Chemically Regulated STING‐Activating Prodrugs of Deoxyribose Cyclic Dinucleotides Elicit Robust Immune Activation and Durable Antitumor Immunity Authors: Zhiqiang Xie, Huimin Liu, Yuchen Yang, Xiaohan Sun, Jingwen Wang, Tianxiang Liu, Baifan Wang, Wenrui Qiu, Yiwen Sun, Yuting Bai, Yao Zhang, Dejun Ma, Rong Xiang, Shibo Li, Zhen Xi Published: 2026 ABSTRACT ABSTRACTDerivatives of deoxyribose cyclic dinucleotides (dCDNs) displayed notable advantageous properties in the activation of STING pathway. The chirality of the phosphotriester of dCDN prodrugs may influence the stability, cellular permeability, bioactivity and duration, which remains inadequately explored through systematic evaluations in cellular and animal models. Herein, we isolated and characterized all three diastereoisomers of alkyne‐conjugated esterase‐sensitive dCDN prodrugs, and meticulously examined their difference in stability, cellular uptake and bioactivity. Within THP1 cells, the (Rp,Rp) diastereoisomer demonstrated the highest level of activation of STING pathway (EC50 = 1.7 nM), and induced more robust and prolonged activation pulse than the (Rp,Sp) and (Sp,Sp) diastereoisomers. In murine models, prodrugs elicited significantly stronger stimulation in the development of an antitumor immune response compared to the parent CDN 3′,3′‐c‐di‐dAMP, as well as the clinically relevant STING agonist ADU‐S100. The (Rp,Rp) diastereoisomer exhibited the most pronounced antitumor activity in the context of intravenous administration, significantly suppressing tumor proliferation, extending the survival with a complete response (CR) rate of 90% in a mouse CT26 tumor model, and establishing long‐lasting tumor‐specific immunological memory. These findings underscore the importance of considering the chirality of phosphotriesters in the development of more effective, safer, and sustainable STING agonist in tumor immunotherapy.Keywords: deoxyribose cyclic dinucleotides, diastereoisomers, prodrugs, STING agonist, tumor immunotherapy Introduction Cyclic GMP‐AMP Synthase (cGAS) and Stimulator of Interferon Genes (STING) are two essential components of the cGAS‐STING pathway in the innate immune system [1, 2]. The sensor cGAS detects cytosolic DNA and catalyzes the conversion of GTP and ATP into 2′,3′‐cyclic GMP‐AMP (2′,3′‐cGAMP), a cyclic dinucleotide (CDN). Subsequently, 2′,3′‐cGAMP binds directly to STING, leading to its activation, and facilitates the generation of type I interferons and pro‐inflammatory cytokines in a TBK1‐IRF3‐dependent way [3, 4, 5, 6]. As STING agonists, 2′,3′‐cGAMP and other CDNs have attracted significant interest in tumor immunotherapy to stimulate the immune system and eliminate malignancies. Additionally, CDNs have been explored as potential vaccine adjuvants that enhance the production of pathogen‐specific antibodies and induce T‐cell responses in mammals [7, 8, 9, 10, 11, 12, 13, 14]. In vivo applications of natural CDNs were limited due to their low structure stability, cell penetration, and bioactivity. Numerous efforts have been made to modify the structures of CDNs at the ribose rings, nucleotide bases, internucleotide phosphodiester or phosphorothioate linkages [15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26]. These strategies have resulted in the development of several synthetic CDN‐based STING agonists, including ADU‐S100 [27], MK‐1454 [22], E7766 [28], SB‐11285 [23], IMSA‐101 [23], and BI1387446 [29]. These STING agonists have demonstrated improved absorption, distribution, metabolism, and excretion (ADME) in clinical studies [23]. However, CDN‐based STING agonists have garnered significant attention due to emerging concerns, including rapid clearance from the body, limited membrane permeability and bioavailability, and unexpected toxicity or side effects. To address the inherent limitations of CDNs, a prodrug approach has been devised wherein the phosphodiester linkage is substituted with a phosphotriester moiety. This modification effectively masks the negative charge, thereby enhancing cellular uptake efficiency. Upon cellular internalization, intracellular enzymatic activity or alterations in pH and redox states may facilitate the reconversion of the phosphotriester back to the native phosphodiester form. The advancement of novel CDN‐based STING agonists has been substantially informed by our earlier work, which demonstrated that S‐acyl‐2‐thioethyl (SATE) and dithioethanol (DTE)‐conjugated CDN prodrugs can elicit potent immunological responses via the STING‐dependent pathway, significantly extending the half‐life of CDNs to 114 h [30, 31, 32]. Furthermore, vinylphosphonate‐modified CDN prodrugs have been engineered by converting phosphodiester linkages into esterase‐sensitive phosphotriesters, thereby promoting intracellular activation [33]. Recent research has also introduced phosphoester CDN prodrugs incorporating biolabile