The 4T1-luc mammary tumor cell line resembling human TNBC metastasis and constitutively expressing firefly luciferase was a kind gift from Prof. Clare Isacke (Breakthrough Breast Cancer Research Centre, London, UK). 4T1-luc cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 100 U/ml penicillin and 100 μg/ml streptomycin (Thermo Fisher Scientific, Waltham, MA, USA) in culture flasks at 37°C and 5% CO₂. Cell cultures were negative for mycoplasma and bacterial contamination based on PlasmoTest^TM results (Invivogen, San Diego, USA).

The eMSCs and eEpSCs, respectively, were isolated from venous blood collected from the vena jugularis and epidermal skin sample of a donor horse. Animal manipulations necessary for this isolation were approved by an independent ethics committee approved by the Flemish government (recognition number: LA1700607; EC 2018-002 and 2014-001) and Faculty of Veterinary Medicine, Ghent University (EC 2014-020). As described previously, the donor horses were first tested for multiple transmittable diseases (16). A subsequent enzymatic dissociation step on the skin sample was necessary to obtain an eEpSC-containing suspension. The cells were then cultivated in a Good Manufacturing Practice (GMP)-certified production site (BE/GMP/2018/123) according to GMP-guidelines until passage (P) 5. In order to obtain a highly pure eEpSC population, several purification rounds were performed with final plating of the cells on ultralow-attachment plates. The eMSCs and eEpSCs were further characterized on viability, morphology, presence of cell surface markers, population doubling time (acceptance criteria are between 0.7 and 3.0) and, only in the case of eMSCs, trilinear differentiation. Evaluation of the presence (CD29, CD44, CD90 for eMSCs; CD49f, p63 for eEpSCs) and absence (Major Histocompatibility Complex (MHC) II, CD45 and marker for monocytes and macrophages for eMSCs; CD105, MHCI and II for eEpSCs) of specific cell surface markers was accomplished by flow cytometry as previously described (14). Consequently, the cells were cultivated until P10, trypsinized, resuspended in DMEM low glucose with 10% dimethylsulfoxide (DMSO) and stored at -80°C in cryovials until further use.

Mouse experiments were performed according to Good Scientific Practice-principles and approved by the Ethical Committee (EC) of the Faculty of Veterinary Medicine, Ghent University (EC 2020-030 and 2021-018).

Eight-week (w)-old female and male BALB/c mice were mated and pups were weaned 12-14 days (d) post-parturition. One hour (h) after weaning, lactating females were intraductally inoculated in the third mammary gland pair with 5 × 10⁴ 4T1-luc cells suspended in a 100 μl mixture of PBS and Matrigel^® (1:10; Corning, Bedford, MA, USA) under inhalation anesthesia and analgesia as described previously (15). For intravenous (i.v.) treatment, eMSCs or eEpSCs were dissolved in DMEM at 3x10⁵ per 100 μl and injected through the tail vein using a 29G insulin needle. I.v. injections with DMEM only were used as a negative (sham) control. Antibody treatments for depletion of CD4⁺ (clone GK1.5) and CD8α⁺ T-cells (clone YTS169.4) or rat IgG2b isotype controls (clone LTF-2) were purchased from BioXCell (West Lebanon, NH, USA) and diluted (200 μg/100 μl) in InVivoPure dilution buffer (BioXCell) at pH6.5 (for anti-CD4) or pH7 (for anti-CD8α and isotype control) for i.p. administration.

4T1 primary tumor growth was determined through weekly measurement of primary tumor volumes and in vivo imaging. Primary tumor volume measurements relied on the use of a digital caliper to determine length, width and height of the primary tumor for subsequent volume calculation (length x width x height). The in vivo imaging technique to determine primary tumor growth relied on the detection of the 4T1-luc-derived bioluminescence in primary tumors using the IVIS lumina III system (PerkinElmer, Zaventem, Belgium). For the in vivo imaging procedure, mice were i.p. injected with D-luciferin (2 mg/100 μl PBS; Gold Biotechnology, St. Louis, MO) and placed in the IVIS system 10 min later under inhalation anesthesia as described previously (17). For ex vivo imaging, primary tumors, spleens and metastases-bearing organs were quickly isolated from the mice following terminal sedation and sacrification through cervical dislocation, and placed in the IVIS system to capture bioluminescent signals (17). IVIS measurements and subsequent analysis were performed using the living image software 4.7.2.

