Vascular modulation through exercise improves chemotherapy efficacy in Ewing sarcoma

Recent studies in mouse models of cancer have shown that exercise improves tumor vascular function, thereby improving chemotherapy delivery and efficacy. However, the mechanisms underlying this improvement remain unclear and the effect of exercise on Ewing sarcoma (ES), a pediatric bone and soft tissue cancer, is unknown. The effect of exercise on tumor vascular hyperpermeability, which inversely correlates with drug delivery to the tumor, has also not been evaluated. We hypothesized that exercise improves chemotherapy efficacy by enhancing its delivery through improving tumor vascular permeability. We treated ES‐bearing mice with doxorubicin with or without moderate treadmill exercise. Exercise did not significantly alter ES tumor vessel morphology. However, compared to control mice, tumors of exercised mice had significantly reduced hyperpermeability, significantly decreased hypoxia, and higher doxorubicin penetration. Compared to doxorubicin alone, doxorubicin plus exercise inhibited tumor growth more efficiently. We evaluated endothelial cell sphingosine‐1‐phosphate receptors 1 and 2 (S1PR1 and S1PR2) as potential mediators of the improved vascular permeability and increased function afforded by exercise. Relative to tumors from control mice, vessels in tumors from exercised mice had increased S1PR1 and decreased S1PR2 expression. Our results support a model in which exercise remodels ES vasculature to reduce vessel hyperpermeability, potentially via modulation of S1PR1 and S1PR2, thereby improving doxorubicin delivery and inhibiting tumor growth more than doxorubicin alone does. Our data suggest moderate aerobic exercise should be tested in clinical trials as a potentially useful adjuvant to standard chemotherapy for patients with ES.

cancers primarily found in the adult population (melanoma, pancreatic, breast, and prostate cancer), we and others have demonstrated that exercise can remodel tumor vasculature and improve chemotherapy efficacy by increasing drug delivery [3][4][5][6][7][8] as well as oxygenation. In prostate tumors, for example, increased blood flow to the tumor after exercise decreased intratumoral hypoxia, [3][4][5] converting a hypoxic, aggressive tumor phenotype with poorly structured tumor vasculature to a more oxygenated, slower-growing tumor phenotype with organized, mature vasculature. 3,5 These studies suggest that exercise has a potential role as an adjuvant therapy to chemotherapy, yet the mechanisms by which exercise improves tumor vessel function remain unclear. One key aspect of tumor vascular remodeling for improved chemotherapy delivery is the reduction of hyperpermeability. Vessel hyperpermeability causes excessive fluid loss from the tumor vasculature to the interstitial space, resulting in impaired blood flow to the majority of the tumor and subsequent inadequate chemotherapy delivery. 9 Tumor vascular leak is inversely correlated with chemotherapy delivery to the tumor as poorly functioning hyperpermeable tumor vessels cause leakage of drug at the periphery due to the high interstitial fluid pressure of a solid tumor. 9 Vessel hyperpermeability is likely to be impacted by exercise via sphingosine-1-phosphate receptors 1 and 2 (S1PR1 and S1PR2) on the surface of endothelial cells. 10,11 These receptors act as shear stress responsive mechanoreceptors that regulate vessel barrier integrity through their signaling pathways. 12,13 Shear stress is one mechanical stimuli exerted on endothelial cells by blood as it flows, 14 and exercise creates laminar shear stress proportional to its intensity. 15 S1PR1 expression suppresses hyperpermeability through the stabilization of vascular endothelial cadherin at adherens junctions in response to laminar flow. 11,13 S1PR2 expression, which is associated with pathologic endothelium and increased vascular permeability, is upregulated by disturbed blood flow, such as that found in tumors. 12,16 Exercise has been demonstrated to increase S1P ligand expression in the serum of human subjects, 17 but regulation of the receptors in response to exercise has not been evaluated. We hypothesized that regulation of S1PR1 and S1PR2 correlates with tumor vascular remodeling in response to exercise-induced shear stress.
In this study, we used human ES xenografts as a pediatric cancer model in adolescent mice to show that a clinically relevant program of moderate intensity aerobic exercise improves chemotherapy efficacy against the tumor without increasing chemotherapy delivery to healthy organs. Further, we present the modulation of S1PRs as a potential molecular explanation for the improvement in vascular function attributed to exercise, in turn improving chemotherapy delivery and efficacy.

