Immune cell typing across human and mouse tumors

Sixty to seventy percent of patients do not respond to anti-PD-1 therapy due to the complexity of the TME.23 The next wave of co-inhibitory targets, including CD24, CD47, CD155, LAG3, CD276, CD39, CD73, adenosine A2A receptor, etc. is being explored in clinical development. Targeting immune checkpoints in macrophages could restore the phagocytic activity of macrophages and prevent tumor relapse and progression.

Therefore, ICI therapy should be based on the main population of infiltrating mononuclear cells, such as T cells or macrophages. We first explored the immune landscapes of four types of tumors, including glioma, colon cancer, breast cancer and lung cancer. Tumor microarrays of glioma (n = 88), colon cancer (n = 23), breast cancer (n = 62) and lung cancer (n = 24) samples were used to identify B cells (anti-CD20), CD4+ T cells (anti-CD4), CD8+ T cells (anti-CD8), regulatory T cells (Tregs, anti-Foxp3), macrophages (anti-CD68) and myeloid-derived suppressor cells (MDSCs, anti-Arginase-1) by multiplex immunohistochemistry (mIHC) (Fig. 1a and Supplementary Fig. 1a–d). MDSCs were the most abundant constituent of all four types of tumors (Fig. 1b). Glioma, colon cancer and breast cancer had a higher proportion of macrophages than CD8+ T cells, but a high proportion of CD8+ T cells was observed in lung cancer (Fig. 1b). We further examined the expression of immune checkpoints (CD24, CD47, CD155, HLA-DQB1, LGALS9, CD276, LAG3, ADORA2A, CD73, TIM3, CD39, CD80, CD86, and PD-L1) in various tumors. RNA-seq data from TCGA (http://www.cbioportal.org/) revealed high expression of CD24, CD47, and CD155 in most cancers (Fig. 1c and Supplementary Table 1). The mRNA expression of CD24, CD47 and CD155 was further verified in colon adenocarcinoma (COAD), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD) and stomach adenocarcinoma (STAD) patient samples via cDNA microarray. The results showed that the mRNA expression of CD24, CD47, and CD155 was presented in these four cancer types, which was consistent with TCGA results (Supplementary Fig. 2a). The correlation analysis between the immune checkpoints and the immune cells was further studied. Increased expression of Siglec10, SIRPα, and TIGIT was significantly associated with the infiltration levels of neutrophils and dendritic cells (DCs) in almost all tumor types except THYM. In addition, the expression of Siglec10 and SIRPα had strong associations with macrophages while TIGIT had weak associations with macrophage among almost all tumor types. In addition, TIGIT had high relativity with CD8+ T cells (Supplementary Fig. 2b). High expression of CD24, CD47, and CD155 was associated with short overall survival (OS) in cancers (Fig. 1d). Based on the composition and phenotypic states of intratumoral immune cells in the different tumor types and even the expression of ligands, CD47 and CD24 were rationally selected as targets to restore macrophage-mediated phagocytosis in macrophage-dominated tumors, while CD155 was used as a target to prevent T-cell exhaustion in CD8+ T-cell-dominated tumors.

Fig. 1

Immune cell typing across human and mouse TMEs. a, b The immune landscape was identified by mIHC staining of human glioma, colon cancer, breast cancer and lung cancer tumor microarrays. c, d Heatmap of immune checkpoint molecules in tumors to matched normal expression ratios (log2 (TPM + 1)) (c) and OS of patients with tumors (d). e The expression of CD24, CD47 and CD155 in MC38, 4T1, and CT26 tumor cells was detected by flow cytometry. f C57BL/6J mice were subcutaneously inoculated with 1 × 106 MC38 cells. BALB/c mice were subcutaneously inoculated with 1 × 106 4T1 cells or 1.5 × 106 CT26 cells. When tumor sizes reached ~100 mm3, single-cell suspensions were prepared from the mouse tumors prior to analyzing the composition of immune cells by flow cytometry. Data are represented as mean ± SD

