Replenishment of myeloid-derived suppressor cells (MDSCs) overrides CR-mediated protection against tumor growth in a murine model of triple-negative breast cancer

Abstract

Caloric restriction (CR) is the leading non-pharmacological intervention to delay induced and spontaneous tumors in pre-clinical models. These effects of CR are largely attributed to canonical inhibition of pro-growth pathways. However, our recent data suggest that CR impairs primary tumor growth and cancer progression in the murine 4T1 model of triple negative breast cancer (TNBC), at least in part, through reduced frequency of the myeloid-derived suppressor cells (MDSC). In the present study, we sought to determine whether injection of excess MDSCs could block regression in 4T1 tumor growth and metastatic spread in BALB/cJ female mice undergoing daily CR. Our findings show that MDSC injection impeded CR-mediated protection against tumor growth without increasing lung metastatic burden. Overall, these results reveal that CR can slow cancer progression by affecting immune suppressive cells.

Impact statement: Inoculation of MDSCs from donor mice effectively impedes the ability of calorie restriction to protect against primary tumor growth without impacting lung metastatic burden in recipient animals.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-022-00635-y.

Introduction

Breast cancer is the leading cause of cancer-related mortality in women worldwide [1]. Triple-negative breast cancer (TNBC) is the most aggressive form, accounting for approximately 20% of all cases, and is associated with earlier age of onset and higher metastatic frequency [2–4]. Chemotherapy is the most common treatment option for patients with TNBC because these tumors lack estrogen and progesterone receptors, are HER2 negative, and do not respond to hormone therapy drugs [5]; however, chemotherapy has many off-target effects, including activation of inflammatory pathways [6] and heightened risk of future metastases [7]. Thus, treatment options that can reduce these deleterious side effects are urgently needed.

Like other types of cancer, TNBC relies on myeloid-derived suppressor cells (MDSCs) to block anti-tumor immunity [8, 9]. Tumor-derived signals induce recruitment and expansion of MDSCs from the bone marrow to the tumor, which, in turn, promote immune tolerance and facilitate tumor growth and metastatic spread [10]. Unfortunately, chemotherapy is associated with upregulation of circulating MDSCs [10], a prognostic marker for poor overall survival [10–13].

Caloric restriction (CR) is the most studied dietary intervention shown to increase lifespan and delay disease onset [14–17], including cancer [18]. Downregulation of the canonical insulin/IGF-1 signaling pathway by CR opposes the rapid growth of cancer cells driven by oncogenic mTOR signaling [19]. In the clinical setting, CR can blunt the negative side effects associated with chemotherapy [6] and can reduce metastases [20–22]. However, the immunological impact of CR is less well-understood. Prior work has found that CR and CR-like interventions result in the downregulation of tumor regulatory T cells, including MDSCs, and concurrent expansion of cytotoxic tumor-infiltrating lymphocytes (TILs) [21, 22].

In the present study, we utilized a 4T1 murine model of TNBC to determine if transfusion of excess MDSCs was capable of blocking regression in tumor growth and metastatic spread in BALB/cJ female mice undergoing daily CR. Our findings show that replenishment of MDSCs can reduce CR’s ability to delay primary tumor growth but not its capacity to curb lung metastasis. Additionally, CR tilted the immunological tumor environment by promoting TIL accumulation, a previously identified biomarker of positive response to more targeted immunotherapeutic approaches.

Methods

All animal protocols were approved by the Institutional Animal Care and Use Committee (277-TGB-2024) of the NIA and performed under the Guide for the Care and Use of Laboratory Animals. The ARRIVE 2.0 guidelines for animal research reporting have been followed.

Animals and diet

Twelve-week-old BALB/cJ female mice were purchased from Jackson Laboratories (Bar Harbor, ME; strain #:000,651; RRID:IMSR_JAX:000,651) and were single-housed in duplex cages (#15 Single Housed Duplexed Cages; dimensions 22.2 × 30.8 × 16.24 cm; Thoren Caging System, Hazelton, PA) upon arrival. Autoclaved corncob bedding and nestlets were provided for enrichment at the NIA Biomedical Research Center (Baltimore, MD). Low-velocity HEPA-filtered air via sealed shelf plenums, located directly above the cages, was supplied via air supply orifices above the cage filter top. Upon arrival until study completion, mice were fed AIN-93G Purified Rodent Diet (catalog #102,423, Dyets, Bethlehem, PA) with ad libitum access to water. Baltimore City tap water was treated by reverse osmosis and then hyper-chlorinated to 2–3 ppm. Cages were changed weekly in a biological safety cabinet or sooner if needed. Animal rooms were maintained at 22.2 ± 1 °C and 30–70% humidity with 12-h light cycles. Body weight and food consumption were measured twice weekly. Mice with unlimited food access had their chow placed in the hopper, while those on daily CR feeding regimen had their daily food allotment placed on the floor of the cage between 4:00 and 6:00 pm. Mice were euthanized by CO₂ inhalation, followed by cardiac puncture. Upon euthanasia, serum and tissue were collected between 8:00 am and 12:00 pm, approximately 20 h after the last meal for the CR-fed mice.

Daily CR feeding regimen

Two weeks prior to study initiation, food consumption by mice maintained on AIN-93G was recorded twice per week to determine baseline feeding. Mice were randomized into the treatment groups, and baseline feeding for the ad libitum mice was averaged to calculate the daily food allotment for CR. Over a 2-week period, a stepwise decrease in food intake was conducted, until 20% CR was achieved and maintained for the duration of the study.

Cell culture and injection

Murine breast cancer 4T1 cells were purchased from ATCC (catalog no. CRL2539, Manassas, VA). Cells were free of mycoplasma, as shown using a Mycoplasma Detection Kit (Lonza, Rockville, MD), and were cultured in RPMI (catalog no. 118–156-101, VWR, Gaithersburg, MD) supplemented with 10% FBS (catalog no. 89510186, VWR) 2 mM L-glutamine, 10 mM HEPES buffer, pH 8, 1% sodium pyruvate, 1% non-essential amino acids, and 50 μg/mL penicillin–streptomycin (catalog no. 11875101, Gibco, Waltham, MA) (complete RPMI medium). Cells were grown at 37 °C in humidified 5% CO₂ incubators. Immediately prior to injection, cells were resuspended in sterile PBS and 100 μL were injected (1 × 10⁵ cells per mouse) subcutaneously in the fourth mammary gland on the right side of BALB/cJ female mice. Tumor size was measured with a digital vernier caliper.