protecting groups, which enhance cellular permeability and yield improved bioactivity, with EC50 values ranging from approximately 20 to 30 nM in cellular assays [34, 35]. Additionally, chemically inducible CDNs (iCDNs) have been developed through the integration of bioorthogonal caging groups at phosphodiester sites, enabling efficient cellular uptake alongside spatiotemporally controlled activation of STING [36]. Collectively, these strategies provide robust platforms for optimizing CDN delivery and facilitating precise modulation of STING signaling, thereby advancing targeted immunotherapeutic interventions. All these examples demonstrate the feasibility of utilizing the dCDN prodrug strategy to achieve higher activity. However, the chirality of the phosphotriester may influence the stability, delivery, and bioactivity, which has received little attention. For instance, the induction of type I IFNs in THP1 cells by the (Rp,Rp) stereoisomer of the bis‐phosphorothioate CDN 2′,3′‐cdiAM(PS)2 was significantly stronger than that induced by its corresponding (Rp,Sp) stereoisomer [16]. Given these differences, the effect of chirality on phosphotriesters should be systematically evaluated in cellular and mouse models to identify potential optimization strategies for the future design of more effective, durable, and safer STING agonist. In this study, we synthesized and purified all three diastereoisomers [(Rp,Rp), (Sp,Sp), and (Rp,Sp)] of an alkyne‐modified, esterase‐sensitive dCDN prodrug via deoxynucleoside substitution and the phosphotriester approach. The antitumor effects of all three dCDN prodrug diastereoisomers were comprehensively characterized through in vitro and in vivo assays. Notably, the (Rp,Rp) diastereoisomer exhibited the highest activity, with an EC50 of 1.7 nM for stimulating STING pathway. Our results indicate that these phosphotriester diastereoisomers exhibit increased stability and bioactivity in regulating immune cells, elicit a robust antitumor immune response to remodel the tumor microenvironment (TME) in tumor‐bearing mouse models. In addition, the enhanced cellular uptake by tumor cells activated the STING pathway within these cells, inducing strong immunogenic cell death (ICD). This subsequently promoted more robust antigen endocytosis and presentation by dendritic cells or macrophages as well as increased T cell infiltration. These results demonstrate the feasibility and effectiveness of using a novel diastereoisomer‐optimized design strategy for dCDN prodrugs to enhance cancer immunotherapy. Results and Discussion Synthesis and Isolation of Three Alkyne‐Modified dCDN Prodrug Diastereoisomers In the design of CDN prodrugs, converting phosphodiesters to phosphotriesters has proven to be an effective strategy [30, 31, 32, 34, 35]. In our previous work, we have demonstrated that SATE serves as an excellent protective group for phosphotriesters [30, 32]. Following cleavage of the thioester by intracellular esterases, such as carboxylesterases [37], the exposed mercaptoethyl group is readily broken down through intermolecular nucleophilic displacement. The conversion of phosphotriesters back to phosphodiesters results in the release of the parent CDN. The alkyne group not only facilitates further functionalization through click chemistry but also, due to its distinct polarizability, has been utilized in drug design to influence the electrical structure, affinity, and activity of pharmacological molecules [38, 39]. Here, we introduced an alkyne group to the thioester, and replaced ribose with deoxyribose in the CDN backbone to enhance the stability of the prodrug. This novel alkyne‐containing dCDN prodrug has three diastereoisomers: (Rp,Rp), (Sp,Sp), and (Rp,Sp). One was asymmetric, with one phosphate ester protective group oriented down and the other up, while the other two were symmetric, with both protective groups oriented either up or down. The designed prodrug was synthesized as summarized in Figure 1A. 2,2‐Dimethylbut‐3‐ynoic acid 2 was converted to compound 4 through acyl chlorination and esterification with mercaptoethanol. Compound 5 was obtained by coupling compound 4 with 2′‐deoxynucleoside phosphoramidite 1 using the activator ETT in dry acetonitrile, followed by oxidation with TBHP. To produce intermediate 6, the cyanoethyl group of compound 5 was removed with tert‐butylamine. By treating compound 5 with 6% dichloroacetic acid (DCA) in methylene chloride, the 5′‐DMTr protective group was removed, yielding intermediate 7. Subsequently, the equivalent linear dinucleotides 8 were produced by coupling compounds 6 and 7 in the presence of MSNT, followed by treatment with 6% DCA. After the cyanoethyl group of intermediate 8 was removed with tert‐butylamine, the cyclic dinucleotide phosphotriester 9 was obtained through cyclization by MSNT. Compound 9 was further treated with 10% diisopropylamine (DIA) in methanol to produce dCDN prodrug 10, achieving an overall yield of 14.3%. Synthesis, cellular uptake and metabolism of diastereoisomers of prodrug 10. (A) Synthesis of prodrug 10. Reagents and conditions: (a) (COCl)2, DMF, DCM (78.8%); (b) Et3N, DCM (86.2%); (c) 1. ETT, MeCN; 2. TBHP (94.0%); (d) t‐BuNH2, MeCN (100%); (e) 6% DCA, DCM (86.4%); (f) 1. MSNT, pyridine; 2. 