SN38 was bought from Tokyo Chemical Industry (TCI) Co., Ltd (Tokyo, Japan). Anti-OX40 antibody was purchased from Bio X Cell (NH, USA). Rat IgG (Catalog: SP032) and Cell Counting Kit-8 (CCK-8) were bought from Solarbio Science & Technology Co., Ltd (Beijing, China). The 4-arm PEG-ONH₂ (Mw = 10 kDa) was purchased from Jenkem Technology Co., Ltd (Beijing, China). ELISA kits (IFN-β, IL-6, INF-γ, TNF-α) were purchased from Elabscience Biotechnology Co., Ltd (Wuhan, China). Antibodies used for Western Blot were bought from Cell Signaling Technology Co., Ltd (MA, USA). BCA protein assay kit was obtained from Thermal Fisher Scientific Co., Ltd (MA, USA). GM-CSF and IL-4 were obtained from Beyotime Biotechnology Co., Ltd (Shanghai, China). All antibodies used for flow cytometry were obtained from BD Biosciences Co., Ltd (San Jose, CA, USA) and BioLegend Co., Ltd (San Diego, CA, USA).

The murine breast cancer cells (E0771 and 4T1) were obtained from BeNa Culture Collection (Beijing, China). Bone marrow derived dendritic cells (BMDCs) were harvested according to Son’s method (29). All the cells incubated at 37°C in culture medium at an atmosphere of 5% CO₂.

Female C57BL/6 and BALB/c mice were bought from Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). Female SD rats were obtained from Liaoning Changsheng biotechnology Co., Ltd (Liaoning, China). E0771 cells (1 × 10⁶ cells per mouse) were injected subcutaneously into the right flank of female C57BL/6 mice (6–8 weeks) to generate the subcutaneous E0771 tumor mouse model.

The SiGel was prepared by cross-linking 4-arm PEG-ONH₂ and ODEX through oxmide bonds at a weight ratio of 6% (w/w) based on our previously reported methods (14, 15). 400 μg SN38 was dissolved in DMSO (10 μL), and the solution was added to 4-arm PEG-ONH₂ solution. Then the obtained hydrogel was named as SiGel@SN38. Similarly, aOX40 (15 μg) in solution was solubilized in 4-arm PEG-ONH₂ solution to obtain SiGel@aOX40. The combinations of SN38 and aOX40 was designated as SiGel@SN38/aOX40.

The morphological characteristics of the lyophilized SiGel was assessed by scanning electron microscope (SEM, JSM-7000F, JEOL Ltd, Tokyo, Japan). Rheological analysis was performed according to our previously reported test methodology (15).

The safety of the SiGel was evaluated by randomly assigning female C57BL/6 mice to two groups: the SiGel implant group and the PBS (surgery-only) control group. Peripheral blood samples were collected on days three and seven post-surgery. Serum IL-6 levels and complete blood counts were measured using Elisa assay kits, following the manufacturer’s instructions.

To evaluate in vivo degradation, female C57BL/6 mice were subcutaneously injected with SiGel into their flanks. Mice were euthanized at predetermined time points (7, 14, and 21 days), and the remaining SiGel samples were collected and documented through imaging.

For the in vitro degradation study, SiGel was immersed in PBS buffer (pH 7.4) and incubated at 37°C with continuous shaking at 90 rpm. The residual weight of SiGel was measured at predetermined time points.

SiGel loaded with SN38 or IgG-Cy5 (a fluorescent analog representing the aOX40 antibody) was incubated in 2 mL of PBS (pH 7.4) at 37°C with constant shaking at 90 rpm. At predetermined intervals, the release medium was collected and replaced with fresh PBS. The concentration of SN38 was quantified using UV-Vis spectroscopy at 378 nm, while fluorescence spectroscopy ex = 649 nm, em = 670 nm) was used to measure the release of IgG-Cy5.

The cytotoxic effects of free SN38 and SiGel@SN38 on E0771 cells, 4T1 cells, and BMDCs were assessed in vitro using CCK-8 assay kits. Briefly, tumor cells (4000 cells per well) and BMDCs (15000 cells per well) were seeded in 96-well plates and incubated overnight. Then the medium was removed and replaced with new media containing medicines. Following a 12-hour or 48-hour incubation, 20 μL of CCK-8 solution was added to each well. After a 2-hour incubation, absorbance was measured using a TECAN microplate reader. Cell viability (%) was calculated as the percentage of treated cells relative to the untreated control group.

