Abstract Objective Targeting CD47 for cancer immunotherapy has been studied in many clinical trials for the treatment of patients with advanced tumors. However, this therapeutic approach is often hampered by on-target side effects, physical barriers, and immunosuppressive tumor microenvironment (TME). Methods To improve therapeutic efficacy while minimizing toxicities, we engineered an oncolytic vaccinia virus (OVV) encoding an anti-CD47 nanobody (OVV-αCD47nb). We demonstrated the specific binding activity of αCD47nb secreted from the virus-infected cells to CD47 and that both secreted αCD47nb and OVV-αCD47nb blocked the “don’t eat me” signal of macrophages. Results Intratumorally injected OVV-αCD47nb continuously releases the αCD47nb in tumor tissues, thereby conferring superior systemic activity against breast and colon tumor cells and prolonging survival compared with OVV control. Furthermore, treatment with OVV-αCD47nb also remodeled the TME, as shown by increased T cell infiltration, CD8^+ T cell activation and tumor-associated macrophages polarization, significantly enhancing innate and adoptive immunity. Additionally, the inclusion of programmed cell death protein-1 inhibiting boosted the anticancer efficacy of OVV-αCD47nb and raised the full response rate in tumor-bearing animals. Conclusion Overall, our findings highlight the therapeutic potential of OVV-αCD47nb for breast and colon cancer, and demonstrate its ability to modulate the immune cell profiles within tumors. This has established a rationale for further exploring OVV-αCD47nb as a potential therapy in the clinic. Keywords: Oncolytic virus, Macrophage, Immunotherapy __________________________________________________________________ WHAT IS ALREADY KNOWN ON THIS TOPIC * Targeting CD47 for cancer immunotherapy has been studied in many clinical trials for the treatment of patients with advanced tumors. However, this therapeutic approach is often hampered by on-target side effects, physical barriers, and immunosuppressive tumor microenvironment. WHAT THIS STUDY ADDS * Our results demonstrate the potential of an oncolytic vaccinia virus (OVV) encoding an anti-CD47 nanobody (OVV-αCD47nb) as a safe and effective treatment that can promote systemic antitumor immunity by increasing T cell infiltration, CD8^+ T cell activation and tumor-associated macrophages polarization, significantly enhancing innate and adoptive immunity. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY * The findings will facilitate the understanding of the mechanism of action of an OVV-expressing CD47 inhibitor and encourage the clinical translation of OVV-αCD47nb virotherapy. Introduction Cancer immunotherapies, including immune checkpoint blockade and oncolytic virus (OV), have recently shown impressive results in the treatment of multiple metastatic and treatment-refractory cancers.[37]1,[38]3 However, their application in solid tumors is often constrained by several factors, including response heterogeneity where a few patients achieve a favorable response but others never do, tumor relapse due to different immune escape mechanisms, the existence of tumor microenvironment (TME), and the emergence of systemic toxicities and immune-related adverse events.[39]^1 4 5 One crucial mechanism of tumor evasion involves the overexpression of CD47, an immunosuppressive signaling molecule.[40]^6 7 CD47, an integrin-associated protein, is commonly expressed in a variety of normal cell types. However, its high expression has been detected in a wide array of human tumors, including hematopoietic malignancies and solid tumors such as ovarian, breast, bladder, colon, liver, pancreatic, lung, and head and neck squamous cell carcinoma.[41]^8 9 CD47 is also expressed on tumor-initiating cells.[42]^10 In cancer cells, CD47 serves as a mechanism of immune evasion via binding to signal-regulatory protein α (SIRPα), thereby inducing an inhibitory signaling pathway that enables tumor cells to evade phagocytosis by macrophages.[43]^11 Moreover, overexpression of CD47 is associated with poor clinical outcomes in a variety of cancers,[44]12,[45]15 highlighting the potential for developing therapeutics targeting the CD47-SIRPα phagocytic pathway. Recently, anti-CD47 monoclonal antibodies (mAbs) and other CD47 blockage agents have demonstrated preclinical activity against lymphoma, leukemia, and many different solid tumors.[46]^16 Based on preclinical data, so far, more than 10 anti-CD47 mAbs are being tested clinically for the treatment of a variety of hematological cancers and advanced solid tumors.[47]^17 Among them, magrolimab (Hu5F9), the first-in-class anti-CD47 antibody, demonstrated an objective response rate of 50% in heavily pretreated patients with non-Hodgkin’s lymphoma when combined with rituximab.[48]^18 Clinical trials investigating Hu5F9, in combination with hypomethylating agent azacitidine have also reported encouraging response rates in patients with both high-risk myelodysplastic syndromes and untreated acute myeloid leukemia .[49]^16 17 19 However, low response rates were observed in patients with advanced solid tumors when anti-CD47 mAbs were used as monotherapy.[50]^20 21 In addition, the therapeutic utility of these mAbs is compromised due to significant toxicities such as anemia and thrombocytopenia as well as subsequent target-mediate clearance, highlighting the need to develop new therapeutic strategies.[51]^22 OV represents a potentially unique therapeutic approach for solid tumors because it can not only selectively infect and kill cancer cells but also disrupt the immunosuppressive TME via stromal remodeling and anti-angiogensis.[52]22,[53]25 OV-induced immunogenic cell death results in the release of tumor-associated antigen, danger-derived and virus-derived pathogen-associated molecular patterns as well as proinflammatory cytokines and chemokines, in turn promoting systemic anticancer immunity.[54]^22 Consequently, oncolytic virotherapy is therefore a valuable component of a combinatorial immunotherapeutic approach. Moreover, OVs can be engineered to express transgenes, including mAbs for immune checkpoint inhibition or macrophage phagocytosis activation, which may overcome barriers of poor tumor accumulation and penetration of antibodies.[55]26,[56]30 Among the various OV platforms, oncolytic vaccinia virus (OVV) is one of the most extensively investigated. In addition to causing direct lysis of tumor cells and indirect induction of innate and adaptive antitumor immunity, OVV can also be used as a delivery vector to encode multiple therapeutic genes.