6% DCA, DCM (43.3%); (g) 1. t‐BuNH2, MeCN; 2. MSNT, pyridine (62.5%); (h) 10% DIA, MeOH (64.9%). (B) Comparison of phosphorus NMR spectra of the diastereoisomers. (C) Comparison of proton NMR spectra of the diastereoisomers. (D‐F) THP1 cells were treated with 10 µM prodrugs for the indicated period of time, and the concentrations of the prodrugs or parent CDN 3′,3′‐c‐di‐dAMP in cellular extracts were determined by LC‐MS/MS. Data points in (D‐F) represent mean ± SD (n = 3). Since there are two sets of signals corresponding to units A and B of the dinucleotide, the proton and phosphorus NMR spectra demonstrate that (Rp,Sp)‐10 is an asymmetric diastereoisomer (Figure 1B,C). The predicted chemical shift differences were compared with the configurations of the other two symmetric diastereoisomers (Tables S1 and S2). In (Sp,Sp)‐10, the phosphate ester protective groups are in the β‐position (up), whereas in (Rp,Rp)‐10, they are in the α‐position (down) (Figure S2). This distinction is supported by the larger chemical shifts of hydrogen H1′ and H3′ in (Rp,Rp)‐10 compared to those in (Sp,Sp)‐10. The preparative reverse phase HPLC was employed to separate the three diastereoisomers of prodrug 10, yielding a ratio of 75 (Rp,Sp): 20 (Rp,Rp): 5 (Sp,Sp). Prodrugs Improved Stability, Cell Permeation, and Release of Parent CDN We first assessed the prodrugs’ stability in 20% fetal bovine serum (FBS). Clinically relevant STING agonist ADU‐S100 displayed a half‐life of 67 h. Meanwhile, (Rp,Rp)‐10 exhibited a longer half‐life period of 115 h compared to (Rp,Sp)‐10’s 62 h and (Sp,Sp)‐10’s 74 h, indicating that the (Rp,Rp) configuration have higher stability in serum (Table S3). In terms of enzymatic stability, all three diastereoisomers hardly experienced any degradation by snake venom phosphodiesterase (SVPD) at 2 h, while the percentage of surviving ADU‐S100 under the same condition was 33.2% (Figure S3A). Higher enzymatic stability was also demonstrated by all three diastereoisomers in ectonucleotide pyrophosphatase/phosphodiesterase family member 1 (ENPP1) enzymatic cleavage resistance assay at 72 h (Figure S3B), suggesting that the protective phosphotriester group can effectively shield the parent CDN from degradation by phosphodiesterase. Hydrolysis of the thioester by intracellular esterases and the following removal of ethylene sulfide results in the release of parent CDN (Figure S4). In order to examine the release of parent CDN from the prodrugs after cellular uptake, we conducted an in vitro cleavage experiment by incubating prodrugs with THP1 human leukemia monocyte cells for the specified duration. The concentration of prodrugs and the parent CDN 3′,3′‐c‐di‐dAMP was analyzed by LC‐MS/MS. Prodrug 10’s three diastereoisomers all entered THP1 cells quickly and were effectively converted to their parent CDN 3′,3′‐c‐di‐dAMP (Figure 1D–F). While intracellular levels of (Rp,Sp)‐10 and (Sp,Sp)‐10 peaked at 4 h, with corresponding levels of parent 3′,3′‐c‐di‐dAMP peaking at 8 h, the intracellular concentration of (Rp,Rp)‐10 peaked at 2 h, accompanied by higher level of parent 3′,3′‐c‐di‐dAMP peaking at 6 h. These findings suggested that prodrug 10’s three diastereoisomers could increase cell permeability and break down fast to release parent CDN into the cytoplasm. The peak concentration of the parent CDN achieved by (Rp,Rp)‐10 esterolysis was also higher than the other two. Consequently, as compared to the other two diastereoisomers, (Rp,Rp)‐10 demonstrated the most effective absorption and metabolism to supply parent CDN, which would be advantageous in preserving stronger long‐lasting activity in vivo. Prodrugs Exhibited Increased Efficacy in Activating the STING Pathway When the parent CDN is released into the cytoplasm, it binds to and activates STING, thereby triggering the expression of downstream genes regulated by interferon regulatory factors (IRFs) [40]. We assessed the STING‐stimulating activity of these prodrugs in THP1‐Lucia reporter cell line, in which the expression of Lucia luciferase is driven by an IRF‐inducible promoter. The compounds 3′,3′‐c‐di‐dAMP and ADU‐S100 exhibited respective EC50 values of 41.3 µM and 2.9 µM (Figure 2A). Notably, all three prodrug diastereoisomers demonstrated exceptionally high levels of activity: (Rp,Sp)‐10, (Rp,Rp)‐10, and (Sp,Sp)‐10 had respective EC50 values of 7.8, 1.7, and 36.8 nM. More importantly, the maximum level of inductions achieved by prodrugs were two to three times higher than those by ADU‐S100 and 3′,3′‐c‐di‐dAMP. These