The 4T1 murine breast cancer cell line was kindly gifted by Fred Miller (Wayne State University) and cultured at 37°C and 5% C0₂ in Dulbecco’s modified Eagle medium glutaMAX (DMEM GlutaMAX; Gibco, Thermo Fisher Scientific, cat. no. 10566016, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific) and 1% penicillin-streptomycin (100 U/mL, Gibco, Thermo Fisher Scientific).

8-15 weeks-old female BALB/cAnNRj mice were purchased from Janvier. All mouse experiments were conducted in compliance with the ARRIVE guidelines and approved by the Danish Animal Experiments Inspectorate (2015-15-0201-00656 and 2020-15-0201-00596). Mice were housed 2 mice/cage in standard housing cages with enrichment (nest materials, gnawing sticks, cardboard/plastic tunnels) and ad libitum food and water in a temperature and humidity-controlled room with a 12:12-h light-dark cycle. After arrival, the mice were acclimatized for at least a week. For exercise interventions, mice were giving access to running wheels (Starr Life Science, diameter 12 cm or Mouse Mag Wheels, The Columbus Instruments Starr Life Science, diameter 9.2 cm). The number of wheel rotations was monitored, and the running distances (km/mouse) calculated by converting wheel rotations to kilometers and dividing the results by two, since each cage contained two mice to prevent isolation-induced stress.

Prior to all mouse studies, the cells were tested negative for mycoplasma.

For orthotopic tumor growth studies, mice were randomly assigned to 3 groups (n=14, repeated twice) and either got access to running wheels 5 weeks prior to a tumor induction (EX group), after primary tumor removal (Post-sur EX group), or not at all (Control group) ( Figure 1A ). Orthotopic tumors were induced by injecting 4x10⁵ cells in volume of 50 µl phosphate buffer solution (PBS)/mouse into the mammary fat pad and surgically resected once the primary tumors reached a size of 8-10 mm. 2-2.5% isoflurane was used to anesthetize the mice in both procedures, and prior to the tumor resection the mice were administered 0.05-1 mg/kg Buprenorphine subcutaneously and a mixture of lidocaine (5 mg/kg) and bupivacaine (1 mg/kg) around the tumor. Additionally, after the surgical intervention mice receive analgesia via the drinking water (6 mg/L Buprenorphine) for 48 hours. Tumor growth was determined by measuring the length and width of tumor and calculating volume using the following formula: volume (mm³) = (length (mm) * width (mm)² * π)/6. An hour prior to resection, mice received an intraperitoneal injection with pimonidazole (60 mg/kg dosage using 10 mg/ml dilution, Hypoxyprobe™ Kit, Hypoxyprobe, Inc., HPI Catalog # HP1-XXX, Burlington, MA, USA).

The resected tumors were weighed, their volume determined and divided into pieces for further examinations. In a blinded manner, mice were assessed for clinical signs of metastasis and euthanized once the humane endpoints were reached. The lungs of mice were collected and weighed and processed for further examinations. Body weight was measured at set-up, tumor inoculation, tumor resection, 2-3 times weekly while monitoring for metastatic disease, and when euthanized. During the two studies, 20 mice had to be sacrificed due to surgical complications, wounds, or regrowth of primary tumors (n=6 (Control), n=10 (Post-sur EX group), and n=3 (EX group)). These mice were as a result censored from the survival analysis.

For experimental lung metastasis studies mice were randomly assigned to 4 groups (n=8, repeated twice) and either got access to running wheels 5 weeks prior to metastasis induction (EX groups, Pre EX group), after metastasis induction (Post EX group), or not at all (Control group) ( Figure 2A ). Lung metastasis was induced by injecting 3x10⁵ cells in volume of 200 µl phosphate buffer solution (PBS)/mouse into the tail vein. Mice were weighed and euthanized after 12 days, where their lungs were collected and weighed.

For surgical stress studies, mice were randomly assigned to 3 groups (n=6, repeated three times). All groups were induced with metastatic disease (3x10⁵ cells in volume of 200 µl PBS/mouse injected into the tail vein), but while the Control and Post-sur EX groups underwent surgical resection of the tissue around the 4^th mammary fat pad to mimic a tumor resection, the sham group was only anaesthetized. The mice received the same anesthesia and analgesia as previously described. A day after the surgery the EX group got access to running wheels ( Figure 2A ). Mice were weighed and euthanized 14 days after 4T1 inoculation, where their lungs were also collected and weighed. All mice used for the manuscript were euthanized by cervical dislocation or decapitation.