[57]^26 30 31 In this work, we constructed a novel OVV that encodes an anti-CD47 nanobody (OVV-αCD47nb) and explored if arming OVV with αCD47nb could enhance the killing effect against 4T1 breast tumor cells and CT26 colon tumor cells. We assessed the effect of OVV-αCD47nb in preclinical tumor models and elucidated the underlying mechanism using single-cell RNA-sequencing (scRNA-seq) and mass cytometry (CyTOF) to evaluate functional and genetic changes of intratumoral immune cell populations of BALB/c mouse model following the treatment of OVV-αCD47nb. Results Construction and characterization of OVV encoding a nanobody against murine CD47 First, we engineered a novel OVV armed with a high-affinity anti-mouse CD47 nanobody (αCD47nb) that had been reported to be a minimal antigen binding domain (V[HH]) of heavy chain-only antibody and bind tumor cell-surface CD47,[58]^32 which is termed “OVV-αCD47nb” ([59]figure 1A). A control OVV expressing the marker green fluorescent protein (GFP) (OVV-GFP) was also established. We then compared the infectivity of OVV-αCD47nb to parental OVV by detecting GFP expression. In a panel of mouse tumor cell lines including 4T1 breast cancer, CT26 colorectal cancer, and EL4 lymphoma cell line, similar GFP surface expression, in terms of % GFP^+ cells, was observed when cells were treated with OVV-αCD47nb and control OVV ([60]figure 1B), respectively, indicating that arming of OVV with an αCD47nb did not affect viral infection efficiency. Next, we examined the replication of OVV-αCD47nb, which determines the ability of OVV to trigger oncolytic cell death. After being infected with OVV-αCD47nb or OVV-GFP at a multiplicity of infection (MOI) of 0.1, Hela-S3 was cultured for the indicated times, and then the virus titer was assessed by the TCID50 method. Consistent with our previous results,[61]^31 arming the OVV by engineering them to produce αCD47nb did not influence the viral production ([62]figure 1C). We also verified the expression of αCD47nb in the 4T1 cells infected with OVV-αCD47nb and secretion of the therapeutic αCD47nb proteins that were quantified by quantitative PCR (qPCR) and western blot assay in the supernatant of OVV-αCD47nb-treated cells ([63]figure 1D and E and [64]online supplemental figure S1A and B). Figure 1. Generation and characterization of OVV-αCD47nb. (A) Schematic representation of OVV-αCD47nb. V[HH] domain of anti-CD47 is inserted into the vaccinia virus TK gene under the control of the promoter Pse/1. (B) GFP expression level in tumor cells following OVV/OVV-αCD47nb infection at MOI=5, measured by flow cytometry. (C) Viral production of OVV-αCD47nb. Tumor cell line HeLa-S3 was infected with OVV or OVV-αCD47nb at MOI=0.1. At indicated time points, cell extracts were harvested and titrated by an TCID50-based method. (D) Anti-CD47 nanobody mRNA expression level in 4T1 cells following OVV/OVV-αCD47nb infection at MOI=2, measured by real-time PCR. (E) Anti-CD47 nanobody protein expression level in 4T1 cells following OVV/OVV-αCD47nb infection at MOI=0.1, measured by Western blot. (F) 4T1 and CT26 cells were incubated with OVV/OVV-αCD47nb supernatants and αCD47nb binding was detected by flow cytometry. (G) CellTracker-labeled mouse macrophage cells (red) were incubated with CFSE-labeled 4T1 cells (green) (E:T=1:4) and the indicated supernatants or antibody, and examined by immunofluorescence microscopy to detect phagocytosis (yellow). (H) Mouse macrophage cells were incubated with 4T1 cells (E:T=1:4) and the indicated supernatants or antibody, then examined by flow cytometry to detect phagocytosis. The significance p values were derived from a two-way ANOVA and compared with OVV treatment. *p<0.05, **p<0.01, ***p<0.001. αCD47nb, anti-CD47 nanobody; ANOVA, analysis of variance; GFP, green fluorescent protein; MOI, multiplicity of infection; mRNA, messenger RNA; OVV, oncolytic vaccinia virus. [65]Figure 1 [66]Open in a new tab Supernatants of OVV-αCD47nb infected cells show antitumor activity in vitro CD47 is widely expressed in human solid tumors, playing a crucial role in antiphagocytic capacity.[67]8,[68]11 Flow cytometry analysis revealed high CD47 expression on various mouse tumor cell lines, including 4T1 and CT26 cells ([69]online supplemental figure S1C). To confirm the binding ability of secreted αCD47nb to CD47, we incubated mouse tumor cells 4T1 and CT26 with the supernatant of OVV-αCD47nb-infected cells, followed by the addition of FITC-labeled anti-FLAG antibody. Flow cytometric analyses showed that αCD47nb was able to bind to surface CD47 on 4T1 and CT26 cells ([70]figure 1F). We next assessed the ability of αCD47nb to induce the phagocytosis of 4T1 cells by mouse bone marrow-derived macrophages (BMDMs). In contrast to cells treated with a control medium or supernatant of OVV-infected cells, 4T1 cells treated with the anti-CD47 antibody Hu5F9 and supernatant of OVV-αCD47nb-infected cells were efficiently phagocytosed by BMDMs isolated from BALB/c mice ([71]figure 1G). In agreement with these results, flow cytometry data also demonstrated that both Hu5F9 and the supernatant harvested from OVV-αCD47nb-infected cells induced a significantly higher level of mouse macrophages phagocytosis against 4T1 cells compared with vehicle control ([72]figure 1H). To further confirm the therapeutic potential, specific antibody-dependent cell cytotoxicity mediated by secreted αCD47nb compared with anti-CD47nb protein was measured by using real-time cell analysis (RTCA) against 4T1 cells. As expected, after exposure to different amounts of supernatant of OVV-αCD47nb-infected cells, we observed a dose-dependent lytic effect on 4T1 cells, indicating that the OVV-encoded αCD47nb can mediate functional cytotoxicity of macrophages ([73]online supplemental figure S1D). OVV-αCD47nb kills different cell types and exhibits superior tumor inhibition Following confirmation of the functionality of the secreted αCD47nb protein from OVV-infected tumor cells, we aimed to validate the in vivo anticancer activity in BALB/c mice bearing 4T1 breast tumor, CT26 and MC38 colon tumor, respectively. In all models, tumor-bearing mice were enrolled in the study when the average tumor volumes reached 50–80 mm^3 and were divided into four groups, phosphate-buffered saline (PBS), anti-CD47nb, OVV, and OVV-αCD47nb according to the experimental schedule. We observed that OVV or anti-CD47nb protein led to slow and insignificant delays in tumor growth in comparison to untreated tumors in all models. Whereas, OVV-αCD47nb was significantly more effective than anti-CD47nb or OVV ([74]figure 2A and B). We next assessed the biodistribution of intratumourally administered OVV-αCD47nb. The 4T1 mice received three doses of 5×10^7 pfu OVV-αCD47nb on days 1, 4, and 7. After treatment, the mice were randomly selected to be sacrificed (n=4) at different time points and then viral gene expression in normal and tumor tissues was analyzed using qPCR. The viral gene expression in tumors reached its peak 3 days after OVV-αCD47nb treatment, with a mean of 2.55×10^6 copies/µg genomic DNA. A second peak of the viral gene expression was also observed 10 days after therapy, and then the viral gene copy number fell to low or undetectable levels by day 17 ([75]figure 