tris, pH [7.4], 400 mM NaCl, 10% glycerol, 0.1% DMSO) for another 120 s. The binding affinity of h/mSTING-LBD toward honokiol was calculated using Octet software (Version 9.0.0.49) (Sartorius).

Crystallization

To obtain the APO crystals of hSTING-LBD, purified hSTING-LBD protein (5 mg mL⁻¹) was mixed with equal volume (1 μL: 1 μL) of reservoir solution (0.16 M calcium acetate, 0.08 M sodium cacodylate, pH [6.5], 14.4% w/v PEG 8000, 20% v/v glycerol) and then the growth of crystals was induced by hanging-drop vapor diffusion. Next, to crystallize the hSTING-LBD in complex with honokiol, APO crystals of hSTING-LBD were incubated with reservoir solution containing 5 mM honokiol for 12 h at 16°C. The resulting crystals were cryoprotected by reservoir solution supplemented with 20% glycerol, then snap-frozen in liquid nitrogen.

Structure determination

All the diffraction datasets were collected at BL02U1 beamline of the Shanghai Synchrotron Radiation Facility (SSRF). Structural data were indexed, integrated, and scaled with the XDS program suite (Version ASCII). The structure was solved by molecular replacement using the deposited STING structure (PDB code: 4EF5) as a searching model. Model building and refinement were performed by the programs COOT (Version 0.8.9.2) and PHENIX (Version 1.18.2–3874), respectively. Structure figures were generated using the Pymol (Version 1.3) and COOT software. Statistical data of the diffraction and refinement are shown in Table S2.

Hematological tests

Whole blood from orbital veins of tumor-bearing mice was drawn onto blood collection tubes at different time points post honokiol administration. For routine blood tests, the whole blood samples (⁓ 200 μL) without addition of anticoagulants were analyzed using the ADVIA 2120i hematology system (Siemens). For blood biochemical tests, serum samples derived from the whole blood were analyzed using the ADVIA chemistry XPT system (Siemens).

Enzyme-linked immunosorbent assay (ELISA)

Serum levels of cytokines were measured using ELISA kits (Jiangsu Meibiao Biotechnology #MB-6317A for IFN-β, #MB-2868A for TNF-α, and #MB-2899A for IL-6) following the manufacturer’s protocols. Briefly, serum and standard samples (50 μL) were added into the wells precoated with capture antibodies, and then incubated for 1 h at room temperature. The wells were washed twice with PBS, incubated with HRP-conjugated detection antibodies (50 μL) for another 1 h at room temperature. The plate was washed 4 times with PBS and the signal was detected by adding 50 μL TMB solution into each well for 15–30 min. HRP-catalytic reaction was terminated by adding 50 μL stop solution and the plates were read at absorbance of 450 nm using a BioTek microplate reader. Cytokine concentration in each sample was calculated according to the standard curve constructed by the standard samples.

Isothermal titration calorimetry (ITC) assay

To determine whether honokiol competes the binding of 2′3′-cGAMP to STING, the purified hSTING LBD protein and 2′3'-cGAMP were diluted into the same buffer (20 mM HEPES pH [7.4], 150 mM NaCl, 1 mM DTT). ITC assay was performed using the MicroCal ITC-200 (GE Healthcare) at 25°C. hSTING LBD protein (50 μM) was mixed with or without 100 μM honokiol, and then 250 μL of the sample was loaded into the thermostated sample cell. 2′3′-cGAMP (1 mM) was injected stepwise over 20 injections with 120 s space apart and stirring at 750 rpm. Heat change was recorded and binding affinity data were analyzed using the Origin 7.0 software (Origin Laboratory).

Measurement of honokiol levels by LC-MS

For extraction of plasma honokiol, plasma (10 μL) isolated from tumor-bearing mice was mixed with 120 μL of ice-cold methanol, vortexed for 30 s, and then cleared by centrifugation at 14, 000 × g for 10 min at 4°C. For extraction of tumor honokiol, tumor tissues (50 mg) were soaked in 500 μL extraction solution (methanol and water ratio 8: 2) and pulverized by cryomilling. After keeping on ice for 10 min, the samples centrifuged at 14, 000 × g for 10 min at 4°C and the supernatants were collected. Next, the solvent in the extraction samples was removed by vacuum evaporation, and the dried extracts were dissolved in methanol with optimized volumes (100 μL for serum extracts and 500 μL for tumor extracts).

To detect honokiol, the samples were separated by an Agilent ZORBAX 300SB-C8 column (3.5 μm, 2.1 mm × 50 mm) at a column temperature of 25°C. The mobile phase was composed of solution A (0.1% formic acid in water) and solution B (0.1% formic acid in acetonitrile). A total volume of 10 μL sample was injected into the column and separated at flow rate of 0.2 mL min⁻¹ following the program: 99% A and 1% B during 0–7 min, 85% A and 15% B during 7–13 min, and 99% A and 1% B during 13–15 min. HRAM data were acquired using the liquid chromatography/Quadrupole-TOF mass spectrometer (Agilent 6530) operated at a negative mode. The amount of honokiol in samples were calculated according to the standard curve constructed by honokiol samples with a concentration gradient.

Measurement of 2′3′-cGAMP levels by LC-MS

A total number of 1 × 10⁷ of THP-1 and RAW264.7 cells plated in 10-cm dishes and without or with poly (dA:dT) transfection were washed twice with ice-cold PBS, and then incubated with 1 mL extraction solution (80% methanol in water) for 5 min at 4°C. The cell lysates harvested by a scraper were vortexed at 1, 000 rpm for 10 min at 4°C, then centrifuged at 15, 000 × g for 20 min at 4°C. The supernatant (980 μL) was collected, dried by vacuum evaporation using a refrigerated CentriVap Concentrator (Labconco), and then reconstituted in 100 μL methanol.

To detect 2′3′-cGAMP by LC-MS analysis, the samples (10 μL) were loaded onto an Agilent ZORBAX 300SB-C8 column (3.5 μm, 2.1 mm × 50 mm) at a column temperature of 25°C, and then separated at flow rate of 0.2 mL min⁻¹ following the program: 99% A (0.1% formic acid in water) and 1% B (0.1% formic acid in acetonitrile) during 0–2 min, 85% A and 15% B during 2–5 min, and 99% A and 1% B during 5–8 min. HRAM data were obtained using the liquid chromatography/Quadrupole-TOF mass spectrometer (Agilent 6530) operated at a negative mode. The amount of 2′3′-cGAMP in each sample was calculated according to the standard curve generated by loading serial concentrations of 2′3′-cGAMP and then normalized to the cell number.

Quantification and statistical analysis

All immunoblotting experiments were independently repeated at least three times and a representative result was shown. Sample numbers (n) and experimental repeats are indicated in figure legends. All data are presented as mean ± SD or mean ± SEM. Statistical analyses were performed using the GraphPad Prism 8.0.2 and Rstudio 4.1.3. The two-tailed Student’s t test and one-way ANOVA were used to compare paired and multiple variables, respectively. Survival curves were compared using the two-tailed log-rank test. p values less than 0.05 were considered statistically significant.

Published: July 15, 2025