MDSC isolation and injection

MDSCs were collected from two cohorts of “donor” AL-fed BALB/cJ female mice. 4T1 injections in donor mice were staggered by 8 days to ensure maximum MDSC collection. MDSCs were isolated from spleens of donor mice on day 28 after implantation of 4T1 tumor cells. Following mechanical grinding, Ly6C/Ly6G (Gr1⁺) PE-tagged cells (catalog no. 108408, BioLegend, San Diego, CA) were isolated from PBMCs and splenic cell suspension via magnetic sorting (Miltenyi Biotec, Cambridge, MA) using anti-PE microbeads (catalog no. 130–048-801, Miltenyi Biotec) and then injected (1 × 10⁷ cells) via tail vein into recipient mice 2 and 10 days after 4T1 tumor cell implantation.

Glucose, insulin, and IGF-1 measurements

Blood glucose was measured in whole blood with the Bayer Breeze2 handheld glucometer (Bayer, Mishawaka, IN). Whole blood was collected via cardiac puncture immediately post-euthanasia in prepared microtubes containing serum gel with clotting activator (catalog no. 101093–958, VWR) and centrifuged at 8000 rpm for 6 min at 4 °C. Insulin and IGF-1 serum concentrations were determined using the Mouse Insulin ELISA kit (CrystalChem, Elk Grove Village, IL) and Mouse/Rat IGF-1 Quantikine ELISA kit (R&D Systems, Minneapolis, MN), respectively.

HOMA calculation

Insulin resistance was calculated using the blood glucose and insulin values entered in the HOMA2 Calculator software available from the Oxford Centre for Diabetes, Endocrinology, and Metabolism, Diabetes Trials Unit Website (http://www.dtu.ox.ac.uk/).

India ink staining

India ink (1:10 solution of India ink in sterile PBS) was intratracheally injected. The lungs were fixed overnight in Fekete solution (300 mL 70% ethanol, 30 mL 37% formaldehyde, and 5 mL glacial acetic acid). The total number of white nodules was counted against the black background of the lung. Each nodule was also measured and sorted based on size ranging from 0.5 < 1 (small), 1 < 1.5 (medium), and > 1.5 mm (large). The lungs were de-identified to prevent bias in the counts.

Flow cytometry

Immune cells from spleen, blood, lymph nodes, and primary tumor were collected, processed, and isolated. Briefly, blood was collected in heparin-coated tubes, spleen and lymph nodes were collected in complete RPMI medium, and tumors were collected in ice-cold gentleMACS™ dissociation tubes (catalog no. 130–093-237, Miltenyi Biotec). Blood was incubated for 10 min at room temperature in ACK (ammonium-chloride-potassium) lysis buffer (3 × blood volume). Spleen and lymph nodes underwent mechanical disaggregation in a cell strainer (70 μM, catalog no. 07–201-431, Fisher Scientific, Waltham, MA), whereas tumors were minced and underwent enzymatic digestion (catalog no. 130–096-730, Miltenyi Biotec). Cell suspensions were centrifuged at 1500 rpm for 5 min at 20 °C. The supernatant was discarded, and cell pellets were resuspended in a minimal volume of prewarmed complete RPMI medium. An aliquot was removed for cell counting to achieve a final dilution of 4 × 10⁶ cells/mL. Cells were stained according to manufacturer’s instructions (eBioscience, San Diego, CA).

The following flow cytometric antibodies and their isotype controls were used: CD11b BUV661 or BV605 (clone M1/70), Gr1 PB (RB6-8C5), iNOS Pe-Cy7 (CXNFT), Ly6G PerCP Cy5.5 (1A8), Ly6C BV605 (HK1.4), Foxp3 PE (NRRF-30), CD4 BV605 or BUV661 (GK1.5), CD8 BUV395 (53–6.7), CD44 Fitc, IFNγ PE Cy7 (XMG1.2), CTLA PerCP Cy5.5 (UC10-4B9), TIM3 BV421 (RMT3-23), LAG3 APC (C9B7W), and PD-1 BV510 (29F.1A12) (BioLegend, BD Biosciences, or Life Technologies, San Diego, CA). Fixability dye eFluor™ 780 (eBioscience) was used to stain dead cells. Flow cytometry data were collected on a Cytoflex platform (Beckman Coulter, Eldersburg, MD) and analyzed with the CytExpert software (Beckman Coulter).

MDSC functional assays

On day 23 after 4T1 tumor cell implantation, blood from AL- or CR-fed mice was collected in heparin-coated tubes and incubated for 10 min at room temperature in ACK lysis buffer (3 × blood volume), followed by centrifugation at 1500 rpm for 5 min at 20 °C. The supernatant was discarded, and the pellet was resuspended in complete RPMI medium. MDSC isolation was conducted using autoMACS pro-separator (Miltenyi Biotec) coupled with anti-PE microbeads (catalog no. 130–048-801, Miltenyi Biotec) to isolate Ly6G/Ly6C (Gr1⁺) cells (catalog no. 108408, BioLegend) for transfer and suppression assays. ROS production was assessed in specific MDSC populations by flow cytometry using the following three stains: 4-amino-5-methylamino-2′,7′-difluorofluorescein diacetate (DAF-FM DA) measures nitric oxide production (catalog no. D23844, Thermo Fisher); 6-carboxy-2′,7′-dichlorodihydrofluorescein (DC-FDA) measures hydrogen peroxide generation (catalog no. C2938, Thermo Fisher); and diacetate dihydroethidium (DHE) measures superoxide production (catalog no. D11347, Thermo Fisher). Additional surface staining was also performed on isolated cells. Total population and specific subsets of MDSCs were detected via flow cytometry using the following: CD11b BV650 (clone M1/70) (catalog no. 563402, BD Biosciences), Ly6C BV510 (HK1.4) (catalog no. 128033, BioLegend), and Ly6G (1A8) (catalog no. 560602, BD Biosciences). Flow cytometry data were collected on a Cytoflex platform (Beckman Coulter) and analyzed with the CytExpert software (Beckman Coulter).

