High-Fat Diet-Induced Inflammation Accelerates Prostate Cancer Growth via IL6 Signaling

Tumor-promoting inflammation and avoidance of the immune system were reported to be some of the new hallmarks of cancer (1). Inflammation-mediated cancer progression is accelerated by macrophages, mast cells, and granulocytes via various cytokines and is suppressed by T cells (2). Myeloid-derived suppressor cells (MDSCs), which play a role in suppressing multiple immune effectors and T cells, are linked with inflammation and cancer (3). Inflammation can also promote the initiation and progression of prostate cancer (4). Macrophage polarization plays an important role in tissue inflammation and regeneration (5). Recently, alternatively activated macrophages (M2 macrophages) were suggested to have pro-tumor functions (6).

Prostate cancer is linked with dietary habits (7). A high-fat diet (HFD) causes obesity and chronic inflammation (8, 9), and epidemiologic studies have shown that HFD could be associated with progression and survival of prostate cancer (10, 11). Studies have also shown HFD induced tumor progression via adipose-secretory cytokines or chemokines in TRAMP mice and within a xenograft mouse model of the prostate cancer cell line LNCaP (12–14). However, in a xenograft model of patient-derived prostate tumor tissue using immunodeficient mice, HFD was reported not to induce tumor progression (15). These results suggest that HFD might accelerate tumor progression of prostate cancer via interactions with immune responses including various cytokines. Although HFD increases serum proinflammatory cytokines and promotes prostate cancer progression in TRAMP mice (16), it remains unclear whether the tumor progression resulted from these cytokines.

It was reported that a genetically engineered autochthonous mouse model (Pb-Cre⁺;Pten(fl/fl)), which corresponds to somatic mutations of human prostate cancer (17), was valuable for elucidating the interaction of prostate cancer with the tumor microenvironment (18).

The aim of this study was to identify the mechanisms of the associations of prostate cancer with inflammation and immune responses by administration of an HFD and celecoxib, which has anti-inflammatory functions, to Pb-Cre⁺;Pten(fl/fl) model mice.

In the C57BL/6 genetic background, prostate-specific Pten-knockout mice (Pten-deficient model mice [Pb-Cre⁺;Pten(fl/fl)]) were generated by crossing Pten^loxP/loxP mice (Pten flox mice [Pten(fl/fl)]) with ARR2Probasin-Cre [Pb-Cre] mice, wherein the Cre recombinase is under the control of a modified rat prostate-specific probasin promoter. All mice were bred and maintained in the Institute of Experimental Animal Sciences of Osaka University Medical School. All animal experiments were approved by the Osaka University Animal Research Committee and conducted according to relevant regulatory standards.

All mice were fed either a control diet (CD; total energy: 3590 kcal/kg, 12.5% energy from fat, 25.7% protein, 61.8% carbohydrate; MF, Oriental Yeast, Tokyo) or high-fat diet (HFD; total energy: 5062 kcal/kg, 62.2% energy from fat, 18.2% protein, 19.6% carbohydrate; HFD-60, Oriental Yeast, Tokyo). The HFD was started at 5 weeks of age.

Celecoxib (CAS169590-42-5, Tokyo Chemical Industry) was dissolved in the drinking water (calculated as 8 mg/kg/day), and administration was started at 5 weeks of age.

Intraperitoneal injections of 500 μg of MR16-1 (rat anti-IL6 receptor antibody, Chugai Pharma) and control rat IgG (Wako) were performed weekly (19), and injections were started at 6 weeks of age.

All mice were sacrificed at 22 weeks of age for evaluation of tumor progression and local immune cells.

The ratios of Ki67-positive or pSTAT3-positive cells to tumor cells in the model mice were counted at a magnification of ×400 under light microscopy, and the averages of the ratios in three different random areas were calculated. CD11b-positive and CD206-positive cell counts in the human specimens were evaluated in ×200 magnified images of tumor areas in which the positive cells were most infiltrated. The ratios of pSTAT3-positive cells to tumor cells in the human specimens were evaluated in ×400 magnified images of tumor areas in which the ratios were at maximum. The IHC study using human radical prostatectomy specimens was approved by the institutional review board of UAB Hospital, and the protocol has an approved waiver of authorization and informed consent (IRB #X140724007).