The murine cancer cell lines MC38, 4T1, and CT26 were chosen to investigate targeted therapies. MC38, 4T1, and CT26 cells all showed high expression of CD47 and CD155, but only 4T1 cells expressed CD24 (Fig. 1e). Furthermore, the landscapes of infiltrated immune cell types in the three tumors were different, with a high abundance of monocytic MDSC (mMDSC) in MC38 and CT26 tumors and a high abundance of granulocytic MDSC (gMDSC) in 4T1 tumors. In addition, tumor-associated macrophages (TAMs) accounted for a large proportion of immune cells in 4T1 and MC38 tumors, but CD8+ T cells and NK cells predominated in CT26 tumors (Fig. 1f). Although highly abundance of MDSCs (including gMDSC and mMDSC) were found in TME, but macrophage-targeting strategies show better therapeutic efficacy than MDSCs.24 Therefore, using SIRPα-Fc to block CD47 or Siglec10-Fc to block CD24 in macrophage-dominated MC38 and 4T1 tumors or TIGIT-Fc to block CD155 in CD8+ T-cell-dominated CT26 tumors may be a rational strategy for cancer immunotherapy.

Generation and characterization of OAd-SIRPα-Fc, OAd-Siglec10-Fc, and OAd-TIGIT-Fc

Tumor selectivity was conferred to replication-competent OAds by inserting a modified hTERT (mhTERT) promoter to drive the expression of the E1 gene in which a 24-bp sequence in the E1A region and an E1B55-kD viral protein in the E1B region were deleted (Fig. 2a).25,26 The mhTERT promoter produced significantly higher luciferase gene activity than the wild-type hTERT promoter (wt-hTERT) in mouse tumor cells (GL261, MC38, LL/2, 4T1, and CT26) but a low level of luciferase gene activity in normal mouse 3T3-L1 cells (Supplementary Fig. 3).

Fig. 2

Generation and characterization of OAd-SIRPα-Fc, OAd-Siglec10-Fc, and OAd-TIGIT-Fc. a Schematic representation of OAd structures. b Transmission electron microscopy view of OAd-SIRPα-Fc, OAd-Siglec10-Fc, and OAd-TIGIT-Fc (scale bar: 50 nm). c The oncolytic potency of OAd-SIRPα-Fc, OAd-Siglec10-Fc, and OAd-TIGIT-Fc was evaluated against MC38, 4T1, CT26, LL/2, and GL261 tumor cells. Nontumor 3T3-L1 cells were used as a negative control. Ad is a replication-deficient adenovirus without the mhTERT promoter and E1A/E1B genes. d The expression and secretion of SIRPα-Fc, Siglec10-Fc, and TIGIT-Fc into the supernatant from the indicated OAd-infected MC38 tumor cells were detected by western blotting under reducing conditions. e Wild-type cells or ligand-knockdown cells were incubated with purified SIRPα-Fc, Siglec10-Fc and TIGIT-Fc from corresponding virus-infected tumor cells and then stained with anti-IgG Fc for flow cytometric detection. IgG acted as a negative control. f, g HEK293 cells were infected with OAd-SIRPα-Fc or OAd-Siglec10-Fc for 72 h, and then the supernatant was collected. pHrodo (red)-labeled MC38 (f) or 4T1 cells (g) were incubated with supernatants containing SIRPα-Fc or Siglec10-Fc for 1 h and then cocultured with CDFA-SE-labeled M1-BMDMs. Colocalization of the two cell types demonstrated phagocytosis. The supernatant from OAd-null-infected HEK293 cells was used as a control. (***p < 0.001)

