Environmental activation of a hypothalamic BDNF-adipocyte IL-15 axis regulates adipose-natural killer cells

Abstract

Physical and social environments influence immune homeostasis within adipose tissue, yet the mechanisms remain poorly defined. We report that an enriched environment (EE) housing modulates the immune cell population in white adipose tissue of mice including an increase in the abundance of natural killer (NK) cells. EE upregulates the expression of IL-15 and its receptor IL-15Rα specifically within mature adipocytes. Mechanistically, we show that hypothalamic brain-derived neurotrophic factor (BDNF) upregulates IL-15 production in adipocytes via sympathetic β-adrenergic signaling. Overexpressing BDNF mediated by recombinant adeno-associated virus (rAAV) vector in the hypothalamus expands adipose NK cells. Conversely, inhibition of hypothalamic BDNF signaling via gene transfer of a dominant negative TrkB receptor suppresses adipose NK cells. In white adipose tissue, overexpression of IL-15 using an adipocyte-specific rAAV vector stimulates adipose NK cells and inhibits the progression of subcutaneous melanoma, whereas local IL-15 knockdown blocks the EE effect. These results suggest that bio-behavioral factors regulate adipose NK cells via a hypothalamic BDNF-sympathoneural-adipocyte IL-15 axis. Targeting this pathway may have therapeutic significance for cancer.

Introduction

Environmental factors and lifestyle impact health and contribute to an individual’s risk for cancer as well as prognosis. Epidemiologic studies indicate a connection between psychological stress and cancer progression (1, 2). Our work on environmental enrichment (EE), a housing environment boosting mental health, has revealed anti-cancer and anti-obesity phenotypes, which are not induced by exercise alone (3). Our studies and studies from other groups support the notion that components of EE—physical, social, and cognitive stimulations, work together to exert beneficial effects of EE (3–6). We have elucidated one key underlying mechanism: the activation of a specific neuroendocrine axis, the hypothalamic-sympathoneural-adipocyte (HSA) axis. Along this brain-fat axis, the key mediator in the hypothalamus is brain-derived neurotrophic factor (BDNF) whose upregulation preferentially elevates sympathetic drive to the white adipose tissue (WAT). The subsequent WAT remodeling includes the stimulation of VEGF contributing to the induction of beige cells, elevated energy balance, and resistance to obesity, and the suppression of leptin expression and release (3, 4, 7, 8). Leptin is thought to be a link between obesity and increased cancer risk (9). We have demonstrated that the robust drop of leptin is critical for the anticancer effect of EE (3, 7, 10). Furthermore, hypothalamic BDNF mediates the EE-induced immune modulation in secondary lymphoid tissue resulting in an anti-cancer effect characterized by an increase in the proportion of CD8 cytotoxic T lymphocytes and a decrease in the proportion of CD4 T helper cells (11). Additionally, EE enhances natural killer (NK) cell maturation in secondary lymphoid tissues that is mediated at least in part by hypothalamic BDNF (12). Other labs have reported significant anticancer effects of EE in additional solid tumor models including breast cancer, pancreatic cancer, colon cancer, glioma, and have elucidated additional mechanisms beyond the HSA axis (8, 13–15).

There is increasing interest in the roles of the adipose immune microenvironment, because the composition of immune population has significant consequences for both normal adipose tissue physiology and for the tumor immune microenvironment and cancer progression (16). Many different immune cell populations are found in adipose tissue, including myeloid and lymphoid cells. Both adaptive and innate lymphocytes play important roles in adipocyte physiology and response to stress such as high fat diet (HFD) and cold exposure (17–22). Recently, several studies have demonstrated that NK cells play a role in adipose tissue homeostasis and the initiation of insulin resistance in obese state (23, 24). However, little is known about how one’s physical and social environment may influence the adipose tissue immune microenvironment. Given the profound remodeling of the adipose tissue by an EE, we investigated how it regulates WAT immune populations, particularly on NK cells.

Material and Methods

Animal experiments.

Male C57BL/6 mice were purchased from Charles River, and randomly assigned to each intervention. The number of mice used in each experiment was based on previous publications (3, 4, 7, 8, 11, 12) and preliminary studies. All use of animals was approved by, and in accordance with the Ohio State University Animal Care and Use Committee.

EE Protocol.

Male 3-week-old C57BL/6 mice (Charles River) were housed in groups in cages of 1.5 m × 1.5 m × 1.0 m (10–20 mice per cage) supplemented with running wheels, tunnels, igloos, huts, retreats, wood toys, a maze, and nesting material (25). We housed control mice under standard laboratory (SE) conditions (5 mice per cage). Mice were housed in temperature (22–23 °C) and humidity-controlled rooms with food and water ad libitum. We fed the mice with normal chow diet (NCD, 11% fat, caloric density 3.4kcal/g, Teklad).

