Effective delivery of STING agonist using exosomes suppresses tumor growth and enhances antitumor immunity

The Stimulator of Interferon Genes (STING) pathway is implicated in the innate immune response and is important in both oncogenesis and cancer treatment. Specifically, activation of the cytosolic DNA sensor STING in antigen-presenting cells (APCs) induces a type I interferon response and cytokine production that facilitates antitumor immune therapy. However, use of STING agonists (STINGa) as a cancer therapeutic has been limited by unfavorable pharmacological properties and targeting inefficiency due to rapid clearance and limited uptake into the cytosol. Exosomes, a class of extracellular vesicles shed by all cells are under consideration for their use as effective carriers of drugs owing to their innate ability to be taken up by cells and their biocompatibility for optimal drug biodistribution. Therefore, we engineered exosomes to deliver the STING agonist cyclic GMP-AMP (iExoSTINGa), to exploit their favorable pharmacokinetics and pharmacodynamics. Selective targeting of the STING pathway in APCs with iExoSTINGa was associated with superior potency compared with STINGa alone in suppressing B16F10 tumor growth. Moreover, iExoSTINGa showed superior uptake of STINGa into dendritic cells compared with STINGa alone, which led to increased accumulation of activated CD8+ T-cells and an antitumor immune response. Our study highlights the potential of exosomes in general, and iExoSTINGa specifically, in enhancing cancer therapy outcomes.

The Stimulator of Interferon Genes (STING) pathway is implicated in the innate immune response and is important in both oncogenesis and cancer treatment. Specifically, activation of the cytosolic DNA sensor STING in antigen-presenting cells (APCs) induces a type I interferon response and cytokine production that facilitates antitumor immune therapy. However, use of STING agonists (STINGa) as a cancer therapeutic has been limited by unfavorable pharmacological properties and targeting inefficiency due to rapid clearance and limited uptake into the cytosol. Exosomes, a class of extracellular vesicles shed by all cells are under consideration for their use as effective carriers of drugs owing to their innate ability to be taken up by cells and their biocompatibility for optimal drug biodistribution. Therefore, we engineered exosomes to deliver the STING agonist cyclic GMP-AMP (iExo STINGa ), to exploit their favorable pharmacokinetics and pharmacodynamics. Selective targeting of the STING pathway in APCs with iExo STINGa was associated with superior potency compared with STINGa alone in suppressing B16F10 tumor growth. Moreover, iExo STINGa showed superior uptake of STINGa into dendritic cells compared with STINGa alone, which led to increased accumulation of activated CD8 + T-cells and an antitumor immune response. Our study highlights the potential of exosomes in general, and iExo STINGa specifically, in enhancing cancer therapy outcomes.
The success of immune checkpoint inhibitors in invasive cancer has renewed interest in harnessing the immune control of cancer for clinical benefit (1). Recent focus on enhancing antitumor responses, in particular for patients who remain refractory to immune checkpoint blockade, has energized the study of therapeutics that polarize the tumor microenvironment and boosting T cell response using alternative pathways. Promoting an antitumor immune microenvironment relies in part on sustained activation of T cells to eradicate cancer cells, but also on the engagement of the innate immune system (2,3).
Dendritic cells (DCs) bridge innate and adaptative response, and DNA released from genomic unstable cancer cells elicits DCs' type I interferon signaling, activating naïve T cells and promoting antitumor responses (4,5). A critical pathway in DCs sensing cytosolic DNA, in part serving as viral infection police, is the stimulator of interferon genes (STING) pathway. Cytosolic DNA is converted to cyclic GMP-AMP (cGAMP) by cGAS (cyclic GMP-AMP synthase). The STING protein on the endoplasmic reticulum of DCs responds to cGAMP by activating the transcription of type I interferon and cytokines, which in turn activates and primes antigen-specific CD8 + T cells (6,7). In the context of tumors, STING activation can also influence the immune microenvironment by limiting the accumulation of MDSCs and Tregs and by promoting M1macrophage polarization (4). Activation of the STING pathway promotes innate and adaptive immune cell infiltration in tumors, but also exerts antitumor effects by triggering apoptosis, inducing autophagy, and suppressing cell cycle progression of cancer cells, and by promoting vascular normalization in endothelial cells (8). Developing anticancer therapeutics by enhancing STING signaling in tumors thus may generate clinical benefit by impacting both cancer cells and the tumor microenvironment.
The utility of STING agonist (STINGa), namely cGAMP and other cyclic dinucleotides (CDNs), in the treatment of cancer is in early phase of clinical testing (7), and ongoing efforts are directed toward overcoming its unfavorable pharmacological profile and poor bioavailability (9)(10)(11)(12). Despite extensive preclinical studies indicating antitumor efficacy of STINGa, notably in synergy with other immune modulators, initial efforts in the development of pharmacological STINGa resulted in marginal benefit in patients, prompting the development of agonists with enhanced stability (13). CDNs are targeted for degradation by phosphodiesterases, in circulation and on the cell surface, severely limiting the half-life of STINGa (6,13). In addition, cGAMP and emerging modified STINGa are hydrophilic and negatively charged, rendering them largely nonpenetrating and limiting their cellular uptake (8,14).
The use of nanocarriers for cytosolic delivery of STINGa would presumably enhance target engagement and minimize their rapid clearance (6,15). Exosomes are shed by all cells and are abundant in circulation and other biological fluids (16). They originate from the double invagination of the plasma membrane and released as 40-150 nm, lipid-bilayer vesicles with a surface that, in part, mimics the surface of the cells they came from (16). Toward this end, exosomes, a unique class of extracellular vesicles, are natural nanocarriers that demonstrate an efficient uptake by DCs and other APCs, with potential privilege from immune clearance (17)(18)(19)(20)(21)(22)(23)(24). Notably, T-cells-derived exosomes containing genomic and mitochondrial DNA were reported to stimulate the STING pathway in DCs (25).
We recently described the use of exosomes as a carrier for the delivery of siRNA therapeutic payload, targeting oncogenic Kras in pancreatic cancer (26,27). We identified that exosomes were readily taken up by cancer cells, enabling superior siRNA-mediated targeting of oncogenic Kras compared with synthetic liposomes (27). CD47 on their surface limited their clearance from the circulation via phagocytosis, when exogenously administered to tumor bearing mice, enhancing antitumor responses (27). Exosomes likely have multiple features enabling them as natural nanocarriers for cancer therapeutics and are under active study (24). Taking advantage of our exosomes production platform (26), we tested the underlying biology and antitumor efficacy of engineered exosomes containing STINGa (iExo STINGa ).