OAd-SIRPα-Fc, OAd-Siglec10-Fc and OAd-TIGIT-Fc were engineered by introducing the murine soluble SIRPα, Siglec10 or TIGIT extracellular domains, respectively, fused with IgG1 Fc by the CMV promoter into the ∆E3 region of OAd-null, which had oncolytic and immune checkpoint blockade functions (Fig. 2a). The hexagonal structures and their fiber dots on the surface of purified OAds were clearly visible by transmission electron microscopy (Fig. 2b). These engineered OAds were able to selectively kill multiple mouse tumor cell lines, demonstrating that the insertion of transgene cassettes did not interfere with the infection or replication of OAds in vitro (Fig. 2c). Immunoblot analysis showed that tumor cells infected with OAd-SIRPα-Fc, OAd-Siglec10-Fc or OAd-TIGIT-Fc efficiently secreted the corresponding fusion protein as a dimer into the supernatant in vitro (Fig. 2d and Supplementary Fig. 4). To determine the binding affinities of secreted SIRPα-Fc, Siglec10-Fc and TIGIT-Fc for their ligands, we purified these proteins from the supernatants of corresponding virus-infected tumor cells and constructed CD47-knockdown MC38 cells, CD24-knockdown 4T1 cells and CD155-knockdown CT26 cells using shRNA (Supplementary Fig. 5). Purified SIRPα-Fc was capable of binding to CD47+ MC38 cells but not to CD47-knockdown cells (Fig. 2e). Similarly, the results of the Siglec10-Fc and TIGIT-Fc binding assays demonstrated specific affinity for the corresponding ligand (Fig. 2e). Furthermore, the specific binding of purified SIRPα-Fc and Siglec10-Fc to CD47 or CD24, respectively, enhanced macrophage‐mediated tumor cell phagocytosis in vitro (Fig. 2f, g). Taken together, these data demonstrate that the armed OAd-SIRPα-Fc, OAd-Siglec10-Fc and OAd-TIGIT-Fc can selectively lyse tumor cells and secrete high levels of functional fusion proteins.

Precise antitumor activities against primary tumors

Our results demonstrated that MC38 and 4T1 tumors were rich in macrophages, while CT26 tumors were rich in CD8+ T cells and NK cells. Moreover, OAd-SIRPα-Fc and OAd-Siglec10-Fc were used to target macrophages. OAd-TIGIT-Fc was employed to target CD8+ T cells and NK cells. To characterize the precise effects of OAds, we evaluated antitumor activity in three tumor models established with immunocompetent mice treated with an intratumoral injection of OAd-null, OAd-SIRPα-Fc, OAd-Siglec10-Fc, or OAd-TIGIT-Fc. PBS acted as a control (Fig. 3a).

Fig. 3

Assessment of the functional states of tumor-infiltrating TAMs and T cells in the MC38 model. a Schematic flow diagram of the precision treatment strategy with different OAds for different tumor types. b–d Tumor-bearing mice were intratumorally injected with 50 μL of OAds (1 × 108 pfu per tumor) for the MC38 (b) and CT26 models (d) or with OAds (3 × 108 pfu per tumor) for the 4T1 model (c) on days 1, 4, 7, 10 and 13. The treatment regimens for PBS, Ad and OAd-null in the corresponding models were consistent with those for the armed OAds. Tumor volume was monitored. e–g MC38 tumor-bearing mice were treated with PBS, OAd-null, OAd-SIRPα-Fc, OAd-Siglec10-Fc, or OAd-TIGIT-Fc. Two days after the third OAd dose, tumor issues were profiled by scRNA-seq. UMAP plot of all single cells in the MC38 model and histogram indicating the proportions of cell clusters in tumor tissues (f). UMAP plots of identified macrophages and frequencies of macrophage subsets from the scRNA-seq analysis (e). UMAP plots of identified T cells and frequencies of T-cell subsets from the scRNA-seq analysis (g). h Violin plots showing comparisons of C1qc+ and Spp1+ score levels among macrophage subclusters. Heatmap of enriched KEGG pathways in TAM clusters. i Relative average expression of canonical marker genes across different T-cell clusters. j Developmental trajectory of CD8+ T cells inferred by Monocle 2. (*p < 0.05, **p < 0.01, ***p < 0.001)