Isolation of stromal vascular fraction (SVF) from adipose tissue.

The adipose tissue was dissected and put into 12-well plate with Kreb-Ringer HEPES buffer (pH: 7.4). We minced the tissue into small pieces, and then added collagenase (sigma, C6885) at a concentration of 1mg/ml. Incubate and shake the plate at 37 °C, 120 pm for 45 min. Collect the mixture into tube with 5 mL Kreb-Ringer HEPES buffer, and spin at 1200 rpm for 10 min. Adipocytes were collected in the top layer. 1 mL RBC lysis buffer were added to lyse red blood cells. Spin down the cells and wash the cells with PBS once. And then re-suspend SVF pellet with PBS buffer and filter SVF cells using 70 μM strainer.

Flow cytometry.

We mechanically dissociated spleen through a 100 μm strainer to obtain single cell suspensions. Spleen cell suspension or SVF cells were stained with antibodies for 20 minutes on ice in the dark. Conjugated antibodies were purchased from BD Biosciences for CD3-FITC (Cat. #553062, 145-2C11), CD19-FITC (Cat. #553785, 15995), GR1-FITC (Cat. #553127, RB6-8C5), CD4-FITC (Cat. #553047, RM4-5), NK1.1-PE (Cat. #557391, PK136), CD11b-PE (Cat. #553311, 63324), CD11c-PECy7 (Cat. #558079, HL3), GR1-AF700 (Cat. #557979, RB6-8C5), CD19-AF700 (Cat. #557958, 1D3), CD25-APC (Cat. #557192, PC61), CD27-PE (Cat. # 558754, LG.3A10) and CD4-APCH7 (Cat. # 560181, GK1.5). Conjugated antibodies from Biolegend were purchased for CD206-FITC (Cat. #14704, C068C2), CD25-PE (Cat. #101904, 3C7), IL33R-PerCP-Cy5-5 (Cat. #145312, B207230), CD278-PECy7 (Cat. #313520, B213626), TCRβ-PECy7 (Cat. #109222, H57-597), CD45-APCH7 (Cat. #103116, 30-F11), CD127-V450 (Cat. #135024, A7R34), F480-V450 (Cat. #123132, BM8), CD45.2-FITC (Cat. # 109806, 104), NK1.1-PercpCy5 (Cat. #108728, PK136), CD3-PECy7 (Cat. # 100220, 17A2), CD25-APC (Cat. # 102012, PC61) and CD8a-BV421 (Cat. # 100738, 53-6.7). Conjugated antibodies from eBioscience were purchased for CD170-PerCP-Cy5-5 (Ref. #46-1702-80, 1RNM44N), TCRδγ-PerCP-Cy5-5 (Ref. #46-5711-82, eBioGL3), Foxp3-APC (Ref. #17-5773-80B, FJK-16s) and CD19-AF700 (Ref. #56-01-9382, EBIO1D3), CD1d (Ref. #12-0011-82, 1B1). Cells were stained with sytox and flow cytometry was performed using a BD LSRII, and results were analyzed using Flowjo.

rAAV Vector Construction and Packaging.

The rAAV plasmid contains a vector expression cassette consisting of the CMV enhancer and chicken β-actin (CBA) promoter, woodchuck post-transcriptional regulatory element (WPRE) and bovine growth hormone poly-A flanked by AAV2 inverted terminal repeats. Transgenes: Human BDNF, destabilized YFP, or TrkB.T1 (3) was inserted into the multiple cloning sites between the CBA promoter and WPRE sequence. rAAV serotype 1 vectors were packaged, and purified as described elsewhere (26).

microRNA targeting Il15.

We used microRNA to target mouse IL15. Two targeting sequences in the mouse Il15 coding sequence were cloned into the Block-iT PolII miR RNAi expression vector (pcDNA6.2-Gw/miR, Invitrogen). The knockdown efficiency was determined by co-transfecting HEK293 cells with a standard AAV expression plasmid containing mouse Il15 as transgene. Both miR constructs inhibited IL15 expression confirmed by qRT-PCR and ELISA. The miR-IL15_2 (mature miR seq: TGCAACTGGGATGAAAGTCAC) was chosen for in vivo experiments. This miR-IL15 and a scrambled miR (miRscr, targeting no known gene, Invitrogen, Inc.) were subcloned to the standard AAV plasmid and Rec2 serotype vectors were generated as described previously (27, 28).

Liver-restricting Rec2 vector expressing Il15.

Mouse Il15 cDNA was subcloned to a novel AAV plasmid of dual cassettes that restricts off-target transduction in liver (29). An empty vector containing no transgene was used as control. Rec2 serotype vectors were packaged to generate adipose-targeting liver-restricting AAV vector: AS/Rec2-IL15 and AS/Rec2-empty.

rAAV mediated BDNF or TrkB.T1 overexpression in hypothalamus.

