Drawing inspiration from the hypoxia-immune microenvironment relationship, the initial investigation involved an evaluation of The Cancer Genome Atlas (TCGA) database’s pan-cancer data to determine the hypoxia status. ( Figure 1A ) It was revealed that the TCGA skin cutaneous melanoma (SKCM) tumor samples exhibited elevated Buffa hypoxia scores (37), suggesting melanoma could be classified among the cancers characterized by a heightened level of hypoxia. Further, 459 SKCM samples were categorized into two groups: ones with high hypoxia scores and the others with low hypoxia scores. Kaplan-Meier survival curves in the TCGA-SKCM cohort indicates that higher Buffa hypoxia scores are associated with a decreased overall survival (OS). ( Figure 1B ) In order to enhance the understanding of the associations between hypoxia and other relevant factors, three machine learning algorithms (LASSO regression, XGBoost, and RandomForest) were employed to ascertain pivotal hypoxia genes based on 199 hypoxia-related genes. Ultimately, six hub genes were identified by intersecting the common variables among the three machine learning algorithms. ( Figure 1C ) The reliability of the new hypoxia scoring approach based on six key genes was demonstrated by the pronounced difference in OS comparing the high hypoxia score group with low score group. ( Figures 1D, E ) Additionally, a strong correlation was identified between high hypoxia scores and tumor, node, metastasis (TNM) staging results of The American Joint Committee on Cancer (AJCC) in melanoma ( Figure 1F ), implying a potential link between heightened hypoxia levels and increased tumor invasiveness and metastatic dissemination (38, 39).

The potential association between hypoxia and the immune microenvironment status has been directly investigated in melanoma. Findings obtained from ESTIMATE Algorithm (40) application revealed a notable inverse relationship between hypoxia and Stroma score, Immune score, and Estimated score, indicating a compromised immune microenvironment infiltration. ( Figures 1G, H ) In order to provide additional evidence regarding the influence of hypoxia on the formation of the immunosuppressive tumor microenvironment, analyses were conducted between hypoxia and various non-cellular constituents within the tumor immune microenvironment. There was a notable correlation observed between higher hypoxia scores and lower expression levels of chemokine, immunostimulator, and major histocompatibility complex (MHC) molecules ( Figures 1I–K ), which play crucial roles in the anti-tumor immune response (41, 42). It was consistently supported by the findings of Gene Set Enrichment Analysis (GSEA), which indicated that the group with high hypoxia scores exhibited significant inhibitions of the chemokine signaling pathway, immune response activation, and MHC protein binding, as compared to the one with low hypoxia scores. ( Figure 1L ) According to these results, hypoxia amelioration presents a viable approach to enhance immune responses against melanoma. Considering the favorable impact of exercise on cardiorespiratory, it is worthwhile to investigate whether the improvement of hypoxia contributes to the mechanism of exercise-induced anti-tumor effects.

In order to ascertain the credibility and dependability of exercise as an anti-tumor treatment, we established B16F10 homograft malignant melanoma model utilizing C57BL/6 mice. These mice were subjected to a 17-day intervention of weight-bearing swimming exercise with varying levels of intensity, while the weight of the load was determined as 5% of their own body weight based on their floating time. ( Supplementary Figure S1A ) Specifically, the mice were allocated in a random manner into three groups (n=6 per group): control, low-intensity swim (LS), and moderate-intensity swim (MS) groups, and then subjected to these exercise interventions with the same exercise cycle (1 day of rest for every 5 days of exercise) and different total exercise durations (0, 6/8/10, and 12/16/20 minutes per day). ( Figure 2A ) The variations in the overall level of physical activity among the distinct groups of mice were evidenced by the levels of lactate dehydrogenase (LDH) activity and lactate concentrations in blood. ( Supplementary Figures S1B, C )