OAd-null treatment markedly reduced tumor sizes compared with PBS treatment in the MC38 and 4T1 models (Fig. 3b, c). OAd-SIRPα-Fc treatment conferred better antitumor activity than OAd-Siglec10-Fc and OAd-TIGIT-Fc in the MC38 model (Fig. 3b), but OAd-Siglec10-Fc showed satisfactory tumor suppression in the 4T1 model (Fig. 3c). However, OAd-Siglec10-Fc and OAd-null had similar efficacies, which was likely caused by the lack of CD24 expression in the CT26 model (Fig. 3d). Compared with the other treatments, OAd-TIGIT-Fc showed the best antitumor activity in the CT26 model (Fig. 3d). These data demonstrated that SIRPα-Fc blocking of CD47 or Siglec10-Fc blocking of CD24 significantly suppressed the growth of tumors with a macrophage-dominated TME, while TIGIT-Fc blocking of CD155 showed promising antitumor activity in CD8+ T-cell- and NK cell-dominated tumors.

To better understand the mechanism associated with precise tumor regression achieved with the different OAd treatments, we performed comprehensive and unbiased scRNA-seq of MC38 and CT26 tumors to examine the cellular transcriptomic changes in the TME. Two days after the third injection, tumors were collected, and scRNA-seq analysis was performed on MC38 and CT26 tumors treated with PBS, OAd-null, OAd-SIRPα-Fc, OAd-Siglec10-Fc, or OAd-TIGIT-Fc. Following gene expression normalization for read depth and mitochondrial read count, we obtained high-quality expression data for 33758 cells from MC38 tumors and 30215 cells from CT26 tumors.

In MC38 tumors, after unbiased cell type classification using Seurat v4, 7 cell clusters were identified based on marker gene expression; these clusters included Non-immune cells, B cells, T cells, macrophages, neutrophils, and NK cells (Fig. 3f and Supplementary Fig. 6a). Compared to PBS-treated tumors, tumors treated with one of the four OAd treatments, especially OAd-SIRPα, had increased proportions of T cells (Fig. 3f). We calculated scores for macrophage, T-cell and NK cell clusters by using the Seurat function AddModuleScore to analyze functional states (Supplementary Table 2). OAd-SIRPα-Fc and OAd-Siglec10-Fc showed higher C1qc+ scores than OAd-null (Supplementary Fig. 7a). We did not find a significant difference in the C1qc+ score between the OAd-null and OAd-TIGIT groups. However, OAd-SIRPα-Fc showed a lower Spp1+ score than OAd-null. Furthermore, OAds treatments showed higher cytotoxicity scores for the T-cell and NK cell clusters than PBS treatment (Supplementary Fig. 7b, c). However, OAd-SIRPα-Fc showed a lower exhaustion score for the T-cell and NK cell clusters (Supplementary Fig. 7b, c). These data indicated improvements in antitumor activity following OAd-SIRPα-Fc treatment.

Based on the expression of canonical markers, we annotated macrophages into three subtypes (TAM-C1, TAM-C2 and TAM-C3) (Fig. 3e and Supplementary Fig. 6b). TAM-C2 was enriched in OAds treatments, especially OAd-SIRPα-Fc, compared with PBS treatment. TAM-C1 was markedly decreased in the OAd-SIRPα-Fc group compared with the other four groups (Fig. 3e). To better understand the roles of these populations, we further calculated C1qc and Spp1 gene signatures (Fig. 3h). TAM-C1 showed a high Spp1+ score, while TAM-C2 showed a high C1qc+ score. Interestingly, we noticed significant enrichment of gene expression signatures in the proinflammatory phenotype, such as antigen processing and presentation, HIF-1 signaling, TNF-α signaling, and IL17 signaling, in TAM-C2 compared to TAM-C1 (Fig. 3h). Importantly, our data show that the lysosomal and phagosomal pathways of TAM-C2 were more enriched than those of TAM-C1, which possessed the characteristics of M2 macrophages indicated by enrichment of gene expression signatures of oxidative phosphorylation.27