Single-cell suspensions were stained with conjugated antibodies after exposure to CD16/CD32 (BD Pharmingen) to block nonspecific binding, according to the manufacturers' instructions. 7-AAD (BD Pharmingen) for dead cell discrimination was added to samples prior to data acquisition. Flow cytometric analysis was performed on a FACSCanto II instrument (BD Biosciences), and data were analyzed with Flow Jo (Tree Star Inc.).

MDSCs, B cells, T cells, and macrophages were defined as “CD45⁺, CD11b⁺, Ly6G and Ly6C⁺”, “CD45⁺, CD19⁺”, “CD45⁺, CD3e⁺”, and “CD45⁺, F4/80⁺”, respectively. The CD8/CD4 ratio and M2/M1 macrophage ratio were defined as CD8a⁺/CD4⁺ of T cells and CD206⁺/MHC class II⁺ of macrophages, respectively.

Quantitative RT-PCR was performed with a Thermal Cycler Dice Real Time System (TP800, TaKaRa) using SYBR premix Ex Taq Π (TaKaRa). PCR conditions were 95°C for 30 seconds followed by 40 cycles at 95°C for 5 seconds and 60°C for 30 seconds for each gene-specific primer. The primers used in quantitative RT-PCR are indicated in Supplementary Table S1.

Relative expression levels were determined by normalization to Gapdh using the ΔΔC_t method. PCR reactions for each sample were carried out in triplicate.

Immunofluorescence stains were performed on 4-μm sections of formalin-fixed, paraffin-embedded tissue. Primary and secondary antibodies were applied before mounting with ProLong Gold antifade reagent with DAPI (Invitrogen). Confocal images were obtained by using an FV1000 microscope (Olympus).

Data are expressed as the mean ± SD. Comparison between two groups were made using the Mann–Whitney U test because the Shapiro Wilk W test revealed that the data were non-normally distributed. The Bonferroni correction was used to adjust for multiple tests. Statistical significance was determined as P < 0.05. All analyses were performed by JMP Pro 13 (SAS Institute).

More details can be found in Supplementary Data.

The prostate weights of the HFD-fed model mice were significantly higher than those of the CD-fed model mice (P = 0.004; Fig. 1A and B), whereas there was no marked difference in the glandular structures between the CD-fed and HFD-fed model mice (Fig. 1C). Undifferentiated tumor existed in a small fraction of the HFD-fed model mice (9.1%, 1 of 11 mice). Tumor proliferative capacity was evaluated by staining with Ki67 (Fig. 1D). The ratio of Ki67-positive cells to tumor cells in the HFD-fed model mice was significantly higher than that in the CD-fed model mice (P = 0.003; Fig. 1E). The comparisons of normal and Pten-deficient model mice and the changes of HFD-fed normal mice were described in Supplementary Data. Body weights of the HFD-fed model mice were significantly higher than those of the CD-fed model mice (P = 0.012; Supplementary Fig. S2A).

There were no significant differences in the fractions of B cells, T cells, macrophages, and CD8/CD4 ratio between the CD-fed and HFD-fed model mice (Fig. 2H). MDSCs of the HFD-fed model mice were significantly higher than those of the CD-fed model mice (P = 0.038; Fig. 1F and G). The M2/M1 macrophage ratio of the HFD-fed model mice was significantly higher than that of the CD-fed model mice (P = 0.008; Fig. 1H and I).

To evaluate the changes of tumor growth and local immune cells after the suppression of inflammation, we administered celecoxib (cyclooxygenase-2 inhibitor) to the CD- and HFD-fed model mice.

There was no significant difference in the body weights of the HFD-fed model mice between those with and without celecoxib (Supplementary Fig. S2A). Administration of celecoxib to the CD-fed model mice did not significantly change the prostate weights, whereas the prostate weights in the HFD-fed model mice administered celecoxib were significantly lower than those of the HFD-fed model mice without celecoxib (P = 0.002; Fig. 1A and B). There was no marked difference in the glandular structures of the HFD-fed mice between those with and without celecoxib (Fig. 1C). Undifferentiated tumor did not exist in HFD-fed model mice with celecoxib. The ratio of Ki67-positive cells to tumor cells was significantly lower in the HFD-fed model mice administered celecoxib than in those without celecoxib (P = 0.031; Fig. 1D and E).