Male C57BL/6 mice, 6 weeks of age, were purchased from Charles River and randomly assigned to receive AAV1-BDNF, AAV1-TrkB.T1 or AAV1-YFP (n=5 per group). Mice were anaesthetized with a single dose of ketamine/xylazine (100 mg/kg and 20 mg/kg; i.p.) and secured via ear bars and incisor bar on a Kopf stereotaxic frame. A mid-line incision was made through the scalp to reveal the skull and two small holes were drilled into the skull with a dental drill above the injection sites (−0.8AP, ±0.3ML, −5.0DV; mm from bregma). rAAV vectors (AAV1-BDNF, AAV1-TrkB.T1 or AAV1-YFP, 1×10¹⁰ vg per site) were injected bilaterally into the hypothalamus at a rate of 0.1μL/min using a 10μL Hamilton syringe attached to Micro4 Micro Syringe Pump Controller (World Precision Instruments Inc., Sarasota, USA). At the end of infusion, the syringe was slowly removed from the brain and the scalp was sutured. Animals were placed back into a clean cage and carefully monitored post-surgery until fully recovered from anesthesia. We monitored body weight every 5–7 days. Mice were maintained on NCD in SE until the end of the study (4 weeks after surgery).

miR-IL15 mediated IL15 knockdown in eWAT.

Male C57BL/6 mice, 6 weeks of age were housed in SE. Rec2-miR-IL15 or Rec2-miR-scr was injected intraperitoneally in 100 μL of AAV dilution buffer (2 × 10¹⁰ vg per mouse). These mice were randomly assigned to SE or EE housing (n=7–8 per group). Tissues were collected and analyzed three weeks later.

IL15 transgenic mice.

IL-15 Tg mice overexpressing Il15 were generated and maintained as described previously (30). These mice (n=5 per group) were housed in SE and subjected to indirect calorimetry at the age of two months and glucose tolerance test at the age of three months.

Propranolol Experiment.

We randomly assigned 20 male C57BL/6 mice, 3 weeks of age, to live in an EE or SE supplied with propranolol (Roxane Laboratories, Inc. Cat# 0054-3727-63) in drinking water (0.5 g/L) for two weeks.

CL316243 experiment.

Male C57BL/6 mice, 3 weeks of age were housed in SE. β3-AR agonist CL316243 (Sigma) was i.p. injected for 7 consecutive days (1mg/kg/d, n=5 per group).

Western blot.

Adipose tissues or adipocytes were homogenized in RIPA buffer containing Protease Inhibitor Cocktail Set III (Calbiochem). Blots were incubated overnight at 4°C with the following primary antibodies: GAPDH (1:500, CB1001; Calbiochem), tubulin (1:1000, Cat#2144, Cell Signaling) and IL15 (1:500, PA5-34511, Thermo Fisher Scientific). ImageJ was used to quantify the bands of western blot.

B16-F10 Melanoma experiment.

Male C57BL/6 mice, 6 weeks of age, were randomized to receive AS/Rec2-IL15 or AS/Rec2-empty (2 × 10¹⁰ vg per mouse, i.p.) and housed in SE. Metabolic studies were performed at 4~5 weeks after AAV injection. After a week interval, B16-F10 melanoma cells (ATCC) were subcutaneously implanted on the flank (1×10⁵ cells/mouse, n =10 per group). The mice were sacrificed on day 18 after tumor implantation.

Quantitative RT-PCR.

Hypothalamus was block dissected from mouse brain. Tissue was sonicated, and total RNA was isolated using RNeasy Kit plus RNase-free DNase treatment (Qiagen). cDNA was reverse transcribed using TaqMan Reverse Transcription Reagent (Applied Biosystems). Quantitative PCR was carried out on StepOnePlus Real-Time PCR System (Applied Biosystems) using Power SYBR Green PCR Master Mix (Applied Biosystems). Primers were designed to detect mouse mRNA include: Bdnf, Adrb2, Adrb3, Lep, Ppargc1a, Il6, Il33, Il15, Il15ra, Il2rg. Primer sequences are available on request. Relative gene expression was quantified using the 2^−ΔΔCT method (31).

Statistical analysis.

Data are expressed as mean ± s.e.m. Data were determined to be normally or approximately normally distributed according to the Shapiro-Wilk test. Means between two groups were compared with two-tailed Student’s t-tests. For multiple comparisons, two-way ANOVAs with pairwise comparisons were used to determine statistical significance. P values less than 0.05 were considered significant, and the level of significance was indicated as *p < 0.05, **p < 0.01, and ***p < 0.001.