2C). There was a weak expression of the viral gene in normal tissues that included the liver, kidney, and lung at 10 days, and in the liver, kidney as well as spleen at 14 days after OVV-αCD47nb injection. Evaluation of the liver and kidney function of 4T1 tumor-bearing mice treated with OVV-αCD47nb included serum biochemical analysis of aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, total bilirubin, direct bilirubin, thiobarbituric acid, glucose challenge, blood urea nitrogen, creatinine, and uric acid ([76]online supplemental figure S2). To better understand the mechanisms for the superior potency of OVV-αCD47nb, we performed another animal experiment and examined the existence of αCD47nb in tumor cells in vivo. Tumors were harvested from 4T1-bearing mice that received a single dose of OVV-αCD47nb or OVV (1×10^8 pfu/per mouse), respectively, and then were killed at the indicated times. Single-cell suspensions were prepared from harvested tumors and the presence of CD47 on the surface of tumor cells was detected by flow cytometry using a myc-tag αCD47nb. As shown in [77]figure 2D, tumor cells prepared at different time points expressed a high level of CD47, as evidenced by the detection of myc^+ cells. Whereas, tumor cells treated with OVV-αCD47nb experienced a fast and significant decline of CD47 expression at 56 hours post-treatment, with the nadir at day 7, and recovery to the expression level in OVV-treated tumor cells by day 18. These data indicated that secreted αCD47nb by OVV-αCD47nb infected cells prevented the binding of myc-tag αCD47nb to CD47 antigen on binding to CD47 on the surface of tumor cells. Figure 2. Antitumor activity of OVV-αCD47nb in vivo. (A) Tumor volume was measured every 2 or 3 days. Error bars represent SD. Once the tumor volume exceeded 2000 mm^3, the mouse was considered dead. (B) Kaplan-Meier survival analysis of tumor-bearing mice treated with PBS (control), OVV, or OVV- αCD47nb. (C) The content of viral DNA in each organ tissue was determined by qPCR in indicated time. (D) αCD47nb secretion by OVV-αCD47nb detection in vivo. 4T1-bearing BALB/c mice were treated with 1×10^8 pfu OVV or OVV-αCD47nb in tumor, then tumors were taken out after virus injection at indication time points. Incubate Myc-labeled αCD47nb protein with tumor cells and detect the binding by flow cytometry. (E–G) 4T1 models were established as previously described and the single-cell suspensions were preprepared 2 days post the last viral injection. (E) Flow cytometric analysis of the proportions of CD3^+ T cells, CD8^+ T cells, CD4^+ T cells in 4T1 tumors. (F) Flow cytometric analysis of the expression of granzyme B, IFN-γ and CD107A on CD8+T cells in 4T1 tumor tissues. (G) Flow cytometric analysis of the proportions of F4/80^+CD11b^+ macrophages, CD80^+CD206^− M1 macrophages and CD80^+CD206^+ M2 macrophages in 4T1 model. Error bars represent SD. ns, not significant; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. αCD47nb, anti-CD47 nanobody; OVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; qPCR, quantitative PCR. [78]Figure 2 [79]Open in a new tab Intratumoral administration of OVV-αCD47nb induced infiltration of immune cells and enhanced anti-PD-1-mediated antitumor efficacy We evaluated the infiltration and activation of immune cells 1 day after the last intravenous injection of OVV-αCD47nb in the 4T1 subcutaneous tumor model. Flow cytometry analysis showed that treatment with OVV-αCD47nb significantly increased the infiltration of CD3^+ T cells and CD8^+ T cells compared with either OVV or PBS, as well as the treatment with OVV-αCD47nb and OVV significantly increased the infiltration of CD4^+ T cells compared with PBS ([80]figure 2E). A similar increase in lymphocyte infiltration was also observed in a CT26 colon cancer model ([81]online supplemental figure S3A). We then evaluated the expression of activation markers (IFN-γ, granzyme B and CD107A) on CD8^+ T cells. Injection of OVV-αCD47nb significantly increased the composition of IFN-γ^+, granzyme B^+ and CD107A^+ cells in the CD8^+ T cells ([82]figure 2F). A similar increase of CD8^+ T activity was also observed in CT26 cancer ([83]online supplemental figure S3B). Besides, OVV-αCD47nb significantly increased the infiltration of macrophages in the 4T1 model as evidenced by flow cytometry. Moreover, OVV-αCD47nb reshaped macrophage polarization from M2 to M1 ([84]figure 2G). Similarly, in a CT26 colon cancer model, OVV-αCD47nb also significantly increased the infiltration of macrophages and polarization of tumor-associated macrophages (TAMs) ([85]online supplemental figure S3C). Altogether, these results indicate that OVV-αCD47nb can reshape the TME by recruiting immune cells, activating the tumor-infiltrating CD8^+ T cells and polarizing TAMs. To assess the tumor-specific T cell responses triggered by OVV-αCD47nb, the percentage of gp70 tetramer-specific CD8^+ T cells in the spleen was monitored by flow cytometry analysis. Following OVV-αCD47nb-treatment, the proportion of tumor-reactive gp70 tetramer-specific CD8^+ T cells rose to a high level of∼8.8% ([86]online supplemental figure S4), indicating the induction of tumor-specific T-cell immunity. Furthermore, we postulated that OVV-αCD47nb might improve the therapeutic outcomes of immune checkpoint inhibitors such as programmed cell death protein-1 (PD-1) antibody. In a breast 4T1 carcinoma subcutaneous tumor model, OVV-αCD47nb robustly enhanced the antitumor efficacy of αPD-1 compared with each single therapy or αPD-1 in combination with OVV ([87]online supplemental figure S5A), which resulted in over 40% (2 out of 5 mice) complete response ([88]online supplemental figure S5B), and the significantly prolonged survival ([89]online supplemental figure S5C). Systemic tumor control effects and inhibition of metastases by OVV-αCD47nb Abscopal effects characterized by the regression of metastases outside the field of intratumoral administration of anticancer agents are necessary for systemic tumor control, which was demonstrated in immunotherapies such as immunostimulatory mAbs, immune cells, and genetically engineered OVs.