MDSCs in the prostatic tissues of the HFD-fed model mice administered celecoxib were also significantly lower than those of the HFD-fed model mice without celecoxib (P = 0.044; Fig. 1F and G). The M2/M1 macrophage ratio of the HFD-fed model mice administered celecoxib tended to be lower than that in the mice without celecoxib (P = 0.076; Fig. 1H and I). There were no significant differences in the fractions of B cells, T cells, macrophages, and the CD8/CD4 ratio after administration of celecoxib to the HFD-fed model mice (Supplementary Fig. S2H).

To elucidate the mechanism of HFD-induced tumor growth and changes of immune cells, the expressions of cytokines associated with inflammation and polarization of macrophages in the prostatic tissues of the model mice were evaluated (Fig. 2A; Supplementary Fig. S3A). HFD significantly increased the expression of Il6 in the prostatic tissues of the model mice (P = 0.024), and celecoxib significantly decreased this expression in the HFD-fed model mice (P = 0.048). The HFD also significantly increased the expressions of Il1b, Il13, and Il17a (P = 0.018, 0.014, 0.018, respectively).

IL6 was reported to have pro-tumor functions (20). IHC analyses revealed that stromal immune cells but not cancer cells expressed IL6 in the HFD-fed model mice (Fig. 2B). Immunofluorescent analyses using both anti-F4/80 antibody and anti-CD68 antibody as macrophage markers showed that almost all of the IL6-positive cells were macrophages (Fig. 2C). Because the IL6 signal is transduced via phosphorylation of signal transducer and activator of transcription 3 (STAT3), pSTAT3 was immunohistochemically analyzed (Fig. 2D). HFD significantly increased the ratio of pSTAT3-positive cells to tumor cells in the model mice (P = 0.007), and celecoxib significantly suppressed the ratio in the HFD-fed model mice (P = 0.017; Fig. 2E).

There were no differences in the gene expressions of COX2 (Ptgs2) in the prostatic tissues between the CD-fed model mice, HFD-fed model mice, and HFD-fed model mice with celecoxib (Supplementary Fig. S3B). IHC analyses revealed that COX2 was expressed mainly in tumor cells (Supplementary Fig. S3C).

The results of analyses of gene and protein expressions in the prostatic tissues suggested that the IL6 signal could be a key regulator of HFD-induced tumor growth. To test this hypothesis, MR16-1 (anti-IL6 receptor antibody) was administered to the model mice to inhibit the IL6 signaling.

Administration of MR16-1 to the CD-fed model mice caused no significant change in the prostate weights, whereas the prostate weights of the HFD-fed model mice with administration of MR16-1 were significantly lower than those of the HFD-fed model mice with control IgG (P = 0.048; Fig. 3A and B). There was no marked difference in the glandular structures of the HFD-fed model mice between those with control IgG and MR16-1 (Fig. 3C). Undifferentiated tumor existed in a small fraction of the HFD-fed model mice with control IgG (10.0%, 1 of 10 mice), whereas it did not exist in HFD-fed model mice with MR16-1. The ratio of Ki67-positive cells to tumor cells was significantly lower in the HFD-fed model mice with administration of MR16-1 than in those with control IgG (P = 0.012; Fig. 3D and E).

MDSCs in the prostatic tissues of the HFD-fed model mice with administration of MR16-1 were significantly lower than those of the HFD-fed model mice with control IgG (P = 0.037; Fig. 3F; Supplementary Fig. S3D). There was no significant difference in the M2/M1 macrophage ratio of the HFD-fed model mice between MR16-1 and control IgG (Fig. 3G; Supplementary Fig. S3E).

The ratio of pSTAT3-positive cells to tumor cells was significantly lower in the HFD-fed model mice with MR16-1 than in those with control IgG (P = 0.022; Fig. 3H and I).