Results

EE increases NK cell abundance in subcutaneous and visceral white adipose tissue

Immediately weaned, male C57BL/B6 mice were randomly assigned to live in either standard laboratory housing (SE) or EE housing. Although SE-housed and EE-housed mice weighed the same at 2 weeks (Figure 1A), EE reduced the mass of inguinal WAT (iWAT, a surrogate for subcutaneous fat) and epididymal WAT (eWAT, a surrogate for visceral fat), both absolutely (Figure 1B) by ~40% and relative to total body weight (Figure 1C). These EE-induced changes in WAT are associated with increased basal metabolic rate, insulin sensitivity, and lipolysis (4, 5).

Figure 1. EE modulates adipose immune microenvironment.

A. Body weight. B. iWAT and eWAT mass. C. Relative fat mass calibrated to body weight. D. SVF cell numbers per fat depot. E. Immune cell abundance of SVF. F. Fold change of immune cell composition in iWAT. G. Total NK cell numbers in WAT. H. Fold change of immune cell composition in eWAT. I. Gating strategy of NK cells. J. Representative flow images of iWAT. Data are mean ± SEM. n=5 per group. * P<0.05, ** P<0.01, *** P<0.001.

The stromal vascular fraction (SVF) of adipose tissue contains both non-immune cells, such as endothelial cells, fibroblasts and preadipocytes, and immune cells including myeloid cells, adaptive and innate lymphocytes (32). EE decreased the number of SVF cells in eWAT but not in iWAT (Figure 1D). EE did not change the percentage of CD45⁺ immune cells within the SVF of either eWAT or iWAT (Figure 1E).

We performed a flow cytometry-based screening to identify cellular changes in the WAT immune microenvironment. EE increased the abundance of NK cells in SVF of both iWAT and eWAT defined as CD45⁺ LIN (CD3 CD19)⁻ NK1.1⁺ (Figure 1F, H). EE also increased the abundance of innate lymphoid cells (ILCs), defined as LIN⁻ CD127⁺ in iWAT but not in eWAT. The NK cell numbers relative to fat mass from iWAT were two folds higher in EE mice than SE mice (Figure 1G). The NK cell number from eWAT displayed similar increase (Figure 1G).The EE effects on myeloid populations were dependent on the fat depots. EE showed no effects on myeloid populations in iWAT whereas it reduced macrophage and M2-polarized macrophage abundance while increased eosinophils in eWAT (Figure 1F, H). CD8⁺ T cell abundance was decreased in both iWAT and eWAT (Figure 1F, H). A representative flow cytometry gating schematic for defining NK cells is shown (Figure 1I, J).

EE alters adipocyte gene expression

To elucidate the mechanism of increased adipose NK cells in EE mice, we profiled gene expression of adipocytes after a 2-weeks exposure of EE. EE significantly upregulated β3-adrenergic receptor (Adrb3) gene expression in iWAT (Figure 2A) and eWAT adipocytes (Figure 2C), consistent with the activation of the HSA axis (3, 4). In contrast, the downregulation of leptin (Lep) or upregulation of PGC-1α (Ppargc1a)—a key regulator of mitochondrial metabolism, which are associated with EE at 4 weeks or longer (4, 5), were not observed at the 2-week interval (Figure 2A, C).

Figure 2. EE induces gene expression changes in adipose tissue.

A. iWAT adipocyte. B. iWAT SVF C. eWAT adipocyte. D. eWAT SVF. E. Western blot of iWAT. Each number represents an individual mouse. F. Quantification of western blot. Data are mean±SEM. n=5 per group in A-D, n=4 per group in E-F. * P<0.05, ** P<0.01, *** P<0.001.

EE induced the upregulation of IL-15 (Il15) and IL-15Rα (Il15ra), both of which support NK cell homeostasis (33–35), in mature adipocytes isolated from both iWAT and eWAT, while the IL-2/15γ_c (Il2rg) was not changed (Figure 2A, C). Immunoblot confirmed the increase of IL-15 protein levels in adipocytes (Figure 2E, F). No change of IL-15 and β-ARs was observed in the SVF (Figure 2B, D).

Hypothalamic BDNF is a key brain mediator of EE effects on adipose NK cells

To test whether the BDNF-driven HSA axis mediates the effect of EE on adipose NK cells, we selectively overexpressed BDNF using a bilateral hypothalamic injection of a recombinant adeno-associated virus (rAAV) vector or a rAAV-YFP control vector as previously described (3, 11) and housed the mice in SE. At 4-week post rAAV injection, mice receiving rAAV-BDNF had decreased body weight and adiposity compared to rAAV-YFP control (Figure 3A, B), consistent with the activation of the HSA axis (4). rAAV-BDNF-treated mice showed a greater proportion of adipose NK cells and higher expression of adipocyte Il15 and Il15ra mRNA when compared to YFP controls (Figure 3C–E). B cells were decreased, and eosinophils were increased. Hypothalamic overexpression of BDNF in SE-house mice was thus sufficient to replicate the cytokine and NK cellular phenotype observed in adipose tissue of EE-housed mice.