[90]^33 Given that breast cancer diffuses preferentially to the lung, brain, bone and liver,[91]^34 we next investigate the potential of OVV-αCD47nb to inhibit metastasis to the lung in vivo. After establishing a tumor model by subcutaneous injection of 4T1 cells into BALB/c mice, we administered intratumorally OVV-αCD47nb at a high dose level, with five doses of 1×10^8 pfu ([92]figure 3A). The mice were euthanized at day 8 after the final viral injection, and then both gross observation and examination on H&E-stained lung tissues were performed to evaluate metastatic progression. Unlike lung tissues in the normal mice group, those in the PBS group showed widespread pulmonary nodules, indicating tumor metastasis and invasion. There was no statistically significant effect on the lung metastases although reduction of pulmonary nodules was observed in animals that were treated with OVV. However, OVV-αCD47nb showed significantly decreased pulmonary nodules compared with the PBS and OVV ([93]figure 3B). Furthermore, results obtained from H&E staining of the lung tissues confirmed a marked decline in the metastatic area ([94]figure 3C). To further confirm the generation of systemic antitumor activity by the OVV-αCD47nb, we established a bilateral flank tumor model, in which 4T1 tumors were implanted by injecting 4T1 cells in the left flank on day 0 and in the right flank on day 3. Mice were enrolled in the study when left tumors reached 50 mm^3, and were randomly assigned to receive three doses of 5×10^7 pfu virus (OVV and OVV-αCD47nb), which were injected intratumoral into the left tumors only, or intraperitoneally administration of anti-CD47nb (0.25 mg/kg). Local virus treatments and intratumoral administration of anti-CD47nb enabled therapeutic control on the left tumor. Importantly, the OVV-αCD47nb showed strong tumor control capability compared with the OVV (p=0.0099 at day 18; OVV-αCD47nb vs OVV). In the right flank, only OVV-αCD47nb and anti-CD47nb showed tumor growth delay with 2/5 mice in each group remaining tumor-free long-term ([95]figure 3D). Overall, these data suggest that arming OVV with αCD47nb efficiently elicits specific antitumor responses in non-injected distant and metastatic tumors, which might contribute to the survival advantage ([96]figure 3E). Figure 3. Systemic tumor control and metastases inhibition activity of OVV-αCD47nb (A) Schematic representation of the treatment schedule for 4T1-bearing BALB/c mice metastases model. 2×10^6 4T1 cells were engrafted in situ. Once the tumor reaches 50–100 mm^3, five doses of 1×10^8 pfu of OVV/OVV-αCD47nb were injected intratumoral. PBS injected mice served as controls. 17 days later, mice were sacrificed and their lungs were taken out for photograph (B) and H&E staining (C). The arrows are pointing the dark dots which representing metastatic sites, and the numbers were counted out on the right panel (C). (D) Tumor growth of a bilateral flank tumor model, which was conducted by injecting1×10^5 4T1 cells in the left flank on day 0 and in the right flank on day 3. When the tumors reached 50 mm^3, three doses of 1×10^8 pfu of OVV/OVV-αCD47nb were injected intratumoral into the left tumors only, or intraperitoneally injected anti-CD47nb (0.25 mg/kg). PBS injected mice served as controls. (E) Kaplan-Meier survival curves of the experiments described in (D) and compared by two-side log-rank test, *p<0.05. αCD47nb, anti-CD47 nanobody; OVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline. [97]Figure 3 [98]Open in a new tab scRNA-seq and CyTOF analysis of remodeling of the tumor immune microenvironment induced by OVV-αCD47nb As both OVs and CD47 blockage have been reported to enhance antitumor efficacy through remodeling of the TME,[99]2235,[100]38 we next performed scRNA-seq analysis and CyTOF to characterize the immune landscape in the collected tumor tissues and the potential impact of OVV-αCD47nb on it. scRNA-seq samples were collected from the tumors of 4T1-bearing mice that were treated with OVV or OVV-αCD47nb as described in the experimental procedure ([101]figure 4A). We used unsupervised clustering data analysis to separate a total of 98,667 single cells into 10 clusters based on their markers ([102]online supplemental table 1). Among non-immune clusters, we identified epithelial cells, fibroblasts, endothelial cells, mural cells, osteoclasts, and erythrocytes. The identified immune cell populations consisted of mononuclear phagocytes (MPs), neutrophils, T cells, and mast cells. The majority of cells in the whole dataset of immune cells were MPs, which are significantly increased in the OVV-αCD47nb treated group ([103]figure 4B and C). We further analyzed a total number of 18,803 MP cells into six subgroups ([104]online supplemental table 2): Macrophages, monocytes, conventional type 1 dendritic cells, conventional type 2 dendritic cells, migratory dendritic cells and plasmacytoid dendritic cells. Among these subgroups, above 90% of cells are macrophages ([105]figure 4D). As we know, macrophages can be roughly classified into the antitumor M1 phenotype and the protumor M2 phenotype. We analyzed the transcripts associated with M1 or M2 macrophages and compared these data between OVV and OVV-αCD47nb treated groups ([106]figure 4E). It shows that the majority of M1-related genes are overexpressed in OVV-αCD47nb treated group while M2-related genes are more likely overexpressed in OVV-treated group, especially the marker genes: M1-related genes Cd68 and Cd86 are significantly higher expressed in OVV-αCD47nb treated group, the p values are 0.012 and less than 0.0001 respectively; M2 related marker genes Arg1 and Mrc1(encode CD206 protein) are lower expressed, the p values are both less than 0.0001 respectively ([107]figure 4F). We also used GO (Gene Ontology) analysis to compare the pathway enriched in the treatment with OVV-αCD47nb over OVV ([108]online supplemental table 3). To note that antigen processing and presentation involved pathways are particularly enriched in the OVV-αCD47nb treated group ([109]figure 4G). Figure 4. Single-cell sequencing of tumor-associated cells. (A) Schematic view of the essential input experiment. 1×10^5 4T1 cells were injected into the right flank of BALB/c mice, when the tumor reached 50 mm^3, three doses of 5×10^7 pfu of OVV/ OVV-αCD47nb were injected intratumoral. PBS injected mice served as controls. The tumors were then dissociated into single cells for the following experiments. (B) Cell clusters from 10x Genomics scRNA-seq analysis visualized by Uniform Manifold Approximation and Projection (UMAP). Colors indicate clusters of various cell types. (C) The percentage of mononuclear phagocytes cells in PBS, OVV and OVV-αCD47nb treated groups respectively. (D) Cell clusters further analyzed from mononuclear phagocytes cells and visualized by UMAP. (E) Heatmap displaying the relative expression of M1-related genes and M2-related genes from cluster macrophage of MPs in the treatment with OVV-αCD47nb versus OVV. (F) Violin plots depicting the expression of marker genes in macrophages. (G) GO pathway analysis of macrophage cells from OVV-αCD47nb treated group versus OVV, with −log(10) p value for each pathway. *p<0.05; **p<0.01, ***p<0.001; ****p<0.0001, wilcox.text. αCD47nb, anti-CD47 nanobody; BP, biological process; CC, cellular component; CyTOF, mass cytometry; ECs, endothelial cells; GO, gene ontology; MF, molecular function; MHC, major histocompatibility complex; MPs, mononuclear phagocytes; OVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; scRNA, single-cell RNA. [110]Figure 4 [111]Open in a new tab Further, we analyzed the intratumoral T cells from CyTOF data. CyTOF analysis revealed 