T cell proliferation and arginase activity assay

Spleens isolated from naive BALB/cJ female mice were used to isolate CD3⁺ T cells. Spleens were mechanically disaggregated, and cell suspension was centrifuged at 1500 rpm for 5 min at 20 °C. The supernatant was discarded, and the pellet was resuspended in complete RPMI medium and labeled with cell proliferation dye (catalog no. 65–0842-90, eBioscience). CD3⁺ cells were isolated using a CD3⁺ T cell enrichment column (catalog no. MTCC-10, R&D Systems, Minneapolis, MN) and subsequently activated (catalog no. 11452D, Thermo Fisher) prior to being plated in complete RPMI medium. To assess the suppression of T cell proliferation by MDSCs, total MDSCs were isolated as described above, that is, MDSCs were isolated from whole blood of AL- or CR-fed mice on day 23 after 4T1 injection via magnetic separation for Ly6C/Ly6G (Gr1⁺) cells. These isolated Gr1⁺ cells were then added to the naïve CD3⁺ activated T cells. The individual cell types were differentiated during flow cytometry by using the following antibodies: CD11b BV650 (clone M1/70) (catalog no. 563402, BD Biosciences) for MDSC detection, CD8 BUV395 (53–6.7) (catalog no. 565968, BD Biosciences), and CD4BV605 (GK1.5) (catalog no. 743156, BD Biosciences) for target cell detection (in this case, the effectors are the MDSCs).

Following co-incubation, arginase activity was measured in the cell pellet following manufacturer’s instructions (catalog no. MAK112, Sigma-Aldrich, St-Louis, MO).

Mouse multiplex cytokine assay

Serum was collected 28 days after 4T1 implantation and CCL5, CCL2, and IL-1B levels were measured using mouse cytokine 23-plex assay following manufacturer’s instructions (catalog no. M60009RDPD, Bio-Rad).

CD8⁺antibody depletion

Three days before and on days 3 and 10 post-implantation (defined as day 0), CD8α⁺ antibody (catalog no. BP0004-1, clone 53–6.7, BioXCell, Lebanon, NH) or isotype-matched IgG control antibody (catalog no. BP0089, clone 2A3, BioXCell) was administered by intraperitoneal injection (200 μg per mouse).

Statistical analysis

No statistical methods were used to predetermine sample size. The investigators were blinded to group allocation during recording of body weight and food consumption, organ weight (liver, spleen), tumor sizing and weight, scoring of lung metastases, biochemical markers, and flow cytometry. No data were excluded from analysis. CR treatment in the 4T1 tumor mouse model was reproduced in nine iterations while the MDSC replenishment experiment was done once. All data are presented as means ± standard error of mean (SEM) unless otherwise stated. GraphPad Prism v8.4.2 (GraphPad Software, San Diego, CA) was used to create graphs and for statistical data analysis. Two-way ANOVA with Tukey post hoc analysis was used for all comparisons to assess the impact of diet (AL vs. CR) alone, injection (PBS vs. MDSC) alone, and the interaction “diet × injection.” P value ≤ 0.05 was considered significant.

Results

Tumor burden triggers remodeling of the immune environment via upregulation of MDSCs

Elevated frequency and numbers of MDSCs in breast cancer are linked to poor clinical prognosis [23], as they trigger a cascade of immune remodeling, including activation of pro-tumorigenic regulatory T cells (Foxp3⁺ Tregs) [24] and suppression of the cytotoxic (CD8⁺) T cell response [8]. Here, we ascertained whether the number of MDSCs, defined by the co-expression of CD11b and Gr1 surface markers, would differ between 16-week-old female BALB/cJ naïve mice and mice harboring 4T1 TNBC tumors. A significant upregulation of MDSCs in peripheral tissues (e.g., spleen, blood, and lymph nodes) was observed in tumor-bearing mice (Supplemental Fig. 1a–c). Under inflammatory conditions and cancer, MDSCs initiate metabolic reprogramming by promoting the differentiation of Tregs from naïve T cells [25]. We found a significant accumulation of CD4⁺Foxp3⁺ Tregs (Supplemental Fig. 1d–f) and CD8⁺Foxp3⁺ Tregs (Supplemental Fig. 1g–i) in peripheral tissues of 4T1 tumor-bearing mice. Upregulation of these subsets has been implicated in poor survivorship [26, 27] and advanced metastatic burden [28]. Conversely, the levels of antitumor effector cells such as CD8⁺ (Supplemental Fig. 1j–l) and CD4⁺ (Supplemental Fig. 1m–o) were significantly lower in peripheral tissues of mice harboring 4T1 tumors. Thus, 4T1 tumor burden elicits remodeling of the immune environment, potentially mediated through the upregulation of MDSC levels.

Caloric restriction inhibits MDSC upregulation under 4T1 tumor burden

The association of CR [20, 22, 29] and CR-mimetics [21] with delayed tumor growth and lower MDSC levels led us to assess the impact of 20% daily CR on immunophenotypic remodeling in 4T1 tumor-bearing mice (experimental design, Supplemental Fig. 2a. First, a stepwise reduction in daily caloric intake was carried out to prevent metabolic stress [16]; then, 4T1 tumor cell implantation was initiated once mice reached 20% CR. At 28 days post-tumor implantation, the spleen and lymph nodes, but not blood, of CR-fed mice had significantly lower MDSCs vs. ad libitum (AL)–fed controls (Supplemental Fig. 2b–2d; Supplemental Fig. 3a–3c). The number of CD4⁺Foxp3⁺ Tregs was significantly reduced in the blood, but not the spleen or lymph nodes, of tumor-bearing mice on CR vs. AL (Supplemental Fig. 2e–2g; Supplemental Fig. 3d–3f). A significant decrease in CD8⁺Foxp3⁺ Tregs (Supplemental Fig. 2h–j; Supplemental Fig. 3g–i), combined with an accumulation in effector CD4⁺ (Supplemental Fig. 2k–m; Supplemental Fig. 3j–l) and cytotoxic CD8⁺ T cells (Supplemental Fig. 2n–p; Supplemental Fig. 3j–l), was evident in the blood, spleen, and lymph nodes in response to CR.