Generation and validation of iExo STINGa
We engineered exosomes containing the small-molecule STING agonist cyclic GMP-AMP (cGAMP, STINGa), thereafter referred to as iExo STINGa . Exosomes enriched from the culture supernatant of HEK293T cells were loaded with cGAMP ( Fig. 1A, see Experimental Procedures). Nanoparticle tracking analysis revealed a similar size distribution characteristic of exosomes, and loading of exosomes with STINGa did not alter their size or concentration (Fig. 1, B and C). Both unloaded exosomes (control exosomes, Exo) and iExo STINGa displayed expression of tetraspanin markers characteristic of exosomes (CD9, CD63, and CD81), as evaluated by flow cytometry analysis of surface expression (Fig. 1D). A standard curve employing fluorescein-labeled STINGa was developed, and it was used to estimate that approximately 200 mM STINGa is associated with exosomes (approximately 2% of STINGa, Fig. 1E).
In order to determine whether STINGa may bind to intraluminal STING, we evaluated the level of STING in exosomes. STING protein was not detected in HEK293T cell lysates or exosomes (Fig. S1, A and B) (28)(29)(30). Previous studies identified the folate receptor SLC19A1 as a transporter of STINGa into cells (31). Therefore, we evaluated SLC19A1 protein levels in HEK293T cells and exosomes, and SLC19A1 was not detected in exosomes (Fig. S1, C and D). In agreement with these findings, incubation with folic acid, which competes for folic acid receptors, did not alter the amount of STINGa associated with the exosomes (Fig. 1E). The glutamine/glutamate transporter, SLC38A2, another potential transporter of STINGa was assessed, and it was identified in HEK293T cells but not the exosomes (Fig. S1, E and F). Incubation with glutamine to compete with STINGa for glutamine/glutamate transporters did not alter STINGa content in the exosomes (Fig. 1E). Treatment of exosomes with proteinase K to cleave all surface proteins and ectodomains of transmembrane proteins ( Fig. S1G) that could be potentially involved in STINGa transport did not significantly alter STINGa present in the exosomes (Fig. 1F). Together, these data suggest that STINGa may enter exosomes in a passive manner without involvement in exosomal surface proteins.
To evaluate the association of fluorescein-labeled STINGa with exosomes, small-particle flow cytometry was performed. Fluorescein-STINGa + particles were specifically detected in iExo STINGa when compared with Exo ( Fig. 1, G and H), and treatment with snake venom phosphodiesterase (SVPDE) to cleave extraluminal STINGa did not reduce the number of STINGa + particles (Fig. 1, G and H), suggesting that STINGa is found predominantly within the lumen of exosomes.
Uptake of iExo STINGa in DCs was evaluated using bonemarrow-derived dendritic cells (BMDCs) from wildtype and Sting1 knock out (STING KO ) mice. The uptake of fluorescently labeled STINGa in BMDCs was superior when using iExo STINGa compared with STINGa ( Fig. 1I). BMDCs treated with iExo STINGa also showed increased expression of Ifnb1, Cxcl10, and Il6 when compared with cells treated with STINGa (Fig. 1J). The transcriptional upregulation of these genes, supporting STING pathway activation, was not observed in STING KO BMDCs, indicating a specific target engagement (Fig. 1J).

