Superantigen-fused T cell engagers for tumor antigen-mediated robust T cell activation and tumor cell killing

Abstract

Inadequate T cell activation has severely limited the success of T cell engager (TCE) therapy, especially in solid tumors. Enhancing T cell activity while maintaining the tumor specificity of TCEs is the key to improving their clinical efficacy. However, currently, there needs to be more effective strategies in clinical practice. Here, we design novel superantigen-fused TCEs that display robust tumor antigen-mediated T cell activation effects. These innovative drugs are not only armed with the powerful T cell activation ability of superantigens but also retain the dependence of TCEs on tumor antigens, realizing the ingenious combination of the advantages of two existing drugs. Superantigen-fused TCEs have been preliminarily proven to have good (>30-fold more potent) and specific (>25-fold more potent) antitumor activity in vitro and in vivo. Surprisingly, they can also induce the activation of T cell chemotaxis signals, which may promote T cell infiltration and further provide an additional guarantee for improving TCE efficacy in solid tumors. Overall, this proof-of-concept provides a potential strategy for improving the clinical efficacy of TCEs.

Graphical abstract

Zhou and colleagues significantly enhance the activation capability of TCEs toward T cells through SEA fusion, thereby increasing the activity of TCEs by tens of times while retaining its high tumor antigen specificity.

Introduction

T cell engagers (TCEs) can recruit T cells to eliminate tumors independently of pre-existing tumor-specific T cells, which has proven clinical efficacy in treating some hematologic malignancies.^1,2,3 However, the response rates of TCEs have been variable. Durable responses are typically seen in only a subset of patients.^4,5 T cells are the core effector cells for TCE activity. The activity of T cells is recognized as one of the most critical factors of TCE efficacy. In a recent study, Friedrich et al.⁶ performed comprehensive longitudinal profiling of T cells in patients undergoing BCMA×CD3 TCE therapy. They found that the clinical effect of TCEs was mainly driven by CD8⁺ T cells, while the abundance of exhausted-like T cells was highly associated with clinical response failure. In addition, solid tumors often have a more complex and varied microenvironment, which may result in a more severe inhibition of T cell function, leading to lower response rates of TCEs in these types of tumors.^5,7

Simply bringing two cells into proximity is insufficient to induce T cells to kill tumor cells. In contrast, effective activation of T cells after crosslinking with tumor cells is necessary for tumor killing.^8,9 Although TCEs can provide a primary signal for T cell activation by activating CD3, activation and proliferation are further promoted by additional activation signals. CD3 activation without costimulation signals may also lead to T cell dysfunction or rapid exhaustion.^10,11 To overcome these issues, costimulatory molecules, such as 4-1BB and CD28, were targeted by agonists to optimize the efficacy of TCEs.^12,13,14,15 Wu et al.¹⁴ assembled CD38×CD3 TCEs and CD28 agonists into a trispecific antibody, showing a significant potency advantage over CD38×CD3 TCEs. In addition, immune checkpoint inhibitors are also commonly used to release T cell activity and promote TCE efficacy.^16,17 Nevertheless, costimulatory molecule agonists and immune checkpoint inhibitors have issues with low clinical efficacy and unclear efficacy.^18,19 Whether these strategies can benefit patients more from TCEs remains a concern, as they are still in preclinical research.

Overall, exploring potential T cell activation strategies that can enhance the efficacy of TCEs to expand their clinical application is still necessary. Staphylococcal enterotoxins (SEs) are classical superantigens that have been used clinically as supplementary therapeutic agents for tumor treatment for many years due to their powerful T cell activation ability.^20,21 Our previous study demonstrated that SE-activated T cells exhibited robust tumor cell lysis ability in response to TCEs in vitro. However, it is unknown whether the specificity and in vivo activity of TCEs against tumors were influenced by SEs.²² To further explore the synergistic potential of SEs on TCEs, we used SEA,²³ a subtype of SEs, to design a novel SEA-fused TCE to more closely correlate the activity of SEs and TCEs. Activity and mechanism studies indicated that the SEA-fused TCE vigorously activated T cells to target and kill tumors and this strong antitumor effect was highly dependent on tumor antigens. Moreover, the SEA-mediated TCE activity enhancement strategy was feasible for different targets, which may not lead to decreased TCE indications.

Results

SEA-fused TCEs showed potent antigen-dependent antitumor activity in vitro

Tumor-associated antigens (e.g., HER2 and EGFR) are often expressed in some normal cells, which can cause “on-target off-tumor toxicity,” thus complicating the evaluation of the specificity of SEA-fused TCEs. Therefore, we selected highly tumor-specific NY-ESO-1 as the target. NY-ESO-1 is a type of cancer-testis antigen expressed in various types of cancer, but not in normal tissues except for the testis and placenta.^24,25 Due to the lack of human leukocyte antigen (HLA) expression in the testis and placenta, the HLA-A∗02:01-presented NY-ESO-1 peptide (SLLMWITQC, SLL/A02) is theoretically expressed only in tumor cells.²⁶ Variable domain sequences of 1G4113 (a T cell receptor [TCR] targeting SLL/A02, called tNY here)²⁷ and HXR32 (an antibody targeting CD3ε, called aCD3 here)²⁸ were used to design a SEA-fused TCE (tNY-aCD3/SEA) (Figure 1A). The isotype controls of tNY-aCD3/SEA were engineered by replacing tNY with an irrelevant TCR targeting HLA-A∗02:01-presented TP53 peptide (tCtrl-aCD3/SEA),²⁹ replacing SEA with an irrelevant antibody targeting Claudin18.2 (tNY-aCD3/aCtrl),³⁰ or deleting aCD3 (tNY/SEA). Heterodimeric Fc was assembled through the knob-into-hole method,³¹ and L234A-L235A-P329G mutations in Fc were introduced to eliminate unwanted FcγR-mediated immune effector functions.³² tNY-aCD3/SEA and its controls were successfully produced using standard mammalian expression systems and purified using affinity chromatography (Figure S1A; Tables S1 and S2). After purification, the binding activity of tNY-aCD3/SEA and its controls was evaluated using assays that measured their ability to bind to SLL/A02 or CD3ε. This evaluation is important to confirm that tNY and aCD3 in tNY-aCD3/SEA or its controls can specifically recognize their targets. In contrast, the controls without tNY or aCD3 lost the ability to bind SLL/A02 or Jurkat cells, respectively (Figures 1B, 1C, and S1B).