The ^99mTc labelling of the eMSCs was performed as recently described (18). Briefly, 9 x 10⁵ eMSCs were pelleted and resuspended in saline mixed with SnCl2 and freshly eluted ^99mTcO4 (Sigma Aldrich, US). The preparation was incubated for 30 minutes (min) at room temperature (RT), after which the eMSCs were washed with DMEM and subsequently suspended to a concentration of 3 x 10⁵ eMSCs/100 µl DMEM.

All mice were food deprived for minimum 12 h before injection of the ^99mTc-eMSCs. For preparing the injection line and during injection of the tracer (23.6 ± 5.5 MBq) via the tail vein, mice were put under inhalation anesthesia using isoflurane (5% induction, 2% maintenance). SPECT scans were acquired for each mouse on a U-SPECT-II system (MILabs, The Netherlands) and were immediately followed by a CT acquisition on the same device. For each animal, SPECT/CT scans were performed at 4 different time points following i.v. tracer injection: 1 h, 4 h, 7 h and 22 h. The SPECT acquisitions were acquired using a multi-pinhole collimator containing 75 pinholes with 1 mm pinhole diameter. Five bed positions were acquired to obtain total body mouse SPECT images. The total acquisition time was 45 min (9 bed positions of 5 min/bed position). The following CT acquisition parameters were used to obtain total body mouse CT images: 50 kV tube voltage, 600 µA tube current, 200 projections over 360 degrees, 1 bed position, averaging 1, total acquisition time 3 min and 24 seconds (sec).

The acquired SPECT data were iteratively reconstructed using the Ordered Subsets Expectation Maximization (OSEM) algorithm, delivered by the manufacturer of the U-SPECT-II system. A total of 4 iterations was used in combination with 16 subsets. The reconstructed voxel size was 0.75 mm, the photopeak window for ^99mTc was set to 140 keV ± 20%. The acquired CT projection images were reconstructed by an analytic back projection algorithm, delivered by the manufacturer of the U-SPECT-II system. The reconstructed SPECT and CT images were automatically co-registered, stored as NIfTI format and imported into AMIDE (A Medical Image Data Examiner). Finally, the %ID/g was calculated at each time point for the primary tumor, lungs, liver, spleen and bladder.

Lysates were prepared from isolated primary tumors, axillary lymph nodes, lungs and spleen as previously described (15). Following protein concentration determination, nine cytokines (granulocyte-colony stimulating factor (G-CSF), interferon (IFN)-γ, interleukin (IL)-1β, IL-4, IL-6, IL-10, monocyte chemoattractant protein (MCP)-1, macrophage inflammatory protein (MIP)-2 and tumor necrosis factor (TNF)-α) were quantified in all lysates (25-50 μg protein) using the Luminex Multiplex Assay (ProcartaPlex from Thermo Fisher Scientific) according to the manufacturer’s instructions. Enzyme-linked immunosorbent assay (ELISA) was used to measure transforming growth factor (TGF)-β1 levels (Thermo Fisher Scientific) in the lysates according to the manufacturer’s instructions.

Isolated tissues were fixed in buffered 3.5% formaldehyde for 24 h at RT and embedded in paraffin wax. hematoxylin & eosin (H&E) staining was performed on 5 µm thick deparaffinized sections. For immunohistochemical staining, antigen retrieval was performed on 3-5 μm thick deparaffinized sections either with citrate buffer (pH 6, 10 mM tri-sodium citrate (Santa Cruz Biotechnology, Heidelberg, Germany) for Ki67, CD45, CD163, CD3ϵ, granzyme B), or with Tris-ethylenediaminetetraacetic acid (EDTA) buffer (pH 9, 10mM Tris, 1mM EDTA (Thermo Fisher Scientific) for CD4, CD8a and NCR-1) using a pressurized Decloaking Chamber NxGen (Biocare Medical, CA, USA). As previously described (15, 17), the slides were subsequently incubated on an orbital shaker in a closed microscope box with tris-buffered saline (TBS, Biocare Medical)-wetted tissue paper for all blocking, rinsing and staining steps. For visualization of HRP-positive staining, slides were treated with a 3,3′-diaminobenzidine (DAB)-containing buffer (Dako). Counterstaining with hematoxylin was applied, followed by dehydration and mounting of the slides. Quantification of positive staining was established using either color deconvolution followed by automatic counting in ImageJ. Ki67 proliferation indices were determined using ImageJS (19).