In contrast to the control group, the tumor photographs ( Figure 2B ) as well as tumor weights ( Figure 2C ) demonstrated reductions in tumor sizes among the LS and MS groups, suggesting that both exercise interventions were moderately successful in suppressing tumor growth. Additionally, the corresponding tumor volume curves provided further evidence to substantiate the certain degree of anti-tumor effectiveness, revealing a superior tumor growth inhibition value (TGI) in the MS group (25.92%) compared to the LS group (18.74%). ( Figures 2D–F ) As shown in hematoxylin & eosin (H&E) staining results ( Figure 2G ), the morphological integrity of tumor tissue was partially disturbed in both the LS and MS groups, and the severity of necrosis was notably higher in the MS group. In accordance with this, the results of Ki67 immunohistochemistry ( Figure 2H ), as well as Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining ( Figures 2I, J ), revealed that the exercise intervention effectively suppressed tumor cell proliferation and facilitated apoptosis. Notably, there was hardly any significant difference in the body weights ( Figure 2K ) and food/water intake ( Figures 2L, M ) of the mice in LS/MS group compared with those in control group, demonstrating that the exercise treatments did not adversely affect their survival status. Simultaneously, neither moderate nor low intensities of exercise interventions resulted in substantial hepatic impairment or nephrotoxicity, while the impacts on blood, heart, lung and spleen were nearly inconsequential. ( Supplementary Figures S2 - 5 ) These findings imply that exercise possesses biological safety as a therapeutic modality for cancer.

To investigate the impact of exercise intervention on tumor development in the three groups, we conducted analyses of the tumor microenvironment. As depicted in Figures 3A , C, D , in contrast to the control group, a remarkable augmentation in the quantity of tumor-infiltrating cytotoxic T cells (CD3⁺/CD8⁺) has been observed in both LS and MS groups. Conversely, the population of regulatory T cells (CD4⁺/CD25⁺) has been reduced considerably after exercise administration. ( Figures 3B, E, F and Supplementary Figure S6 ) Notably, these alterations in the T cell infiltration profile, which promote anti-tumor immune responses, were more pronounced in the MS group. In terms of anti-tumor immune cytokines, the exercise intervention groups, particularly MS group, exhibited a noteworthy elevation in interferon-γ (IFN-γ) levels in comparison to the control group, which represents an enhanced immune response (43). ( Figure 3G ) Elevated levels of perforin (PF) and granzyme B, the primary molecules responsible for T-cell-mediated cytotoxicity (44), were detected in both the LS and MS groups, aligning with the previously observed increase in apoptosis following exercise interventions. ( Figures 3H, I ) In contrast to the control group, the mice in LS and MS groups exhibited a significant decrease in interleukin-10 (IL-10) levels within their tumors ( Figure 3J ), potentially indicating an immune microenvironment enhancement and a partial alleviation of immunosuppression (45). Collectively, these data suggest that exercise positively affects the tumor immune microenvironment.

Considering the strong correlation between hypoxic TME and the immune response against tumors ( Figure 1 ), we applied Hypoxyprobe-1 assay to compare the extent of hypoxia between the exercise and control groups. Consistent with our expectations, the exercise intervention led to an evident reduction in hypoxia levels within the tumor tissues of the mice. ( Figures 3K, L ) To delve deeper into the potential involvement of these findings in the anti-tumor mechanism of exercise, we utilized transcriptome sequencing (RNA sequencing) data to analyze the disparities in gene expression between the exercise and non-exercise groups. Volcano plot and hierarchical clustering heatmap presented the expression changes of 961 differentially expressed genes (DEGs) following MS intervention, among which 581 upregulated and 380 downregulated. ( Figures 3M, N ) Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway enrichment analysis showed that DEGs exhibited associations not only with signaling pathways pertaining to immune response and tumor metastasis, but also with metabolic pathways, particularly those related to hypoxia. ( Figure 3O ) The GSEA enrichment analysis exposed an inhibitory signal enrichment within hypoxia gene sets. ( Figure 3P ) Additionally, suppression features were enriched in the Glycolysis/Gluconeogenesis pathway ( Figure 3Q ), which is known to be positively regulated by hypoxia (46). Furthermore, there was also a prominent enrichment of suppressive signatures exhibited in gene sets related to oxidative phosphorylation and respiratory chain complex IV ( Figures 3R, S ), both of which are directly associated with oxygen utilization and its cellular level (47, 48). These results from RNA sequencing further confirmed the proactive role of exercise in hypoxia mitigation at transcriptional level. Collectively, our findings provide evidence that exercise enhances tumor immune infiltration and intratumoral immune responses through the modulation of the hypoxic TME, thereby exerting its anti-tumor efficacy.