Figure 1.

Design and in vitro antitumor activity of SEA-fused TCEs

(A) The configuration of tNY-aCD3/SEA. HXR32 is an anti-CD3ε scFv, SEA was fused to the N terminus of one chain of the Fc arms, and the β and α chain of SLL/A02 targeted TCR was fused to the N terminus of CH1 and CL, respectively. (B) Binding affinity of tNY-aCD3/SEA and its isotype controls to SLL/A02. Analyzed by ELISA (n = 1). (C) Binding affinity of tNY-aCD3/SEA and its isotype controls to Jurkat cells (CD3ε⁺). Analyzed by flow cytometry (n = 1). (D) Cell morphology (400×) of A375-SLL_H cells after coincubation with PBMCs (donor 1) and 10 nM tNY-aCD3/SEA or tNY-aCD3/aCtrl for 72 h. A375-SLL_H cells, black arrow; PBMCs, purple arrow; T cells accumulated and expanded after activation, white arrow. (E–G) T cell-dependent cytotoxicity (donor 2) of tNY-aCD3/SEA and its isotype controls on tumor cells. tNY-aCD3/SEA or its isotype controls redirected PBMCs to lysis A375-SLL_H cells in vitro (E). Comparison of antitumor activity of tNY-aCD3/SEA on homologous tumor cells with different antigen expression levels (F and G). The cytotoxicity was indicated by the degree of tumor cell apoptosis. Apoptotic tumor cells were labeled with Annexin V-633 and PI, and then detected by flow cytometry (n = 3). The statistical differences are shown next to the data points, and the statistical differences with different groups are color coded accordingly. ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To quickly estimate whether SEA can enhance the potency of TCEs, we performed a microscopic examination of T cell proliferation and tumor cell survival after treatment with tNY-aCD3/SEA or tNY-aCD3/aCtrl. Incubation of peripheral blood mononuclear cells (PBMCs) and A375-SLL_H cells (SLL/A02⁺) with tNY-aCD3/SEA-stimulated T cell proliferation to a large extent. It led to the lysis of almost all the tumor cells. In contrast, incubation with tNY-aCD3/aCtrl did not significantly increase T cell numbers and resulted in less tumor cell lysis, which suggested that SEA maintained its superantigen activity in this fusion protein, leading to the improved antitumor activity of TCEs (Figure 1D). Then, we performed in vitro cytotoxicity assays to determine the activity and specificity of tNY-aCD3/SEA (Video S1). tNY-aCD3/SEA (∼40 pM EC₅₀) was more than 30-fold more potent than tNY-aCD3/aCtrl (>1,200 pM EC₅₀) against A375-SLL_H cells (Figure 1E), indicating that SEA markedly enhanced the potency of TCEs. In addition, tNY-aCD3/SEA was more than 25-fold more potent than tCtrl-aCD3/SEA (>1,000 pM EC₅₀) and more than 19-fold more potent than tNY/SEA (>780 pM EC₅₀) against A375-SLL_H cells (Figure 1E), indicating that the antitumor activity of tNY-aCD3/SEA depended to a certain extent on the tumor antigen and CD3. Furthermore, the superior potency of tNY-aCD3/SEA compared with its control was also verified in different tumors (Figures S2A and S2C). These findings suggest that tNY-aCD3/SEA has potent antitumor activity and that each component of the fusion protein contributes to its effectiveness.

Video S1. The antitumor activity and specificity of tNY-aCD3/SEA against A375-SLLSLL_H cells

Morphology (400×) of A375-SLLSLL_H cells (GFP⁺) after co coincubation with PBMCs (donor 1) and tNY-aCD3/SEA or its isotype controls for different times (400×).

We further explored the tumor specificity of tNY-aCD3/SEA to preliminarily evaluate the safety hazards caused by SEA and CD3-induced non-specific T cell activation. We found that the ability of tNY-aCD3/SEA to redirect PBMCs to kill homologous tumor cells was significantly positively correlated with the expression level of SLL/A02 on the tumor cell surface (Figures 1F and 1G; Table S3). Although the efficacy of tNY-aCD3/SEA against A375-SLL_H cells (∼40 pM EC₅₀) with high SLL/A02 expression was much higher than that against A375 cells (∼1,634 pM EC₅₀) with deficient SLL/A02 expression, and the efficacy of tNY-aCD3/SEA against SLL/A02-positive K562-SLL cells (∼0.6 nM EC₅₀) was also much higher than that against SLL/A02-negative K562-Ctrl cells (∼10.3 nM EC₅₀), tNY-aCD3/SEA at concentrations of 1 nM and above may exhibit antigen-independent antitumor activity on tumor cells (Figures 1F and 1G). To enhance the dependence of tNY-aCD3/SEA potency on tumor antigens, we attempted to optimize the superantigen activity of SEA. According to previous research, we mutated aspartic acid at position 227 of SEA to alanine (D227A) to generate a new SEA-fused TCE called tNY-aCD3/D227A,³³ as well as its controls (Figures S3A–S3D; Table S2). The D227A mutation may affect the ability of SEA to interact with its target, potentially reducing its superantigen activity without completely losing it. However, SEA with the D227A mutation slightly increased TCE-induced tumor apoptosis (Figure S4B; Video S2), and its ability to enhance TCE-induced T cell cytotoxicity to tumors was limited (Figure S4A). Thus, exploring moderate activity SEA variants may be a promising approach for developing novel SEA-fused TCEs with improved efficacy and safety profiles.

Video S2. The antitumor activity and specificity of tNY-aCD3/D227A against A375-SLLSLL_H cells

Morphology (400×) of A375-SLLSLL_H cells (GFP⁺) after coincubation with PBMCs (donor 1) and tNY-aCD3/D227A or its isotype controls for different times (400×).