Following anesthesia and disinfection of the skin using an iodine solution (Povidone-Iodine, Ecuphar), a small incision (5-10 mm) was made in the ventral midline caudal to the xiphoid of the abdomen of the 4T1 tumor-bearing mice. A small stab incision was made in the linea alba and lengthened by blunt opening of the scissors. The spleen was extoriated by lifting the omentum and splenic vessels were cauterized at the hilum. Analgesia (PBS-diluted buprenorphine) was provided in the abdomen prior to closure, and mice recovered under an infrared lamp after surgery. Mice undergoing sham splenectomy also received similar surgical incisions and the spleen was also exteriorized, but, after inspection, carefully placed back in the abdominal cavity.

Primary tumors and spleens were isolated from the 4T1 tumor-bearing mice and stored in MACS tissue storage solution (Miltenyi Biotec, Leiden, The Netherlands) or directly placed in DMEM used for 4T1 cell culture. Briefly, primary tumors were transferred into a tumor dissociation enzyme mix (Miltenyi Biotec) and processed on a gentleMACS Dissociator, followed by incubation for enzymatic digestion and filtering through a 70 μm strainer. Splenic fragments were aspirated and pushed through a 22G ¼ needle, after which a 70 μm strainer was applied and blood cell lysis was performed by applying ACK (Ammonium-Chloride-Potassium) lysing buffer as described previously (17). The splenic and primary tumor cell solution was subsequently stained with Viobility 488/520 Fixable dye (Miltenyi Biotec) for exclusion of dead cells. Next, a cocktail composed of anti-mouse REAfinity antibodies (all from Miltenyi Biotec) diluted in FACS buffer (containing PBS with 1% bovine serum albumin (BSA), 2.5mM EDTA and 0.01% sodium azide) was applied and analyses were performed on a flow cytometer. Data were processed using CytExpert v2.0.0.153 software (Beckman Coulter, Inc., California, USA).

Statistics were performed using Prism (Graphpad). Data normalization was performed through log¹⁰ normalization when appropriate. P-values were calculated by Student’s t-tests or Analysis of Variance (ANOVA) tests followed by Newman-Keuls post-hoc test for multiple comparisons.

Upon their intraductal injection, 4T1 cells already break through the epithelial barrier by 2 w post-inoculation (p.i.), mimicking the progression from ductal carcinoma in situ (DCIS) to invasive carcinoma (IC) observed in TNBC patients ( Figure 1A ). This initial transition was set as the starting point for systemic treatment with a single (1x) or multiple (3x) doses of eMSCs.

First, the homing locations of eMSCs following their tail vein injection were identified through ^99mTc-labeling and SPECT/CT tracing. The imaging data showed no uptake of ^99mTc-eMSCs for uniformity in 4T1 primary tumors during kinetic measurement ( Supplementary Figures 1A, B ). Instead, the majority of ^99mTc-eMSCs migrated to the liver and spleen after 1, 4 and 7 h. Although the lungs were also a homing location for ^99mTc-eMSCs, these signals were 8 and 4 times lower at all measured time points than those in liver and spleen, respectively. The signals in the bladder confirmed the urinary excretion of ^99mTc. At 22 h after injection, the ^99mTc-eMSCs signals in liver, spleen, lungs and bladder decreased due to radioactive decay.

Second, the potential anti-4T1 tumor effects of eMSC treatment were evaluated by monitoring 4T1 tumor progression until 6 w p.i. of the 4T1 cells, i.e. at 4 w after injection of the eMSCs. This treatment did not cause side effects as body weight and temperature of all eMSC-treated 4T1 tumor-bearing mice were comparable to their sham-treated counterparts (DMEM as negative control), with minimal variation across the 6 w study period ( Figures 1B, C ). Additional observations that highlighted the safety of the xenogeneic treatment included the absence of altered behavior (such as auto-mutilation or abnormal reaction to stimuli) and the absence of local skin rash or liver necrosis that could indicate a host-versus-graft response. Lactation allows for efficient and surgery-free intraductal inoculation of the mammary gland, but its cessation is associated with a drop in body weight after 1 w p.i. ( Figure 1B ). Primary tumor volume measurements showed that a single dose of eMSCs already significantly decreased tumor progression by 6 w p.i., compared to the sham treatment ( Figure 1D ). This reduction in tumor growth was also confirmed by in vivo imaging at the primary tumor site ( Figures 1E, F ). Based on PCR equine DNA data, eMSCs could not be detected at that time in several organs (including liver, kidneys, spleen, axillary lymph nodes, lungs and primary tumor; Supplementary Figure 1C ). Comparable Ki67 staining and associated proliferation indexes from eMSC- and sham-treated tumors demonstrated that the tumor growth reduction was not due to a decrease in 4T1 cell proliferation ( Figure 1G ). In addition, the treatment effect of eMSCs could not be further enhanced by increasing the dosing frequency, as weekly dosing of eMSCs from 2 to 4 w p.i. of 4T1 cells induced similar tumor growth reduction and Ki67 proliferation index increase as a single dose at 2 w p.i., also without negatively affecting animal welfare ( Figures 1B-G ).