15 T-cell clusters ([112]figure 5A and [113]online supplemental table 4). Compared with controls, there were significantly lower levels of three kinds of T-cell subpopulation associated with the OVV, which included PD-1^+CD86^lowCD8^+ T cells (C09), PD-1^lowCD86^−CD8^+ T cells (C10) and BST2^+V-domain immunoglobulin (Ig suppressor of T cell activation (VISTA)^+CD4^+ T cells (C11). We also observed that treatment with OVV or OVV-αCD47nb led to a significant decrease in the numbers of PD-1^+CD86^lowCD8^+ T cells (C08) and PD-1^−Ki67^−CD4^+ T cells (C14). Moreover, treatment with OVV but not OVV-αCD47nb was associated with significant increases in Ly6C^+Ki67^−CD8^+ T cells (C03), PD-1^lowLy6C^+CD11c^midKi67^midCD8^+ T cells (C04), PD-1^lowLy6C^+CD11c^−Ki67^midCD8^+ T cells (C05), and PD-1^lowLy6C^+CD11c^−Ki67^lowCD8^+ T cell subsets (C06). Interestingly, a significant change was seen in Treg (C13), the proportion of which decreased after OVV treatment ([114]figure 5B). In addition, the marker protein CD45 was expressed much higher in the OVV-αCD47nb treated group ([115]figure 5C). Figure 5. CyTOF analysis of tumor-associated cells (A) CyTOF analysis of CD3^+ T-cell clusters in three experimental groups virtualized by t-distributed stochastic neighbor embedding (tSNE) plot, each color represented individual cluster. (B) The frequencies of these T-cell clusters described previously in (A) in three groups. (C) tSNE analysis of intratumoral CD45^+ cells from across all treatment groups. (D) tSNE plot shows clusters identified by macrophage marker gene CD11b^+, each color represented individual cluster. (E) The frequencies of these macrophage clusters described previously in (D) in three groups. (F–I) tSNE plot and Boxplot of marker genes iNOS, CD11b, CD163 and CD172a (SIRPα). P values were derived from paired t-test and unpaired t-test, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. αCD47nb, anti-CD47 nanobody; CyTOF, mass cytometry; iNOS, nitric oxide synthase 2; OVV, oncolytic vaccinia virus; PBS, phosphate-buffered saline; SIRPα, signal-regulatory protein α. [116]Figure 5 [117]Open in a new tab We also investigated changes in macrophages in response to the treatment by CyTOF. 60 cell clusters were identified by cell type specific markers ([118]figure 5D and [119]online supplemental table 5). OVV-αCD47nb treatment was associated with a decrease in macrophages that expressed CCR2, a CCL2 receptor, which is critical for TAM development.[120]^39 Treatment with either OVV-αCD47nb or OVV led to an increase in the frequency of CCR2^− macrophages. OVV increased M1 macrophage frequency in the tumor tissues on average when compared with those in controls ([121]figure 5E). We also noticed that the biomarker of M1 type macrophage[122]^40 iNOS (nitric oxide synthase 2) and CD11b expressed higher in OVV-αCD47nb than others, while the classical M2 type macrophage marker[123]^41 CD163 expressed lower ([124]figure 5F), which is consistent with the data from scRNA-seq. Discussion Both oncolytic virotherapy and anti-CD47 therapy have shown promise as novel classes of drugs to treat advanced tumors. However, despite a substantial investigation into these therapeutic approach, clinical trials have shown that only a small subset of patients benefits from treatment with these agents due to multiple mechanisms of immune evasion, including the complexity and uniqueness of the TME.[125]^42 Recent clinical trials showed that combination treatment with OV and PD-1 antibodies appeared to have greater efficacy than either OV or PD-1 blockage therapy.[126]^43 44 In preclinical models, another combination strategy, in which OV such as oncolytic herpes viruses and myxoma viruses is modified by the integration of gene encoding anti-CD47 antibody so that the OV infects cancer cells and locally releases the anti-CD47 antibody, has been shown to improve cancer treatment via enhancing innate immune response, as well as performing its known oncolytic function and modulation of the TME.[127]^45 In this study, we generated a novel OVV that expresses an αCD47nb, termed OVV-αCD47nb, and demonstrated the specific binding activity of secreted αCD47nb to CD47 on the surface of tumor cells. Both secreted αCD47nb and OVV-αCD47nb blocked the “don’t eat me” signal of macrophages. OVV was selected to construct the OVV-αCD47nb on the basis of its features that including the ability to ensure fast, efficient dissemination of the virus, lack of promoting disease in healthy humans, and induction of immunogenic tumor cell death that is beneficial to the development of adaptive anticancer immunity.[128]^46 It was reported that Hu5F9, a humanized mAb targeting CD47, combined with rituximab showed promising activity in patients with B-cell lymphoma.[129]^18 However, systemic administration of anti-CD47 antibodies demonstrates little to no efficacy across multiple cancer types, including solid tumors.[130]^21 47 48 Other limitations of anti-CD47 antibodies include difficulty in penetrating solid tumors and systemic off-target toxicities that might lead to notable anemia and thrombocytopenia.[131]^48 The ability of OVs to specifically infect cancer cells makes them ideal carriers that are able to deliver transgenes into the TME, thereby enhancing their antitumor efficacy.[132]^31 49 In this study, we demonstrate that OVV-αCD47nb exhibits superior antitumor efficacy compared with either OVV or anti-CD47nb in preclinical breast tumor and colon tumor models. Moreover, intratumoural administration of OVV-αCD47nb led to a significant decrease in the number of metastases and inhibited the growth of distant tumor, thereby conferring survival advantage compared with OVV and anti-CD47nb. Biodistribution studies showed that viral replication could be detected at least for 2 weeks and suggested the presence of αCD47nb within tumor tissues for the full observation period of up to 17 days post first injection of OVV-αCD47nb. These results will help elucidate the mechanisms underlying the enhanced antitumor ability of OVV-αCD47nb. Recently, a few studies have highlighted the ability of OVVs to increase tumor-specific effector and memory T cells, which can boost the killing of the virus at the site of viral injection, and induce systemic anticancer activity.[133]^2 50 51 Here we show that OVV or OVV-αCD47nb induced an increase in CD3^+ T-cell infiltration in intratumoral regions. Using CyTOF analysis, we observed that OVV or OVV-αCD47nb treatment not only led to endogenous CD8^+ T-cell toward a less exhaustion phenotype and upregulated markers of T cell activation (granzyme B and IFNγ), but also significantly decreased the frequency of Treg subset. Whereas, the tumors from the control showed a high frequency of PD-1^+CD86^lowCD8^+ T cells and BST2^+VISTA^+CD4^+ T cells. VISTA, a potent negative regulator of T-cell function is overexpressed within the TME.