SEA can efficiently enhance the ability of TCEs to activate T cells

Indeed, the activation level of T cells is directly related to the activity of TCEs. Therefore, assaying tNY-aCD3/SEA and its controls for their ability to activate T cells is an essential step in evaluating the potential of SEA to enhance TCE activity. Measuring the expression of activation markers on T cells is a commonly used assay for T cell activation. As shown in Figure 2A, tNY-aCD3/SEA significantly increased CD69 (early marker) and CD25 (intermediate or late marker) expression on T cells in a dose-dependent manner and more efficiently than its controls when PBMCs were cocultured with A375-SLL_H cells for 48 h. The effect of SEA on the enhancement of TCE-induced T cell activation was alleviated at high concentrations of tNY-aCD3/SEA, as measured by the proportion of activated T cells. However, this phenomenon was not apparent when measured by the CD25 expression level of single T cells (Figures 2A and S5A). Increased expression of CD25, also known as the IL-2 receptor α chain, potentially enhances the responsiveness of T cells to IL-2 and promotes further activation and proliferation of T cells.³⁴ In addition, the proportion of activated T cells after treatment with tNY-aCD3/SEA or its controls for 72 h (Figure S5A) was similar to that at 48 h. It seems that T cell activation may have reached a plateau by 48 h. However, the CD25 expression level of T cells was still slightly upregulated between 48 and 72 h (Figures 2A and S5A), which suggested that individual T cells may still be undergoing further activation and proliferation. Superantigens activate T cells by linking the TCR to major histocompatibility complex class II (MHC class II), which presents antigens to CD4⁺ T cells, but CD8⁺ T cells are currently accepted as the primary effector cells of TCEs. Therefore, we investigated whether there was a preference for the activation of CD4⁺ T cells by tNY-aCD3/SEA, and found that the activation of CD8⁺ T cells was consistent with that of total T cells (Figures S5B and S5C), indicating unbiased activation of CD4⁺ and CD8⁺ T cells by tNY-aCD3/SEA. After activation, T cells undergo a process of clonal expansion, resulting in the generation of a large number of daughter cells. T cell proliferation was observed after incubation with tNY-aCD3/SEA and A375-SLL_H cells for 72 h. It became more apparent at 96 h (Figures 2B and 2C), which was consistent with the expected timeline of T cell activation and proliferation. The proliferation of T cells increased significantly and continuously within 148 h (Figures 2D and S5D), which was beneficial for T cells to execute tumor-killing activity lastingly. Unlike the undifferentiated T cell activation observed above, lower proliferation of CD8⁺ T cells compared with total T cells induced by lower concentrations of tNY-aCD3/SEA was observed (Figure S5E). Finally, we detected a significant increase in T cells expressing the anti-apoptotic protein Bcl after treatment with tNY-aCD3/SEA (Figure S5F), which indicated that the sustained expansion of T cells induced by tNY-aCD3/SEA may be due in part to the increased expression of Bcl in these cells. The controls had specific effects on the activation and proliferation of T cells. However, they were significantly weaker than those of tNY-aCD3/SEA, indicating that the combination of all components of tNY-aCD3/SEA led to a synergistic effect on T cell activation and proliferation (Figures 2 and S5).

Figure 2.

T cell activation and proliferation in response to the stimulation of tNY-aCD3/SEA or its isotype controls

(A) T cell (CD3⁺, donor 1) activation mediated by tNY-aCD3/SEA or its isotype controls in the presence of A375-SLL_H cells for 48 h is marked as CD69 or CD25 positive (n = 3). (B) Mean fold expansion of T cells (donor 1) mediated by tNY-aCD3/SEA or its isotype controls in the presence of A375-SLL_H cells for 72 or 96 h (n = 3). (C) Ratio of T cells (donor 1) with different fold expansion at 96 h (n = 3). (D) Mean fold expansion of T cells (donor 1) mediated by tNY-aCD3/SEA or tNY-aCD3/aCtrl in the presence of A375-SLL_H cells for 96 or 148 h (n = 3). ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Measuring cytokine production by T cells is another common assay for T cell activation. tNY-aCD3/SEA activated T cells to release a large amount of IFN-γ and IL-2, far exceeding the levels induced by its controls (Figures 3A and 3B). Interestingly, tNY-aCD3/aCtrl was more inclined to induce T cells to release IFN-γ, while tCtrl-aCD3/SEA and tNY/SEA were more inclined to induce T cells to release IL-2 (Figures 3A and 3B). This finding implied that different components of tNY-aCD3/SEA can induce T cells to release different cytokines, which may have important implications for the antitumor activity of tNY-aCD3/SEA. Granzyme B, a core effector molecule for T cells to execute tumor killing, is released by T cells in response to recognition of target cells and then enters the target cell and activates a cascade of events that leads to apoptosis, or programmed cell death. We found that the proportion of granzyme B-positive T cells was increased in a time- and dose-dependent manner (Figure 3C) after coincubation with tNY-aCD3/SEA and A375-SLL_H cells. More importantly, granzyme B was efficiently delivered to A375-SLL_H cells to ensure specific tumor killing (Figure 3C), and the timing of granzyme B upregulation in tumor cells was consistent with the time of significant tumor cell death observed in the cytotoxicity assays in vitro. The ability of the controls to stimulate T cells to express granzyme B was far less than that of tNY-aCD3/SEA, and granzyme B could only be detected in A375-SLL_H cells when the concentration of the controls was 2 nM (Figure 3D), which explained the antigen-independent activity of the controls at concentrations of 1 nM and above (Figures 1F and 1G).

Figure 3.

Cytokine release of T cells in response to the stimulation of tNY-aCD3/SEA or its isotype controls

(A and B) IFN-γ (A) and IL-2 (B) release of PBMCs (donor 1) mediated by tNY-aCD3/SEA or its isotype controls in the presence of A375-SLL_H cells for 48 or 72 h (n = 3). (C) Upregulation of granzyme B⁺ T cell (CD3⁺) or A375-SLL_H cell (GFP⁺) ratio mediated by tNY-aCD3/SEA for 24 or 40 h (n = 3) (donor 2). (D) Upregulation of granzyme B⁺ T cell (CD3⁺) or A375-SLL_H cell (GFP⁺) ratio mediated by tNY-aCD3/SEA or its isotype controls for 40 h (n = 3). ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Overall, SEA can effectively enhance the ability of TCEs to induce T cell activation, but this effect can be almost completely removed by the D227A mutation (Figure S6). The results are consistent with the antitumor activity of tNY-aCD3/SEA and its controls in vitro. To further explore the mechanism by which SEA enhanced the potency of TCEs, we performed mRNA sequencing of PBMCs after coincubation with A375-SLL_H cells and 200 pM tNY-aCD3/SEA or tNY-aCD3/aCtrl. As shown in Figure 4A, tNY-aCD3/SEA induced more changes in mRNA expression than tNY-aCD3/aCtrl. Immune-activating molecules such as CD69, CD25, and granzyme B were significantly upregulated and immunosuppressive molecules such as VSIG4 were significantly downregulated in tNY-aCD3/SEA-treated PBMCs, which did not occur in tNY-aCD3/aCtrl-treated PBMCs. In addition, we observed the upregulation of multiple chemokines (Figure 4A), which may attract T cells to enter the tumor sites and further enhance the therapeutic effect of TCEs in solid tumors. To identify enriched gene sets, we conducted an over-representation analysis using the hypergeometric test implemented in the enricher library, including gene ontology and KEGG pathway annotations. Consistent with the previous results, SEA significantly enhanced the sensitivity of TCE-activated T cells to various cytokines, especially chemokines (Figure 4B). All the results of T cell activation assays illustrated that SEA achieved its synergistic effect on TCEs by enhancing T cell activity at multiple levels. In contrast, SEA-fused TCEs maintained high tumor-antigen specificity.

