Elevated leptin during diet-induced obesity reduces the efficacy of tumor immunotherapy

Abstract

Various malignancies are reproducibly cured in mouse models, but most cancer immunotherapies show objective responses in a fraction of treated patients. One reason for this disconnect may be the use of young, lean mice lacking immune altering co-morbidities present in cancer patients. While many cancer patients are overweight or obese, the effect of obesity on antitumor immunity is understudied in preclinical tumor models. We examined the effect of obesity on two immunotherapeutic models: systemic anti-CTLA-4 mAb and intratumoral delivery of a TRAIL-encoding adenovirus plus CpG. Both therapies were effective in lean mice, but neither provided a survival benefit to diet-induced obese (DIO) BALB/c mice. Interestingly, tumor-bearing leptin deficient (ob/ob) obese BALB/c mice did respond to treatment. Moreover, reducing systemic leptin with soluble leptin receptor:Fc restored the antitumor response in DIO mice. These data demonstrate the potential of targeting leptin to improve tumor immunotherapy when immune modulating co-morbidities are present.

Introduction

Obese humans face an array of health issues, including an increased risk for cancer. Cancer-associated morbidity and mortality are also greater in obese persons. The reason for this obesity-cancer link is multifactorial, but generalized immune system dysfunction is a significant contributor (1). Increased cancer risk during obesity has also been ascribed to elevated production of tumor-promoting hormones and growth factors by adipocytes. As weight increases and develops into obesity, visceral adipose tissue actively contributes to a state of systemic, low-level inflammation. Growing evidence suggests chronic inflammation, resulting from obesity and/or tumor growth, plays a salient role in tumor cell survival/proliferation and immune suppression (2–4). The obesity epidemic makes understanding how this common co-morbidity influences the immune response to cancer imperative (5). This is especially true for cancers that highly correlate with obesity, such as renal cell carcinoma (RCC), which affects >60,000 and kills >14,000 people in the U.S. annually, making it the second most lethal urologic cancer (6–9).

Herein we describe the impact of obesity on two immunotherapy strategies. We have previously described an experimental therapy using a recombinant adenovirus encoding TNF-related apoptosis-inducing ligand (TRAIL) in combination with a TLR agonist (CpG) in an orthotopic mouse model of RCC (10–12). Delivered directly into the primary tumor-bearing kidney, Ad5-TRAIL/CpG immunotherapy stimulates an abscopal CD8⁺ T cell-mediated antitumor response that significantly enhances survival in lean but not diet-induced obese (DIO) mice (10, 13). These data implied that some obesity-related factor was responsible for immunotherapy failure. The present study extends our previous work by evaluating immunotherapeutic activity in a second model of obesity caused by the genetic loss of leptin using ob/ob mice, which spontaneously become obese without dietary intervention (14), as well as using a systemically administered checkpoint inhibitor (anti-CTLA-4 mAb). Tumor growth was blunted and overall survival enhanced in lean and ob/ob obese mice after local or systemic antitumor immunotherapy, whereas DIO mice did not demonstrate any benefit with either treatment. The different therapeutic responses between diet-induced and leptin deficient (ob/ob) obesity led us posit the elevated leptin in DIO mice was a contributor to therapeutic failure. Indeed, reducing the amount of leptin in DIO mice by systemic delivery of recombinant leptin receptor:Fc (ObR:Fc) restored immunotherapy efficacy. These data collectively indicate the importance of incorporating clinically-relevant co-morbidities into preclinical tumor models to broaden the testing of immunotherapeutic protocols. Moreover, our data have identified leptin as a potential therapeutic target for neutralization to enhance immunotherapy efficacy in obese cancer patients.

Materials and Methods

Animals and Diets

Female BALB/c mice (7–8 weeks old; Charles River) were given standard chow or high fat feed (HFF; Research Diets #12492) for 20 weeks ad libitum. Mice in the HFF group whose body weight was > 3 S.D. above the mean weight of the standard chow group were defined diet-induced obese (DIO), while mice whose body weight was < 3 S.D. above the mean weight of the standard chow group were defined obese resistant (10). BALB/c mice with a heterozygous deletion of the leptin gene (ob/+) were bred to produce leptin deficient (ob/ob) offspring. Male and female littermates (ob/+ or +/+) were used as controls. ob/ob and littermate controls were fed standard chow from weaning throughout the duration of experiments. Body composition analysis was performed using an Echo-MRI 3-in-1 (Echo Medical Systems LLC, Houston, TX).