Animals and experimental protocol
All animal experiments were approved by MD Anderson's Institu-  Figure 1A). Tumors were harvested and fixed in formalin or frozen in optimal cutting temperature (OCT) compound. The experiment was also conducted with TC71 cells (2 × 10 6 ).
For the assessment of doxorubicin delivery, 20 mice were injected with A673 cells (2.5 × 10 6 ). Ten tumor-bearing mice were treated with moderate-intensity treadmill running for 2 weeks (10 exercise sessions) and 10 served as controls. Seventy-two hours after the final exercise session, all exercised mice and eight of the 10 control mice were injected with a bolus dose of doxorubicin. The remaining two mice served as negative controls for doxorubicin spectrophotometry and did not receive doxorubicin ( Figure 1B). The experiment was also conducted with TC71 cells (2 × 10 6 ).

Immunofluorescence staining
Paraffin-embedded slides were deparaffinized using xylenes and ethanol and then antigen retrieval was performed in 20 µg/mL pro-

Vessel function quantification
Mice received an intravenous injection of 1 mg of high molecular weight (2 × 10 6 kDa) dextran-fluorescein isothiocyanate (FITC; Sigma-Aldrich) into the lateral tail vein immediately prior to euthanasia.
Dextran-FITC was visualized by fluorescence microscopy.
To quantify the vessel leak represented by dextran staining, we averaged the area of dextran immunofluorescence in five random 10× or 20× magnification photographs of each slide to obtain one value for each tumor. Individual averages for all tumors within a treatment group were then averaged to determine the group average and SEM. The dextran-positive area was then compared against the CD31-positive area to identify vessel leaks.

Quantitative polymerase chain reaction (qPCR)
Tumor  verify the presence of one gene-specific peak and the absence of a primer dimer. All samples were amplified in triplicates, and the mean mRNA was used for further analysis.

Doxorubicin quantification
Mice in the control and exercise-only groups ( Figure 1A) were injected with 10 mg/kg doxorubicin via the tail vein 72 hours after the last exercise session ( Figure 1B). Thirty minutes later, the mice were euthanized and tissues were immediately harvested, weighed, and homogenized in acid isopropanol/Triton-X, as described previously. 19 Doxorubicin fluorescence was read using a spectrophotometer ( = 470) and compared to a standard curve. Values were then normalized against the weight of the tissue input and the background reading of a tumor from a control mouse that received no doxorubicin.

Exercise does not significantly alter vessel morphology
To investigate whether moderate treadmill exercise morphologically changes ES tumor vasculature, we first compared the tumor vessels from tumor-bearing exercised or nonexercised mice (Figures 2A and 3A). Neither A673 nor TC71 tumors of exercised or control mice differed significantly in open lumens, total vessels, microvessel densities, or numbers of elongated vessels (Figures 2B and 3B).
Pericytes are associated with vascular remodeling and function. 20 To investigate whether exercise changed their prevalence we evaluated desmin and -SMA in tumors of exercised or nonexercised mice.
Both A673 and TC71 tumors from exercised mice had significantly greater desmin-positive vessels than tumors from the control mice did (P = .0007, Figure 2C and P = .0198, Figure 3C). The number of -SMA-positive vessels was also significantly greater in A673 tumors ( Figure 2D) and approached significance in TC71 tumors (P = .0891; Figure 3D) of exercised compared to control mice. Critical mediators of blood vessel permeability are S1PR1 and S1PR2, expressed on the surface of endothelial cells and pericytes.

Exercise improves tumor hypoxia and chemotherapy delivery and efficacy
We hypothesized that improved vessel permeability would allow for blood flow to a greater area of the tumor. Thus, we evaluated tumor hypoxia as a surrogate for blood delivery. The mRNA levels of HIF1were 41.28% lower (P = .26; Figure 5A) and those of CAIX were significantly lower by 51.23% (P = .03; Figure 5B) in A673 tumors from exercised mice relative to tumors from nonexercised mice.
Improved blood flow in tumors of exercised mice compared to nonexercised mice suggested that exercise could increase delivery of chemotherapy to the tumor. Both A673 and TC71 tumors from exercised mice had significantly higher doxorubicin levels than those in tumors of nonexercised mice after a single dose of the drug (P = .0365, Figure 5C; P = .0058, Figure 5E). Importantly, doxorubicin levels in non-tumor-bearing organs (heart, liver, muscle, and lungs) of exercised and nonexercised mice did not differ significantly ( Figures 5D   and 5F).