Part of the tumor tissue and isolated lungs tissue were fixed in 10% neutral buffered formalin overnight at 4 °C and then processed for standard paraffin embedment. Paraffin-embedded tissue was sliced to 5 µM tissue slides and mounted on glass. All lung slides were stained with hematoxylin and eosin, scanned on a Hamamatsu NanoZoom slide scanner, and quantified for metastatic disease using NDP.view2 software by determining the number and size of the metastatic lung lesions.

Additionally, immunohistochemistry stainings were conducted on tumor (n=12-13/group) and lung sections (n=6-7/group) from the orthotopic tumor studies, where the tissue sections were de-paraffinized, rehydrated, and heated with antigen retrieval buffer.

After the antigen retrieval treatment, the tumor section was stained to assess tumor hypoxia via the hypoxia marker pimonidazole hydrochloride, which mice were injected with one hour prior to having their tumor removed. The slides were washed in water, incubated for 10 min with 3% H₂O₂, and washed in Tris-buffered saline with tween (TBS-T). The tumor sections were encircled with a waterproof pen, blocked for 60 min, and incubated overnight with the primary antibody, i.e. anti-pimonidazole mouse IgG1 monoclonal antibody (H Hypoxyprobe™ Kit, Hypoxyprobe Inc.) at 4°C. The tissue sections were then washed in TBS-T buffer, and incubated for 45 min at RT with the DAKO HRP Mouse antibody (Agilent Technologies), followed by a washing step with TBS-T buffer and the addition of DAB Chromogen (Peroxidase) (Vector Laboratories) as per kit instructions. Lastly, the tissue sections were counterstained with Mayer’s hematoxylin (Sigma-Aldrich), dehydrated, mounted with DPX, and scanned on the Hamamatsu NanoZoomer-XR (Hamamatsu) whole slide scanner (×40 magnification). Tumor hypoxia was quantified via ImageJ (24) by manually measuring the total area of both tumor and hypoxia and calculating the fraction of the tumor affected by hypoxia.

The lung sections were stained for Ki-67, as Ki-67 is a prognostic marker associated with breast cancer cells’ proliferative potential and an indicator of prognosis (25). Following antigen retrieval, the lung sections were washed with PBS, encircled using a PAP pen (Dako, Denmark), incubated with 5% donkey control serum (D9663, Sigma-Aldrich, Merck Life Science A/S, Denmark) for 10 min at RT, incubated with primary antibody against Ki-67 (1:1,000-1:2000, ab15580; Abcam, San Diego, USA) at 4 °C overnight. The next day, the sections were washed twice in PBS and incubated with secondary antibody using fluorescence Donkey anti rabbit Alexa 546 for 1 hr at RT (Invitrogen, Taastrup, Denmark, diluted in PBS 1:1000) and counterstained with DAPI before being mounted. Images of the slides was captured with a x40 magnification on a Zeiss Axiovert 220 Apotome system. The images were processed using the Axiovision program (Carl Zeiss) and all images were imported, and the threshold was set for all. The MetaMorph microscopy automation and the ImageJ analysis software (24) were used for automatic nuclei counting and for detection of Ki-67 stained tumor cells. The total number of DAPI stained tumor cells was estimated by automatic nuclei counting. The number of Ki-67 stained cells were counted manually and the fraction of tumor cells expressing the Ki-67 antigen was determined.

The quantifications (metastatic lesions, hypoxia, Ki-67 stainings) were performed while blinded.

Frozen and digested tumor and lung tissue were washed once in FACS buffer (PBS + 2% FBS) and incubated with FC-block and live-dead staining for 15 minutes at RT. Afterwards, cells were washed once, followed by antibody staining ( Supplementary Table 1 ) for 30 minutes at 4°C and followed by fixation in 2% paraformaldehyde for 10 minutes at RT. Cells were stored in FACS buffer until acquisition within the next 3 hours.