Figure 4.

mRNA sequencing of PBMCs (donor 1) after coincubation with A375-SLL_H and 200 pM tNY-aCD3/SEA or tNY-aCD3/aCtrl

(A) Differential expression analysis. The x and y axes represent log-transformed fold change and log-transformed p values, respectively. We filtered differentially expressed genes at a threshold of fold change at 1.5 and p value at 0.05 (gray line). Upregulated, downregulated, and no-different genes are labeled in red, blue, and gray, respectively. The figure also showed the number of up- and downregulated genes. (B) Functional enrichment analysis of differentially expressed genes. The width of the bars represents the odd ratio of gene sets. The color of the bars represents log-transformed p values.

SEA significantly improved the antitumor activity of TCEs in vivo

To further investigate SEA’s ability to enhance TCE’s antitumor activity, xenograft mouse models were used to evaluate the in vivo antitumor activity of tNY-aCD3/SEA and its controls. Immune-deficient SCID/beige mice were inoculated subcutaneously with A375-based tumor cells and PBMCs. These mice were treated intravenously with tNY-aCD3/SEA or its controls at days 0, 3, and 6.

First, the A375-SLL_H xenograft model was used to evaluate the antitumor activity of tNY-aCD3/SEA and its controls. Tumor growth was markedly suppressed in the presence of tNY-aCD3/SEA, and 4/5 mice treated with 100 μg/kg and 2/5 mice treated with 25 μg/kg tNY-aCD3/SEA survived without tumor burden during the observation period (160 days, Figure 5A). In contrast, lethal progression was observed in all mice in the saline and control groups. However, the overall survival time of the mice was slightly prolonged in the presence of 100 μg/kg tCtrl-aCD3/SEA (Figure 5A). To further explore whether the in vivo antitumor activity of tNY-aCD3/SEA was tumor antigen dependent, we compared the ability of tNY-aCD3/SEA and tCtrl-aCD3/SEA to inhibit the growth of transplanted tumors in a separate trial. As shown in Figure 5B, the potency of tNY-aCD3/SEA in vivo reemerged and was still significantly superior to that of tCtrl-aCD3/SEA. Although there was no statistically significant difference in tumor size and weight between the tCtrl-aCD3/SEA group and the saline group, it must be noted that tCtrl-aCD3/SEA may cause non-specific T cell activation to inhibit tumor growth to a certain extent (Figure 5B). In addition, the potency of tNY/SEA in vivo was surprisingly good (Figure 5B). However, its activity in vitro was not different from that of other controls. Nevertheless, we believed that the activity of tNY/SEA was dependent on the tumor antigen because of its better activity than tCtrl-aCD3/SEA. Besides, SEA with the D227A mutation did not enhance TCE’s in vivo antitumor activity (Figure S7), which was consistent with the results of the in vitro activity evaluation.

Figure 5.

Effect of tNY-aCD3/SEA or its isotype controls on tumor growth in mouse xenograft models

In vivo efficacy was evaluated in SCID/beige mice coinoculated with tumor cells and PBMCs (donor 1) subcutaneously. The first dose was administered 1 h after tumor transplantation, and the day was recorded as day 0. Tumor volume was quantitated at various time points, and mice were euthanized when the tumor volume was larger than 600 mm³. (A) Growth inhibition of A375-SLL_H xenograft and survival rate of the mice after treatment with tNY-aCD3/SEA or its isotype controls. The saline group and the tNY-aCD3/aCtrl 100 μg/kg group each contained 4 mice, while the other groups each contained 5 mice. (B) Growth inhibition of A375-SLL_H xenograft after treatment with 50 μg/kg tNY-aCD3/SEA or its isotype controls (n = 5). All mice were euthanized on day 40, and the tumors were removed, photographed, and weighed. Tumor only group: SCID/beige mice inoculated with tumor cells subcutaneously. ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (C) Growth inhibition of A375-SLL_L xenograft and survival rate of the mice after treatment with 100 μg/kg tNY-aCD3/SEA or its isotype controls (n = 4).

Next, we assessed the antitumor activity and specificity of tNY-aCD3/SEA in the A375-SLL_L xenograft model. As expected, tNY-aCD3/SEA also demonstrated superior antitumor activity compared with its controls, as demonstrated by the significant inhibition of tumor growth and tumor-free survival in half of the mice within 200 days after treatment with tNY-aCD3/SEA, while all mice in the control groups exhibited lethal tumor progression (Figure 5C). Interestingly, the difference in antitumor activity between tNY/SEA and tCtrl-aCD3/SEA was precisely the opposite in A375-SLL_L and A375-SLL_H xenograft models (Figures 5B and 5C). The antitumor activity of tCtrl-aCD3/SEA was not affected by antigen expression. Hence, we speculated that the lower expression level of SLL/A02 in A375-SLL_L cells limited the antitumor activity of tNY/SEA.

The design strategy of SEA-fused TCEs applied different targets

To investigate whether the enhancement of TCE potency by SEA applies different targets, we used an RMF/A02 targeted antibody (aWT1) to generate aWT1-aCD3/SEA.³⁵ aWT1 and aCD3 in aWT1-aCD3/SEA or the isotype controls held specific affinity to their cognate antigens, and the irrelevant antibodies in the isotype controls erased their ability to recognize the corresponding target antigens (Figure S8; Tables S4 and S5). The in vitro cytotoxicity was assayed for BV173 cells and SKOV3 cells that naturally expressed WT1 and HLA-A∗02:01 (Table S6). aWT1-aCD3/SEA showed dose-dependent activity to redirect PBMCs for the killing of both tumor cells, while the potency of its isotype controls was significantly weaker than that of aWT1-aCD3/SEA (Figure 6A). Surprisingly, tCtrl-aCD3/SEA was more effective against SKOV3 cells than against the other tumor cells here (Figures 1E, 6A, S2B, and S2C). It is possible that there was a more significant number of T cells in the PBMCs that recognize SKOV3 antigens compared with other tumor cells (e.g., BV173 and A375). The cytotoxic activity of these T cells against SKOV3 cells can be enhanced by tCtrl-aCD3/SEA, leading to the destruction of more SKOV3 cells.