Leptin Quantitation and Neutralization

Serum leptin was measured by ELISA (Chrystal Chem Inc, Downers Grove, IL). Soluble mouse leptin receptor:Fc (ObR:Fc) (15) was purchased from R&D Systems (Minneapolis, MN).

Tumor Challenge

The murine renal adenocarcinoma cell line Renca (ATCC, CRL-2947) was transfected with the Sleeping Beauty transposon system to confer stable expression of firefly Luciferase and maintained in RPMI supplemented with 10% fetal bovine serum, 100 U/mL penicillin, 100 µg/mL streptomycin, 1mM sodium pyruvate, 1x non-essential amino acids, 50 µM 2-Mercaptoethanol, 10 mM HEPES, and 0.05 µg/mL puromycin. Intrarenal (IR) tumor challenge was performed as described (10–12). On d 7, mice were injected in the tumor-bearing kidney with sterile PBS or Ad5-TRAIL (10⁹ pfu) (16) and CpG ODN 1826 (5’-tccatgacgttcctgacgtt-3’, 100 μg; IDT, Coralville, IA). Anti-CTLA4 mAb or control Ig (10 mg/kg i.p., clones 9D9 and MCP-11, respectively; BioXcell, West Lebanon, NH) was given on d 4, 8, 11, 15, and 18 after subcutaneous (s.c.) tumor challenge (17).

DC activation

Splenocytes were cultured for 24 h in RPMI media alone or with CpG (10 μg/ml), and then stained for CD8α DC (I-A^d+CD11c⁺CD8α⁺ cells) or plasmacytoid DC (pDC; I-A^d+CD11c⁺B220⁺CD317⁺ cells) and analyzed by flow cytometry to determine co-stimulatory molecule expression.

Intravascular staining

To identify the CD8⁺ T cells among the CD45.2⁺ leukocytes located in the tumor-bearing kidney tissue, intravascular staining was done using anti-CD45.2-PE (BioLegend, San Diego, CA) (18). Tumor-bearing kidneys were harvested, manually disrupted, and digested with DNaseI (15 µg/ml; Sigma) and Liberase Blendzyme 3 (0.026 Wuensch unites/ml; Roche). After blocking with anti-CD16/32 in normal mouse serum, cells were stained (anti-CD45.2-BV650, CD3-v500, CD4-AF700, CD8-BUV395; eBioscience, San Diego, CA or BioLegend) and analyzed using multi-parameter flow cytometry on a BD LSR Fortessa (BD Biosciences, San Diego, CA) and FlowJo software (TreeStar Inc., Ashland, OR).

Statistics

Statistical comparisons of body weight composition, serum leptin, mean fluorescence intensity (MFI), and tumor-infiltrating CD8⁺ T cells were performed using the Student’s t-test. Survival data was compared by Log-rank (Mantel-Cox) analysis. All tests were performed using Prism 7 (Graph Pad Software, Inc). p values < 0.05 were considered statistically significant.

Results and Discussion

Diet-induced vs. genetic models of obesity in mice

Murine obesity can be modeled using a diet of high-fat feed (HFF) or leptin gene (ob/ob) deficiency. Slowly progressing and clinically relevant DIO using HFF is marked by increased leptin production and systemic inflammation (19–21). Consistent with our previous data (10), ~50% of BALB/c mice on HFF for 20 weeks became obese (Fig. 1A). In contrast, all leptin-deficient ob/ob BALB/c mice became obese relative to ob/+ littermates. Furthermore, body composition analysis demonstrated the elevated body weight resulted in increased body fat in the DIO and ob/ob mice compared to their lean counterparts (Fig. 1B). Serum leptin was expectedly increased in DIO mice compared to mice fed standard chow, and no leptin was detected in the serum of ob/ob mice (Fig. 1C). The serum leptin levels in the lean and DIO mice are consistent with amounts seen in lean and obese humans (22). These data define the starting parameters of the lean and obese (DIO and ob/ob) mice.

Figure 1. Body weights, fat accumulation, and serum leptin levels in lean and obese mice.