PD-1/PD-L1 blockade is currently regarded as a foremost therapeutic approach for melanoma; nevertheless, it encounters challenges such as inadequate immune response and the emergence of drug resistance due to the intricate and suppressive characteristics of TME (49–51). The encouraging outcomes of exercise interventions on tumor immune responses prompt us to investigate the potential synergistic and sensitizing effects of exercise on PD-1/PD-L1 blockade by means of TME modifying. The B16F10 homograft malignant melanoma models were subjected to random assignment into three distinct groups: the PBS-treated group (Control), Anti-PD-1 monotherapy group (PD-1), and Exercise-Anti-PD-1 combination therapy group (MS+PD-1). The moderate-intensity swim treatment with superior anti-tumor performance was chosen for the exercise intervention, and the duration and frequency were consistent with the aforementioned details. Anti-PD-1 administration was performed via intraperitoneal injection with a dosage of 5 mg/kg every 2 days. ( Figure 4A ) The combination of exercise and Anti-PD-1 treatment exhibited remarkable in vivo therapeutic effects against tumors. In contrast to the control and Anti-PD-1 monotreatment cohorts, mice receiving combined treatment exhibited evident reduction in tumor weight and volume, indicating superior suppression of tumor growth. ( Figures 4B–D ) The efficacy of the combination therapy has also been supported by H&E staining from a pathological perspective. ( Figure 4E ) At the cellular level, the results of Ki67 immunofluorescence also indicated that tumor cell proliferation was suppressed with the intervention of exercise combined with Anti-PD-1. ( Figure 4F , Supplementary Figure S10A ) Meanwhile, the administration of combination therapy resulted in a remarkably higher quantity of apoptotic tumor cells compared to the administration of single Anti-PD-1, providing further evidence of the enhanced anti-tumor efficacy achieved through the exercise intervention in conjunction with immunotherapy. ( Figures 4G, H , Supplementary Figure S10B ).

The noteworthy tumor growth suppression achieved by the exercise intervention combined with Anti-PD-1 emphasizes our need to investigate alterations within TME. Encouragingly, cytotoxic T cell infiltration exhibited a substantial increase in the tumor of the combined treating group, as opposed to Anti-PD-1 monotherapy and control groups. ( Figures 4I–K ) Meanwhile, a statistically notable decrease was found in the quantity of regulatory T cells that infiltrated the tumors of combination treatment group. ( Figures 4L–N , Supplementary Figure S7 ) These observations potentially contribute to the immunosuppressive TME alteration. Prompted by this, we examined the presence of hypoxia in tumors across three cohorts. The data revealed that tumors derived from mice subjected to a combination of exercise and Anti-PD-1 treatment exhibited reduced hypoxia levels, reinforcing the association between hypoxia mitigation and immune microenvironment remodeling. ( Figures 5A, B ) Benefiting from the diminished degree of TME hypoxia, the levels of IFN-γ, perforin, and granzyme B within tumor tissues of mice subjected to combined treatment presented a noteworthy increase, and the levels of IL-10 experienced down-regulation when compared to control and mono-immunotherapy groups. ( Figures 5C–F ).

It is worth noting that hardly any alarming deleterious consequence has been observed in the tumor-bearing mice subjected to Exercise-Anti-PD-1 combined treatment. Throughout the treatment period and upon its culmination, the mice receiving combined therapy did not manifest abnormal weight reduction and maintained relatively consistent consumption of food and water in comparison to the mice of two remaining groups. ( Figures 5G–I ) Based on the evaluations of hepatotoxicity and nephrotoxicity, no further detrimental impacts on the standard hepatic and renal functioning were observed. ( Figures 5J–P ) Furthermore, the integration of exercise with immunotherapy did not cause evident adverse effects on the blood, heart, lung, and spleen. ( Supplementary Figures S8 , 9 ) Taken together, these data presented in our study offer partial validation for the guideline’s assertion that there is a low incidence of exercise-related adverse events during treatment (52). Consequently, our results demonstrate that exercise, with a favorable biosafety profile, exerts synergistic anti-tumor effects with PD-1 blockade in a manner that improves TME hypoxia and augments the immune response. ( Figure 5Q )