Immune remodeling of the chemokine milieu in breast tumor cells is associated with advanced disease prognosis. Indeed, increased levels of CCL5 (RANTES) and CCL2 (MCP-1) are significant predictors of disease progression [30], as well as suppression of cytotoxic T cell activity in breast cancer [31], and Treg recruitment to the site of inflammation [32]. Moreover, the enhanced production of IL-1β by 4T1 tumor cells enables the recruitment of MDSCs in the tumor microenvironment [25]. Daily 20% CR was associated with significant reduction in the serum levels of CCL5 (Supplemental Fig. 2q), CCL2 (Supplemental Fig. 2r), and IL-1β (Supplemental Fig. 2s) collected at 28 days post-tumor implantation, indicative of a less-favorable tumor microenvironment in response to CR.

MDSC transfusion results in metabolic perturbation in CR-fed mice

The anti-tumoral action of CR has been attributed to the impairment in mTORC1 activity [33]; however, less is known about the role of CR in tumor-driven immune remodeling. Specifically, we sought to ascertain whether replenishment with ectopic MDSCs from BALB/cJ donor mice counteracts CR’s ability to delay tumor progression in syngeneic recipient animals.

Two sets of “donor” mice underwent 4T1 tumor cell implantation and were fed AL for 28 days prior to MDSC isolation (experimental layout, Fig. 1a). No differences in tumor growth profile (Supplemental Fig. 4a and 4b), average body weight (Supplemental Fig. 4c), and average caloric intake (Supplemental Fig. 4d) were found between the two groups.

Fig. 1.

Excess MDSCs block CR-mediated improvement in metabolic markers. (a) Experimental layout. Two sets of 16-week-old BALB/cJ donor female mice on an ad libitum (AL) diet were injected with 4T1 tumor cells, with an 8-day gap between the first set (− 26 days) and second set (− 18 days) of donor mice. Prior to tumor implantation, recipient mice remained on AL feeding or underwent a stepwise reduction in caloric intake until 20% caloric restriction (CR) was achieved. At day 2 and day 10 following 4T1 cell implantation, recipient mice were injected with PBS (control) or MDSCs isolated from “donor” mice. (b) Average body weight and (c) average caloric intake over the course of the study. (d) Blood glucose, (e) serum insulin levels, (f) the homeostatic model assessment of insulin resistance (HOMA2-IR), and (g) serum IGF-1 levels at day 28. (h) Liver and (i) spleen weight. The data is presented as scatter plots with mean ± SEM, with two-way ANOVA used to determine statistical significance (Supplemental Table 1). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 compared to AL-fed mice (AL-PBS) (black symbols), AL-MDSC mice (gray symbols), or CR-PBS mice (red symbols), with n = 10–14 mice per treatment group

Concurrently, another two sets of mice were maintained on AL or 20% CR for 14 days and then were subjected to 4T1 tumor cell implantation (experimental layout, Fig. 1a). Two days post-implantation, “recipient” mice were administered either PBS (AL-PBS, CR-PBS) or isolated MDSCs (1.0 × 10⁶ cells) from “donor” mice (AL-MDSC, CR-MDSC); the same protocol was repeated 10 days post-4T1 cell implantation (Fig. 1a). At the conclusion of the experiment on day 28, recipient CR-PBS and CR-MDSC mice showed a similar decrease in average body weight (Fig. 1b), average caloric intake (Fig. 1c), and fasting blood glucose (Fig. 1d) compared to AL-PBS (control) and AL-MDSC. Levels of insulin (Fig. 1e) and HOMA2-IR (Fig. 1f) were significantly elevated in AL-MDSC mice, suggesting the synergistic relationship between MDSC levels and elevated serum insulin levels [34]. Serum levels of IGF-1 were significantly lower in CR-PBS vs. AL-PBS mice, whereas higher IGF-1 levels were observed both in AL- and CR-fed mice subjected to MDSC transfusion (Fig. 1g). A similar profile in liver weight was observed among the experimental groups, with the liver of CR-PBS mice weighing significantly less than AL-PBS controls. In addition, MDSC replenishment was associated with enlarged liver in recipient mice on both dietary regimens (Fig. 1h). The significant drop in spleen weight in the two groups of CR-fed mice (Fig. 1i) suggests the absence of splenomegaly, e.g., de novo splenic accumulation of tumor-promoting immune cells post-MDSC [35].

Administration of MDSCs impairs CR’s ability to delay primary tumor growth

The reduction in the endogenous MDSC population in CR-fed mice (Supplemental Fig. 2b and 2d) led us to ascertain whether transfusion of MDSCs eliminates the tumor-protective effect of CR in recipient animals (experimental layout, Fig. 1a). Tumor growth rate was significantly slower in CR-fed mice vs. AL-fed controls (Fig. 2 a and b). Within treatment groups, AL-MDSC infusion caused an upward trend in the tumor growth compared to AL-PBS mice, while the delayed growth observed in response to CR was partially abrogated in animals transfused with MDSCs, resulting in significantly larger tumors in CR-MDSC mice compared to CR-PBS animals (Fig. 2 a and b). At 28 days post-tumor implantation, only CR-PBS mice showed a significant decrease in the size (Fig. 2c) and mass (Fig. 2d) of the tumor compared to AL-PBS controls. Overall, CR-fed mice injected with MDSCs had significantly larger tumors compared to CR-PBS mice, suggesting that the beneficial response of CR was blunted by ectopic MDSC administration (Fig. 2a–d).

Fig. 2.

Excess MDSCs eliminate CR-mediated delay in primary tumor growth. (a) Growth rate of primary tumor. (b) Primary tumor area at day 28. (c) Tumor mass. (d) Representative images depicting the appearance of white masses in India ink–stained lungs, indicative of metastases. (e) Total number of lung metastases. (f) Lung metastases were scored in a blinded fashion and divided into three groups based on size: 0.5 < 1 mm, 1 < 1.5 mm, and > 1.5 mm. The larger tumor indicates more advanced metastatic growth. Most of the data are presented as scatter plots with mean ± SEM, with two-way ANOVA used to determine statistical significance (Supplemental Table 2). *p < 0.05; **p < 0.01; and ***p < 0.001 compared to AL-PBS mice (black symbols), AL-MDSC mice (gray symbols), or CR-PBS mice (red symbols), with n = 10–14 mice per treatment group

The syngeneic 4T1 breast cancer model mirrors the aggressive metastatic spread found in TNBC patients [36]. Therefore, we assessed whether transfused MDSCs accelerate the metastatic spread of 4T1 tumor cells in recipient mice and examined the potential of CR in lowering the number and size of lung metastases under these conditions. At 28 days post-tumor implantation, there was significant reduction in the total number (Fig. 2 d and e) and size (Fig. 2f) of lung metastatic nodules in CR-fed mice, irrespective of having been the recipient of MDSCs or not.