Figure 3. Hypothalamic BDNF regulates adipose NK cells. A-D. Hypothalamic overexpression of BDNF.

A. Body weight. B. Relative iWAT mass calibrated to body weight. C. Fold change of immune cell composition in iWAT, calculated as %CD45 (BDNF-YFP)/YFP. D. NK cell abundance in iWAT. E. iWAT adipocyte gene expression 4 weeks after AAV-BDNF injection and housed in SE. F-J. Inhibition of hypothalamic BDNF signaling. F. Body weight. G. Relative iWAT mass calibrated to body weight. H. Fold change of immune cell composition in iWAT, calculated as %CD45 (TrkB.T1-YFP)/YFP. I. NK cell abundance in iWAT. J. iWAT adipocyte gene expression 3 weeks after AAV-TrkB.T1 injection and housed in SE. Data are mean±SEM. n=5 per group. * P<0.05, ** P<0.01, *** P<0.001.

Conversely, we used a dominant-negative truncated form of the high-affinity BDNF receptor (TrkB.T1) to specifically inhibit BDNF signaling in the hypothalamus (4). Mice receiving a hypothalamic injection of rAAV-TrkB.T1 increased weight and adiposity when compared to YFP-injected mice (Figure 3F, G). TrkB.T1-treated mice had a reduced proportion of adipose NK cells and lower expression of adipocyte Il15 and Il15ra mRNA when compared to YFP controls (Figure 3H–J). B cells were increased, and eosinophils were decreased. These data support the hypothesis that the EE effect on adipose NK cells is mediated by hypothalamic BDNF.

EE requires the sympathetic nervous system (SNS) to regulate adipose NK cells.

Preferential elevation of sympathetic tone to the WAT links hypothalamic BDNF to the remodeling of WAT (3, 4, 8). We used the β-blocker propranolol to examine whether an intact SNS was required for the EE effect on adipose NK cells. Mice receiving propranolol in their drinking water were randomly assigned to EE or SE for 2 weeks. Body weight and adiposity were similar between the two groups (Figure 4A, B). However, propranolol prevented the EE-induced increase of NK cells and upregulation of adipocyte Il15 and Il15ra expression (Figure 4C, D).

Figure 4. Sympathetic signaling regulates adipose NK cells. A-D. Propranolol inhibits EE regulation of adipose NK cells.

A. Body weight. B. Relative iWAT mass calibrated to body weight. C. Fold change of immune cell composition in iWAT, calculated as %CD45 (EE-SE)/SE. D. iWAT adipocyte gene expression in mice treated with propranolol and housed in EE for 2 weeks. E-H. CL-316243 mimics EE regulation of adipose NK cells. E. Body weight. F. Relative iWAT mass calibrated to body weight. G. Fold change of immune cell composition in iWAT, calculated as %CD45 (CL-PBS)/PBS. H. iWAT adipocyte gene expression in mice treated with CL-316243 for 7 days. Data are mean±SEM. n=5 per group. * P<0.05, ** P<0.01, *** P<0.001.

β1- and β2-ARs are widely expressed among tissues including immune cells. β3-AR is more selectively expressed in rodent adipose tissue (36), where it mediates SNS-driven, metabolic remodeling of WAT (37–40). Thus, we used the β3-AR specific agonist, CL-316243, to test whether β3-AR activation replicated the EE-induced increase of adipose NK cells. SE-housed mice were injected with either PBS as a vehicle control or CL-316243 for 7 days. CL-316243 treatment had no significant effects on body weight or iWAT adiposity (Figure 4E, F) whereas reduced eWAT mass by ~30%. CL-316243 induced a robust increase of adipocyte Il15 and Il15ra expression as well as higher abundance of adipose NK cells in iWAT, thereby mimicking the EE effect on NK cell homeostasis (Figure 4G, H). We also noted that a concordant decrease of total T cells in the adipose tissue as result of CL-316243 treatment. Thus, an intact SNS is necessary for the EE-induced increase in adipose NK cells and adipocyte-synthesized IL-15. Stimulation of the β3-AR is sufficient to recapitulate this effect.

Knockdown of adipocyte IL-15 eliminates the NK cell modulations associated with EE

IL-15 supports NK cell development and homeostasis (33–35), and the cytokine is presented in trans to NK cells from dendritic cells (41–43). The adipocyte may be another source of IL-15 (44). Given the correlation between upregulation of adipocyte-derived Il15 and Il15ra and the change in adipose NK cells, we hypothesized that adipocyte-derived IL-15 directly mediates the increase of adipose NK cells seen in the EE. To test this hypothesis, we generated a microRNA against murine Il15 (miR-Il15). We also generated a control microRNA targeting a scrambled sequence (miR-scr) against no known genes (3). We then used a dual-cassette rAAV expression system coupled with an engineered hybrid serotype Rec2 (AS/Rec2 vector) to achieve efficient and selective transduction of adipocytes while restricting off-target transduction of liver by intraperitoneal (i.p.) administration (29, 45).