F I G U R E 2
Exercise does not significantly alter Ewing Sarcoma vessel morphology but increases mural cell density in A673 tumors. After A673 tumors were measurable, mice were randomized into no exercise or treadmill exercise groups. Tumors were harvested 21 days later. (A) Representative images of CD31 immunofluorescence (red) with DAPI staining to visualize nuclei (blue) in A673 tumors from nonexercised mice and exercised mice; (B) microvessel densities (MVD), numbers of visible lumens, total numbers of vessels, and numbers of vessels >100 µm (elongated) were counted in five random sections/tumor and averaged to obtain a value per tumor in nonexercised (black circles) and exercised (gray squares) mice. Bars represent means ± standard error of the mean (SEM), n = 6-7. Mural cell density was also measured on five random sections/tumor and averaged. (C) Mean desmin:CD31 ratio ± SEM for individual A673 tumors, n = 8-9, **P = .00007; (D) mean -SMA:CD31 ratio ± SEM for individual A673 tumors, n = 6, **P = .0006 To further investigate the effect of exercise on chemotherapy delivery to ES, we used moderate-intensity daily treadmill running and treated tumor-bearing mice with doxorubicin, exercise, or doxorubicin combined with exercise ( Figure 1A). Experiments resulted similarly each time; pooled data are shown in Figures 5G and 5H. In a single representative experiment of A673 or TC71, mice treated with doxorubicin + exercise had smaller tumors than those treated with doxorubicin alone (43.9% or 58.4%, P = .04, respectively). Tumors from mice treated with the combination of exercise and doxorubicin were significantly smaller than tumors from mice treated with doxorubicin alone in both A673 and TC71 models (P = .02 and P < .0001, respectively; Figures 5G and 5H).

DISCUSSION
Our study uniquely demonstrates that exercise reduces tumor vascular permeability and improves oxygenation in a mouse model of pediatric cancer. Exercise improved doxorubicin delivery, thus enhancing its efficacy and resulting in greater inhibition of ES growth, but did not increase chemotherapy delivery to non-tumor-bearing sites. This study also suggests that modulation of S1PR1 and S1PR2 expression by exercise may be partially responsible for improved tumor vascular function.
The tumor vascular response to exercise appears to be partially dependent on tumor type. Others have demonstrated that exercise changes microvessel density, vessel length, or vessel patency; these were unchanged in our model. 3