The improvement in immune markers within peripheral tissues of mice on CR is impeded by the presence of transfused MDSCs

Cancer progression occurs via evasion of immune detection partly through the upregulation and activation of MDSCs, which causes increased production of immunosuppressive cytokines, recruitment of Tregs, and inhibition of the anti-tumor cytotoxic T cell response (CD8⁺ and CD4⁺) [8]. The blunting of MDSC activation by CR [37, 38] (Supplemental Fig. 2b and 2d; Supplemental Fig. 3a and 3c) led us to investigate whether the immune protection conferred by CR can be overridden by MDSC supplementation. Peripheral tissues (spleen, blood, and lymph nodes) of recipient mice were collected 28 days post-tumor implantation. AL-MDSC mice had a significant increase in splenic MDSCs (Fig. 3a; Supplemental Fig. 5a), whereas the reduction in total MDSCs, evident in both the spleen and lymph nodes of CR-PBS mice, was largely abolished in the spleen of CR-MDSC mice (Fig. 3a and c; Supplemental Fig. 5a and 5c). No significant change in total MDSCs was evident in the peripheral blood of any groups of recipient mice (Fig. 3b; Supplemental Fig. 5b).

Fig. 3.

Immune profile in peripheral tissues. Frequency of immune cells detected in spleen (a, d, g, j, m, p, s, v), blood (b, e, h, k, n, q, t, w), lymph nodes (c, f, i, l, o, r, u, x). (a–c) Percentage of CD11b⁺Gr1⁺ (MDSCs); (d–f) MDSCs expressing iNOS; (g–l) MDSC subsets (g–i) granulocytic and (j–l) monocytic; (m–o) Foxp3⁺CD4⁺ T regulatory cells; (p–r) Foxp3⁺CD8⁺ T regulatory cells; (s–u) Effector CD4⁺ cells and (v–x) effector CD8⁺ cells. The data are presented as scatter plots representing mean ± SEM, with two-way ANOVA used to determine statistical significance (Supplemental Table 3); *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001 compared to AL-PBS mice (black symbols), AL-MDSC mice (gray symbols), or CR-PBS mice (red symbols), with n = 8–10 mice per treatment group

Activated MDSCs consume large amounts of arginine via the action of both arginase I and inducible nitric oxidase synthase (iNOS), leading to the inhibition of T cell–mediated cytotoxicity [39]. Here, the proportion of iNOS expressing MDSC cells was significantly lower in the spleen and blood of CR-PBS and CR-MDSC mice compared to AL-PBS and AL-MDSC mice (Fig. 3 d and e; Supplemental Fig. 5g and 5h). In contrast, the spleen contained a higher percentage of iNOS expressing MDSC cells in both AL and CR mice transfused with MDSCs (Fig. 3d; Supplemental Fig. 5d). Thus, CR reduces the expression of iNOS in MDSCs of tumor-bearing mice, which, in turn, could diminish their immunosuppressive potency.

MDSCs consist of two main subsets, including polymorphonuclear (PMN-MDSC) and monocytic (M-MDSC). PMN-MDSCs, which express Lys6G^high, are the predominant form in peripheral tissues and suppress immune function via generation of free radicals. M-MDSCs, which express Lys6C^high, are enriched within the tumor microenvironment and possess greater immunosuppressive capacity due to heightened arginase activity and enhanced production of nitric oxide (NO) and cytokines [40]. A marked increase in the population of PMN-MDSCs was observed in all three tissues of AL-MDSC vs. AL-PBS, whereas CR significantly reduced the number of splenic PMN-MDSCs, irrespective of MDSC supplementation (Fig. 3g–i; Supplemental Fig. 5g–i). No significant change in the frequency of M-MDSCs was noted in any of the experimental groups (Fig. 3j–l; Supplemental Fig. 5j–l).

Upregulation of Tregs by activated MDSCs hinders effector T cell proliferation (CD4⁺ and CD8⁺) [24]. Notably, expansion of Foxp3⁺CD4⁺ Tregs is evident in a pre-clinical model of breast cancer [41] and is often linked to poor clinical prognosis [42]. Here, the number of Foxp3⁺CD4⁺ Tregs was significantly higher in the spleen and peripheral blood of CR-MDSC vs. CR-PBS mice (Fig. 3m–o; Supplemental Fig. 5m-o). CR-PBS mice had significantly lower level of Foxp3⁺CD8⁺ Tregs in all peripheral tissues compared to AL-PBS controls (Fig. 3p–r; Supplemental Fig. 5p-r). Upon MDSC replenishment, the population of Foxp3⁺CD8⁺ Tregs remained unchanged in AL-MDSC recipient mice vs. AL-PBS controls, whereas a significant upregulation in this subset of Tregs occurred in the spleen of CR-MDSC vs. CR-PBS mice (Fig. 3p; Supplemental Fig. 5p). These findings align with clinical outcomes, wherein Foxp3⁺CD8⁺ Tregs have been found in greater frequency in patients with advanced breast cancer [43].

Cytotoxic T cells (CD4⁺ and CD8⁺) are the predominant immune defense against cancer growth and one which can be significantly dampened by MDSC proliferation. A significant increase in effector CD4 + T cells (Fig. 3s–u; Supplemental Fig. 5s-u) and CD8⁺ T cells (Fig. 3v–x; Supplemental Fig. 5s-u) was observed in all peripheral tissues of CR-PBS mice vs. AL-PBS controls. However, this increase of protective T cells with CR was significantly blunted by MDSC administration, resulting in significantly fewer CD4⁺ and CD8⁺ T cells in CR-MDSC mice compared to CR-PBS mice (Fig. 3s–x; Supplemental Fig. 5s–u). These findings suggest that CR blocks MDSC expansion, thereby enabling the antitumorigenic functions of cytotoxic T cells.