[134]^52 53 Taken together, the results obtained in our study suggest that effective T-cell-mediated immune responses contribute to enhanced antitumor efficacy. CD47 is highly upregulated in different types of solid tumors, which enable cancer cells to evade phagocytosis by macrophages and promote the cancer stem cell phenotype.[135]^7 54 OVV-αCD47nb selectively infected cancer cells and replicated inside cancer cells, followed by high intratumoral CD47 transgene expression, leading to better antitumor function. In the 4T1 model, the most significant finding was that treatment with OVV-αCD47nb induced an increased MP and CD11b positive macrophages population within tumors, and could lead to upregulation of antitumor M1 related genes and proteins, such as Cd68, Cd86 and iNOS. Furthermore, both OVV-αCD47nb and OVV inhibited CD206-positive M2 macrophages. These data suggest functionally activated macrophages and TAM re-education play a strong antitumor effect on mediating our OVV-expressing αCD47nb. In summary, we have developed an OVV platform that provides intratumoral delivery of αCD47nb. Our findings suggest the therapeutic potential of OVV-αCD47nb for breast and colon cancer, and show that OVV-αCD47nb improves the immune cells profiles within tumors. This has established a rationale for further exploring OVV-αCD47nb as a potential therapy in the clinic. Materials and methods Cell lines HEK293, Hela-S3, 4T1, EL4, CT26 and MC38 cell lines were purchased from the American Type Culture Collection (ATCC; Manassas, USA). Dulbecco’s modified Eagle’s medium (DMEM; Cat# 11965092, Gibco-Thermo Fisher Scientific, USA) was used to cultivate HEK293, Hela-S3, 4T1, and CT26 cells. 10% fetal bovine serum (FBS; Cat# 16000044, Gibco) was added as a supplement. RPMI 1640 media (Cat# 11875093, Gibco) supplemented with 10% FBS (Cat# 16000044, Gibco) was used to cultivate EL4 cells. Hela-S3 cells were grown in suspension in spinner flasks (Cat# TCB002002, JetBiofil, Guangzhou, China) using a serum-free medium (Cat# H740KJ, Basalmedia, Shanghai, China). Every cell was cultured in an environment containing 5% CO2 at 37°C. Oncolytic vaccinia viruses As previously mentioned, the control VV came from our laboratory deposit.[136]^31 GenScript (Nanjing, China) produced the αCD47nb gene fragment, which was then subcloned into the shuttle plasmid pVV-control to create the recombinant plasmid pVV-αCD47nb. This plasmid contains a synthesized early/later promoter (pSE/L) that drives the expression of αCD47nb, and a p7.5K early/later promoter of VV that drives the expression of EGFP (enhanced GFP; reporter gene) and GPT (guanine-hypoxanthine phosphoribosyl transferase; screening gene). Using a western reserve strain of VV (WR-VV; Cat# VR-1354; ATCC) and the shuttle plasmid pVV-αCD47nb, homologous recombination was used to create OVV-αCD47nb. In summary, HEK293 cells were transfected with pVV-αCD47nb using the Lipofectamine 3000 transfection reagent (Cat# L3000015, Thermo Fisher, USA) after being infected with WR-VV for 2 hours at a MOI of 1. The EGFP-positive plaques were selected and sown in plates containing Hela-S3 cells 48 hours later. The growth of WR-VV was inhibited using the conditional DMEM media containing 250 µg/mL xanthine (Cat# A601197, Sangon Biotech, Shanghai, China), 25 µg/mL mycophenolic acid (Cat# A600640, Sangon), and 15 µg/mL hypoxanthine (Cat# A500336, Sangon). Both PCR and DNA sequencing were used to verify that the recombinant virus was no longer contaminated with WR-VV after many picking and planting cycles. Next, Hela-S3 cells were used to progressively enlarge the isolated virus in 6-well plates, cell culture dishes, and cell culture spinner flasks. A TCID50 technique was used to determine the viral titer. The following is the calculating formula: virus titer=0.7×10×10^(1+S) (D−0.5), where D is the total of the EGFP positive ratios in each dilution and S is log[10] (dilution). Western blot analysis In 6-well plates, tumor cells were planted at a density of 5×10^5 cells per well, and OVV-αCD47nb or OVV at a MOI of 1. The cell culture supernatants were collected after a 48-hour incubation period. 10 μL of the supernatants were removed and combined in equal amounts with a 2×loading buffer (Cat# P0015, Beyotime, Shanghai, China). After 5 min of heating at 100°C, the protein samples were placed onto an SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) gel for electrophoresis. A semidry membrane transfer device was used to move the protein from the PAGE gel onto a polyvinylidene fluoride (PVDF) membrane (Cat# K5MA6539B, Merck Millipore, Germany) following electrophoresis. After that, the PVDF membrane was treated for 1 hour at room temperature with goat anti-mouse IgG (H+L) from HuaAn Biotechnology that had been labeled with horseradish peroxidase and mouse anti-Flag antibody (Abcam, ab205606). The protein bands were visible using an upgraded chemiluminescent kit (Cat# FD8030, FDbio-Femto, Hangzhou, China) following the incubation. Viral replication HeLa-S3 cells were plated in 24-well plates with 1×10^4 cells per and incubated at 37°C in an environment containing 5% CO2. OVV-αCD47nb and OVV were introduced to the wells at a MOI of 0.1 after the cells had grown to>90% confluency. Following a 24, 48, 72, and 96-hour culture period, the cells underwent three cycles of freeze-thaw lysing before being centrifuged for 10 min at 3000×g. After harvesting the supernatants, the viral titer was ascertained using the TCID50 technique. The fold change was computed in relation to the 12-hour viral titer. Binding assays The 4T1 and CT26 tumor cells, which are CD47^+, were used in binding experiments. The OVV-infected supernatants were incubated on ice for 1 hour with target cells (1×10^5). Using flow cytometry and the APC anti-DYKDDDDK Tag antibody (Cat# 637308, BioLegend, USA), αCD47nb binding was assessed. Phagocytosis The macrophages were subjected to phagocytosis analysis as previously reported.