(A). Mice fed standard chow or high fat feed (HFF) for 20 weeks were weighed. Mice with a body weight 3x the S.D. of the standard chow animals (indicated by the dashed line) were considered obese, those that did not reach this threshold were considered obese resistant. ob/+ and ob/ob were fed a standard chow for the duration of the experiments. All ob/ob mice become obese without dietary intervention (dashed line indicates 3x the s.d. of the ob/+ littermates). (B) Body composition analysis revealed the increased weight in DIO and ob/ob mice was due to fat accumulation. (C) Serum leptin was determined in standard chow and HFF mice after 20 weeks and ob/+ and ob/ob mice at 16–22 weeks of age. *** p < 0.005, **** p < 0.0001, n.d. – not detected. Results are representative from at least 3 independent experiments, where at least 5 mice were in each group.

Immunotherapy is effective in lean and ob/ob obese mice, but not DIO mice

Intratumoral delivery of Ad5-TRAIL/CpG in an orthotopic mouse model of RCC (10, 13, 16, 23) resulted in a significant survival advantage compared to PBS treatment in lean mice, but no difference in survival was noted among PBS- and Ad5-TRAIL/CpG-treated DIO mice (Fig. 2A). In contrast, tumor-bearing ob/ob mice displayed increased survival after Ad5-TRAIL/CpG therapy, paralleling the response seen in the ob/+ lean littermates (Fig. 2B). Lean, ob/+, and ob/ob mice bearing s.c. Renca tumors also demonstrated enhanced overall survival when given systemic anti-CTLA-4 mAb therapy (17), whereas DIO mice failed to respond (Fig. 2C-D). Importantly, both Ad5-TRAIL/CpG and anti-CTLA-4 mAb therapies significantly increased survival in HFF obese resistant mice co-housed with HFF DIO mice (Supplemental Fig. 1), which provides an important internal control to exclude such variables as an effect of diet or microbiome differences as contributing factors. These data suggest obesity alone is not enough to alter the efficacy of immunotherapeutic treatment of tumor-bearing mice.

Figure 2. Lean and obese ob/ob mice, but not DIO mice, respond to Ad5-TRAIL/CpG therapy.

(A-B) Age-matched standard chow fed (“lean”), DIO, ob/+, and ob/ob mice were implanted intrarenally (IR) with 2×10⁵ Renca cells. PBS or Ad5-TRAIL (10⁹ pfu)/CpG (100 µg) was then injected IR into the tumor-bearing kidney on d 7. Survival data are shown for lean and DIO mice (A) and ob/+ and ob/ob mice (B). (C-D) Lean, DIO, ob/+, and ob/ob mice were implanted subcutaneously (s.c.) with 2×10⁵ Renca cells, followed by intraperitoneal (i.p.) injection (10 mg/kg) of anti-CTLA-4 mAb or control Ig (cIg) on d 4, 8, 11, 15, and 18. Survival data are shown for lean and DIO mice (C) and ob/+ and ob/ob mice (D). Differences between PBS- and Ad5-TRAIL/CpG-treated mice or cIg- and anti-CLTA-4-treated mice in each group were determined using the Log-rank (Mantel-Cox) test. Data presented are combined from 3 independent experiments, where at least 5 mice were used in each group

An effective antitumor immune response requires proper antigen (Ag) presentation by Ag-presenting cells (APC) to CD8⁺ T cells. Dysfunctional APC would result in poor CD8⁺ T cell priming and reduced antitumor response. CD8α⁺ DC and plasmacytoid DC (pDC) are needed for Ad5-TRAIL/CpG therapeutic efficacy (23), prompting us to test the responsiveness of these DC populations from DIO and ob/ob mice to CpG stimulation. We noted increased CD86, CD80, and CD40 expression on the surface of both CD8α⁺ DC and pDC from spleens of non-tumor bearing lean and ob/+ mice after CpG stimulation (Fig. 3A and data not shown), as well as DC from ob/ob mice. In contrast, co-stimulatory molecule expression on either CD8α⁺ DC or pDC from DIO mice after CpG stimulation was not significantly different from that seen on unstimulated cells.