Receptor surface expression was acquired using an LSRFortessa equipped with 3 lasers (488nm, 640nm, 405nm) maintained by the flow cytometry core facility at Copenhagen University using FACSDiva software v.8.01 (BD Biosciences, USA). Analyses of blinded samples (gating strategy in Supplementary Figure 1 ) and compensation were performed in FlowJo v.10.6.1 (BD Biosciences, USA). Gating was based on Fluorescence minus one (FMO) controls for each parameter. Cells were defined as follows: all cells (DAPI^-, CD45⁺), T cells (CD3⁺), CD4 T cells (CD3⁺ CD4⁺), CD8 T cells (CD3⁺ CD8⁺), NK cells (CD3^- CD24^- SSC^low CD49^dim/hi Nkp46^dim/hi), Neutrophils (Ly6G⁺), Macrophages (Ly6G^- SSC^hi F4/80⁺) and Dendritic cells (Ly6G^- F4/80^- MHC-II⁺ CD11c⁺ CD11b^+/-).

Statistical analyses were performed in Prism 10 (version 10.2.2). Statistical significance was defined as a p-value < 0.05 throughout. The survival of mice subjected to surgical removal of mammary fat pad tumors with or without access to running wheels was analyzed with Kaplan-Meier analysis. The survival curves were compared using a Log-rank (Mantal-Cox) test. The median time it took for mice to develop clinical signs of metastatic disease that necessitated euthanasia was examined by performing a Kaplan-Meier survival analysis combined with a Log-rank (Mantal-Cox) test on data from mice that developed metastatic disease with “Pair-wise” comparisons of the survival curves. To examine if the tumor growth of mice without access to running wheels differed from mice with access, the data was log transformed with a natural logarithm and analyzed by fitting a nonlinear regression model using the exponential growth with log(population) equation. The analysis revealed that one curve fitted both data sets (i.e. sedentary and exercising mice). For analyses of two groups, parametric data was analyzed with an unpaired t-test or Welch's t test, while nonparametric data was analyzed with a Mann Whitney test. When analyzing the difference between multiple groups, parametric data was analyzed with an ordinary one-way ANOVA paired with Holm-Šídák’s multiple comparisons test/Šídák’s multiple comparisons test and nonparametric data with a Kruskal-Wallis test with Dunn’s multiple comparisons test. Furthermore, a two-way ANOVA was used to analyze the difference in macrophages and neutrophils between mice with and without metastatic disease.

We investigated the effect of voluntary exercise on tumor development and metastasis in a group of immunocompetent mice with induced TNBC. We used 4T1 murine breast cancer cells due to the high clinical relevance of the experimental model, which includes easy orthotopic transplantation in the tissue of origin (mammary fat pad) and spontaneous development of metastatic disease with tumor cell dissemination patterns similar to that of human mammary cancer (26). Specifically, we combined an orthotopic tumor transplantation with a subsequent surgical removal of the primary tumor to mimic the clinical setting, where primary tumors are surgically removed, while potential metastatic lesions remain, and to allow the development of metastatic disease to become the experimental endpoint (27).

Mice were given access to running wheels either 5 weeks prior to tumor induction with syngeneic 4T1 breast cancer cells (EX group), or after primary tumor removal (Post-sur EX group), or not at all (Control group) ( Figure 1A ). We observed no differences in the running distance between the different exercise groups ( Supplementary Figures 2A, B ), nor differences in the food intake or body weights across all groups as recorded throughout the experiment ( Supplementary Figures 2C, D , respectively).

Exercise did not significantly affect the primary tumor growth rate ( Supplementary Figure 3B ). Furthermore, there was no significant difference in the time it took for the tumors to reach a size that required their removal ( Supplementary Figure 3C ), nor in the weight of the resected primary tumors ( Supplementary Figure 3D ).

In contrast, we observed a striking difference in survival, whereby the introduction of voluntary wheel running after surgical removal of the primary tumor led to increased median survival of the mice ( Figure 1B ), which was significantly different from the mice with continuous access to running wheels (p= 0.03). Furthermore, the median time it took for mice to develop clinical signs of metastatic disease that necessitated euthanasia was significantly delayed in the Post-sur EX group compared to both the Control group (p=0.02) and the EX group (p= 0.03) ( Figure 1C ).