Ex vivo imaging was used to evaluate metastases and observed a decreased metastatic growth in axillary lymph nodes ( Figures 2A, B ) and lungs ( Figures 2C, D ) with both single and multiple dosing regimens, although only the metastatic decrease with the single dose treatment reached statistical significance. This discrepancy in statistical significance for the 3x eMSCs treatment group due to slightly higher luminescent signals is likely the result of background signals derived from residual circulating 4T1 cells in lymphatic vessels connected to the lymph node and in blood that flows through the lung. We reported that such background signals can similarly hamper metastatic measurements through ex vivo imaging in the liver (20), necessitating confirmation of metastatic reduction on tissue slides. H&E histology confirmed the presence of smaller and fewer metastases in axillary lymph nodes and lungs in eMSC- compared to sham-treated mice ( Figures 2E, F ).

Splenomegaly has been associated with tumor progression and leukemoid reactions in the 4T1 model for TNBC, as reviewed by our group (21). The weight of the spleen in all tumor-bearing groups was significantly increased compared to a healthy spleen, and eMSC treatment did not reduce the spleen weight despite the decrease in both primary tumor growth and metastases in eMSC-treated mice ( Figures 2G, H ).

In order to evaluate whether the eMSC-mediated anti-4T1 tumor effect could be replicated by other equine stem cell types, eEpSCs were intravenously injected ( Supplementary Figure 2A ). They established a comparable reduction in primary tumor growth at 6 w p.i. following either 1 or 3 doses, without toxicity or change in tumor cell proliferation based on Ki67 staining ( Supplementary Figures 2B-G ). Further in line with eMSCs, single and multiple eEpSC treatment doses provided comparable significantly reduced axillary lymph node and lung metastases without causing reduced splenomegaly ( Supplementary Figures 3A-H ).

To investigate immunological changes that accompany the eMSC-mediated disease reduction, tumor-associated cyto-/chemokine profiles were analyzed at 6 w p.i. in target organs, including primary tumor, axillary lymph nodes, lungs and spleen. Screening of 10 cyto-/chemokines in these tissues of eMSC- compared to sham-treated mice showed that eMSC treatment increases several pro-inflammatory/anti-tumorigenic cyto-/chemokines in primary tumors treated with one or multiple eMSC doses compared to sham-treated controls. However, only the increase in G-CSF levels upon 1x and 3x eMSC treatment compared to sham treatment, IL-6 levels upon 1x eMSC treatment compared to sham treatment, and MIP-2 levels upon 3x eMSC compared to 1x eMSC and sham treatment was significant ( Supplementary Figure 4A ). In contrast, cytokine profile changes induced by eMSC treatment in all other investigated organs were not statistically significant ( Supplementary Figures 4B-D ). Overall, immune cell abundance did not significantly change upon treatment with either a single or multiple eMSC doses based on CD45 immunostaining on primary tumor tissue at 6 w p.i. ( Figure 3 ). Similar findings were obtained for anti-inflammatory M2 tumor-associated macrophages (TAMs) at the tumor margin based on CD163 staining ( Figure 3 ). On the other hand, T-cells significantly increased in eMSC-treated primary tumors based on CD3ϵ staining, further specified as an increase in both CD4⁺ and CD8α⁺ T-cells ( Figure 3 ). Moreover, staining for granzyme B, as marker for T-cell and NK(-T) cell activation, identified that stimulation of cytotoxicity accompanied the increased T-cell influx in eMSC- compared to sham-treated mice ( Figure 3 ).