Figure 6.

Tumor cytotoxicity and T cell activation elicited by SEA-fused TCEs

(A) T cell-dependent cytotoxicity (donor 2) of aWT1-aCD3/SEA and its isotype controls on BV173 cells and SKOV3 cells. The cytotoxicity was indicated by the degree of BV173 cell apoptosis or SKOV3 cell lysis. Apoptotic tumor cells were labeled with Annexin V-633 and PI, and then detected by flow cytometry (n = 3). Cell lysis was determined by LDH released level (n = 3). (B) T cell activation (donor 2) mediated by aWT1-aCD3/SEA or its isotype controls in the presence of SKOV3 cells for 48 h (n = 3). (C) IFN-γ and IL-2 secretion of PBMCs (donor 2) mediated by aWT1-aCD3/SEA or its isotype controls in the presence of SKOV3 cells for 72 h (n = 3). ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (D) T cell-dependent cytotoxicity (donor 5) of aWT1-aCD3/SEA and its isotype controls on SKOV3 cells, A375 cells, and K562 cells. Cell lysis was determined by LDH released level (n = 3). Normalized cell lysis means the ratio of OD₄₉₀ in the treated group to OD₄₉₀ in the blank group. ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

As expected, we observed aWT1-aCD3/SEA dose-dependent T cell activation and cytokine release in the presence of SKOV3 cells, and aWT1-aCD3/SEA had a more robust ability than the controls to activate T cells to release cytokines at the same concentration (Figures 6B and 6C). In addition, we found that coincubation of tCtrl-aCD3/SEA with SKOV3 cells induced T cells to release more IFN-γ than coincubation with A375-SLL_H cells (Figures 3A and 6C), which was consistent with the stronger in vitro activity of tCtrl-aCD3/SEA against SKOV3 cells compared with that against A375-SLL_H cells. Due to the lack of tumor recognition arms in tCtrl-aCD3/SEA, its antitumor activity or ability to activate T cells is theoretically not affected by the type of tumor. This evidence suggested that different tumor cells may cause varying degrees of T cell immune response, ultimately affecting the activity of SEA-fused TCEs.

In addition to SLL/A02 and RMF/A02 with high tumor specificity, we also investigated whether the TCE-SEA fusion strategy applies to some tumor-associated antigens with relatively low tumor specificity. We used a HER2-targeted antibody (trastuzumab, aHER2) to generate aHER2-aCD3/SEA and, as expected, aHER2 and aCD3 in aHER2-aCD3/SEA or the isotype controls held specific affinity to their cognate antigens, and the irrelevant antibodies in the isotype controls erased their ability to recognize the corresponding target antigens (Figure S9, Tables S7 and S8). We performed in vitro cytotoxicity on SKOV3 cells (highly HER2-positive, Table S9), A375 cells (low HER2-positive, Table S9), and K562 cells (HER2-negative, Table S9), and found that aHER2-aCD3/SEA showed dose-dependent activity to redirect PBMCs for the killing of both HER2-positive tumor cells, whereas it had little effect on K562 cells (Figure 6D). Similar to the activities of the two SEA-fused TCEs described above, aHER2-aCD3/SEA showed a significant increase in activity over its isotype controls (Figure 6D). More interestingly, whereas conventional TCEs (aHER2-aCD3/aCtrl) showed an approximately 13.5-fold increase in activity against SKOV3 cells (∼17.0 pM EC₅₀) compared with A375 cells (∼229.1 pM EC₅₀), aHER2-aCD3/SEA showed an approximately 50.3-fold increase in activity in SKOV3 cells (∼0.73 pM EC₅₀) compared with A375 cells (∼36.69 pM EC₅₀), suggesting that an increase in HER2 expression may have a more pronounced elevating effect on aHER2-aCD3/SEA activity (Figure 6D). It is well known that tumor-associated antigens, such as HER2, are highly expressed in tumor cells, but also have certain expression levels in some normal cells. Thus, aHER2-aCD3/SEA may have a more tremendous difference in activity between tumor cells and HER2-positive normal cells than aHER2-aCD3/aCtrl, which is potentially valuable for enhancing the therapeutic window.

Nevertheless, the fusion of SEA also increased the risk of TCEs activating T cells before reaching the tumor tissues, as we observed that aCD3/SEA-induced T cells to secrete more IFN-γ and TNF-α than aCD3 alone (Figures S10A and S10B). Fortunately, detached from the tumor antigen, aHER2-aCD3/SEA did not significantly increase the release of IL-6, a significant contributor to the cytokine storm, over conventional TCEs (Figure S10C). SEA exerted some activity in detachment from the tumor, which also suggests that its activity does not arise exclusively from the action of MHC class II molecules on the surface of the tumor cells, and we found that the majority of tumor cells we used for the evaluation of the activity were MHC class II negative (Figure S11).

Finally, we re-evaluated the effects of SEA on TCE activity and safety hazards in human CD3ε transgenic mice (Figure S12). Compared with aHER2-aCD3/aCtrl, aHER2-aCD3/SEA can better inhibit the growth of transplanted 4T1-HER2 (Table S9) tumors and is accompanied by more significant upregulation of IL-6 levels (Figures 7A, 7B, and S13) but, still and all, there were no significant differences in body weight or body temperature among the groups of mice, suggesting that the SEA-induced increase in cytokine may be tolerable (Figures 7C and 7D).

Figure 7.