results are consistent with the prodrugs' excellent stability and effectiveness in attaining higher intracellular concentrations of the parent CDN. Prodrugs maintained superior biological activity to stimulate STING pathway. (A) The EC50 values of three diastereoisomers of CDN prodrug 10 and control CDNs after 24 h treatment were assayed in THP1‐Lucia. (B–F) The EC50 of control CDNs and three diastereoisomers of prodrug 10 at indicated time points. (G) Knockout of STING or cGAS in THP1 was confirmed by western blot. (H–J) The expression and phosphorylation levels of STING, TBK1 and IRF3 in WT, STING‐KO or cGAS‐KO THP1 cells treated with 1 µM control CDNs or prodrugs for 6 h were shown by western blot. Data points in (A‐F) represent mean ± SD (n = 3). To better understand the differences in STING activation among the prodrugs, we analyzed the EC50 values of the diastereoisomers at various time points. ADU‐S100 was able to activate STING from the time point of 12 h with an EC50 of 8.0 µM (Figure 2B), while 3′,3′‐c‐di‐dAMP did not activate STING until 24 h, with an EC50 of 56.1 µM (Figure 2C). Comparatively, both (Rp,Sp)‐10 and (Rp,Rp)‐10 demonstrated activation as early as 4 h, with EC50 values of 32.8 nM and 3.0 nM, respectively (Figure 2D–E). (Sp,Sp)‐10 showed no activity within the first 4 h, and its EC50 was 56 nM at 8 h (Figure 2F). At 96 h, the EC50 of (Rp,Rp)‐10 was 14.1 nM, indicating its ability to maintain high activity over an extended period (Figure 2E). Meanwhile, the EC50 values of (Rp,Sp)‐10 and (Sp,Sp)‐10 at 96 h were 106.8 nM and 84.7 nM, respectively, both lower than that of (Rp,Rp)‐10 (Figure 2E,F). These findings suggest that, compared to the other diastereoisomers and control CDNs, (Rp,Rp)‐10 elicited a significantly stronger activation response and sustained a longer duration of activation in THP1 cells. Following the activation of STING by CDNs, STING translocates from the endoplasmic reticulum (ER) to Golgi apparatus, whereas STING oligomers activate TBK1, which subsequently phosphorylates itself, STING, and IRF3 [41]. We examined the phosphorylation levels of STING, TBK1, and IRF3 in wild‐type, STING knockout, and cGAS knockout THP1 cells after drug treatment for 6 h (Figure 2G–J). The target specificity of the CDNs was confirmed by the observation that all three diastereoisomers effectively promoted robust phosphorylation of STING, TBK1, and IRF3 in wild‐type and cGAS knockout THP1 cells, but not in STING knockout cells. Furthermore, cells treated with 3′,3′‐c‐di‐dAMP and ADU‐S100 did not exhibit observable phosphorylation at this time point, which is consistent with their inability to induce STING activation prior to 12 h, as shown in Figure 2B,C. Prodrugs Effectively Promote the Maturation of Dendritic Cells STING activation initiates downstream gene transcription, leading to the production of type I interferons and pro‐inflammatory cytokines, such as CXCL10, IL‐6, and TNF‐α, in antigen‐presenting cells [40, 42]. The mRNA levels of IFN‐β, CXCL10, IL‐6, and TNF‐α were analyzed using RT‐qPCR in THP1 cells treated with prodrugs or control CDNs. Treatment with 1 µM of prodrugs for 4 h resulted in significantly higher mRNA levels of IFN‐β, CXCL10, IL‐6, and TNF‐α compared to 100 µM of 3′,3′‐di‐dAMP and 10 µM of ADU‐S100. The expression levels of IFN‐β, CXCL10, IL‐6, and TNF‐α induced by 1 µM (Rp,Rp)‐10 were approximately 12.5, 214.7, 19.2, and 46.6‐fold higher, respectively, compared to 10 µM of ADU‐S100 (Figure 3A–D). Prodrugs promoted dendritic cell maturation and induced macrophage repolarization. (A‐D) mRNA levels of IFN‐β, CXCL10, IL‐6, and TNF‐α in THP1 cells after 4 h treatment with 1 µM prodrug, 100 µM 3′,3′‐di‐dAMP or 10 µM ADU‐S100 were quantified by RT‐qPCR and normalized to DMSO. (E‐G) The secreted cytokines in the culture medium of BMDCs treated with 1 µM control and prodrugs for 24 h were quantified by ELISA. (H, I) The percentage of CD80+ CD86+ cells in drug‐treated BMDCs was analyzed by flow cytometry. (J–L) mRNA levels of IFN‐β, CXCL10, and IL‐6 in RAW264.7 cells treated with 1 µM control CDNs or prodrugs for 24 h were quantified by RT‐qPCR. (M) The expressions of CD86 in drug‐treated RAW264.7 cells were quantified by flow cytometry. (N‐O) The expressions of CD86 or CD206 in BMDM cells treated with control CDNs and prodrugs were analyzed by flow cytometry. Data points represent mean ± SD (n = 3). Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. We isolated bone marrow‐derived dendritic cells (BMDCs) from mice and treated them with 1 µM of either control or CDN prodrugs for 24 h. Treatment with all three diastereoisomers of prodrug 10 resulted in a 1.7 to 7.2‐fold increase in cytokine secretion levels. In contrast, ADU‐S100 did not influence the secretion of IFN‐β, TNF‐α, and IL‐6, while the parent CDN 3′,3′‐di‐dAMP exhibited only a mild increase under the tested conditions, as measured by ELISA (Figure 3E–G). Furthermore, all three diastereoisomers significantly increased the expression of costimulatory markers CD86 and CD80 in BMDCs, increasing their levels by 1.5 to 1.6‐fold and 1.4 to 1.6‐fold, respectively, compared to DMSO and ADU‐S100 (Figure 3H,I). These findings suggest that the prodrugs can effectively activate the STING signaling pathway in dendritic cells and promote the maturation of BMDCs. Prodrugs Induce Robust Macrophage Repolarization It has been reported that STING activation is capable of regulating macrophage polarization [43]. We examined the effects of STING activation induced by prodrugs in RAW 264.7 mouse mononuclear macrophage cells. The prodrugs induced a high level of production of IFN‐β, CXCL10, and IL‐6 at 6 h, which subsequently declined to baseline levels by 24 h. In contrast, ADU‐S100 and 3′,3′‐c‐di‐dAMP exhibited only minimal effects after 6 hours’ treatment, with a slight increase observed at 24 h (Figure 3J–L). This indicates that prodrugs can induce a rapid and robust stimulation of the STING pathway in macrophages. However, the duration of the stimulation in RAW 264.7 is significantly shorter than that in THP1 cells, suggesting differing regulatory mechanisms of the STING pathway in these cell lines. Notably, treatment with all three diastereoisomers of prodrug 10 resulted in a 2.9 to 4.0‐fold increase in CD86 expression compared to 3′,3′‐c‐di‐dAMP and ADU‐S100, indicating stronger activation of RAW 264.7 cells (Figure 3M). Next, we examined the effects of prodrugs on bone marrow‐derived macrophages (BMDMs) isolated from mice. BMDMs treated with all three diastereoisomers exhibited a significant increase in the surface expression of CD86, a well‐established marker of M1‐polarized macrophages (Figure 3N), along with a marked decrease in the surface expression of CD206, a classic hallmark of M2‐polarized macrophages, particularly in the case of (Rp,Rp)‐10 (Figure 3O). These results indicate that all three diastereoisomers possess the ability to stimulate the repolarization of macrophages from M2 to M1, thereby reducing immune suppression and contributing to the development of an antitumor microenvironment. Prodrugs Exhibit Significant Antitumor Effects When Administered Intratumorally Previous studies have reported that the dosage of ADU‐S100 in mouse tumor models ranges from 1 to 5 mg/kg [44, 45, 46, 47, 48, 49]. While lower dosages tend to produce fewer undesirable inflammatory effects, a dosage of 1 mg/kg often results in only a modest therapeutic impact [44]. We investigated the in vivo effects of prodrugs on cancer immunotherapy. We evaluated the therapeutic effects of the prodrugs in a CT26 mouse colon carcinoma model. On days 9, 12, and 15, prodrugs and control CDNs were administered intratumorally to BALB/c mice bearing CT26 tumors at a dosage of 0.5 mg/kg (Figure 4A). The prodrugs significantly suppressed tumor development in treated mice more efficiently (Figure 4B,C). In contrast to the 40% CR observed with ADU‐S100 and the 80% CR with the parent CDN 3′,3′‐c‐di‐dAMP, all three diastereoisomers of prodrug 10 exhibited a remarkable 100% CR rate, with all mice (n = 10) remaining tumor‐free at the study endpoint, and notably extended the survival period of the mice (Figures 4B, D and S5). Furthermore, the consistent increase in body weight among the mice indicated that all three diastereoisomers were safe and well tolerated (Figure S6). Intratumoral administration of prodrugs eliminates established CT26 tumors. (A) Illustration of tumor inoculation and drug administration. (B) Spider plots of individual tumor growth curves and (C) Mean tumor volume of mice treated with control CDNs or prodrugs. (D) Kaplan‐Meier survival curves of drug‐treated mice (n = 10). (E‐F) Expressions of MHCII+ in CD11b+ macrophages in CT26 tumor tissues were analyzed by flow cytometry (n = 4). (G) Immunostaining of CD8+, GZMB+, or apoptotic cells (TUNEL) in tumor tissues from drug‐treated mice. Tumor‐free survivors after (Rp,Rp)‐10 treated were rechallenged with CT26 tumors were monitored for tumor growth (H) and survival (I) (n = 5). Data points represent mean + SD in (C) and (H), or mean ± SD in (F). Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Survival curves in (D) and (I) were compared using log‐rank (Mantel‐Cox) test. In a separate study involving drug‐treated CT26 tumor‐bearing mice, we collected tumor tissues and analyzed macrophage and T cell activation. We observed an increase in M1 macrophage activation from 40% to 70% in the tumor tissue of mice treated with (Rp,Rp)‐10, which may explain its rapid response in tumor elimination (Figure 4E,F). In contrast to ADU‐S100 and 3′,3′‐c‐di‐dAMP, we found increased infiltration of CD8+ T cells in prodrug‐treated CT26 tumors (Figures 4G and S7). This was accompanied by a higher proportion of CD69+ in CD8+ T cells (Figure S8). Furthermore, we observed elevated levels of granzyme B and enhanced apoptosis induction, particularly in tumors treated with (Rp,Rp)‐10 (Figures 4G and S7), indicating that the intratumoral administration of (Rp,Rp)‐10 induced an effective phagocytic macrophage response and a cytotoxic T cell response for tumor elimination. To further evaluate the effects of immunological memory, mice that had survived after treatment with prodrugs 90 days after the initial CT26 tumor inoculation were subjected to a rechallenge with the CT26 tumor. Notably, 100% of the survivors from the (Rp,Rp)‐10 group exhibited resistance to the CT26 tumor rechallenge during the 60‐day observation period, in stark contrast to the rapid tumor growth observed in treatment‐naïve mice (Figure 4H, I). These findings suggest the development of a long‐lasting tumor‐specific immunological memory in mice treated with prodrugs. Prodrugs Effectively Suppress Tumor Growth When Intravenously Administrated In clinical trials, most STING agonists are injected directly into tumors due to their suboptimal drug‐like properties and rapid enzymatic degradation [15, 50]. However, metastatic tumors are not amenable to intratumoral injection. To address this issue, we evaluated the therapeutic effect of prodrugs delivered intravenously. BABL/c mice bearing CT26 tumors received intravenous treatment on days 9, 12, and 15 with 1 mg/kg of either prodrug, ADU‐S100, or 3′,3′‐c‐di‐dAMP (Figure 5A). (Rp,Rp)‐10 exhibited exceptional antitumor efficacy, significantly slowing tumor growth and extending animal survival to a 90% CR (Figure 5B,C). This response was superior to that of (Rp,Sp)‐10, which achieved an 80% CR rate, and (Sp,Sp)‐10, which had a 60% CR rate (Figure 5C). In contrast, treatment with ADU‐S100 and 3′,3′‐c‐di‐dAMP resulted in response rates of only 10% and 30%, respectively. Notably, in at least 60% of the mice, all three prodrugs administered intravenously significantly reduced the growth of CT26 tumors and eliminated established tumors (Figure 5D). Within 60 days, the survival rate in the (Rp,Rp)‐10 group increased dramatically to 90% (Figure 5E). Systemic administration of prodrugs eliminates established CT26 tumors. (A) Illustration of tumor inoculation and drug administration. (B) Representative photos of tumors from mice treated with control CDNs and prodrugs by intravenous administration on day 22. (C) Spider plots of individual tumor growth curves and (D) Mean tumor volume of mice treated with different drugs by intravenous administration. (E) Kaplan‐Meier survival curves and (F) The time‐course monitoring of body weight of the mice treated with different drugs by intravenous administration (n = 10). Data points represent mean + SD in (D) and (F). Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Survival curves in (E) were compared using log‐rank (Mantel‐Cox) test. Although the overall efficacy of prodrug 10 in intravenous therapy at this dosage is not as high as that observed with intratumoral administration, all three prodrugs were safe and well‐tolerated during intravenous therapy, as evidenced by the steady increase in body weight of the mice (Figure 5F). To assess the toxicity of the prodrugs, 1 mg/kg of control CDNs and prodrugs were administered intravenously to non‐tumor‐bearing mice. H&E staining, body weight monitoring, and blood biochemical analyses indicated no significant adverse effects of the prodrugs on the health of the mice (Table S4 and Figure S9). Prodrugs Induce Strong Immunogenic Tumor Cell Death As illustrated in Figure 6A–C, CT26 cells treated with (Rp,Rp)‐10 exhibited peak phosphorylation of STING, TBK1, and IRF3, along with increased production of IFN‐β and CXCL10, 2 to 4 h post‐treatment. This finding indicates that CDN prodrugs can activate the STING pathway in this tumor cell line, although the induced response is not as robust as that observed in antigen‐presenting cells. Prodrugs induce strong immunogenic tumor cell death. (A) Expression and phosphorylation levels of STING, TBK1, and IRF3 in CT26 cells treated with (Rp,Rp)‐10 for indicated time were assayed by western blot. (B, C) mRNA levels of IFN‐β and CXCL10 in CT26 cells treated with (Rp,Rp)‐10 for indicated time were quantified by RT‐qPCR. (D) ATP release, (E‐F) CRT expression, and (H) Membrane expression of HMGB1 in CT26 cells treated with control CDNs and prodrugs were assayed by immunostaining or flow cytometry. (G) CRT expression in CT26 tumor tissues from drug‐treated mice was assayed by immunostaining. (I) ROS level in CT26 cells treated with control CDNs and prodrugs was assayed by immunostaining. Data points in (D) and (F) represent mean ± SD (n = 3). Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. We conducted transcriptome analysis on CT26 cells and found that pathways upregulated by (Rp,Rp)‐10 treatment included the production of Type I interferon, autophagy regulation, the apoptotic signaling pathway, and antigen processing and presentation (Figure S10). Previous studies have suggested that STING activation in tumor cells may induce immunogenic cell death (ICD), which can trigger antigen‐presenting cells to initiate an adaptive immune response [51]. A hallmark of ICD is the release of damage‐associated molecular patterns (DAMPs), including ATP, HMGB1, and calreticulin (CRT), among others [52]. In fact, CT26 cells treated with (Rp,Rp)‐10 and (Rp,Sp)‐10 released more ATP into the culture medium (Figure 6D). Additionally, both flow cytometry analysis and confocal imaging of CT26 cell cultures and tumor tissues revealed significantly higher levels of CRT expression on the plasma membrane surface in prodrug‐treated CT26 cells as compared to control CDNs (Figure 6E–G). Confocal microscopy also demonstrated a dramatic decrease in intracellular HMGB1 levels in prodrug‐treated cells (Figure 6H), which is consistent with STING‐activation‐induced HMGB1 release from tumor cells. It has been reported that STING signaling in tumor cells can regulate the generation of reactive oxygen species (ROS) [53], and our transcriptome analysis corroborated this finding (Figure S10). In CT26 cells treated with prodrugs or control CDNs, the intracellular ROS production was assessed using the fluorescent probe DCFH‐DA. The (Rp,Rp)‐10 diastereoisomer induced a significantly higher level of intracellular ROS in CT26 cells compared to the other treatments, as shown in the confocal images (Figure 6I). Consequently, the diastereoisomers of prodrug 10 may elicit more potent immunogenic cell death, thereby contributing to the activation of APCs and the development of an antigen‐specific T cell response. Prodrugs Were Effective in Invoking an Antitumor Microenvironment The therapeutic efficacy of the prodrugs was subsequently evaluated using a tumor model of inadequately immunogenic B16‐F10 melanoma. After C57BL/6J mice were inoculated subcutaneously in the flank with 2×105 B16‐F10 cells, we intratumorally administered three doses of 1 mg/kg of prodrugs or control CDNs at three‐day intervals (Figure 7A). The (Rp,Rp)‐10 prodrug significantly delayed tumor growth, eliminated B16‐F10 tumors in 50% of the mice (n = 8), and prolonged the survival time of the mice. In contrast, ADU‐S100 and 3′,3′‐c‐di‐dAMP demonstrated CR rates of 0% and 10%, respectively (Figure 7B–D). All three diastereoisomers significantly extended the lifespan of the mice compared to the control CDNs (Figure 7E). Intratumoral administration of prodrugs eliminates established B16‐F10 tumors. (A) Illustration of tumor inoculation and drug administration. (B) Representative photos of B16‐F10 tumors on day 18 from mice treated with control CDNs and prodrugs by intratumoral administration. (C) Spider plots of individual tumor growth curves and (D) Mean tumor volume of mice treated with different drugs by intratumoral administration. (E) The survival period of the mice treated with different drugs by intratumoral administration. (E) Kaplan‐Meier survival curves and (F) The time‐course monitoring of body weight of the mice treated with different drugs by intratumoral administration (n = 10). Data points represent mean + SD in (D) and (F). Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Survival curves in (E) were compared using log‐rank (Mantel‐Cox) test. In comparison to the parent CDN 3′,3′‐c‐di‐dAMP and ADU‐S100, BMDCs treated with all three diastereoisomers of prodrug 10 demonstrated a significantly enhanced T‐cell response, as evidenced by increased IFN‐γ production and cytotoxicity of activated T cells in the in vitro assay (Figure S11). Additionally, we observed improved microenvironment remodeling in the CT26 model (Figure 4E–G). Consequently, we collected tumor tissues, TDLNs and spleens from treated mice 24 h after the third injection in the B16‐F10 model. All three prodrug‐treated mice exhibited smaller tumors, particularly those treated with (Rp,Rp)‐10 and (Rp,Sp)‐10 (Figure S12). The percentage of CD8+ CD11c+ DC in the TDLNs of treated mice increased from 12% in the vehicle group to 22% in the (Rp,Rp)‐10 group. In contract, ADU‐S100 and 3′,3′‐c‐di‐dAMP only increased to 13% and 15%, respectively (Figure S13). Furthermore, (Rp,Rp)‐10 induced greater activation of M1 macrophages compared to the other drugs, suggesting that (Rp,Rp)‐10 can stimulate a stronger antigen‐presenting cell response (Figure S14). Tumor tissues from the (Rp,Rp)‐10 group displayed a lower percentage of Treg (Foxp3+ CD4+ T cells) compared to the ADU‐S100 group, suggesting a less immunosuppressive T cell response (Figure 8A). The prodrug‐treated tumors exhibited a higher ratio of CD8+/CD4+ T cells, a decrease in the CD4+ T cell population, and an increase in the CD8+ T cell population (Figure 8B), which serves as a crucial indicator of the enhanced efficacy of immunotherapy. Furthermore, the percentages