Upon treatment with single or multiple doses of eEpSCs, similar to eMSC treatment, several pro-inflammatory cyto-/chemokines were increased in primary tumors compared to sham-treated controls, although only the increase in MCP-1 levels upon 1x eEpSC treatment was significant ( Supplementary Figure 5A ). Also similar to eMSC treatment, cytokine profiles were not significantly changed by eEpSC treatment in other investigated organs, except for a significant increase in splenic MIP-2 levels upon 1x eEpSC treatment ( Supplementary Figures 5B-D ). The numbers of CD45⁺ pan-leukocytes and CD163⁺ M2 TAMs were again unaltered, whereas CD3ϵ⁺ and more specifically the CD4⁺ and CD8α⁺ T-cells along with the cytotoxic marker granzyme B again significantly increased compared to sham-treated 4T1 primary tumors based on immunohistochemistry ( Supplementary Figure 6 ).

Since eMSCs were primarily trapped inside the spleen upon their systemic injection, it was investigated whether the spleen plays a role in the observed primary tumor immune effects and eMSC treatment effect. Tumor-bearing mice underwent either splenectomy or sham splenectomy at 1 w p.i. of 4T1 cells followed by either eMSC or sham (DMEM) treatment at 2 w p.i., and tumor progression was monitored until 6 w p.i. ( Figure 4A ).

Body weight and temperature of all tumor-bearing mice showed minimal variation across the 6 w study period, indicating that neither of the evaluated treatments nor the (sham) splenectomy negatively impacted animal health ( Supplementary Figures 7A, B ). Based on primary tumor volume ( Figure 4B ) and in vivo bioluminescence ( Supplementary Figures 7C, D ), splenectomy + sham treatment significantly reduced tumor progression compared to sham splenectomy + sham treatment in tumor-bearing mice. This result identifies the spleen as tumor-stimulating organ, probably due to its production of immunosuppressive myeloid cells (21). Moreover, eMSC treatment provided a significant additional reduction in primary tumor volume ( Figure 4B ) and bioluminescence imaging at 6 w p.i. in the splenectomized tumor-bearing mice ( Supplementary Figures 7C, D ).

The 4T1-derived bioluminescence was also significantly decreased in metastatic organs i.e. axillary lymph nodes and lungs following eMSC treatment, and prior splenectomy did not affect this reduction ( Supplementary Figures 7E-H ). Although the splenectomy + eMSC treatment showed an additional reduction in all metastatic signals compared to splenectomy + sham treatment, this metastatic decrease was only statistically significant in the lungs. Moreover, the similar increase in spleen weight in the sham-splenectomized mice treated with either eMSCs or sham ( Supplementary Figure 7I ), corroborated previous findings ( Figures 2G, H ). It was investigated whether this splenomegaly was also associated with similar percentages of splenic immune cells by performing flow cytometric immunophenotyping ( Supplementary Figure 8 ). Overall, percentages of total splenic CD45⁺ leukocytes and also more specific splenic innate immune cells (including CD45⁺CD11b⁺Ly6C^intLy6G⁺ polymorphonuclear (PMN)-myeloid-derived suppressor cells (MDSCs), CD45⁺CD11b⁺Ly6C^hiLy6G^- monocytic (M)-MDSCs, CD45⁺CD11b⁺F4/80⁺ TAMs and CD45⁺CD11c⁺ dendritic cells (DCs)) ( Supplementary Figures 9A, B ) as well as splenic adaptive immune cells (including CD3ϵ⁺ T-cells and more specific CD45⁺CD3ϵ⁺CD4⁺CD8α^- and CD45⁺CD3ϵ⁺CD4^-CD8α⁺ T-cell subpopulations, CD45⁺CD19⁺ B-cells, CD45⁺CD3ϵ^-NKp46⁺ natural killer (NK) cells and CD45⁺CD3ϵ⁺NKp46⁺ NK-T cells) ( Supplementary Figures 9C, D ) did not significantly differ between all treatments.

Complementary immunophenotyping on primary tumors showed that (sham) splenectomy followed by eMSC treatment did not affect the overall percentage of CD45⁺ leukocytes, nor that of innate immune cells (i.e. PMN- and M-MDSCs, TAMs and DCs) ( Figures 4C, D ). In marked contrast, primary tumors receiving eMSC treatment after sham splenectomy showed a significant increase in percentages of CD3ϵ⁺ T-cells and specific CD4⁺ and CD8α⁺ T-cell subpopulations as well as in percentages of NK and NK-T cells but not in that of B-cells compared to sham treatment ( Figures 4E, F ). A significant increase in percentages of CD3ϵ⁺ T-cells and both T-cell subpopulations as well as NK-T cells was also observed in sham-treated splenectomized mice ( Figures 4E, F ), indicative for enhanced T-cell-mediated anti-4T1 tumor responses upon splenectomy. However, eMSC treatment further enhanced the T-cell-mediated adaptive immunity and the percentage of NK(-T) cells in splenectomized mice ( Figure 4F ). Significantly increased primary tumor staining for CD4, CD8α and NCR-1 as NK(-T) cell marker upon eMSC compared to sham treatment in all operated mice corroborated the flow cytometry results ( Supplementary Figure 10 ). Furthermore, granzyme B staining in sham splenectomized mice was significantly enhanced by eMSC compared to sham treatment ( Suppl. Figure 10 ). Splenectomy also significantly increased granzyme B staining compared to sham splenectomy, and the eMSC treatment further enhanced these staining levels ( Supplementary Figure 10 ).