F I G U R E 4
Exercise reduces tumor vessel permeability, increases tumor S1PR1, and decreases tumor S1PR2. Ewing-sarcoma-tumor-bearing mice were divided into four groups-control, exercise, doxorubicin, or doxo + exer. Tumors were harvested approximately 21 days later. Immediately prior to euthanasia, mice in each group were injected with high molecular weight (2000 kDa) dextran-FITC. (A) Representative images of dextran-FITC (green) and CD31 (red) immunofluorescence in A673 tumor vasculature of nonexercised mice compared to exercised mice; arrow indicates a dextran-positive vessel. (B) Mean dextran:CD31 ratio ± standard error of the mean (SEM) for individual A673 tumors, n = 6-7, *P = .02, and TC71 tumors, n = 6-7, P = .067; (C) representative images of CD31 (green) and S1PR1 (red) immunofluorescence with DAPI staining to visualize nuclei (blue) in TC71 tumors; (D) mean S1PR1:CD31 ratio of immunofluorescence ± SEM for individual A673 tumors, n = 4-6, *P = .04; (E) representative images of CD31 (green) and S1PR2 (red) immunofluorescence on TC71 tumors with DAPI staining to visualize nuclei (blue); (F) mean S1PR2:CD31 ratio of immunofluorescence ± SEM for individual A673 tumors, n = 4-6; (G) mRNA levels of S1PR1 and S1PR2 in the lungs and (H) muscle in TC71 tumor-bearing mice relative fold change compared to GAPDH housekeeping gene in tumor homogenates assessed by qPCR, mean ± SEM for individual tumors, n = 7-8 F I G U R E 5 Exercise reduces hypoxia, increases doxorubicin delivery specifically to the tumor, and improves chemotherapy efficacy. (A, B) mRNA levels of HIF-1 and carbonic anhydrase IX relative fold change compared to 18S housekeeping gene in tumor homogenates as assessed by qPCR, mean ± standard error of the mean (SEM) for individual tumors, n = 6-8, *P = .03; (C) spectrophotometry quantification of doxorubicin in A673 tumors harvested from mice after 2 weeks of exercise, mean ± SEM for individual tumors, n = 7-9, *P = .0365, and of (D) non-tumor-bearing organs of nonexercised and exercised mice with A673 tumors, mean ± SEM, n = 7-9; (E) spectrophotometry quantification of doxorubicin in TC71 tumors harvested from mice after 2 weeks of exercise, mean ± SEM for individual tumors, n = 8-10, **P = .0058, and of (F) non-tumor-bearing organs of nonexercised and exercised mice with TC71 tumors, mean ± SEM, n = 8-10; (G, H) volumes of A673 (n = 5-7) and TC71 (n = 4-5) tumors were measured on indicated days, mean ± SEM, *P = .02 and *P = < .0001, respectively spaces and create blood lakes. 21  In the present study, we demonstrated that exercise additionally reduces excessive tumor vascular leak. This is important because with reduced vessel hyperpermeability chemotherapy can more efficiently reach the tumor, as permeability is inversely correlated with drug delivery to tumors. 22 A reduction in vessel permeability would allow for greater delivery of not only chemotherapy but also blood, providing a mechanism for the work of others who have demonstrated that exercise increases blood perfusion and oxygenation of tumors. 3,5,8 In our mouse models of ES, we also found reduced tumor hypoxia after exercise in one tumor model, A673, but not in the other, TC71.
We suspect the lack of difference in the TC71 model is due to the extensive necrosis noted in the tumors of all four treatment groups.
Because tumor vasculature is dysfunctional, tumor blood perfusion is heterogeneous, resulting in tissue hypoxia. 23 The reduction of intratumoral hypoxia through exercise has been demonstrated in mouse models of prostate 4,5 and breast cancers. 3 Tumor hypoxia contributes to a more aggressive tumor phenotype 24,25 and greater metastatic potential [26][27][28][29] as well as chemotherapy and radiation resistance. 30,31 Thus, hypoxia reduction after exercise may have positive implications for patient outcomes.
Similar to our own findings, exercise has been found to increase drug delivery to the tumor. [3][4][5][6][7][8] In the present study, we found that exercise does not, however, increase chemotherapy delivery to non-tumorbearing organs, including the heart, lungs, liver, and skeletal muscle.
The clinical implications of these data are relevant to pediatric cancer survivors since survivors often die at younger ages than their peers owing to treatment-related late effects, such as heart failure or pulmonary disease. 32 Shear stress has been proposed to mediate exercise-induced improvements in tumor vascular function. 3-6 S1PR1 and S1PR2 regulate endothelium in vascular disease 10,33,34 and cancer [35][36][37][38] and are shear stress responsive. 12,13 S1PR1 signaling promotes tight intercellular junction assembly to appropriately modulate vascular permeability 13,38 and induces vessel maturation. 37 Conversely, S1PR2 signaling increases vascular permeability 10,39 and is inhibited by laminar shear stress. 12 Here, we show that S1PR1 and S1PR2 on tumor vasculature can be exploited with exercise, as exercise increased S1PR1 and decreased S1PR2. Consistent with increased doxorubicin delivery to tumors but not to other organs, exercise did not impact S1PR1 and S1PR2 expression in the lungs (where S1PR expression is prevalent 40,41 )  Clinical trials are needed to determine whether exercise has the same impact in patients as it does in mice.
Future clinical trials may include biomarkers to assess the vascular response to exercise. Previously, we demonstrated that thrombospondin-1 (TSP-1) is a key anti-angiogenic protein in the normalization of vasculature induced by exercise via the calcineurinnuclear factor of T-cell (NFAT)-TSP-1 signaling. 6 S1PR1 has been shown to induce calcium mobilization, enhancing NFAT-TSP-1 activity. 43 TSP-1 may serve as a potential serum biomarker for exercise-induced vascular remodeling, and future work will aim to elucidate the relationship between S1PR1 and TSP-1 in response to exercise.
The impact of the immune system was not accounted for in our mouse models that utilized nude male mice. ES xenografts cannot grow in immunocompetent mice and there are no transgenic models of ES.
However, further work is warranted in an immunocompetent model since the immune system is implicated in exercise-induced changes in growth and sex hormones, 44 which may impact tumor growth. This work is also limited by using subcutaneous models. Although subcutaneous tumor models have proven to be useful for evaluating molecular mediators of tumor vasculature [45][46][47] and novel therapies for the treatment of ES, 48-50 the findings reported here should be strengthened using orthotopic ES tumors. Additionally, our studies provided a limited number of days of exercise in a nonvariable fashion with controlled settings on a treadmill. This is not unique to our study 3-8 since it is necessary to demonstrate that the effects of exercise can be replicated. Exercise effects may be even more robust in prolonged doses of chronic exercise.
In the present study, we determined that moderate aerobic exercise remodeled ES tumor vasculature and modulated the S1PR1 and S1PR2 balance, thereby improving chemotherapy delivery and inhibiting tumor burden more than chemotherapy alone did in mice. Notably, moderate aerobic exercise is feasible in adult and pediatric patients with cancer undergoing treatment. [51][52][53] Our preclinical data suggest that clinical trials testing moderate aerobic exercise as a potential method to improve chemotherapy delivery and efficacy against ES should be performed.

DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.