We injected mice with AS/Rec2-miR-Il15 or AS/Rec2-miR-scr (2× 10¹⁰ vg per mouse, i.p.) and then randomly assigned half of vector group to live in EE or SE. After 4-weeks EE exposure, eWAT mass was lower in EE mice regardless of the injection with miR-Il15 or miR-scr compared to their counterparts in the SE housing (Figure 5A, B). The reduction of the serum leptin level is a reliable readout for EE-induced metabolic adaptations (3, 4, 8). Both miR-Il15 and miR-scr mice living in EE showed a significant decrease in the serum leptin level (Figure 5C). Furthermore, miR-Il15 did not interfere with the EE-induced upregulation of Bdnf in the hypothalamus (Figure 5D). However, miR-Il15 gene delivery to adipocytes blocked EE-induced upregulation of adipocyte Il15 mRNA (Figure 5E) and attenuated the upregulation of adipocyte Il15ra mRNA (Figure 5F). In addition, adipocyte Il15 knockdown reduced the adipose NK cell proportion in the SE-housed mice (miR-Il15/SE versus miR-scr/SE) and prevented the increase of adipose NK cells in the EE-housed mice (miR-Il15/EE versus miR-Il15/SE) (Figure 5G). Of note, adipocyte Il15 knockdown had no significant effect on adipose T cell percentage in CD45+ cells (SE/miR-scr: 15.43% ± 0.82%, EE/miR-scr: 14.72% ± 1.04%, SE/miR-Il15: 14.38% ± 0.84%, EE/miR-Il15: 12.23% ± 1.17%, n=7–8 per group).

Figure 5. Adipose IL-15 knockdown inhibits EE modulation of adipose NK cells.

A. eWAT mass after i.p. injection of rAAV vectors and EE housing. B. Relative eWAT mass calibrated to body weight. C. Serum concentration of leptin. D. Hypothalamic Bdnf expression. E. eWAT adipocyte Il15 expression. F. eWAT adipocyte Il15ra expression. G. eWAT NK cells abundance. Data are mean±SEM. n=7–8 per group. * P<0.05, ** P<0.01, *** P<0.001, # P=0.07.

Adipocyte overexpression of IL-15 selectively increases adipose NK cells

To conduct a gain-of-function study on adipocyte-derived IL-15, we generated Rec2 vectors to deliver Il15 preferentially to the visceral fat (AS/Rec2-IL15) and used a vector containing the same expression cassette but lacking a transgene as control (AS/Rec2-empty). SE-housed mice received either AS/Rec2-IL15 or AS/Rec2-empty (2 × 10¹⁰ vg per mouse, i.p.). At 3-weeks post rAAV injection, AS/Rec2-Il15 treatment did not lead to any difference in body weight (Figure 6A), thymus, spleen, iWAT, or eWAT mass (Figure 6B) compared to mice receiving AS/Rec2-empty. No significant changes of eWAT SVF cell number (Figure 6C), serum levels of IL-15 or leptin (Figure 6D), or eWAT T cell percentage in CD45+ cells (AS/Rec2-empty: 9.53% ± 1.4%, AS/Rec2-Il-15: 8% ± 0.9%, n=4 per group) were observed.

Figure 6. Adipose overexpression of IL-15 stimulates locally adipose NK cells.

A. Body and B. Tissue weight. C. SVF cell number. D. Serum concentration of IL-15 and leptin. E. eWAT adipocyte gene expression. F. Western blot of IL-15 from eWAT adipocytes and quantification. G. NK cell abundance in blood, spleen and eWAT. H. Representative flow images NK cells from eWAT. Data are mean±SEM. n=4 per group. * P<0.05, ** P<0.01, *** P<0.001.

AS/Rec2-IL15 injected mice did not show changes in adipocyte gene expression of Il15ra, Adrb3, Lep, or Ppargc1a (Figure 6E), but did displayed a robust overexpression of Il15 in eWAT adipocytes determined by immunoblot and qRT-PCR (Figure 6E, F) and a two-fold increase in adipose NK cells in the absence of any significant change in blood or spleen NK cells (Figure 6G, H). No off-target Il15 overexpression was found in liver (Supplementary Figure). Thus, rAAV-mediated overexpression of Il15 in visceral fat primarily led to a local increase in adipose NK cells.