We further performed unsupervised clustering of T cells and obtained 12 clusters: CD3-C1 to CD3-C4, CD4-C1 to CD4-C2, and CD8-C1 to CD8-C6 (Fig. 3g and Supplementary Fig. 6c). The abundance of CD3-C1, CD4-C2 and CD8-C5 was significantly decreased in the OAd treatments compared to the PBS treatment. However, the relative percentage of CD8-C2 in the OAd treatment groups was increased compared with that in the PBS group. Interestingly, the abundance of CD8-C6 was increased with only OAd-SIRPα-Fc treatment compared with other treatments. Based on the expression of canonical markers, we identified naive CD3 cells (CD3-C1; TCF7+ and SELL+), Tregs (CD4-C2; Foxp3+, TIGIT+, and CTLA4+), and terminally exhausted T cells (CD8-C5; PDCD1+, LAG3+, HAVCR2+, and CD96+), indicating that the immune microenvironment in PBS-treated tumors was skewed toward a tolerogenic milieu but that the immunosuppressive TME was relieved in the OAd groups (Fig. 3i). Importantly, CD8-C2 showed high expression levels of GZMA, GZMK and CCL5, representing cytotoxic T cells with a high cytotoxicity T-cell gene signature score but a low exhausted T-cell gene signature score (Supplementary Fig. 7d, e). Similarly, CD8-C6 showed high expression levels of PDCD1, LAG3, HAVCR2 and CD96, representing terminally exhausted T cells with a high cytotoxicity T-cell gene signature score but a low exhausted T-cell gene signature score. CD8-C5 and CD8-C6 exhibited an activation-coupled exhaustion program. CD8+ T cells in the CD8-C1 showed high expression levels of IFNG, CCL4, and CCL3 (Fig. 3i), thus representing activated non-circulating tissue-resident memory T cells/effector memory T cells.28

CD8+ T cells in the CD8-C3 and CD8-C4 clusters showed high expression levels of some cell proliferation marker genes, such as MKi67, PCNA, CDK1, Top2a, and CCNA2, indicating that these clusters represented proliferating T cells.29

Finally, we constructed a developmental trajectory of CD8+ T cells that was associated with the cytotoxicity and exhaustion scores of CD8+ T-cell clusters (Fig. 3j). Along the trajectory, T cells exhibited an exhaustion status with almost no activation in the PBS group. T cells in the OAd-TIGIT-Fc group exhibited increasing cytotoxic activity, ultimately followed by exhaustion. However, T cells in the OAd-null, OAd-SIRPα-Fc and OAd-Siglec10-Fc groups exhibited gradually increasing cytotoxic activity, which was accompanied by gradually increasing exhaustion. Moreover, some of the T cells in the OAd-SIRPα-Fc group developed into a memory population (Fig. 3j). These data demonstrated that OAds, especially OAd-SIRPα-Fc, enhanced the activation of proinflammatory TAMs and cytotoxic CD8+ T cells and alleviated immunosuppression in the MC38 tumor model.

In the analysis of CT26 tumor scRNA-seq, we obtained similar results (Supplementary Results, Supplementary Fig. 8 and Supplementary Fig. 9). These data demonstrated that OAd-TIGIT-Fc treatment mainly enhanced the activation of cytotoxic CD8+ T cells, and alleviated immunosuppression in the CT26 tumor model.

Enhanced antitumor activities and activated immune cells in primary tumors

Our results demonstrated that OAd-SIRPα-Fc, OAd-Siglec10-Fc, and OAd-TIGIT-Fc showed precise antitumor efficacy against MC38, 4T1, and CT26 tumors, respectively.

Furthermore, their antitumor effects were assessed in these three different subcutaneous tumor models by analyzing tumor inhibition, survival time, and immunocyte infiltration in the TME. For the MC38, 4T1 and CT26 tumor models (Fig. 4a–c), intratumoral injection of OAd-null, OAd-SIRPα-Fc, OAd-Siglec10-Fc, or OAd-TIGIT-Fc in the corresponding model was significantly more potent than that of OAd-null regarding tumor growth inhibition (Fig. 4d–f) and survival prolongation (Fig. 4g–i), suggesting that SIRPα-Fc, Siglec10-Fc, and TIGIT-Fc were important contributors to the antitumor effect in addition to exerting direct oncolytic activity.