To assess whether CD4⁺ and CD8α⁺ T-cells are drivers in the eMSC treatment effect, these subpopulations were each separately depleted in 4T1 tumor-bearing mice through specific antibodies followed by eMSC or sham-treatment ( Figure 5A ). Due to antibody toxicity, monitoring of tumor progression was limited to 4 w p.i. of 4T1 cells and 2 w after eMSC injection. Body weight and temperature of all treated tumor-bearing mice showed minimal variation across the 4 w study period ( Supplementary Figures 11A, B ). In the isotype control group, eMSCs significantly reduced tumor volumes compared to sham treatment at 4 w p.i. ( Figure 5B ). Depletion showed that CD4⁺ and, even to a higher extent, CD8α⁺ T-cells were necessary for eMSC-mediated growth reduction. Although a trend towards increased tumor growth was visible with anti-CD8α both upon sham and eMSC treatment, statistical analysis showed no significant differences in tumor volume compared to anti-IgG control ( Figure 5B ). These data indicate that T-cells are not able to effectively combat disease progression in 4T1 tumor-bearing mice. Again corroborating previous findings ( Figures 2G, H ), splenomegaly was not affected by either eMSC or sham treatment ( Supplementary Figure 11C ). In line with the primary tumor growth, a trend towards increased spleen weight was detectable with anti-CD8α both upon sham and eMSC treatment, but no statistically significant difference was found compared to anti-IgG control spleen weights ( Supplementary Figure 11C ).

Similar to 4T1 primary tumors in (sham) splenectomized mice receiving eMSC or sham treatment, immunophenotyping showed that percentages of CD45⁺ leukocytes and innate immune cells (i.e. PMN- and M-MDSCs, TAMs and DCs) were not significantly altered in any of the groups ( Figures 5C, D ). In contrast, CD4⁺ and CD8α⁺ T-cell depletion showed differential effects on the total CD3ϵ⁺ T-cell percentage, with anti-CD4 depletion not affecting and anti-CD8α depletion preventing an eMSC-mediated T-cell percentage increase ( Figure 5E ). This indicates that CD8α⁺ T-cells are the most important T-cell subset for the eMSC-mediated CD3ϵ⁺ adaptive immunity increase. As a control for successful depletion, CD4⁺ and CD8α⁺ T-cells were not detectable in primary tumors of mice upon anti-CD4 and anti-CD8α depletion, respectively ( Figure 5F ), and this depletion was corroborated in blood samples ( Supplementary Figure 12 ). In eMSC-treated mice, primary tumor CD8α⁺ T-cell percentages significantly increased, but, similarly to total CD3ϵ⁺ T-cell percentages, CD4⁺ T-cell depletion did not affect this increase ( Figure 5F ). In marked contrast, CD8α depletion inhibited the significant eMSC-mediated increase in primary tumor CD4⁺ T-cell percentages ( Figure 5F ). The percentages of NK(-T) cells also significantly increased by eMSC treatment and neither isotype control, CD4⁺ or CD8α⁺ T-cell depletion impacted this increase ( Figure 5F ). Immunophenotyping was corroborated by the unaffected eMSC-mediated increase in CD3ϵ and CD8α staining upon anti-CD4 depletion and the inhibited eMSC-mediated increase in CD4 staining upon anti-CD8α depletion in primary tumors ( Supplementary Figure 13 ). Treatment with eMSCs also significantly increased NK(-T) cells as confirmed through NCR-1 staining ( Supplementary Figure 13 ). Due to the unaffected increase in NK(-T) cells, granzyme B staining remained significantly increased upon eMSC treatment, and neither CD4⁺ or CD8α⁺ T-cell depletion had an impact on this increase ( Supplementary Figure 13 ).