of activated CD4+ and CD8+ T cells, as indicated by CD69 expression, in both TDLNs and spleens were significantly higher in the prodrug‐treated mice than in those treated with ADU‐S100, indicating a stronger activation of T helper cells and cytotoxic T cells (Figure 8C–F). Collectively, these results strongly demonstrate that the CDN‐based prodrugs developed in this study promote the establishment of an immunogenic antitumor microenvironment. Prodrugs were effective in invoking antitumor T cell response. Tumors, spleens, TDLNs were collected from mice B16‐F10 tumor model 24 h after the third intratumoral administration of control CDNs or prodrugs. (A) Foxp3 expression in Treg cells in tumor tissues (n = 3), (B) CD4+ or CD8+ T cells in tumor tissues (n = 3), (C) CD69 expression in CD4+ T cells in spleens (n = 6), (D) CD69 expression in CD8+ T cells in spleen (n = 6), (E) CD69 expression in CD4+ T cells in TDLNs (n = 6), and (F) CD69 expression in CD8+ T cells in TDLNs (n = 6), were analyzed by flow cytometry. Data points represent mean ± SD. Statistical significance was determined by one‐way analysis of variance (ANOVA). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Conclusion In summary, we synthesized all three diastereoisomers (Rp,Rp), (Sp,Sp), (Rp,Sp) of a deoxyribose CDN prodrug containing an alkynyl group and showed that the introduction of a phosphotriester significantly enhanced the prodrug resistance to phosphodiesterase and masked the negative charge of the phosphodiester group, thereby improving cellular uptake efficiency. Cellular assays showed that all diastereoisomers outperformed ADU‐S100 and the parent CDN in terms of cellular uptake and the release of the parent CDN to trigger STING phosphorylation. In THP1, BMDC, and RAW264.7 cells, all three diastereoisomeric prodrugs exhibited significantly higher STING activation, as evidenced by the strong expression of IFN‐β, CXCL10, IL‐6, and TNF‐α, compared to ADU‐S100 and the parent dCDN. Among them, the (Rp,Rp)‐10 diastereoisomer demonstrated the most robust STING activation. In mouse models, the diastereoisomers of prodrug 10, particularly (Rp,Rp)‐10, promoted the recruitment and maturing of dendritic cells, macrophage polarization, and immunogenic cell death within tumors. Furthermore, (Rp,Rp)‐10 also stimulated Treg cell suppression, antigen‐specific cytotoxic T cell generation, and the development of immunological memory, contributing to reshape the TME in anticancer immunotherapy. Therefore, considering the effect of the phosphotriester chirality will be highly beneficial for developing more effective, durable, and safer STING agonist in the future. It is noteworthy that these phosphotriester prodrugs have been shown to be safe, effective, and well tolerated when administered intratumorally and systemically in mouse models, which is beneficial for the development of STING agonist‐based monotherapies or combination drug therapies. Author Contributions Zhiqiang Xie: conceptualization, methodology, investigation, validation, formal analysis, visualization, writing – original draft, writing – review and editing, software, data curation. Huimin Liu: conceptualization, methodology, software, formal analysis, validation, investigation, visualization, writing – original draft, writing – review and editing, data curation. Yuchen Yang: investigation. Xiaohan Sun: investigation. Jingwen Wang: investigation. Tianxiang Liu: software. Baifan Wang: software. Wenrui Qiu: investigation. Yiwen Sun: investigation. Yuting Bai: investigation. Yao Zhang: investigation. Dejun Ma: supervision. Rong Xiang: supervision. Shibo Li: supervision, funding acquisition, writing – original draft, writing – review and editing, project administration, resources, conceptualization. Zhen Xi: conceptualization, supervision, resources, project administration, funding acquisition, writing – original draft, writing – review and editing. Conflicts of Interest The authors declare no conflicts of interest. Supporting information The authors have cited additional references within the Supporting Information [54].Supporting File: anie73280‐sup‐0001‐SuppMat.pdf. Acknowledgments This work was supported by National Key Research and Development Program of China (Grant No. 2023YFD1700502), National Natural Science Foundation of China (Grant No. 32270579), Frontiers Science Center for New Organic Matter, Nankai University (Grant No. 63181206), the Haihe Laboratory of Sustainable Chemical Transformations (Grant No. YYJC202102). Contributor Information Shibo Li, Email: lishibo@tjnu.edu.cn. Zhen Xi, Email: zhenxi@nankai.edu.cn. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. References Associated Data Supplementary Materials The authors have cited additional references within the Supporting Information [54].Supporting File: anie73280‐sup‐0001‐SuppMat.pdf. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.