Induction of T-cell activation with iExo STINGa enables systemic antitumor response
To ascertain the underlying mechanism associated with the antitumor activity of iExo STINGa , the immune microenvironment STING agonist delivery by exosomes  STING agonist delivery by exosomes of iExo STINGa -treated tumors was evaluated. A significant increase in CD45 + , CD3 + , and CD8 + immune cells was observed in iExo STINGa -treated tumors compared with untreated, Exotreated control tumors (Fig. 3, A-C). CD4 + and proliferating (Ki67 + ) CD4 + T cells were unchanged, but proliferating CD8 + T cells were increased in iExo STINGa -treated tumors (Fig. 3, C-E).
Indication of a systemic influence on immune control of tumor with iExo STINGa therapy was evidenced with a reduction in contralateral tumor growth when ipsilateral tumors are treated.
Mice were implanted with tumors on both flanks, and ipsilateral tumors were treated as described above (and shown in Fig. 4A).
Tumor growth suppression in the ipsilateral (receiving intratumoral treatment) tumor was significant in iExo STINGa -treated mice when compared with control and superior to STINGatreated mice (Fig. 4, B-D). A significant tumor growth suppression was also evident in the contralateral tumor in iExo STINGa -treated mice when compared with control mice, whereas STINGa did not significantly suppress contralateral tumor growth (Fig. 4, B-D). The suppression of ipsilateral tumors with iExo STINGa was associated with a significant increase in proliferating CD4 + and CD8 + T cells when compared with STINGatreated tumors and control tumors (Fig. 4E, Fig. S3, A and B).
To confirm the in vivo specificity of iExo STINGa on the STING pathway, we treated B16F10 tumor-bearing STING KO mice with 25 μg iExo STINGa or 25 μg STINGa. Tumor growth was not significantly different when B16F10 tumors are implanted in STING KO mice compared with wildtype mice (Fig. S3C). Tumor volumes measurements indicated that 25 μg iExo STINGa or 25 μg STINGa failed to suppress tumor growth on the STING KO background (Fig. 4, F and G, Fig. S3D). These results collectively support that the antitumor response exerted by iExo STINGa is realized by engaging its specific target (STING pathway). Finally, we tested the stability of by iExo STINGa in stimulating BMDCs. Storage of iExo STINGa at -20 C for 2 weeks or -80 C for 1 month before thaw and use did not significantly impair increased expression of Ifnb1, Cxcl10, and Il6, when compared with freshly prepared iExo STINGa (Fig. S3, E-G).