To confirm that the HFD-induced inflammation observed in the Pten-deficient model mice also exists in humans, tumor-infiltrating MDSCs and M2 macrophages and phosphorylation of STAT3 in tumor cells of prostate cancer were evaluated by immunohistochemistry for CD11b, CD206, and pSTAT3, respectively, using radical prostatectomy specimens of “non-obese” (BMI <26 kg/m²), “overweight and obese” (26 kg/m²≤ BMI <35 kg/m²) and “severely obese” (BMI ≥35 kg/m²) patients. The characteristics and prostate cancer grade group based on classifications from the 2014 ISUP (International Society of Urological Pathology) Consensus Conference are summarized in Table 1. There were no significant differences in age, race, and Gleason grade group of the two groups.

The tumor-infiltrating CD11b-positive cell count of the “overweight and obese” patients was significantly higher than that of the “non-obese” patients (P = 0.003; Fig. 4A and B). There were no significant differences in the tumor-infiltrating CD206-positive cell counts (P = 0.098; Fig. 4C and D) or in the ratio of pSTAT3-positive cells to tumor cells between the three groups (Fig. 4E and F). Multiple regression analysis showed that the tumor-infiltrating CD11b-positive cell count was significantly associated with the BMI groups (P = 0.010), but not with age, race or Gleason grade group.

The numbers of MDSCs, macrophages, and mast cells in the mice with prostate cancer were significantly higher than those in the mice with normal prostate. It was reported that local MDSCs, which promote immunosuppression, expanded during tumor progression in the Pten-deficient model mice (21). MDSCs and the M2/M1 macrophage ratio of the HFD-fed model mice were significantly higher than those of the CD-fed model mice. These results suggest that local MDSCs and M2 macrophages play roles in HFD-induced tumor growth and that local mast cells might be associated with inflammation and tumor initiation. HFD has been shown to alter the tumor immune microenvironment, including macrophages, and promote disease progression in pancreatic and breast cancer mouse models (22). In addition, administration of carcinogens to rats can induce infiltrations of mast cells and macrophages in prostatic tissue (23). In human studies, inflammation and immune cells are also linked to prostate cancer. We reported that increases of mast cells and macrophages in prostate biopsy specimens predict a poor prognosis (24, 25) and that peripheral monocyte count reflecting tumor-infiltrating macrophages was a predictive factor of adverse pathology in radical prostatectomy specimens (26).

Celecoxib suppressed tumor growth of the HFD-fed model mice but not the CD-fed model mice, which suggested that HFD-induced tumor growth was associated with local inflammation. The expansion of tumor-infiltrating CD11b-positive cells was noted in the obese patients with prostate cancer. Although celecoxib had no effect on improving survival in patients with locally advanced or metastatic prostate cancer in the STAMPEDE trial (27), our results suggested that celecoxib may have therapeutic benefits in specific subgroups with local inflammation in prostate cancer such as obese patients. However, we have no available data about obesity in the STAMPEDE trial. In our study, the human (50 kg body weight) dose of celecoxib equivalent to the 8 mg/kg/d dosing in mouse was 400 mg/d, which has been used clinically. Celecoxib suppresses the synthesis of prostaglandin E2 from arachidonic acids, which are parts of metabolites derived from ω6-polyunsaturated fatty acids (PUFA) contained in an HFD. Dietary ω6-PUFAs promoted tumorigenesis of model mice compared with ω3-PUFAs (28) and changed the phenotype of macrophages in a mouse allograft model of prostate cancer (29). Celecoxib was reported to suppress MDSCs and M2 macrophages (30). The gene expressions of COX2 (Ptgs2) were not associated with HFD-induced tumor progression, which was suppressed by celecoxib. Our results suggest that the local expression of COX2 is not an indication for treatment of prostate cancer with celecoxib.

The gene expressions of Il6 and Il13 in the prostatic tissues of the HFD-fed model mice were significantly higher than those in the CD-fed model mice, and IL6 was secreted mainly by local macrophages. In human prostate cancer, IL6 expression was also restricted to the prostate stromal component (31). IL6 and IL13 were respectively reported to be one of the inducers of MDSCs (32). IL6 signaling in tumor cells results in tumor progression transduced via phosphorylation of STAT3 (20), and IL13 promotes macrophage polarization to M2 (33). Inhibition of IL6 signaling suppressed HFD-induced tumor growth and local expansion of MDSCs.