In vivo efficacy and cytokine storm evaluation of aHER2-aCD3/SEA and aHER2-aCD3/aCtrl in human CD3ε transgenic mice

4T1-HER2 cells (2 × 10⁶) were subcutaneously engrafted into the right armpit of BALB/c-hCD3E mice and intravenously with saline, aHER2-aCD3/SEA (200 μg/kg), or aHER2-aCD3/Ctrl (200 μg/kg) (n = 3). The first dose was administered 1 h after tumor transplantation, and the day was recorded as day 1. (A) The growth of transplanted 4T1-HER2 tumors. (B) The levels of mouse IL-6 in serum was determined by ELISA. (C and D) Body temperature (C) and weight (D) monitoring. The arrows mark the time of administration. ns ≥ 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Discussion

TCEs are designed to address the issue of insufficient tumor-specific T cells for immunotherapy by indiscriminately bridging T cells and tumor cells.^1,2 In some hematologic malignancies, TCEs have achieved remarkable clinical responses. However, resistance appears even in tumors without target antigen loss.^4,6 Even worse, most patients burdened with solid tumors fail to respond to TCE therapy.^5,7 Here, we report a novel type of SEA-fused TCE to enhance TCE’s efficacy in treating tumors, especially solid tumors.

We showed that the potency of TCEs can be increased more than 30-fold by SEA (∼40 pM EC₅₀ versus >1,200 pM EC₅₀), which most likely benefited from potent SEA-triggered T cell activation. As a superantigen, SEA can simultaneously bind MHC class II molecules and particular Vβ chains of TCRs, which intensely induces T cell activation and proliferation.³⁶ Nevertheless, SEA has a bias for T cell activation. On the one hand, CD4 but not CD8 has been reported to assist in T cell activation in response to SEA, resulting in predominant CD4⁺ T cell proliferation.³⁷ We found that tNY/SEA can effectively induce PBMCs to secrete IL-2, mainly produced by activated TH1 cells (CD4⁺) to enhance cellular immunity further.³⁴ Therefore, even though SEA may have a more substantial direct activation effect on CD4⁺ T cells, we observed that the T cell activation ultimately caused by SEA-fused TCEs did not strongly prefer CD4⁺ T cells. On the other hand, SEA selectively recognized T cells bearing TCR Vβ 1, 5.3, 6.3, 6.4, 6.9, 7.4, 9.1, and 23, leading to the direct activation of these T cells.³⁶ TCR Vβ is also one of the main determining factors for the MHC limitation of T cells.³⁸ Researchers recently found that MHC- and TCE-mediated TCR-CD3 activation has a synergistic effect, which was very important for T cells to differentiate into effector phenotypes with solid cytotoxicity.⁶ Hence, SEA-fused TCEs can simultaneously provide both activation signals on T cells expressing SEA-recognized TCR Vβ, thereby inducing these T cells to exhibit stronger antitumor activity. In addition to the more efficient tumor-killing activity of tNY-aCD3/SEA-treated T cells, we were pleasantly surprised to find through mRNA sequencing that these T cells significantly upregulated the expression of various chemokines, which may promote T cell infiltration and further improve the therapeutic effect of TCEs in solid tumors. This finding also coincided with the report that SEB can support CAR T cells against solid tumors by boosting CAR T cell proliferation, activity, and infiltration.³⁹

Although SEA-fused TCEs exhibited potent antigen-dependent activity both in vitro and in xenograft models coinoculated with PBMCs, it still cannot eliminate our concerns about systemic toxicity caused by T cell over-activation. SEA-fused TCEs were ineffective against antigen-negative tumor cells when the concentration was below 1 nM. At the same time, they already exhibited good activity against antigen-positive cells when the concentration reached 100 pM. Of note, we have demonstrated that an increased antigen expression level significantly enhanced the activity of SEA-fused TCEs, and the expression levels of commonly used targets for TCEs in clinical practice are much higher than the targets we have chosen here.⁴⁰ Therefore, SEA-fused TCEs may have a relatively low risk of systemic toxicity when reaching an effective concentration for treating hematologic malignancies. However, SEA-fused TCEs need to overcome barriers such as the blood vessel wall and intercellular matrix before binding to surface targets of solid tumors. When the effective SEA-fused TCE dose is enriched in the solid tumor tissue, peripheral blood T cells will face higher concentrations of SEA-fused TCEs, which may lead to increased non-specific T cell activation and a higher risk of systemic toxicity.^41,42 Because this systemic cannot be studied in immunocompromised mice, we further assessed the risk of SEA-fused TCEs in human CD3ε transgenic mice and fortunately found that SEA did not significantly enhance the safety risk of TCEs in the context of therapeutic activity. Despite this evidence for the positive effect of SEA on TCE efficacy, humans are more sensitive to SEA than mice,⁴³ and it is necessary to evaluate the activity and safety risks of SEA-fused TCEs using models more closely related to the human immune system.

To reduce the risk of adverse effects induced by excessive systemic immune activation, we attenuated the superantigen activity of SEA by introducing the D227A mutation.³³ However, this mutation caused SEA to completely lose its ability to activate T cells after fusion with TCEs. However, Takemura et al.³³ demonstrated that SEA containing this mutation can still exert antitumor activity after fusion with an anti-MUC1 antibody. The molecular format and expression system may affect SEA D227A activity. More importantly, we speculated that the antigen expression level limited the activity of SEA D227A-fused TCEs, as the expression level of MUC1 was significantly higher in tumor cells than that of the targets we used here.⁴⁰ Hence, the D227A mutant SEA may synergistically affect TCEs targeting tumor antigens with high expression levels. In addition, the D227A mutation reduced SEA’s T cell activation ability by weakening its affinity to MHC class II molecules,³³ which implied that MHC class II molecules were essential for the high antitumor activity of SEA-fused TCEs. MHC class II molecules are expressed in professional antigen-presenting cells and some malignant tumors but are often low or not expressed in normal cells.^44,45 That is, tumors with high expression levels of MHC class II molecules may be more sensitive to SEA-fused TCEs, indicating that patients burdened with these tumors are more likely to benefit from therapy with SEA-fused TCEs.

Due to their superior tumor specificity, we chose pHLAs as the targets to validate the feasibility of SEA’s TCE activity optimization. We demonstrated that highly expressed pHLAs (e.g., RMF/A02) have a specific response to SEA-fused TCEs.³⁵ However, the expression of pHLAs is still at a shallow level compared with commonly used TCE targets in clinical practice. Hence, SEA may have a more substantial effect on enhancing the activity of these TCEs. Nevertheless, SEA can theoretically exacerbate the on-target off-tumor toxicity of TCEs mediated by commonly used targets. In the case of HER2, we initially found that SEA can tilt commonly used TCEs in a more effective direction, but the damage of SEA-fused TCEs to normal organs has yet to be verified in antigen transgenic or primate animals.