Fig. 4

Enhanced antitumor activities against primary tumors. C57BL/6J mice were subcutaneously inoculated with 1 × 106 MC38 cells. BALB/c mice were subcutaneously inoculated with 1 × 106 4T1 cells or 1.5 × 106 CT26 cells. When tumor sizes reached ~100 mm3 (counted as day 1), the mice were intratumorally injected with 50 μL of OAd-SIRPα-Fc (1 × 108 pfu per tumor) for the MC38 model (a), OAd-Siglec10-Fc (3 × 108 pfu per tumor) for the 4T1 model (b), and OAd-TIGIT-Fc (1 × 108 pfu per tumor) for the CT26 model (c) on days 1, 4, 7, 10, and 13. The treatment regimens for PBS, Ad or OAd-null in the corresponding models were consistent with those for the armed OAds. Tumor volume was monitored in mice bearing MC38 (d, n = 5), 4T1 (e, n = 7), or CT26 tumors (f, n = 5). Data are represented as mean ± SEM. Survival curves of mice bearing MC38 (g), 4T1 (h), and CT26 tumors (i) (n = 10). j Two days after the third injection, the treated tumors were collected and analyzed by flow cytometry to calculate the percentages of infiltrating CD8+ T cells, Tregs, TAMs and mMDSCs in MC38 tumors. n = 3 mice. Data are represented as mean ± SD. (*p < 0.05, **p < 0.01, ***p < 0.001)

To further demonstrate the remodeling of the tumor immune microenvironment by OVs, we analyzed tumor-infiltrating lymphocytes by flow cytometry after the third intratumoral injection of various OAds (Fig. 4j and Supplementary Fig. 10). We quantified all the tumor-infiltrating immune cells as a percentage of total CD45+ cells in tumors (unless otherwise specified). The results showed that OAd-SIRPα-Fc injection enhanced the proportion of CD8+ T cells in MC38 tumor tissues more significantly than Ad or OAd-null injection (Fig. 4j). Similar increases in CD8+ T cell infiltration in 4T1 or CT26 tumor tissues were observed after OAd-Siglec10-Fc or OAd-TIGIT-Fc treatments (Supplementary Fig. 11). In addition, the percentage of Tregs in CD4+ T cells was significantly reduced after the injection of OAds compared with PBS treatment in all three models (Fig. 4j and Supplementary Fig. 11). Therefore, the injection of OAd-null, OAd-SIRPα-Fc, OAd-Siglec10-Fc or OAd-TIGIT-Fc resulted in a robustly increased CD8+ T cell/Treg ratio (Fig. 4j and Supplementary Fig. 11).

The changes in TAM (CD11b+F4/80+Ly6G−Ly6C−), mMDSC (CD11b+Ly6G−Ly6C+), and gMDSC (CD11b+Ly6G+Ly6C−) infiltration were subsequently analyzed. Intratumoral injections of OAds significantly diminished the tumor infiltration of TAMs in the MC38 and CT26 tumor models (Fig. 4j and Supplementary Fig. 11b). Although the virus injections reduced the percentage of mMDSCs compared with PBS treatment in all three tumor models, the reduction was not associated with “oncolysis” or the secreted fusion proteins (Fig. 4j and Supplementary Fig. 11). However, the percentage of gMDSCs was almost unchanged by the OAds treatments (data not shown). Altogether, these findings demonstrate that intratumoral injections of armed OAds are able to trigger antitumor responses and alter the TME by activating tumor-infiltrating effector T cells and reducing the levels of immunosuppressive cells in tumors.

Enhanced antitumor activities against distant tumors

To test whether a localized intratumoral injection of OAds could induce a systemic antitumor immune response, we established bilateral tumor models with MC38, 4T1, or CT26 cells. MC38 tumor-bearing mice had subcutaneous tumors inoculated into both flanks, followed by five injections of OAds into the tumor in the right flank (Fig. 5a). OAd treatment significantly inhibited tumor growth in both the injected and noninjected tumors compared with Ad therapy in the three bilateral tumor models (Fig. 5b; Supplementary Fig. 12a, b and Supplementary Fig. 13a, b).