Our results suggest that HFD-induced local inflammation accelerates tumor growth of prostate cancer via IL6 signaling (Fig. 5). HFD results in IL6 secretion by local macrophages, and IL6 induces tumor growth via phosphorylation of STAT3 and local expansion of MDSCs in a pro-tumor microenvironment. MDSCs may also play a role in suppressing multiple immune effectors and T cells (3).

MR16-1 (anti-IL6 receptor antibody), which is used to treat human autoimmune diseases, including rheumatoid arthritis (34), was useful in suppressing prostate cancer growth in our HFD-fed model mice. IL6 is a key factor for inflammation and prostate cancer (35). Reduced inflammation delays the accumulation of MDSCs and limits tumor progression (36, 37). Other drugs that have anti-inflammatory functions may be useful for the treatment of prostate cancer, as well as drugs that have anti-inflammatory functions combined with immune checkpoint inhibitors (38).

There were several limitations in this study. The HFD (5,062 kcal/kg) contained more total calories than the CD (3,590 kcal/kg) in our experiment and the difference in calories might have influenced tumor growth. We evaluated the amounts of calories in the diets the CD-fed and HFD-fed mice ingested. The results revealed that the HFD-fed mice ingested about 50% more calories than the CD-fed mice. A high-caloric diet might also have induced tumor growth and inflammation. Because administration of HFD to the model mice was started at 5 weeks of age before the initiation of prostate cancer, HFD may accelerate the initiation of cancer. It remains unclear how HFD induces local inflammation in the prostate. The mediators of the mechanism may be changes of metabolism or gut microbiota (39, 40). Although M1 and M2 macrophages were each separated by one marker (MHC class II and CD206, respectively) in our study, a recent report indicated that the model distinguishing between M1 and M2 macrophages incompletely accounts for anti-tumor and pro-tumor functions (41). Because there are no definitive markers for MDSCs in humans (42), tumor-infiltrating CD11b-positive cells in human prostatectomy specimens represent not only MDSCs but also several subsets of macrophages or neutrophils, or other immune cells. We did not detect which cells were stained with CD11b. Our study neither showed more intratumoral MDSCs was correlated with tumor growth in human patients nor showed that human patients who received celecoxib demonstrated fewer intratumoral MDSCs. The mechanism of HFD-induced tumor progression of prostate cancer may not only be inflammation. Changes in cholesterol in the cell membrane or a deficiency of the protein tyrosine phosphatase 1B with administration of an HFD might result in tumor progression of PTEN-deficient prostate cancer (43, 44).

T. Hayashi reports receiving commercial research grants from JSPS Kakenhi. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Fujita, A. Kawashima, A. Nagahara, M. Uemura

Development of methodology: T. Hayashi, K. Fujita, A. Nagahara, K. Tsujikawa, G.J. Netto

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Hayashi, K. Fujita, K. Jingushi, M.D.C.R. Pena, J.B. Gordetsky, E. Morii, K. Tsujikawa

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Hayashi, K. Fujita, S. Nojima, Y. Hayashi, K. Nakano, K. Jingushi, T. Kato, A. Kawashima, A. Nagahara, J.B. Gordetsky, E. Morii, K. Tsujikawa, G.J. Netto

Writing, review, and/or revision of the manuscript: T. Hayashi, K. Fujita, Y. Ishizuya, Y. Yamamoto, T. Kato, M. Uemura, J.B. Gordetsky, G.J. Netto

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Fujita, S. Nojima, Y. Hayashi, K. Nakano, C. Wang, T. Kinouchi, K. Matsuzaki, A. Nagahara, T. Ujike, K. Tsujikawa

Study supervision: K. Fujita, A. Kawashima, G.J. Netto, N. Nonomura

We thank Tak W. Mak (Medical Biophysics and Immunology, University of Toronto; Campbell Family Institute for Breast Cancer Research, Princess Margaret Cancer Centre), Akira Suzuki (Department of Molecular Genetics, Division of Cancer Genetics, Kobe University Graduate School of Medicine), and Toru Nakano (Department of Pathology, Medical School and Graduate School of Frontier Biosciences, Osaka University) for providing the <Pten(fl/fl)> mice. This work was supported by JSPS KAKENHI grant number JP16K20137.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.