To test this outcome, mice were injected with an IgG (control) or CD8⁺ antibody (experimental design, Supplemental Fig. 6a). Antibody-mediated suppression of CD8⁺ T cells reduced the anticancer properties of CR (Supplemental Fig. 6b and 6c). No difference in bodyweight or food consumption in mice injected with IgG or CD8⁺ antibodies was observed (Supplemental Fig. 6d and 6e). These findings are similar to prior work [44], suggesting the importance of CR in enabling cytotoxic T cell expansion and suppression of tumor growth.

CR decreases the immunosuppressive capacity of MDSCs

The release of immunosuppressive cytokines, L-arginine depletion, and production of reactive oxygen species (ROS) and NO [45] provide MDSCs with multiple means of blocking T cell proliferation. In the present study, we assessed whether CR decreases the immunosuppressive capacity of MDSCs.

The total population of MDSCs and its two subsets, PMN-MDSCs and M-MDSCs, were isolated from peripheral blood of AL-PBS and CR-PBS mice and incubated with fluorescent probes to assess NO production (DAF-FM-diacetate) as well as hydrogen peroxide (H₂O₂) (measured using DCFDA) and superoxide (O₂^.−) generation (measured with DHE) (Fig. 4a–i). Although total MDSCs showed no significant difference in NO production between the two groups, there was a downward trend in PMN-MDSCs (Lys6G⁺) in response to CR (Fig. 4b vs. a). PMN-MDSCs have been reported to be the predominant source of H₂O₂ generation within MDSC subsets [40]; however, PMN-MDSCs were not responsible for the elevated production of H₂O₂ seen in total MDSCs from CR-fed mice (Fig. 4e vs. d). While no difference in O₂^.− production was observed in total MDSCs regardless of diet, PMN-MDSCs isolated from CR-MDSC recipient mice exhibited a significant reduction in O₂^.− generation (Fig. 4h vs. g). No change within the M-MDSC (Lys6C⁺) subset was observed for any of the probes used (Fig. 4c, f, i).

Fig. 4.

Daily CR diminishes MDSC functional capacity. MDSC suppressive capacity against cytotoxic T cell proliferation was measured by the assessment of reactive oxygen species production via flow cytometry in (a, d, g) total MDSC population (CD11b⁺Gr1⁺) and the two MDSC subsets: (b, e, h) granulocytic (Ly6G⁺CD11b⁺Gr1⁺) and (c, f, i) monocytic (Ly6C⁺CD11b⁺Gr1⁺). Measurement of individual reactive oxygen species produced by MDSCs: (a–c) nitric oxide, (d–f) hydrogen peroxide, and (g–i) superoxide. (j) Arginase activity of MDSCs. (k) T cell proliferative capacity in CD8⁺ (left) and CD4⁺ effector cells exposed to MDSCs from AL- or CR-fed mice. Data are presented as scatter plots representing mean ± SEM, with Student’s t-test used to determine significance. *p < 0.05; **p < 0.01; and ***p < 0.001 compared to AL controls (black), with n = 6–7 mice per treatment group

The overt consumption of L-arginine by arginase greatly limits the proliferative capacity of cytotoxic T cells [46]. Indeed, we found that the reduction of arginase activity in the total MDSC population of CR-fed mice (Fig. 4j) was correlated with enhanced proliferation of naïve splenic effector T cells (CD8⁺ and CD4⁺) (Fig. 4k).

MDSC transfusion tilts the primary TME from immunologically “hot” to “cold”

TNBC is a highly heterogenous disease, with some tumors evading immune attack and progress, whereas others are destroyed and regress. These differences are partially attributed to changes within the tumor immunological landscape. Immunologically “hot” tumors arise due to TIL invasion, whereas immunologically “cold” tumors lack TIL infiltration [47, 48]. Prior findings suggest CR may work through tumor microenvironment (TME) remodeling to trigger immune-mediated tumor regression via penetration of CD8⁺ effector T cells [22, 44]. Because MDSCs inhibit TIL proliferation and infiltration in the TME, we sought to determine whether the antitumoral effects of CR stem from MDSC suppression which, in turn, tilts the tumor towards an immunologically ‘hot’ environment.

The occurrence of MDSC downregulation within peripheral tissues led us to assess the status and activation of the total MDSC population within the TME under AL and CR conditions. No significant changes in total MDSCs (Fig. 5a; Supplemental Fig. 7a) or iNOS expression (Fig. 5b; Supplemental Fig. 7b) were found, regardless of diet or MDSC supplementation. M-MDSCs are the predominant MDSC subset within the TME and have higher suppressive capacity than PMN-MDSCs [40]. Compared to AL-PBS controls, the TME from CR-PBS tumors had significantly less PMN-MDSCs (Fig. 5c; Supplemental Fig. 7c), while the TME from AL-MDSC mice exhibited a significant increase in M-MDSCs (Fig. 5d; Supplemental Fig. 7d).

Fig. 5.

Immune profile within the tumor microenvironment. Frequency of immune cells detected in the primary 4T1 tumor by flow cytometry. (a) CD11b⁺Gr1⁺ (MDSCs); (b) iNOS expression in total MDSC population; (c, d) proportion of the granulocytic and monocytic MDSC subsets, respectively. Frequency of (e) Foxp3⁺CD4⁺ T regulatory cells and (f) Foxp3⁺CD8⁺ T regulatory cells. (g) CD19 + B regulatory cells; tumor-infiltrating lymphocytes (h) CD4⁺ and (i) CD8⁺ cells. (j–o) Proportion of different subsets of CD8⁺ T cells (j) CD44^+high, (k) IFNƔ⁺, (l) CTLA⁺, (m) TIM-3⁺, (n) LAG-3⁺, and (o) PD-1⁺. Data are presented as scatter plots representing mean ± SEM, with two-way ANOVA used to determine statistical significance (Supplemental Table 4). *p < 0.05; **p < 0.01 compared to AL-PBS mice (black symbols), AL-MDSC mice (gray symbols), or CR-PBS mice (red symbols), with n = 8–10 mice per treatment group