Fig. 5

Enhanced antitumor activities against untreated distant MC38 tumors. a Mice were subcutaneously inoculated with MC38 tumor cells in both flanks. After establishment of tumors, the right tumor was intratumorally injected with PBS, Ad, OAd-null, or OAd-SIRPα-Fc (1 × 108 pfu per tumor) on days 1, 4, 7, 10, and 13. b Growth of injected tumors and distant tumors in the bilateral MC38 tumor model (n = 10). Data are represented as mean ± SEM. c, d TILs in injected tumors (c) and distant tumors (d) were analyzed by flow cytometry (n = 3). Data are represented as mean ± SD. (*p < 0.05, **p < 0.01, ***p < 0.001)

In the bilateral MC38 tumor model, OAd-SIRPα-Fc treatment achieved greater control of the treated primary tumor and untreated distant tumor than OAd-null therapy (Fig. 5b). Tumor growth suppression of both the treated and contralateral tumors by OAd-Siglec10-Fc was confirmed in 4T1 models (Supplementary Fig. 12a, b). However, there was no difference in tumor inhibition in the CT26 bilateral tumor model between the OAd-null and OAd-TIGIT-Fc treatments (Supplementary Fig. 13a, b). Interestingly, a complete response (CR) of the primary CT26 tumor was observed in 2 of 8 (25%) mice given OAd-null treatment and in 4 of 8 (50%) mice given OAd-TIGIT-Fc treatment (Supplementary Fig. 13a, b). These results supported the potential efficacy of OAd therapy against distant tumors mediated by activating systemic immunity.

To characterize the immunomodulatory effect of intratumoral OAd therapy, infiltrating lymphocytes were analyzed in both tumors of these three bilateral tumor models after the third treatment. The results showed an increased inflammatory response in the virus-injected tumor, with increased infiltration of CD3+ lymphocytes (data not shown). Notably, a significant proportion of CD8+ T cells infiltrated injected tumors treated with OAds, resulting in elevated CD8+ T cells/Treg ratios in the three tumor models. In the MC38 and 4T1 tumor models, the proportion of CD8+ T cells and CD8+ T cells/Treg ratios were also increased quite significantly in the distant tumors (Fig. 5c, d; Supplementary Fig. 12c, d and Supplementary Fig. 13c, d). These results suggested that the systemic antitumor immunity induced by oncolytic virotherapy could suppress tumor growth.

Intratumoral administration of OAd-TIGIT-Fc induces the establishment of long-term antitumor memory

Generation of immune memory is critical for sustained antitumor immunity.30 To assess the immune memory triggering potential of intratumoral administration of OAd-TIGIT-Fc, mice were first challenged with CT26 tumor cells and received OAd-TIGIT-Fc treatment. Then, the mice that achieved a CR after OAd-TIGIT-Fc treatment and age-matched naive mice were rechallenged with CT26 tumor cells 69 days after the first challenge. All age-matched control mice developed tumors that grew quickly, whereas all mice with a CR showed no occurrence of a secondary tumor within 25 days of rechallenge (Fig. 6a–c). This result suggested that long-term antitumor memory was established in mice with CT26 tumors cured by OAd-TIGIT-Fc treatment.

Fig. 6

Intratumoral administration of OAd-TIGIT-Fc induces antitumor memory. a, b BALB/c mice were subcutaneously challenged with 1.5 × 106 CT26 cells (counted as day 1). When tumor sizes reached ~100 mm3, the mice were intratumorally injected with 50 μL of OAd-TIGIT-Fc (1 × 108 pfu per tumor) on days 8, 11, 14, 17, and 20. On day 38, seven OAd-TIGIT-Fc-treated mice achieved a CR. On day 69, age-matched naive mice (n = 7) and mice with a CR (n = 7) were rechallenged with CT26 tumor cells (b). Data are represented as mean ± SEM. Tumor growth for individual mice is shown (c). d Splenic T cells were collected from treatment-naive mice or OAd-TIGIT-Fc-treated mice 45 days after CR achievement to analyze the secretion of IFN-γ in response to CT26 cells in an ELISpot assay (n = 3). Data are represented as mean ± SD. (***p < 0.001)

To further verify the induction of antitumor memory by OAd-TIGIT-Fc therapy, splenic T cells from OAd-TIGIT-Fc-treated mice with a CR were collected 45 days after CR achievement to analyze the immune response against CT26 cells with an IFN-γ enzyme-linked immunosorbent spot (ELISpot) assay. The number of IFN-γ-secreting cells following stimulation with CT26 cells was significantly greater than that following stimulation with MC38 or 4T1 cells, whereas few splenic T cells from treatment-naive mice exhibited IFN-γ secretion (Fig. 6d). These data indicated that OAd therapy accelerated the generation of long-term antitumor memory.