Figure 3. Differential responsiveness of DC from DIO and ob/ob mice to CpG correlates with reduced antitumor T cell response in DIO mice, but not ob/ob mice, after Ad5-TRAIL/CpG therapy.

(A) CpG activates CD8α⁺ DC and pDC from lean and ob/ob mice, but not DIO mice. Splenocytes from lean, DIO, ob/+, and ob/ob mice were cultured in vitro for 24 h in media alone or with CpG 1826 (10 µg/ml). CD86 expression was determined on I-A^d+ CD11c⁺ CD8α⁺ DC and I-A^d+CD11c⁺B220⁺CD317⁺ pDC within the bulk splenocytes. (B) DIO, but not ob/ob mice, have reduced CD8⁺ T cell infiltration into tumor-bearing kidneys after treatment with Ad5-TRAIL/CpG (Ad5-T/CpG). Mice were implanted intrarenally (IR) with Renca (2×10⁵ cells) and then treated with PBS or Ad5-TRAIL/CpG on d 7. Tumor-bearing kidneys were harvested on d 12 and examined for immune cell infiltration using intravascular staining and multi-parameter flow cytometry. The frequency and number of CD8⁺ T cells in the tissue of tumor-bearing kidneys was determined. * p < 0.05, ** p < 0.01, *** p < 0.005, **** p < 0.001, n.s. - not significant. Data presented are representative of at least 3 independent experiments, where 4–5 mice were used in each group in each experiment.

CD8⁺ T cells are also required for the reduced tumor burden after Ad5-TRAIL/CpG administration (16). Using an intravascular staining technique (18) to distinguish tissue-localized CD8⁺ T cells from those in the vasculature, we noted a significant increase in frequency and number of CD8⁺ T cells in the tissue of tumor-bearing kidneys of lean mice after Ad5-TRAIL/CpG injection compared to PBS injection (Fig. 3B). This influx was not observed in DIO mice, but ob/ob mice also had increased CD8⁺ T cell infiltration after Ad5-TRAIL/CpG therapy like the littermate controls. Nearly all of the CD8⁺ T cells detected in the kidney tissue had also upregulated CD11a (Supplemental Fig. 2), indicating they were “Ag experienced” (24) and infiltrating the tumor-bearing kidney in an Ag-specific manner. Collectively, these data suggest DIO mice have functionally impaired DC, correlating with decreased tumor infiltration by CD8⁺ T cells and the reduction in therapeutic efficacy.

Leptin neutralization in DIO mice restores therapeutic efficacy

The two models of obesity used herein have yielded different responses to immunotherapy, and one obvious difference between the two models is the amount of endogenous leptin. DIO mice (and obese humans) have elevated levels of leptin compared to lean controls, whereas ob/ob mice lack leptin entirely (see Fig. 1C). To examine the extent to which leptin was responsible for the disparate results, DIO mice were given a soluble recombinant leptin receptor:Fc fusion protein (ObR:Fc) to reduce the amount of systemic leptin to the levels detected in lean mice (Fig. 4A). Tumor-bearing DIO mice were given ObR:Fc prior to Ad5-TRAIL/CpG or anti-CTLA-4 mAb therapy. While DIO mice failed to show a response to either therapy, ObR:Fc-treated DIO mice now responded to both Ad5-TRAIL/CpG and anti-CTLA-4 mAb therapy, similar to normal weight controls (Fig. 4B-C). Further investigation into the restoration of therapeutic responsiveness after ObR:Fc administration found increased costimulatory molecule expression on CD8α⁺ DC and pDC after CpG stimulation (Fig. 4D-E) and increased CD8⁺ T cell infiltration of tumor-bearing kidneys after Ad5-TRAIL/CpG injection (Fig. 4F). Collectively, these data suggest leptin is one factor driving the lack of immunotherapy efficacy in DIO mice and (in part) explains why ob/ob mice are able to mount an antitumor response despite their profound obesity.

Figure 4. Leptin neutralization restores therapeutic efficacy in DIO mice.