Activation of visceral adipose NK cells by local IL-15 inhibits distal tumor growth

NK cells are able to recognize and lyse malignant cells without prior antigen specificity (46). We asked whether adipose NK cells could be mobilized against the progression of a tumor at distal location. Mice were randomized to receive i.p. injection of AS/Rec2-Il15 or AS/Rec2-empty (2 × 10¹⁰ vg per mouse). Six weeks after rAAV injection, B16 melanoma cells (10⁵ cells per mouse) were implanted subcutaneously on the flank. At 18-days post tumor inoculation, AS/Rec2-Il15-treated mice showed 64% reduction of tumor mass compared to AS/Rec2-empty-treated mice (Figure 7A, B). There was no significant change of serum IL-15 or leptin between the two groups (Figure 7C, D). The decrease in tumor size of AS/Rec2-Il15-treated mice was associated with an increase of NK cells within eWAT, spleen, and tumor compared to the control (Figure 7E).

Figure 7. Stimulation of adipose NK cells via adipocyte overexpressing IL-15 inhibits tumor progression.

A. B16 melanomas dissected 18 days after tumor cell implantation. B. Tumor weight. C. Serum IL-15 level. D. Serum leptin level at sacrifice. E. NK cell abundance in eWAT, spleen, and tumor. Data are mean±SEM. n=7 per group. * P<0.05, ** P<0.01.

Discussion

In this study, we demonstrate that an EE increases NK cell abundance in both subcutaneous and visceral fat depots in lean animals through a hypothalamic-BDNF-sympathoneural-adipocyte-IL-15 axis. Several lines of evidence support the notion of hypothalamic BDNF-adipocyte IL-15 axis. 1. Hypothalamic overexpression of BDNF replicated the EE-induced expansion of adipose NK cells whereas inhibition of BDNF signaling from the hypothalamus via TrkB.T1 reversed this effect. 2. Signaling through the SNS serves as the link between hypothalamic BDNF and WAT. The administration of a β-blocker prevented the expansion of adipose NK cells in the EE, and the use of a β3-specific agonist identified at least one of the adrenergic receptors activating the adipocyte. 3. The key downstream mediator of the NK cell expansion within this axis is adipocyte-derived IL-15. The EE upregulated Il15 and Il15ra expression in the mature adipocyte but not in the SVF. Knockdown of adipocyte Il15 sufficiently blocked EE-induced upregulation of Il15 mRNA and abolished the EE effect on adipose NK cells, yet did not interfere with other changes associated with the EE such as upregulation of hypothalamic BDNF or suppression of leptin. Conversely, overexpression of IL-15 in adipocytes expanded the adipose NK cell population in a SE without affecting blood-circulating or spleen NK cells, body weight, adiposity, circulating leptin, or adipose gene expression. Taken together, these data demonstrate EE-housing shapes the innate immune microenvironment within WAT driven by a defined brain-fat axis.

Although the EE-induced WAT remodeling is controlled by one brain mediator, BDNF, and is driven by SNS signaling, the consequences at the WAT level appear dissociable. To date, three phenomena have been identified to account for the BDNF-mediated WAT remodeling: suppression of leptin expression and release (3, 4), induction of beige cells via upregulation of VEGF (8), and an increase in NK abundance via upregulation of IL-15 as demonstrated in this study. Each of these changes is downstream of adipocyte β-AR signaling but are not interdependent.

EE housing confers on mice a tumor-resistant phenotype, at least in part, through modulation of immune effector cells important for controlling tumor progression (11, 13, 47). Regarding peripheral NK cells, EE acts through the SNS to 1) promote the terminal maturation of NK cells in the bone marrow, spleen, and blood (48); 2) modify NK activating surface receptors (49); 3) increase NK cell infiltration of peripheral tumors (48, 49); 4) enhance spleen NK cell cytotoxicity against cancer cells (3, 48, 50). Our recent study demonstrates that EE promotes peripheral NK recovery after depletion and hypothalamic BDNF mediates the enhancement of NK maturation induced by EE (12). Blockade of sympathetic signaling either with propranolol (3, 48) or sympathectomy with 6-OHDA (49) eliminates these NK modifications and anti-tumor effects. Of note, the SNS blockade also prevents the metabolic phenotype induced by an EE such as suppression of leptin that is also important to the anticancer effect (3, 7). This study reveals how EE regulates NK cell homeostasis in adipose tissue through a hypothalamic BDNF-adipocyte IL-15 axis. These data support the notion that EE induces both metabolic and immune adaptations that may collectively contribute to resistance to cancer onset and progression (3, 7), and hypothalamic BDNF orchestrating these peripheral responses.