Female C57BL/6 mice bearing E0771 tumors (approximately 200–300 mm³) were prepared as described. In order to simulate the clinical state of incomplete tumor resection, approximately 90% of the tumor volume was surgically removed. The mice were then randomly assigned to 6 groups: untreated control (G1), SiGel (hydrogel without drug, G2), SiGel@SN38 (G3), SiGel@aOX40 (G4), Soluble@SN38/aOX40 (G5), and SiGel@SN38/aOX40 (G6). Each treatment was administered via injection into the tumor resection cavity. Body weight and residual tumor volume were continuously monitored.

E0771 or 4T1 tumor cells were cultured in RPMI-1640 (containing 5 μM SN38) for 48 hours. The supernatant from these drug-pretreated tumor cells, designated as conditioned media (CM), was collected after centrifuging at 12,000 rpm for 5 minutes. BMDCs were then incubated for 12 hours in a mixture of 50% CM and 50% fresh RPMI-1640 media. Following incubation, the BMDC supernatant was gathered, centrifuged, and analyzed for cytokine content (IFN-β and IL-6) using ELISA kits.

The mice in each of the six groups had peripheral blood drawn at the end of the treatment. TNF-α and IFN-γ cytokine levels in serum and tumor were measured using ELISA kits.

Tumors from the mice in each of the six groups were gathered at the ending of the trial. The tumors were cut into small pieces and lysed with tumor dissociation buffer (containing collagenase type IV, hyaluronidase, and DNase I). Then the supernatant was collected, filtered through a 300-mesh nylon filter, centrifuged and resuspended. Finally, fluorophore-conjugated antibodies were used to stain the resulting cell suspensions for 40 minutes on ice. A BD FACS Celesta flow cytometer was used to test the samples, and FlowJo software was used to analyze the results.

The supernatant from E0771 cells was collected as previously described and applied to treat BMDCs for 6 hours. Following this treatment, BMDCs were harvested and rinsed with 0.9% NaCl. RIPA lysis buffer was then added, and the samples were incubated on ice for 30 minutes. Proteins were extracted by centrifugation at 12,000 rpm for 5 minutes, and protein concentration was determined using a BCA protein assay kit. The extracted proteins were combined with loading buffer and heated to 100°C for 10 minutes. Protein separation was performed by SDS-PAGE using a Bio-Rad Mini gel electrophoresis system (Bio-Rad, USA), and the proteins were subsequently transferred onto PVDF membranes. Membranes were incubated overnight with specific primary antibodies on a shaker at 4°C. After washing the PVDF membranes five times, they were incubated with HRP-conjugated secondary antibodies for 1 hour. Finally, Western blot images were captured.

The mean ± standard deviation (S.D.) was used to display all data. Two treated groups were compared using two-tailed unpaired Student’s t-tests. P < 0.05 was used to test for significance.

Based on our previous research, we synthesized SiGel by cross-linking 4-arm polyethylene glycol hydroxylamine (4-arm PEG-ONH₂) and oxidized dextran (ODEX) at a 2:1 ratio, maintaining a solid content of 6% to ensure optimal injectability, synchronized degradation, and controlled release of the loaded agents (15, 30). Under these conditions, SiGel exhibited a storage modulus (G’) exceeding 6000 Pa and a loss modulus (G’’) of 35.8 Pa. Additionally, G’ decreased below G′′ at a strain of 370%, indicating the disruption of the hydrogel network at high strain ( Figures 1A, B ). The gelation properties of SiGel were confirmed via the tube inversion test ( Figure 1C ). Additionally, the SiGel also exhibited good shear-thinning characteristics in vitro and could be constantly injected without clogging by using a 26G needle ( Supplementary Figure 1 ). SEM images exhibited that the lyophilized SiGel had a porous and interconnected structure with a diameter of ≈10–30 µm ( Figure 1D ).