Discussion
We report on the efficacy of exosomes as carrier of STINGa for antitumor therapy. STINGa was predominantly found intraluminally in the iExo STINGa and was resistant to SVPDE degradation. The entry of STINGa into exosomes appears to be due to potential passive diffusion through the lipid bilayer. iExo STINGa showed a markedly superior activation of the STINGa pathway and activation of DCs, when compared with STINGa. This is likely as a result of enhanced uptake or retention of iExo STINGa in DCs compared with using STINGa by itself. Our results indicate iExo STINGa showed approximately tenfold increase in uptake by DCs compared with free STINGa. These results support the previously reported efficacy of exosomes in delivering a therapeutic payload into the cytosol of treated cells (16,23,27,32). Liposomes, polymer nanoparticles, and hydrogels have been studied to enhance STINGa cytosolic delivery and stability, with mixed results (6). Nonetheless, these efforts support a potential use of a carrier for STINGa-based therapy to overcome the pharmacological limitations of STINGa. The added benefit of iExo STINGa compared with synthetic carriers may lie in the enhanced stability of exogenously administered exosomes, with lipid and protein compositions that do not elicit phagocytic clearance. Interestingly, the packaging of cGAMP in viral particles, concurrently studied with exosomes, indicated superior transfer of cGAMP in using viral particles (33). This study used exosomes from transfected cells and differs from our approach of engineering exosomes containing cGAMP, but nonetheless supports that exosomes' cargo includes cGAMP. In a related study, exosomes from irradiated cancer cells were shown to transfer dsDNA and stimulate the STING pathway in DCs (34), supporting the exosomes cargo's capacity for STING pathway activation.
Our data also indicate a superior antitumor effect of iExo STINGa treatment compared with STINGa, and the reduced tumor growth with iExo STINGa was associated with an influx of proliferating CD8 + T cells, in agreement with previous studies reporting antitumor response with STING pathway activation (4,35). Despite being administered intratumorally, iExo STINGa treatment showed an antitumor effect on contralateral, noninjected tumors. This was not observed when STINGa was used by itself. The activation of the STING pathway in the iExo STINGa -injected tumors generates abscopal effect that impact non-iExo STINGa -injected tumors in the same mice. Such response with STING pathway activation has been previously reported in the context of radiation and immune checkpoint blockade (13,36). We speculate that such systemic changes include activation of adaptive immunity reaching other tumor sites. Exosomes are now being implicated in adaptive and immune response regulation. Though we did not observe control exosomes (deprived of STINGa cargo) eliciting measurable immune responses at the dosage reported here, it is possible that added benefit could be realized with iExo STINGa generated from a cell source with potential for T-cell activation (37,38). In this study, the robust antitumor response using iExo STINGa as a single agent with systemic effect on secondary tumors supports the potential of iExo STINGa in clinical use via an established GMP-exosomes production platform (26).

Isolation and purification of exosomes
HEK293T cells were grown in T225 flasks with DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37 C/5% CO 2 for 2-3 days until they reached a confluency of 60-70%. The cells were washed twice with PBS (Corning Catalog # 21-040-CV) and fed serum-free medium for 48 h. The media was collected and centrifuged at 800g for 5 min, followed by centrifugation at 2000g for 10 min. The conditioned medium was filtered using a 0.2 μm filter flask (Thermo Fisher) and ultracentrifuged at 100,000g for 3 h at 4 C in a SW 32 Ti rotor (Beckman Coulter). The exosomes were resuspended in PBS, and their concentration and size distribution were evaluated using NanoSight LM10 before storage at -80 C.

Flow cytometry analysis of exosomes
Exosomes (5 × 10 9 quantified by NanoSight analysis) were incubated with 4 μm aldehyde/sulfate latex beads (Invitrogen, A37304) for 2 h at room temperature. This suspension was diluted to 1 ml with PBS, and the reaction was stopped with incubation with 100 mM glycine for 30 min. Exosome-bound beads were blocked with 10% BSA for 1 h and stained with 50 μg/ml mouse IgG1κ isotype control (BD Bioscience, 555746), CD9 (Sigma-Aldrich, SAB4700092), CD63 (BD Bioscience, 556019), and CD81 (BD Bioscience, 555675) in 20 μl of 2% BSA in PBS, and mixed at room temperature for 1 h. Secondary antimouse antibodies with Alexa Fluor 488 (Life Technologies, A21202) or Alexa Fluor 647 (Life Technologies, A31571) were added, and mixed at room temperature for 1 h. Detection of CD9, CD63, and CD81 on beads was analyzed using a BD LSR Fortessa X-20. Positive signal for CD9, CD63, and CD81 was determined based on the signal in isotype control samples using FlowJo (BD Bioscience).
In order to analyze fluorescein-STINGa loaded exosomes, samples were incubated with total exosome isolation reagent from cell culture media (Invitrogen, 4478359) overnight at 4 C, then centrifuged at 15,000 rpm for 1 h. The supernatant was collected and the pellet resuspended in PBS for analysis. Samples were analyzed with an acquisition time of 2 min using a BD LSR Fortessa X-20 cell analyzer equipped with a FSC PMT small-particle detector. Size gating of exosomes was performed based on the SSC-H versus FSC PMT-H distribution of 100 nm FITC beads. Positive fluorescein signal was established based on exosome samples without STINGa using FlowJo.