Combination immunotherapy with anti-PD-1 and OAd-Siglec10-Fc enhanced tumor regression in the 4T1 model

Although OAd-Siglec10-Fc could successfully inhibit the growth of immune “cold” 4T1 tumors by targeting TAMs, further treatment is needed to improve efficacy. As OAd-Siglec10-Fc treatment significantly increased intratumoral CD8+ T cell levels, the antitumor effect elicited by the combination of OAd-Siglec10-Fc and PD-1 blockade was investigated further in the 4T1 model. Mice received three intratumoral injections of OAds or PBS on days 1, 5, and 9 and were intraperitoneally injected with IgG2a or anti-PD-1 on days 3, 7, 11, and 13 (Fig. 7a). 4T1 tumor-bearing mice exhibited almost no response to anti-PD-1 blockade (Fig. 7b, c). Combination therapy with OAd-Siglec10-Fc and anti-PD-1 significantly suppressed the growth of 4T1 tumors compared with OAd-Siglec10-Fc or anti-PD-1 monotherapy (Fig. 7b, c).

Fig. 7

Enhanced antitumor effect of combination therapy with anti-PD-1 and OAd-Siglec10-Fc in the 4T1 model. a The treatment schedule for combination therapy with anti-PD-1 and OAd-Siglec10-Fc in the 4T1 model. BALB/c mice were subcutaneously challenged with 1 × 106 4T1 cells. When tumor sizes reached ~100 mm3, the mice were intratumorally injected with 50 μL of OAds (3 × 108 pfu per tumor) on days 1, 5, and 9 and intraperitoneally injected with 200 μg IgG2a or anti-PD-1 on days 3, 7, 11, and 13. Tumor volumes were monitored every three days (b), and tumor growth for individual mice is shown (c) (n = 5). Data are represented as mean ± SD. d A heatmap showing all upregulated and downregulated genes among the four groups determined by pairwise comparison. e Histogram showing the number of DEGs. f The enriched immune response pathways (n = 3). (*p < 0.05, ***p < 0.001)

We next performed transcriptomic analysis to systematically identify the changes in the gene expression of key molecules after combination therapy. Tumor samples from the PBS, anti-PD-1, OAd-Siglec10-Fc and combination therapy groups collected on day 12 were subjected to RNA-seq. All the differentially expressed genes (DEGs) among the four groups identified by pairwise comparison were analyzed. We observed a tendency for drastic upregulation in the combination therapy group compared to the PBS, anti-PD-1 and OAd-Siglec10-Fc groups (Fig. 7d), with 701, 795, and 435 upregulated genes, respectively (Fig. 7e). For functional analysis, we performed GO enrichment analysis of the upregulated genes in the combination therapy group compared with the PBS, anti-PD-1, and OAd-Siglec10-Fc groups (Fig. 7f). There was significant enrichment in immune-related genes in addition to pathways related to leukocyte migration, leukocyte activation involved in the immune response, lymphocyte proliferation, T-cell differentiation, leukocyte-mediated cytotoxicity, cytokine-mediated signaling, etc. (Fig. 7f). We also selected a few important genes encoding chemokines, chemokine receptors, cytokines, Ifng, interleukin costimulatory molecules, and clusters of differentiation and presented the individual gene expression in heatmaps (Supplementary Fig. 14). This suggested that combination immunotherapy with anti-PD-1 and OAd-Siglec10-Fc significantly enhanced the suppression of tumor growth by activating multiple immune signaling pathways in the 4T1 tumor model.