[137]^55 In summary, after receiving the recommended treatments in 1% FBS medium for 24–48 hours, in vitro differentiated BMDMs were plated on a U-bottom ultralow attachment 96-well plate at 37°C for 20 min to allow them to rest. In order to perform ex vivo phagocytosis of tumor cells using BMDMs, the cells were cultured for 48 hours in 1% FBS media containing a vehicle, cholesterol (10 µg/mL), and LXR-623 (5 µM). After being extracted, these BMDMs were cocultured for an additional 4–12 hours at 37°C with 4T1 (1:2) cells in a U-bottom ultralow attachment 96-well plate in full media. Following cell harvesting, all cells were treated for 10 min at room temperature with cold Trypsin-EDTA to break down surface adhesion. RTCA assays The time-dose response curve of OVV-αCD47nb, OVV and anti-CD47 antibody on the proliferative activity of experimental cell lines was found using RTCA. After seeding cells in a 16-well E-plate with 200 µL of 1×10^4 cells/well, the plates were left at room temperature for half an hour. Using the Agilent xCELLigence RTCA S16 equipment (ACEA Biosciences, San Diego, California, USA), real-time dynamic cell proliferation detection was carried out while the E-plate was positioned on the detection table within an incubator. The new media were switched out after a 24-hour culture period. The supernatants from OVV-αCD47nb or OVV-infected VERO cells were collected 48 hours after OVV-αCD47nb or OVV infection. The collected supernatants were then filtered through a 0.45 µm filter membrane and concentrated through an ultrafiltration tube (Amicon Ultra). After adding the produced gradient viral supernatants to each well, real-time dynamic detection was carried out. Five days after the virus infection, real-time cell index data were still being recorded. Animal experiments All animal experiments were approved by the Ethics Committee for Animal Experimentation from Zhejiang Provincial People’s Hospital. The 4T1 tumor cells were subcutaneously inserted into the right flank of the mice in order to create subcutaneous tumor models. The mice were placed in random groups, treated intratumorally with OVV-αCD47nb, OVV, or PBS, and intraperitoneally with αCD47nb injections after the tumor reached a size of about 50–100 mm^3. Every 2 or 3 days, tumor diameters were measured with a vernier caliper, and the tumor volume was computed using the formula 0.5×length×width^2. For the combination therapy with OVV-αCD47nb and anti-mouse PD-1 antibody (αPD1, Clone RMP1-14, Cat# BE0146, Bio X Cell), a subcutaneous 4T1 model was established as previously described. Each mouse was injected intraperitoneally with 200 µg of αPD1, which was initiated from the next day of viral treatment. Flow cytometry BioLegend provided the following antibodies: PE anti-mouse CD45 (Clone 30-F11, Cat# 103106), APC anti-mouse CD3 (Clone 17A2, Cat# 100236), FITC anti-mouse CD4 (Clone GK1.5, Cat# 100406), PE anti-mouse CD4 (Clone GK1.5, Cat# 100408), FITC anti-mouse CD8α (Clone 53–6.7, Cat# 100706), APC/Fire 750 anti-mouse IFN-γ (Clone XMG1.2, Cat# 505860), Biotin anti-mouse CD107a (Clone 1D4B, Cat# 121603), APC anti-mouse/human CD11b (Clone M1/70, Cat# 101212), Percp anti-mouse F4/80 (Clone BM8, Cat# 123126), Percp anti-mouse CD86 (Clone GL-1, Cat# 105026), APC anti-mouse CD80 (Clone 16–10 A1, Cat# 104713), FITC anti-mouse CD206 (Clone C068C2, Cat# 141704), PE/Fire 700 anti-mouse CD206 (Clone C068C2, Cat# 141741). For the preparation of single-cell suspensions of tumor tissues, mice were anesthetized and sacrificed, and the tumor tissues were harvested and placed in serum-free medium (Cat# H740KJ, Basalmedia) with 0.2% collagenase IV (Cat# C4-22-1G, Sigma-Aldrich, Germany). The tumor tissues were then cut into 1–2 mm pieces, digested for 2 hours, and passed through 70 µm nylon filters (Cat# CSS013070, JetBiofil, Guangzhou, China) to obtain single-cell suspensions. Then, collagenase was removed by centrifugation and the cell pellets were suspended in a serum-free medium and adjusted to 2×107 cells/mL. Data analysis was performed using FlowJo software (Tree Star; Oregon, USA). The tetramer staining analysis through peptide-major histocompatibility complex tetramer tagged using PE was performed to investigate the percentage of antigen-specific CD8^+ T cells. In brief, splenocytes were resuspended with the solution of anti-CD16/32 mAb (BioLegend) for blocking FcR-mediated and non-specific antibody binding. Then, the suspension was incubated for 10 min at 25°C and washed three times using FACS buffer. Subsequently, H-2Ld MuLV gp70 tetramer-SPSYVYHQF-PE (MBL International) was added into the samples and incubated for 20 min at room temperature. Next, fluorescence-labeled antibodies, CD3 (BioLegend) and CD8 (BioLegend) were added and incubated for 20 min on ice. Afterward, the samples were washed two times using FACS buffer. Stained cells were monitored using a flow cytometer (BD FACSCanto) and evaluated using FlowJo software. Quantitative PCR Under liquid nitrogen, frozen tumor samples were broken up with a mortar and pestle. Using the DNA/RNA/protein kit (IBI Scientific), 26 mg of homogenized tissue were separated into RNA and DNA. Thermo Fisher Scientific’s TURBO DNA-free kit was used to treat RNA samples in order to eliminate any remaining genomic DNA. The High-Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific) was used to reverse transcribe 1 µg of RNA. An analysis was carried out in real time using a LightCycler 480 Instrument II (Roche). In the presence of Roche’s SYBR Green I Master, 40 ng of cDNA and 100 ng of DNA were used, respectively, to quantify the αCD47nb transcripts and viral genomes in the tumor. PCR conditions were: 95°C 3 min, 40 cycles of 95°C 10 s, 60°C 30 s and 72°C 30 s, then 4°C 10 s. Viral genome primers were A56F: 5′-: CTGGATCTACACATTCACCGGA-3′ and A56R: 5′- CGGAGTCTCGTCTGTTGTGG −3′ and αCD47nb primers were F: 5′- GGTATTCTTGGTGGCTCTTT −3′ and R: 5′- GTACCATCCCATGTCGTTG −3′. Single-cell RNA sequencing The fresh tumor tissues were isolated from treated mice and were stored in the sCelLiVE Tissue Preservation Solution (Singleron) on ice after the surgery within 30 min. The specimens were washed with Hanks Balanced Salt Solution three times, minced into small pieces, and then digested with 3 mL sCelLiVE Tissue Dissociation Solution (Singleron) by Singleron PythoN Tissue Dissociation System at 37°C for 15 min. The cell suspension was collected and filtered through a 40-micron sterile strainer. Afterward, the GEXSCOPE red blood cell lysis buffer (RCLB, Singleron) was added, and the mixture (Cell: RCLB=1:2 (volume ratio)) was incubated at room temperature for 5–8 min to remove red blood cells. Following a 5 min centrifugation at 300×g 4°C to extract the supernatant, the mixture was gently suspended in PBS. Ultimately, Trypan blue was used to stain the samples, and a microscope was used to assess the cell viability. Using the Singleron Matrix Single Cell Processing System, microwell chips were loaded with single-cell suspensions (2×105 cells/mL) composed of PBS (HyClone). After the barcoding beads are removed from the microwell chip, PCR amplification is performed and the mRNA that the barcoding beads captured is reverse transcribed to produce cDNA. After that, sequencing adapters are used to segment and ligate the amplified cDNA. The GEXSCOPE Single Cell RNA Library Kits (Singleron) procedure was followed in the construction of the scRNA-seq libraries.[138]^56 Individual libraries were diluted to 4 nM, pooled, and sequenced on Illumina NovaSeq 6000 with 150 bp paired end reads. Cell typing, single-cell gene expression and pathway enrichment analysis Gene expression matrices were produced by processing raw readings from scRNA-seq using the CeleSCOPE ([139]https://github.com/singleron-RD/CeleScope) V.1.9.0 pipeline. In a nutshell, low-quality reads were eliminated from the raw reads using CeleSCOPE, and poly-A tail and adapter sequences were trimmed using Cutadapt V.1.17.