Tumor-evoked regulatory B cells (tBregs) are CD19⁺ B cells that suppress T cell activation [49] and induce CD4⁺ transformation into Tregs [25]. To understand whether MDSC transfusion would override the CR-mediated inhibition of Tregs and tBregs within the TME, isolated primary tumors were collected and dissociated to assess sub-populations of immune cells within the TME. AL-MDSC and CR-MDSC mice showed a significant increase in Foxp3⁺CD4⁺ Tregs (Fig. 5e; Supplemental Fig. 7e), but not in Foxp3⁺CD8⁺ Tregs (Fig. 5f; Supplemental Fig. 7f), shedding light on a potential Treg-specific response within the TME. The downregulation of B cells was found only in TME of CR-PBS mice (Fig. 5g; Supplemental Fig. 7 g), a finding similar to the response triggered with a CR-mimetic [50]. Since tBregs are the key inducers of Tregs in mice with 4T1 cells [49] and MDSC replenishment in CR mice increased B cells in the tumor, these results suggest that increased tBreg counts could partly contribute to MDSC-induced immune remodeling in TME.

Effector CD4⁺ cells are critical in priming tumor specific CD8⁺ T cells for expansion [51] and associated tumor eradication [44, 52]. Although a CR-mediated upregulation of CD8⁺ cytotoxic T cells was observed in peripheral tissues of 4T1 tumor-bearing mice (Supplemental Fig. 2), no significant change in CD4⁺ (Fig. 5h; Supplemental Fig. 7 h) or CD8⁺ T cells (Fig. 5i; Supplemental Fig. 7 h) occurred within the TME of these animals. These results are consistent with the idea that the quality of intratumoral CD8⁺ T cells, rather than their density, affects disease-free survival [7]. Invasion of TILs (mainly cytotoxic CD8⁺) within the TME [47, 53], coupled with an increase in memory T cells [38, 54] and IFN-Ɣ production [46], promotes tumor regression in pre-clinical cancer models. Here, CR-PBS mice showed significant upregulation of CD44^high memory CD8⁺ cells (Fig. 5j; Supplemental Fig. 7j) and increased number of IFNƔ-expressing CD8⁺ T cells (Fig. 5k; Supplemental Fig. 7 k), whereas MDSC transfusion abrogated these anti-tumoral properties of CR. These findings suggest a higher tumor-killing capacity of TILs in the TME of CR-fed mice.

Antibodies that target the so-called immune-checkpoints (e.g., T lymphocyte-associated antigen 4 (CTLA4), T cell immunoglobulin and mucin domain 3 (TIM-3), T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT1), lymphocyte activation gene-3 (LAG-3), and programmed death (PD-1)) improve TIL activity and, thus, anticancer responses [55]. Elevation of these immune markers also indicates a “hot” or inflamed TME [47, 48] and is linked with improved survival in the 4T1 pre-clinical model [54] and TNBC patients [47, 56–60]. Noteworthily, MDSC supplementation markedly blocked the significant upregulation of CTLA4-, TIM-3-, LAG-3-, and PD-1-expressing CD8⁺ cells in the TME of CR-PBS mice (Fig. 5l–o; Supplemental Fig. 7 k–n).

Discussion

Caloric restriction (CR) is the leading non-pharmacological intervention shown to slow the progression of induced and spontaneous tumors in pre-clinical models [6, 20, 44] and to lower the risk of breast cancer [61]. CR works by remodeling whole-body metabolism through inhibition of pro-growth pathways; however, less is known about the benefit of CR toward inhibition of pro-tumorigenic immune reprogramming. MDSCs possess an effective array of immune-suppressive mechanisms to block cytotoxic T cell proliferation [46]. MDSC enrichment, detected in a myriad of tumors, enables the tumors to evade immune detection and is associated with poor clinical prognosis [11]. The decrease in MDSC frequency under dietary interventions (e.g., CR) and CR-mimetics is linked with reduced primary tumor growth [20, 21]. Here, the significant drop in MDSC levels that we observed in peripheral tissues of 4T1 tumor-bearing mice fed daily with CR led us to determine whether ectopic administration of MDSCs would eliminate CR’s ability to slow primary tumor growth.

A key finding from this study is that the injection of ectopic MDSCs dampens CR’s ability to delay primary tumor growth, without impacting tumor progression in AL-fed controls. Our findings showing that MDSC transfusion can limit the ability of CR to slow 4T1 tumor growth are concordant with the known reliance of breast cancer on MDSCs to block anti-tumor immunity [46]. Because the administration of MDSCs was initiated early in the tumor growth cycle (e.g., 2 and 10 days after 4T1 cell implantation), the immune remodeling mediated by CR must be occurring early within the tumor initiation and growth period.

A major consequence of TNBC is metastatic relapse, where approximately one third of women that undergo treatment develop metastatic spread [62] to the lungs and lymphatic system [3, 12]. MDSCs have been suggested to facilitate the ideal niche for metastatic expansion into distal tissues [63]. Metabolic dysfunction, as evident in obesity, provides optimal pro-inflammatory conditions for MDSCs to flourish [64]. Conversely, CR and CR-mimetics lower metastatic lung burden in pre-clinical models of breast cancer [20, 65]. Here, we find that CR mitigates metastatic lung invasion even after MDSC transfusion, suggesting that CR may have a broader impact on limiting expansion of pro-tumorigenic immune cells, beyond MDSCs. From a clinical perspective, the ability of CR to prevent metastatic outgrowth is important because patients typically do not die from the primary tumor, but rather from metastatic burden in the periphery [2].