We next investigated the injectability of the SiGel in vivo. SiGel was subcutaneously injected into female BALB/c mice. As shown in Figure 1E , the injected hydrogel quickly formed a bolus upon injection. Furthermore, the SiGel could firmly adhered to the surface of the rat stomach after injection, ensuring the SiGel firmly immobilized at the surgical site ( Figure 1F ).

As depicted in Figures 1G , H , the SiGel experience a swelling at first and degraded gradually for over 15 days. For in vivo degradation assessment, SiGel was subcutaneously injected into the flanks of female C57BL/6 mice. At predetermined intervals (3, 9, 15, and 18 days), the mice were sacrificed, and the residual SiGel samples were collected and photographed. As shown in Figure 1I , J , in vivo degradation continued for > 18 days, indicating the great biodegradability of the SiGel. The sustained-release properties of the prepared SiGel ensure the sustained release of loaded drugs.

Due to its gradual degradation and diffusion mechanism, 89.8% of encapsulated SN38 and 92% of IgG-Cy5 were released in a sustained manner over 18 days ( Figures 1K, L ). The extended and sustained release of SN38 and aOX40 within the surgical area is expected to maximize synergistic antitumor efficacy.

The safety of prepared SiGel is crucial for clinical application. As shown in Figures 1M , N , no significant alterations were observed in white blood cell counts or serum IL-6 levels, indicating a favorable biocompatibility profile. The biosafety of the SiGel was further confirmed through in vitro cytotoxicity assay. As shown in Supplementary Figure 2 , no obvious cytotoxicity was observed in murine 3T3 fibroblasts and human coronary artery endothelial cells (HCAECs) after 48-hour incubation with a high SiGel concentration (40mg/mL), demonstrating excellent cytocompatibility. Additionally, the SiGel-treated mice exhibited no signs of weight loss compared to the control group ( Supplementary Figure 3 ).

First, we determined the cytotoxicity of blank SiGel (hydrogel without drug), free SN38 and SiGel@SN38 (SiGel loaded with SN38) to E0771 cells, 4T1 cells and BMDCs using CCK-8 assay. As expected, blank SiGel exhibited no significant cytotoxicity toward E0771 cells, 4T1 cells and BMDCs, even at high concentration (40mg/mL) ( Supplementary Figure 4 ). In contrast, both free SN38 and SiGel@SN38 displayed dose-dependent cytotoxicity against E0771 cells, 4T1 cells and BMDCs ( Figures 2A–C ). SN38 displayed stronger tumor cell inhibiting effects on E0771 cells than 4T1 cells. Intriguingly, SiGel@SN38 showed a weaker cytotoxic effect than free SN38, which may attribute to the gradual release of SN38 from hydrogel ( Figures 2A–C ).

Dendritic cells (DCs) exert a vital role in stimulating innate and adaptive immune responses (31). To assess the capacity of SiGel@SN38 to activate the STING pathway, we evaluated its effects on type I interferon (IFN) and pro-inflammatory cytokine secretion. STING activation induces type I IFNs and other pro-inflammatory cytokines via phosphorylation of tank-binding kinase 1 (TBK1) and interferon regulatory factor 3 (IRF3), thereby promoting DC activation, antigen presentation, and cytotoxic T-cell recruitment, ultimately enhancing antitumor immunity (32, 33). ELISA assays were conducted to measure IFN-β and IL-6 secretion following SiGel@SN38 treatment ( Figure 2D ). As shown in Figures 2E–H , both free SN38 and SiGel@SN38 treatment significantly increased IFN-β and IL-6 secretion. Western blot analysis further confirmed the activation of STING-related signaling pathways, showing increased phosphorylation of STING, IRF3, and TBK1 within 4 hours of SN38 or SiGel@SN38-conditioned medium (CM) treatment ( Supplementary Figure 5 ). These results presented here confirmed that SiGel@SN38 effectively activates the STING pathway. Next, we further evaluated the ability of SiGel@SN38 to promote DCs maturation and activation using flow cytometry. As shown in Figures 2I–K , BMDCs treated with CM from SiGel@SN38-treated E0771 cells exhibited a significant increase in the expression of the stimulatory molecules CD80 and CD86. Collectively, these findings demonstrate that the sustained release of SN38 from SiGel effectively stimulates the STING pathway and induces a robust antitumor immune response in vitro.