In conclusion, we designed novel SEA-fused TCEs and found that they can effectively and antigen-dependently redirect T cells to kill tumors. The effective activation of T cells is the core factor for SEA to promote TCE activity. Exploring strategies for T cell activation is an effective way to enhance the efficacy of T cell immunotherapies, including TCEs and tumor vaccines. Compared with existing costimulatory molecule agonists and immune checkpoint inhibitors acting on some T cell activation pathways, superantigens may have a broader and stronger activation effect on T cells. However, at the same time, safety hazards also need to be more carefully investigated. In future research, a more in-depth analysis of SEA-fused TCE-induced T cell activation, proliferation, and differentiation is crucial for their clinical application.

Materials and methods

Study design

The main objective of this study was to generate potent TCEs via fusion with superantigens for the preclinical development of T cell therapeutics. We assessed the effectiveness and specificity of the SEA-fused TCE using in vitro and in vivo assays. The number of replicates is reported in the figure legends. For in vivo experiments, the number of mice used was determined using the minimum number of necessary animals to ensure no waste of animal life, and three to five animals were used for each group in every animal experiment. Mice were followed until tumors became too large (A375 > 600 mm³, 4T1 > 1,000 mm³), and no data were excluded. Data analysis was performed in an unblinded manner.

Cell lines and culture conditions

Human embryonic kidney 293F cells (HEK293F) were kindly provided by the Comprehensive AIDS Research Center (Tsinghua University) and rotary cultured in SMM 293-TI medium (Sino Biological, China) supplemented with 0.5% fetal bovine serum (FBS) (Gibco). The human T lymphocyte cell line Jurkat, human melanoma cell line A375, human erythroleukemic cell line K562, human peripheral blood B cell leukemia cell line BV173, and mouse breast cancer cell line 4T1 were purchased from American Type Culture Collection (ATCC), and cultured in RPMI-1640 (Gibco) with 10% FBS and 1% penicillin-streptomycin. A375-SLL_H cells, A375-SLL_L cells (A375 cells transfected with EGFP and SLLMWITQC peptide), K562-SLL cells (K562 cells transfected with HLA-A∗02:01, EGFP, and SLLMWITQC peptide), K562-Ctrl cells (K562 cells transfected with HLA-A∗02:01, EGFP, and MQLMPFGCLL peptide), and 4T1-HER2 cells (4T1 cells transfected with HER2) were constructed and maintained in our laboratory. The human ovarian cancer cell line SKOV3 (ATCC) was cultured in McCoy’s 5A modified medium (Thermo Fisher Scientific) with 10% FBS and 1% penicillin-streptomycin. Human PBMCs were purchased from Saily Bio (China), and rested in RPMI-1640 with 10% FBS for 4–12 h before use. All cells were grown at 37°C with 5% CO₂ and 95% humidity.

Production of SEA-fused TCEs

SEA-fused TCEs and their isotype controls were based on the backbone of the conventional antibody IgG1 in a knob-into-hole format with L234A, L235A, and P329G substitutions to attenuate Fc. In brief, tumor antigen-binding domains and SEA were fused to the N terminus of the two Fc arms. To purify the correctly assembled protein, a 6×His tag was introduced into the C terminus of Fc fused with the tumor antigen-binding arm, and a FLAG tag was introduced into the C terminus of Fc fused with SEA. The anti-CD3 scFv was fused into the 3′ tail of the light chain (Fab) or α chain (TCR) of the tumor antigen-targeting part via a (G₄S)₃ linker. The genes of each subunit were cloned and inserted into the pLVX-puro vector. SEA-fused TCEs and their isotype controls were expressed by transient cotransfection of the corresponding subunits in HEK293F cells, and then purified by nickel chromatography (GE Healthcare) and ANTI-FLAG M1 Agar Affinity Gel (Millipore Sigma, Germany). The purified proteins were analyzed by SDS-PAGE under denaturing conditions after being deglycosylated by PNGase F (Sigma-Aldrich).

Affinity and specificity assay

The binding affinity and specificity of SEA-fused TCEs and their isotype controls to SLL/A02 or RMF/A02 were tested via ELISA. In brief, 96-well EIA/RIA plates (Corning) were coated with streptavidin (2 μg/mL) overnight at 4°C. After blocking, different biotin-labeled SLL/A02 or RMF/A02 (2 μg/mL), generated by refolding as in a previous report,⁴⁶ were added for 1 h at 37°C. Then, increasing amounts of SEA-fused TCEs or their isotype controls were added for 1 h at 37°C, followed by incubation with HRP-conjugated goat anti-human IgG (H + L) (1:1,000 dilution, Beyotime Biotechnology, China) for 1 h at 37°C. The samples finally reacted with the TMB substrate solution and the reaction was stopped by adding 2 M H₂SO₄. The absorbance of the samples at 450 nm was measured using a Model 680 Microplate Reader (Bio-Rad). Washing with PBST was needed between each step.

The binding activity of SEA-fused TCEs and their isotype controls to human CD3ε or HER2 was evaluated by flow cytometry. In brief, Jurkat cells (CD3⁺) or SKOV3 cells (HER2⁺) were incubated with serial dilutions of SEA-fused TCEs or their isotype controls for 30 min at 4°C. After washing with PBS, the cells were incubated with FITC-conjugated goat anti-human IgG (H + L) (1:200 dilution, Beyotime Biotechnology) for 30 min at 4°C. Finally, samples were analyzed by an ACEA NovoCyte flow cytometer (ACEA Biosciences) after washing with PBS.

In vitro cytotoxicity, T cell activation, and cytokine release analysis

Human PBMCs and target tumor cells were incubated at a 4:1 ratio in the presence of serial dilutions of SEA-fused TCEs or their isotype controls. The incubation time was specified in the figure legend for each experiment. For the killing assays, cell viability was measured using an Annexin V, 633 Apoptosis Detection Kit (Dojindo, Japan) or an LDH Cytotoxicity Assay Kit (Beyotime Biotechnology) as indicated by the manufacturer. For the assessment of T cell activation, cell surface staining for CD3 (APC anti-human CD3, Invitrogen) or CD8 (APC anti-human CD8, Invitrogen), CD69 (Super Bright 436 anti-human CD69, Invitrogen), and CD25 (PE anti-human CD25, Invitrogen) was performed according to the suppliers’ indications, and the samples were analyzed using an ACEA NovoCyte flow cytometer. IFN-γ and IL-2 from culture supernatants were analyzed with an IFN gamma Human Uncoated ELISA Kit (Invitrogen) and a Human IL-2 Precoated ELISA Kit (Dakewe, China), respectively, as indicated by the manufacturer. For intracellular granzyme B expression assays, cells were collected and stained for CD3 (APC anti-human CD3), and then the cells were fixed and permeabilized using the eBioscience Intracellular Fixation & Permeabilization Buffer Set (Invitrogen) as indicated by the manufacturer. After permeabilization, PE anti-human Granzyme B (Invitrogen) was used to label granzyme B within the cells. Finally, samples were analyzed using an ACEA NovoCyte flow cytometer.