Quantification of STINGa loading
For quantification of STINGa loaded in exosomes, 10 11 exosomes were treated with 5 mg/ml proteinase K (Qiagen) for 30 min at 37 C followed by 20 min at 60 C. For competition assays, 200 μM of folic acid (Sigma Aldrich) or 100 mM of glutamine (Corning) was added to the exosomes. Fluorescein-STINGa (10 μM) was incubated with exosomes at room temperature for 16 h. Samples were incubated with total exosome isolation reagent from cell culture media overnight at 4 C, then centrifuged at 15,000g for 1 h. The pellet was resuspended in 100 μl of PBS and analyzed on a plate reader (Omega) with an excitation of 485 nm and emission of 520 nm. STINGa concentration in exosomes was calculated based on a standard curve of Fluorescein-STINGa. For verification of proteinase K activity, samples were incubated with 4 μm aldehyde/sulfate latex beads and analyzed by FACS as described above.

Treatment of B16F10 subcutaneous tumors
B16F10 cells (10 6 or 5 × 10 4 cells in 100 μl of PBS, as specific in the figure) were injected subcutaneously into the flank of 8-12-week-old female C57BL/6J mice purchased from the Jackson Laboratory. Tumor volumes were measured every day using digital calipers and calculated using the equation length × width 2 × 0.52. Unless otherwise stated, when the tumors reached a size of approximately 50 mm 3 , the mice were randomly assigned to the distinct treatment groups: untreated, STINGa (25 or 50 μg STINGa, 20 μl), Exo (10 11 exosomes, 20 μl), or iExo STINGa (0.5, 5, 10, 25 or 50 μg STINGa in 10 11 exosomes, 20 μl). Each treatment was administered intratumorally every 48 to 72 h for a total of 3-4 consecutive treatments (as detailed in the figures). Mice were euthanized when a tumor burden endpoint of less than 2000 mm 3 was reached. In specified experiments (Fig. 4), 8-12-week-old female C57BL/6J mice (Jackson Laboratory) were injected subcutaneously with B16F10 cells (5 × 10 4 in 100 μl of PBS) on each of its flank. When the larger of the two tumors reached a volume of 50 mm 3 , intratumoral treatment was initiated. The treated tumor was referred to as the ipsilateral tumor, and the untreated tumor on the opposite flank was referred to as the contralateral tumor. Treatment groups include untreated mice and mice treated with STINGa (10 μg STINGa, 20 μl), Exo (10 11 exosomes, 20 μl), or iExo STINGa (10 μg STINGa in 10 11 exosomes, 20 μl). Each treatment was administered intratumorally every 48-72 h for a total of three consecutive treatments. Mice were euthanized when a tumor burden endpoint of less than 2000 mm 3 was reached. All mice were housed under standard housing conditions at MD Anderson Cancer Center (MDACC) animal facilities, and all animal procedures were reviewed and approved by the MDACC Institutional Animal Care and Use Committee.

Statistical analyses
Statistical analyses were performed in using GraphPad Prism (GraphPad Software) and the respective statistical tests used are indicated in the figure legends. Normal distribution of data was evaluated using Shapiro-Wilk and Kolmogorov-Smirnov tests. An unpaired, two-tailed t-test was performed for comparison of two normally distributed groups, or one-way ANOVA with Bonferroni's multiple comparison test for three or more normally distributed groups. For two groups that were not normally distributed, a Mann-Whitney test was performed. An unpaired t-test with Welch's correction was used for two groups that had significantly different standard deviations. Kruskall-Wallis test with Dunn's multiple comparison test was used to compare three or more groups that were not normally distributed. For comparisons that had significant differences in the standard deviations across three or more groups, Brown-Forsythe and Welch ANOVA with Dunnett's T3 multiple comparison test was performed. Linear regression analyses of averaged tumor volumes over time per experimental group were used to determine if the slopes between two groups were different. The p value reported on the linear regressions informs on the significant difference between the slopes. A p value <0.05 was considered statistically significant. Error bars represented standard error of the mean (S.E.M.).