A unique finding from our study was the metabolic reprogramming evident in tumor-bearing mice after MDSC transfusion. CR lowers pro-growth signals, including a reduction in glucose levels and downregulation of the insulin/IGF-1 signaling pathways [6], two outcomes that dampen cancer cell proliferation [66]. Here, the significant reduction in blood glucose and IGF-1 levels in CR-fed mice was mitigated by ectopic administration of MDSCs. Replenishment of MDSCs in AL-fed mice was associated with metabolic dysregulation, as evidenced by the increase in HOMA2-IR, a marker of insulin resistance, coupled with elevation in circulating insulin and IGF-1 levels. This was not unexpected as MDSCs rely on glycolysis to meet their energy needs [67]. Moreover, direct interaction with cancer cells or cancer-derived signaling molecules can trigger increased dependence of early-stage myeloid cells on glycolysis and development of their immunosuppressive capacities [67]. From a clinical standpoint, high insulin levels have been associated with increased risk for breast cancer in non-diabetic women [68], as well as its recurrence and all-cause mortality [37], suggesting that excess insulin — or remodeling of the metabolic framework by MDSCs — may have a pro-tumorigenic effect. Similarly, increased IGF-1 levels are indicative of poor clinical outcome in TNBC patients [69].

T cell suppression by MDSCs occurs via ROS generation and arginine depletion [46], which, in turn, results in the dampening of pro-inflammatory responses leading to tissue repair [70]. Interestingly, MDSCs from CR-fed mice were less efficient at inhibiting T cell proliferation (CD4⁺ and CD8⁺) than AL-fed controls, potentially due to decreased MDSC numbers [22]. PMN-MDSCs are the dominant subset of the total MDSC population in peripheral tissues [40], exhibiting lower levels of superoxide and hydrogen peroxide generation. MDSCs create a niche favorable for metastatic expansion [71], and patients with advanced TNBC have high levels of MDSCs that are positively associated with increased metastatic burden [10, 23]. Hence, CR may help slow metastatic burden by lowering the suppressive capacity of endogenous MDSCs in peripheral tissues [6, 20, 72]. Our findings show that CR’s ability to suppress metastatic growth occurred regardless of MDSC transfusion. However, it is unclear whether a beneficial effect of CR on metastasis would be observed if MDSC supplementation was performed later in tumor development. Alternatively, we surmise that another CR-mediated immune pathway may be at play in reducing metastatic burden.

To date, chemotherapy is the mainstay of treatment for TNBC due to lack of effective alternatives. However, this treatment results in an unpredictable response, largely attributed to the tumor-immune microenvironment [47]. The positive treatment response to chemotherapy is largely dependent upon TILs (CD8⁺) and expression of pro-cytotoxic markers. The subclonal heterogeneity in TNBC makes it difficult to ascertain whether specific markers on CD8⁺ TILs are associated with good or poor prognosis. Here, our findings show that daily CR upregulates checkpoint markers on TILs, specifically PD-1⁺, CD44^high+, IFNƔ⁺, CTLA⁺, TIM3⁺, and LAG3⁺. All these checkpoint markers have been previously found in other cancers to trigger T cell exhaustion and are associated with poor prognosis [73–75]. However, tumors from TNBC patients with upregulation of checkpoint receptors (PD-1, CTLA, LAG3, and TIM3) show the best overall prognosis [47]. Consistent with this, we observe up-regulation of these markers in response to CR, which is associated with slower rates of tumor growth, and both the effects of CR on immune markers and tumor development are largely negated by MDSC replenishment. This finding highlights the need to delineate the immunological profile to the type of breast cancer (presence or absence of estrogen receptors, progesterone receptors, and/or HER2 protein). Without this distinction, it is difficult to ascertain whether these markers are truly deleterious.

The PD-1/PD-L1 pathway is typically hijacked by many cancers to bypass TIL toxicity [76] and, therefore, targeting this pathway represents one of the leading immunotherapy approaches to combat various types of cancer. Prior findings point to upregulation of this checkpoint inhibitor as a positive prognostic marker for prolonged survival [77]. In TNBC patients, elevation of PD-1 on CD8⁺ TILs is associated with increased disease-free survival, an outcome independent of total CD8⁺ density [78]. Moreover, PD-1-expressing CD8⁺ T cells are capable of producing levels of pro-inflammatory cytokines comparable to effector cells [79]. Although we observed higher PD-1 expression specifically in CR-PBS mice, more work is needed to distinguish the nuances in the immune microenvironment specific to TNBC.

Overall, our findings suggest a novel interplay between dietary interventions (CR) and the immune response. Specifically, MDSC transfusion can limit CR-mediated tumor regression but is unable to block CR protection against lung metastases. Moreover, CR is able to remodel the immune environment in favor of a pro-inflammatory setting, previously shown to be beneficial in long-term survival and improvement in immunotherapy response [47]. More work is needed to better understand the impact of CR on immune remodeling in the presence of chemotherapeutics.

Limitations of the study

Even though the main goal of the present study was to assess if administration of donor MDSCs dampens CR protection against primary tumor growth in recipient animals, we did not ascertain whether MDSC transfusion later in tumor progression (e.g., beyond 10 days after 4T1 cell implantation) would further reduce the protective effect of CR, especially toward metastatic spread. It is important to note that the beneficial immune remodeling observed in response to CR has been studied in the context of TNBC and that different outcomes may occur in other forms of cancer wherein these same changes in immune cells may be considered negative prognostic markers. Therefore, more work is needed to elucidate whether the immune remodeling properties of CR are unique to TBNC or are present in other types of cancer.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

R.d.C., L.C.D.P.-W., M.B. (Monica), and A.B. formulated the experimental design. R.d.C. and L.C.D.P.-W. supervised the experiments. L.C.D.P.-W., M.C., D.C. P.K., S.N., O.B., J.K., S.W., and C.R.S. performed animal work. M.C. carried out various biochemical and molecular analysis of serum. Tumor measurements in mice were performed by L.C.D.P.-W. while M.B. (Monica), M.C., and D.C. carried out the isolation and staining of immune cells. L.C.D.P.-W. and R.d.C. completed the data analysis. L.C.D.P.-W. wrote the original draft and created the figures. Interpretation, review, and final editing was performed by A.B., M.B. (Monica), M.B. (Michel), N.L.P., and R.d.C.

Funding

This study was supported by the Intramural Research Program of the NIA/NIH, and L.C.D.P.W. was supported by the NIH Grant #Fi2GM123963 from the National Institute of General Medical Sciences.

Data availability

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

Code availability

No code was generated in this study.

Declarations

Disclaimer

The funder has no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Ethics approval

All animal protocols were approved by the Animal Care and Use Committee of the National Institute of Aging (NIA), NIH (Protocol Number: 277-TGB-2024).

Competing interests

The authors declare no competing interests.