The importance of IL-15 on the anti-cancer ability of NK cells in EE-housed mice is complex and may depend on the location and type of tumor. One study has reported an important role for brain IL-15 in the regulation of NK activation and inhibition of glioma progression in EE-housed mice (13), whereas other groups have reported no increase in serum or splenic IL-15 (48, 49), and only a slight increase at the site of pancreatic tumor (8). Despite the absence of an increase of systemic IL-15 in EE mice, these studies observed an increase in the frequency of NK cells in the blood and spleen of EE mice when they were challenged with a peripheral tumor (48, 49). In our study, adipocyte-specific overexpression of IL-15 did not alter serum IL-15 or splenic NK cells in the absence of a tumor but did significantly increase splenic NK cells when challenged with a distal tumor. These data suggest that targeting the adipocyte to enhance cellular immunity, either with direct gene transfer of IL-15 or through activation of hypothalamic BDNF, may be a possible strategy for enhancing the antitumor activity of NK cells for cancer treatment. In a separate gene therapy project, we investigated the therapeutic efficacy of an adipose-targeting AAV vector expressing an IL-15/IL-15Rα complex. Intraperitoneal administration of rAAV-IL-15/IL-15Rα complex significantly inhibited the growth of Lewis lung cancer implanted subcutaneously and exerted a significant survival advantage in a B16-F10 melanoma metastasis model. Notably, in the absence of a tumor challenge, adipose IL-15/IL-15Rα complex gene transfer resulted in the expansion of NK cells not only in the visceral fat but also in the spleen (51). In contrast, adipose IL-15 gene transfer with the same vector system, dose, and administration route, did not affect NK cells in the spleen (Figure 6). The difference might be due to enhanced secretion and biodistribution of IL-15/IL-15Rα complex versus IL-15 alone.

One limitation of the current study is lacking direct evidence of the role of adipose NK in EE-induced cancer resistance. We generated adipose-specific IL-15Rα knockout mice by crossing adiponectin-Cre mice with IL-15Rα^F/F mice and subjected them to EE housing. The preliminary data showed some genotype effects on metabolism and cancer progression, likely independent of the decrease of adipose NK cells (data not shown). IL-15Rα is reported to be a determinant of muscle fuel utilization (52). We are currently investigating whether IL-15Rα directly affects adipocyte functions and thereby influencing cancer progression via modulating systemic metabolism independent of its classical role in immune regulation. To delineate the precise roles of adipose resident immune cells including NK cells responding to an EE is challenging because of crosstalk between the adipocytes and various adipose resident immune populations. We are planning alternative approaches such as an inducible knockout system to knockout adipocyte Il15 in adults, or testing our rAAV-miRIl15 for global adipose Il15 knockdown. Studies have shown that EE exerts anticancer effects through multimodal mechanisms including metabolic modulation such as reducing circulating leptin and immune modulation such as enhancing CD8 T cells and NK cells immunity in spleen and lymph node. Because adipose NK contributes a small portion of the global NK pool, it is possible that adipose NK modulation is not essential to the anticancer effect of EE. Nevertheless, the characterization of the hypothalamic BDNF-adipocyte IL-15 axis in this study expands understanding on how physical and social environments regulate adipose resident immune cells.

Myeloid cells also regulate adipose health, and adipose macrophages display significant heterogeneity in their activity and functions (53). The eWAT of EE-housed mice showed a reduction in total macrophages, and a concordant decrease in both M2 abundance (Fig 1H) and IL-6 mRNA expression (Fig 2C). IL-6 is a critical instigator of alternatively-activated macrophages (M2 macrophage) polarization (54). In obese adipose tissue, IL-6 promotes local proliferation of adipose tissue macrophages and contributes to glucose tolerance, whereas M2 macrophages in a lean adipose tissue seems to represent a resting, tissue-resident M2 phenotype characterized by a low proliferation rate (55). In EE-housed mice, the changes in IL-6 and macrophages were observed in visceral but not subcutaneous adipose tissue, which suggests different mechanisms of regulation that remain to be determined.

Moreover, B and T lymphocytes of the adaptive immune system, also regulate adipose tissue health and glucose homeostasis (56). The accumulation of both B cell and CD8 T cells are reported in WAT of diet-induced obese mice (57, 58). In this study, EE decreased B cell abundance in iWAT but not eWAT. This B cell change was also regulated by hypothalamic BDNF and dependent on an intact SNS. In addition, EE led to a relative reduction of CD8 T cells in WAT. We recently reported that EE regulates thymus and thymocyte development mediated by hypothalamic BDNF via the activation of the hypothalamic-pituitary-adrenal axis (59). The functions and the underlying mechanisms of the EE-induced modulation of B and T cells remain to be determined.

In summary, our results demonstrate that a physically, mentally, and socially enhanced lifestyle of an EE regulates the adipose immune microenvironment as part of the adipose remodeling in mice. The combination of pharmacological and genetic mechanistic studies has identified a specific brain-fat axis regulating the increase in adipose NK cells, mediated upstream by hypothalamic BDNF and downstream by at least the β3-AR on adipocytes to upregulate IL-15 and IL15Rα expression. Direct gene transfer of IL-15 to target visceral fat inhibits distal melanoma progression, suggesting therapeutic potential for cancer treatment.