[140]^57 UMI and the cell barcode were retrieved. Next, we mapped reads to the reference genome GRCm38 (Ensembl V.92 annotation) using STAR V.2.6.1a.[141]^58 Using the featureCounts V.2.0.1 program, UMI counts and gene counts of every cell were obtained. These counts were then used to create expression matrix files for further analysis.[142]^58 59 Gene counts less than 200, the top 2% of gene counts, and the top 2% of UMI counts were used to screen the cells. The cells containing more than 20% mitochondria were eliminated. Tumor tissue cells were kept for further analysis after filtration. Dimension reduction and grouping were performed using routines from Seurat V.3.1.2.[143]^60 Next, we selected the top 2000 variable genes for PCA analysis using the Find Variable Features function after normalizing and scaling all gene expressions using the Normalize Data and Scale Data functions. With Find Clusters, we divided the cells into many clusters based on the top 20 principal components. By Harmony, the batch effect between samples was eliminated.[144]^61 Lastly, cells were shown in a two-dimensional space using the Uniform Manifold Approximation and Projection technique. To compare the cell distributions of the two groups, unpaired two-tailed Wilcoxon rank-sum tests were used. Unpaired two-tailed Student’s t-test was used to compare gene expression or gene signatures between two cell groups. Wilcoxon rank-sum tests with paired two-tailed samples were used to compare the cell distributions of groups 1 and 2. R was used for all statistical analysis and presentation. Statistical significance was set at p<0.05, and the statistical tests employed in the figures were indicated in the figure legends. The figures and figure legends displayed the precise value of n as well as what it stands for. We identified the genes expressed in more than 10% of the cells in a cluster and with an average log (Fold Change) value greater than 0.25 as differentially expressed genes (DEGs) using the Seurat Find Markers program based on the Wilcox likelihood-ratio test with default parameters. We used information from the literature to combine the expression of canonical markers found in the DEGs with the cell type annotation of each cluster. The resulting cell type expression was then displayed using heatmaps, dot plots, and violin plots created with the Seurat DoHeatmap/DotPlot/VlnPlot function. Doublet cells were manually eliminated once it was determined that they expressed markers for various cell kinds. Using the SynEcoSys database and the expression of canonical markers identified in the DEGs, the cell type identity of each cluster was ascertained. Using Seurat V.3.1.2 DoHeatmap/DotPlot/VlnPlot, heatmaps, dot plots, and violin plots showing the expression of markers required to identify each type of cell were produced. To investigate the potential function of DEGs, the GO analysis was used with “clusterProfiler” R package V.3.16.1.[145]^62 Pathways with a p adjusted value less than 0.05 were considered as significantly enriched. GO gene sets including molecular function, biological process, and cellular component categories were used as reference. CyTOF data acquisition and analysis The separated tumor tissue cells were stained with a cocktail of surface antibodies for 30 min on ice after being treated in an Fc receptor-blocking solution. After twice washing the cells in FACS buffer (1×PBS+0.5% BSA), the cells were fixed for an overnight period in 200 µL of an intercalation solution (Maxpar Fix and Perm Buffer, Fluidigm) containing 250 nM 191/193Ir. Following fixation, cells were stained with an intracellular antibody cocktail for 30 min on ice before being twice cleaned with FACS solution and then Perm Buffer (eBioscience). After being cleaned and resuspended in deionized water, the cells were collected using a mass cytometer (Helios, Fluidigm) and added to 20% EQ beads (Fluidigm). Using a doublet-filtering approach,[146]^63 each sample’s data were debarcoded from the raw data using distinct mass-tagged barcodes. Bead normalization was used to standardize each.fcs file produced by several batches.[147]^64 Using FlowJo software, manually gate data to remove detritus, dead cells, and doublets, leaving only live, single immune cells. Using the PhenoGraph clustering method,[148]^65 divide all cells into discrete phenotypes according to the amounts of marker expression. On a heatmap of cluster versus marker, annotate each cluster’s cell type based on the pattern of marker expression. To visualize the high-dimensional data in two dimensions and display the distribution of each cluster, marker expression, and the difference between each group or sample type, use the dimensionality reduction algorithm t-SNE. [149]^66 Statistical analyses For all statistical studies, Prism V.8.2.1 (GraphPad Software, California, USA) was used. The statistical differences between the groups were investigated using the analysis of variance. The Kaplan-Meier method was used to generate the survival curve, and the log-rank test was used to assess the statistical significance of the differences between the groups. In every statistical analysis, a value of p<0.05 was considered statistically significant. supplementary material online supplemental file 1 [150]jitc-12-12-s001.docx^ (1,018.3KB, docx) DOI: 10.1136/jitc-2024-009473 online supplemental file 2 [151]jitc-12-12-s002.xlsx^ (8.9KB, xlsx) DOI: 10.1136/jitc-2024-009473 online supplemental file 3 [152]jitc-12-12-s003.xlsx^ (8.7KB, xlsx) DOI: 10.1136/jitc-2024-009473 online supplemental file 4 [153]jitc-12-12-s004.xlsx^ (10.8KB, xlsx) DOI: 10.1136/jitc-2024-009473 online supplemental file 5 [154]jitc-12-12-s005.xlsx^ (8.9KB, xlsx) DOI: 10.1136/jitc-2024-009473 online supplemental file 6 [155]jitc-12-12-s006.xlsx^ (10.5KB, xlsx) DOI: 10.1136/jitc-2024-009473 Footnotes Funding: This work was supported by Foundation of Science Technology Department of Zhejiang Province (LGF22H080012), Zhejiang Provincial Medical Technology Plan Project (No. 2022KY569), National Natural Science Foundation of China (82373308), the Natural Science Foundation of Xiamen, China (3502Z20227251), Xiamen Ocean and Fisheries Development Special Funds Program (22CZP006HJ03). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: All animal procedures and experiments were performed following the guidelines that had been approved by the Animal Care and Use Committee of Zhejiang Provincial People’s Hospital (A202100057). Contributor Information Zengpeng Li, Email: lizengpeng@tio.org.cn. Mengyuan Li, Email: 22018121@zju.edu.cn. Liu Yang, Email: yangliu@hmc.edu.cn. Jie Chen, Email: chenjie@rongu-bio.com. Qian Ye, Email: yeqian@rongu-bio.com. Wenbin Qian, Email: qianwb@zju.edu.cn. Shibing Wang, Email: wangshibing@hmc.edu.cn. Data availability statement Data are available upon reasonable request. References