It is known that hypoxia at the primary tumor can influence metastasis and survival (28) and could likely be reduced by exercise (29). We, therefore, analyzed the resected tumors for hypoxia, but observed no significant differences ( Supplementary Figures 3E, F ). Wennerberg et al. and Garritson et al. previously indicated that voluntary wheel running could lead to a more favorable immune landscape in mice with 4T1 breast cancer by reducing (30) or delaying (31) immune suppression and increasing the activation of NK cells and CD8⁺ T cells (30). We therefore examined, if the differences in metastasis-free survival could be explained by alterations in the immune landscape of the primary tumors by flow cytometry ( Supplementary Figure 1A ). However, we detected no significant differences in immune cell infiltration between the EX group and the Control/Post-sur EX group ( Supplementary Figure 3G ).

Next, we examined the lungs of the mice at the time of termination, as it is the first site of metastatic spread in both human patients (22-77%) and mice transplanted with 4T1 cells (95%) (26). We detected no differences between the groups with regards to the number ( Figures 2B, C ) or the total area of metastatic lung lesions ( Figure 2D ). Therefore, once the mice developed metastatic disease, the endpoint metastatic burden appeared to be similar between the groups. Consistently, there was no change in Ki-67 expression ( Figures 2E, F ), a proliferation marker that is associated with worse disease-free survival and overall survival in patients with resected TNBC (32, 33).

We performed flow cytometry analysis of the lungs ( Supplementary Figure 1B ), and observed no significant differences in NK cells, CD4⁺ and CD8⁺ T cells, dendritic cells, macrophages, or neutrophils between the groups ( Figure 2G ). However, when we compared the mice with metastatic disease to mice without metastatic disease, we noted a significant decrease in the frequency of lung macrophages and increase in the frequency of lung neutrophils in the mice with metastatic disease ( Figure 2G ).

We speculated that the lack of differences between the groups could be explained by the fact that we only compared them at a timepoint, where all the mice were deemed to have clinical signs of metastatic disease that necessitated euthanasia. We, therefore, performed a new experimental metastasis study with intravenous (IV) injections of 4T1 cells, where mice either had access to a running wheel throughout the experiment, prior to induction of metastasis, after the induction of metastasis, or not at all ( Figure 3A ). Furthermore, instead of having a continuous take down, the study was concluded on day 14 and the degree of metastatic disease in the lungs of all mice was evaluated by histological examinations of lungs sections. We observed that mice who ran prior to metastasis induction had a higher running distance ( Supplementary Figures 4A, B ), that food intake was significantly increased in mice that had access to running wheels compared to mice without ( Supplementary Figures 4C, D ), and that body weights of the mice increased throughout the experiment ( Supplementary Figure 4E ). However, to our surprise, we again saw no differences in the number of metastatic lung lesions between the groups ( Figures 3B, C ) or the total area of metastatic lung lesions ( Figure 3D ). In this setup, exercise alone did not abrogate metastatic development.

Given that the orthotopic model also included a surgery, we postulated that this could promote the exercise-mediated increase in metastasis-free survival. Of note, surgical stress has been shown to exacerbate metastatic disease in both animal models and cancer patients (34, 35). We, therefore, performed another IV experimental metastasis study, which included a mock-surgery to induce surgical stress. Specifically, all mice were induced with metastatic disease, and then either only anaesthetized or also subjected to a skin removal surgery similar to that of a primary breast tumor resection. Half of the mice that underwent surgery were given access to running wheels the day after surgery, while the rest were not. After 14 days all mice were taken down and their lungs examined for metastatic disease ( Figure 4A ). When we compared the number of lung lesions in the group undergoing surgery to the group that was only anaesthetized, we observed that the surgery alone group had significantly more metastatic lung lesions ( Figures 4B, C ). However, the number of metastatic lung lesions was significantly reduced if the mice had access to a running wheel following undergoing surgery ( Figures 4B, C ). A similar pattern was observed when we assessed the total area of metastatic lung lesions, such that surgery alone increased the affected lung area. However, access to the running wheel after surgery significantly reduced the total lung area affected by metastatic lesions ( Figure 4D ). These results suggest that the increase in metastasis-free survival observed in the orthotopic model was due to a boost of exercise occurring after but dependent on surgery.