In order to simulate the clinical state of incomplete tumor resection, we established an incomplete tumor resection model by excising ~90% tumor volume when E0771 tumor on the female C57BL/6 mice reached a volume of 200–300 mm³ ( Figure 3A ). The mice were then randomly divided into 6 groups: untreated (G1), SiGel (hydrogel without drug, G2), SiGel@SN38 (G3), SiGel@aOX40 (G4), Soluble@SN38/aOX40 (G5), and SiGel@SN38/aOX40 (G6), with corresponding drug injected into the tumour resection cavity, respectively. As shown in Figures 3B–D , rapid tumor relapse was observed in the untreated (control) group, with a median survival of only 14 days post-surgery. Consistent with the cytotoxicity results, SiGel alone did not show therapeutic effects and the median postoperative survival time was similar to that of untreated group. SiGel@aOX40 treatment exhibited minimal therapeutic effect and fail to supress tumor recurrence and improve postoperative outcome. SiGel@SN38 treatment showed moderate therapeutic effect and partially suppressed tumor growth, with a prolonged postoperative length of 32 days. Notably, SiGel@SN38/aOX40 dramatically suppressed the tumor growth and resulted in complete tumor eradication, with no tumor recurrence over the 120 days surveillance period. Intriguingly, the therapeutic effects of Soluble@SN38/aOX40 were not long-lasting and simply delay the time of the tumor relapse. Ultimately, all mice in the Soluble@SN38/aOX40 group relapsed, demonstrating the importance of encapsulating pharmaceuticals into hydrogels. These results indicated that our designed SiGel and loaded drug combination is necessary for trigging robust and durable tumoricidal immunity following breast cancer surgical resection.

Owing to the trauma caused by surgery and anesthesia, almost all treated mice experienced a body weight loss (≈ 15%) after surgery ( Figure 3E ). Then the treated mice across the groups soon returned to their initial body weight, indicating the safety of these treatments. In addition, AST, ALT, and BUN levels were measured to further assess the system toxicity of various treatments. As shown in Figures 3F , G , no significant liver or kidney dysfunction was observed in treated mice compared to healthy controls. These results demonstrated that our hydrogel-based treatment exhibits a favorable safety profile with minimal and controllable side effects.

To explore the potential mechanisms of the therapeutic effects after various treatments, another subcutaneous E0771 breast cancer model was used to examine the tumor immune microenvironment. In this model, only half of the tumor volume was resected to preserve sufficient tissue for analysis of infiltrating immune cells via flow cytometry at the end of the treatment. Activation of STING pathway could induce DCs maturation and activation, thereby promoting the recruitment of cytotoxic T cells from tumor draining lymph nodes. As expected, the proportion of mature DCs in the SiGel@SN38 and SiGel@SN38/aOX40 groups significantly increased ( Figure 4A ). Of note, the proportion of mature DCs was lower in the Soluble@SN38/aOX40 group, highlighting the necessity of sustained, in situ drug release from the hydrogel. Furthermore, SiGel@SN38/aOX40 treatment displayed the highest proportions of CD4 ⁺ and CD8 ⁺ T cells in comparison with other groups ( Figures 4B, C ). The secretion level of TNF-α and IFN-γ in serum and tumor were detected. As shown in Figures 4D , E , SiGel@SN38/aOX40 treatment significantly enhanced antitumor immunity by elevating pro-inflammatory cytokines. Extensive research has shown that tumor associated macrophages (TAM), the main tumor-infiltrating cells, play pivotal roles in tumor progression. Macrophages can differentiate into M1 phenotype or M2 phenotype. M2 phenotype exhibit pro-tumorigenic activities, whereas M1 phenotype play critical roles in antigen presentation and exert anti-tumorigenic function (34, 35). Reducing M2 macrophages or increasing M1 macrophages is crucial for enhancing T cell-mediated immunity (36). As shown in Figure 4F , SiGel@SN38/aOX40 treatment significantly increased M1 macrophages and reduced M2 macrophages within the tumor, indicating a reprogramming of the tumor immune microenvironment toward immune activation. These results evidenced that SiGel@SN38/aOX40 treatment effectively stimulates strong antitumor immunity in vivo.