(A) Serum leptin in DIO mice was measured before and 3 d after 3 daily doses of soluble mouse leptin receptor:Fc (ObR:Fc; 100 µg/dose i.v.). Serum leptin from lean mice is shown for reference. (B) Lean and DIO mice were implanted intrarenally (IR) with Renca (2×10⁵ cells), followed by PBS or Ad5-TRAIL/CpG (Ad5-T/CpG) treatment on d 7. Some DIO mice received ObR:Fc on d 2–4 (100 µg i.v./q.d.) to neutralize leptin. (C) Lean and DIO mice were implanted subcutaneously (s.c.) with 2×10⁵ Renca cells, followed by intraperitoneal (i.p.) injection (10 mg/kg) of anti-CTLA-4 mAb or control Ig (cIg) on d 4, 8, 11, 15, and 18. Some DIO mice received ObR:Fc on d 1–3 (100 µg i.v./q.d.) to neutralize leptin. Survival results are shown and differences between PBS- and therapy-treated mice in each group were determined using the Log-rank (Mantel-Cox) test. (D-E) Splenocytes from lean, DIO, and DIO mice given ObR:Fc (100 µg i.v./q.d. for 3 d) were cultured for 24 h in media alone or with CpG 1826 (10 µg/ml). CD86 expression was determined on I-A^d+ CD11c⁺ CD8α⁺ DC and I-A^d+CD11c⁺B220⁺CD317⁺ pDC within the bulk splenocytes. (F) Lean and DIO mice were implanted IR with Renca (2×10⁵ cells), followed by PBS or Ad5-TRAIL/CpG (Ad5-T/CpG) treatment on d 7. Some DIO mice received ObR:Fc on d 2–4 (100 µg i.v./q.d.) to neutralize leptin. Tumor-bearing kidneys were harvested on d 12 and examined for immune cell infiltration using intravascular staining and multi-parameter flow cytometry. The number of CD8⁺ T cells in tissue of tumor-bearing kidneys was determined. * p < 0.05, ** p < 0.01, *** p < 0.005, **** p < 0.001, n.s. - not significant. Data in panel A is representative of 2 independent experiments, where 3 mice were used in each group. Data in panels B and C are combined from 3 independent experiments, consisting of 4–6 mice per group in each experiment. Data in panels D-F are representative of 2–3 experiments, where 4–5 mice were used in each group per experiment.

Currently, over one-third of the adult population in the U.S. is obese (5). The increased cancer risk associated with obesity and growing use of cancer immunotherapies make it crucial to understand how obesity can impact antitumor immune responses. We have demonstrated diet-induced obesity in mice compromises the immune response to tumor immunotherapy. DIO mice have less functional DC, reduced tumor infiltration by CD8⁺ T cells, and ultimately an impaired response to therapy. Our data suggest the chronically elevated levels of leptin observed in obesity has a detrimental effect on immunotherapy. Leptin is a 16 kDa protein hormone secreted by adipocytes that signals through its receptor, ObR, in the hypothalamus to signal satiety (25, 26). The absence of leptin, as in ob/ob mice, results in lack of satiation, continuous feeding and ultimately obesity. However, with increasing adiposity, excess leptin can be measured in the circulation resulting in leptin-resistance and overeating (27). Most studies evaluating the role of leptin on immunity have focused on ob/ob mice in which leptin is completely absent or short term, high dose administration of leptin. Our use of DIO mice better recapitulates the chronic overexpression of leptin experienced during human obesity. Leptin is now recognized to have additional roles in multiple systems, including the immune system, as ObR is expressed on all innate and adaptive immune cells (28). Moreover, ObR is expressed on a wide variety of mouse and human tumors, and leptin can have a direct effect on tumor cells – promoting a variety of pro-tumor consequences including cytokine signaling, growth, and invasion (29–31). Understanding the effect of leptin on tumor and immune cells will identify intervention points to enhance immunotherapy efficacy in cancer patients. Further, leptin neutralization using mAb or soluble ObR constructs could serve as a means of increasing the therapeutic “window” in obese cancer patients.

Most preclinical testing of immunotherapies occurs in healthy mice lacking many of the immune modulating co-morbidities present in humans, which may be one reason for the limited clinical success of many promising therapies developed and testing in preclinical animal models. Our data establishes the need for addressing complications of obesity in the design of cancer therapies, and a number of reports have examined the impact of obesity on immunological responses during cancer or infection. Obesity causes a number of changes within the immune system that can reduce the generation and potency systemic immunity, which warrant further investigation and consideration as a mechanism to reduced responsiveness to immunotherapy.