T cell proliferation analysis

Human PBMCs were labeled with CFSE (Invitrogen) according to the manufacturer’s protocol. CFSE-labeled PBMCs were cocultured with tumor cells at a 4:1 ratio in the presence of serial dilutions of SEA-fused TCEs or their isotype controls in 48-well plates. After 72, 96, or 148 h of incubation, cells were stained with PI (Invitrogen) and APC anti-human CD3 or PI and APC anti-human CD8. Finally, the fluorescence intensity of CFSE in T cells was analyzed using an ACEA NovoCyte flow cytometer.

For Bcl expression analysis, PBMCs were cocultured with tumor cells at a 4:1 ratio in the presence of 200 pM tNY-aCD3/SEA or tNY-aCD3/aCtrl. After 72 or 96 h of incubation, cells were collected and stained for CD3 (APC anti-human CD3), and then the cells were fixed and permeabilized using the eBioscience Intracellular Fixation & Permeabilization Buffer Set (Invitrogen) as indicated by the manufacturer. After permeabilization, PE anti-human Bcl-2 (Invitrogen) was used to label Bcl within the cells. Finally, samples were analyzed using an ACEA NovoCyte flow cytometer.

Transcriptomic analysis of PBMCs

Human PBMCs (GFP negative) and A375-SLL_H cells (GFP positive) were incubated at a 4:1 ratio in the presence of 200 pM tNY-aCD3/SEA or tNY-aCD3/aCtrl for 48 h. According to the fluorescence of GFP, PBMCs were isolated from the mixed cells by flow cytometric sorting (Beckman moflo Astrios EQ) for mRNA sequencing. In brief, total RNA was isolated and purified using TRIzol reagent (Invitrogen) and mRNA was purified using Dynabeads Oligo (dT) (Thermo Fisher). Then the mRNA was fragmented into small pieces using a Magnesium RNA Fragmentation Module (NEB) and reverse-transcribed to create the cDNA library. Finally, the cDNA library was performed on an Illumina NovaSeq 6000 (LC-Bio Technology, China) following the vendor’s recommended protocol. To identify enriched gene sets, we conducted an over-representation analysis using the hypergeometric test implemented in the Enrichr library, which includes both gene ontology and KEGG pathway annotations. p values were corrected by the Benjamini-Hochberg procedure.

In vivo antitumor activity evaluation

In brief, 6-week-old female SCID/beige mice (GemPharmatech, China) were housed in a specific pathogen-free facility and were subcutaneously engrafted with tumor cells (A375-SLL_H or A375-SLL_L, 2 × 10⁶ cells per mouse) and PBMCs (2 × 10⁶ cells per mouse) in the right armpit. The mice were randomly divided into several groups and then treated intravenously with tNY-aCD3/SEA or its controls 1 h after engraftment. Each group was treated every 3 days three times (q3d × 3) via the tail vein. From the first administration, tumor length and width were measured every 3–4 days and the tumor volume was calculated according to the formula (tumor volume = length × width²/2). Mice were euthanized when the tumor volume was larger than 600 mm³. In the experiment to measure tumor weight, all mice were euthanized and the tumors were resected to measure their weight when mice with a tumor volume greater than 600 mm³ appeared (Animal Ethics Code: 19699).

Cytokine release analysis without tumor cells

Human PBMCs were cocultured with aCtrl-aCD3/SEA or t aHER2-aCD3/aCtrl for 48 or 72 h. IFN-γ, TNF-α, and IL-6 from culture supernatants were analyzed using an IFN gamma Human Uncoated ELISA Kit, a TNF alpha Human Uncoated ELISA Kit (Invitrogen), and a Human IL-6 Precoated ELISA Kit (Dakewe), respectively, as indicated by the manufacturer.

Analysis of MHC class II expression on tumor cells

Tumor cells were incubated with APC anti-human HLA-DR (BioLegend) for 30 min at 4°C. The samples were analyzed using an ACEA NovoCyte flow cytometer after washing with PBS.

Evaluation of cytokine storm in human CD3ε transgenic mice

Seven-week-old female BALB/c-hCD3E mice (GemPharmatech, China) were housed in a specific pathogen-free facility. Blood samples were collected via retro-orbital bleeds and then red blood cells were lysed to obtain PBMCs, which were then analyzed for the expression of human CD3ε (APC anti-human CD3ε antibody, BioLegend) and mouse CD3 (PE anti-mouse CD3 antibody, BioLegend) using flow cytometry. BALB/c-hCD3E mice were subcutaneously engrafted with 2 × 10⁶ 4T1-HER2 cells per mouse in the right armpit. The mice were randomly divided into three groups (n = 3) and then treated intravenously with saline, aHER2-aCD3/SEA (200 μg/kg), or aHER2-aCD3/Ctrl (200 μg/kg) 1 h after engraftment. Each group was treated every 3 days for a total of three times (q3d × 3). Body temperature and body weight were measured every day, and tumor length and width were measured every 3–5 days. Serum was collected through orbital blood sampling 24 or 48 h post-treatment for subsequent detection of IL-6 levels (Mouse IL-6 Precoated ELISA Kit, Dakewe). When the tumor volume in the saline group exceeded 1,000 mm³, all mice were euthanized and the tumors were removed to measure their weight (Animal Ethics Code: DW202310231504).

Statistical analysis

Graphs were made and statistical analysis was performed using GraphPad Prism (GraphPad Software). Statistical analysis was performed using a two-sided unpaired Student’s t test (two groups) or one-way ANOVA (three groups and above). The results are presented as the mean ± SD. Statistical tests and p values are described in the figure legends. A p value less than 0.05 was considered statistically significant.

Data and code availability

The data supporting this study’s findings are available from the corresponding author upon reasonable request.