Obesity-associated NLRC4 inflammasome activation drives breast cancer progression

Abstract

Obesity is associated with an increased risk of developing breast cancer and is also associated with worse clinical prognosis. The mechanistic link between obesity and breast cancer progression remains unclear, and there has been no development of specific treatments to improve the outcome of obese cancer patients. Here we show that obesity-associated NLRC4 inflammasome activation/ interleukin (IL)-1 signalling promotes breast cancer progression. The tumour microenvironment in the context of obesity induces an increase in tumour-infiltrating myeloid cells with an activated NLRC4 inflammasome that in turn activates IL-1β, which drives disease progression through adipocyte-mediated vascular endothelial growth factor A (VEGFA) expression and angiogenesis. Further studies show that treatment of mice with metformin inhibits obesity-associated tumour progression associated with a marked decrease in angiogenesis. This report provides a causal mechanism by which obesity promotes breast cancer progression and lays out a foundation to block NLRC4 inflammasome activation or IL-1β signalling transduction that may be useful for the treatment of obese cancer patients.

Figure 1. The IL-1β/IL-1R1 axis promotes tumour growth in diet-induced obese mice.

The indicated mice were given either a ND or HFD for 10 weeks, and then the indicated cells were implanted into the #4 mammary gland. See Supplementary Fig. 2b. (a–c) Mice were treated with anti-IL-1R1 antibody or control IgG once tumours were palpable (a,c) or with recombinant IL-1 receptor antagonist (rIL1RA) (b) starting the day of Py8119 cell transplant. Data represent the average tumour volume±s.e.m (a–b: n=5 HFD+IgG, n=5 HFD+anti-IL-1R1, n=5 HFD+anti-IL1RA n=14 ND, n=24 HFD; C: n=5 all groups). (d,e) Data represent the average body weight (d) and Py8119 tumour volume (e) of the indicated mice±s.e.m. (n=5 Il1α^−/− ND, n=4 Il1α^−/− HFD, n=14 WT ND, n=24 WT HFD). (f,g) Data represent the average body weight (f) and tumour volume (g) of the indicated mice±s.e.m. ((e) n=5 Il1β^−/− ND, n=5 Il1β^−/− HFD, n=14 WT ND, n=24 WT HFD; f: n=1 Il1β^−/− ND, n=2 Il1β^−/− HFD). Only mice that had measurable tumours after 5 weeks are represented in g. Two-way ANOVA was used to determine significance. All tumour studies were repeated in a different cohort of animals.

Figure 2. NLRC4 inflammasome promotes tumour growth in diet-induced obese mice.

(a–d) The indicated mice were treated as in Fig. 1. Data represent body weight (a) in Casp1/11^−/− mice versus WT mice) or the average Py8119 tumour volume±s.e.m. (b–d; n=5 Casp1/11^−/− ND, n=5 Casp1/11^−/− HFD, n=5 Nlrp3^−/− ND, n=4 Nlrp3^−/− HFD, n=5 Nlrc4^−/− ND, n=9 Nlrc4^−/−- HFD n=14 WT ND, n=24 WT HFD). Two-way ANOVA was used to determine significance. (e) Py8119 tumour volumes 4 weeks after implantation from the indicated mice from a to d. Indicated P values are from a Student's t-test comparing individual group means. One-way ANOVA was performed across all groups: P<0.0001. (f) Average E0771 tumour volume±s.e.m. from the indicated mice (n=5 WT ND, n=5 WT HFD, n=5 Nlrc4^−/− ND, n=9 Nlrc4^−/− HFD). Two-way ANOVA was used to determine significance. All tumour studies were repeated in a different cohort of animals.

Figure 3. Obesity-induced NLRC4 inflammasome in tumour-infiltrating myeloid cells.

(a) Data represent the number of tumour-infiltrating total (mφ), CD11c⁻ or CD11c⁺ macrophages as a percentage of total cells counted±s.d. (n=5 ND; n=9 HFD). (b) mRNA expression (relative to peptidylprolyl isomerase A gene; Ppia) of the indicated genes in tumours from indicated mice (n=3)±s.d. (c) Western blot analysis for CASP1 (left panels) and CASP11 (right panels) in tumour-infiltrating CD11b⁺ and CD11b⁻ cell populations. Cells were combined from four to five tumours from each group. (d) mRNA fold-change relative to the ND group, using Ppia as reference gene. Mean±s.d. in tumours from the indicated mice (n=3). (e,f) Nlrc4 mRNA expression in tumour-infiltrating CD11b⁺ and CD11b⁻ cell populations from the indicated mice in the indicated tumour model, using Actin beta (Actb) as the reference gene. Cells were combined from four to five tumours from each group, and data are shown in triplicates. (g) Western blot for NLRC4-flag in tumour-infiltrating CD11b⁻ and CD11b⁺ cell populations from DIO mice. (h) Data represent the average number of tumour-infiltrating CD45⁺ cells with CASP1 activation as a percentage of CD45⁺ cells±s.d. (n=5 WT ND, n=5 WT HFD, n=5 Nlrc4^−/− ND, n=4 Nlrc4^−/− HFD). For all panels, Student's t-test was used to determine significance. (i) NLRC4 activation from macrophages drives ODBP. Bone marrow macrophages from WT or Nlrc4^−/− female mice were co-injected with Py8119 cells orthotopically into DIO Nlrc4^−/− female mice. Tumour growth was monitored weekly (n=7, means±s.e.m.). Tumour study was repeated in a different cohort of animals. All other studies represent results from two to three repeats.

Figure 4. NLRC4 inflammasome promotes angiogenesis in diet-induced obese mice.

(a) Left panel: proteome profiler array for angiogenesis proteins using tumour lysates from ND and HFD mice (red boxes indicate the internal reference spots). Right panel: analysis of proteome profiler. Data represent the fold change compared with ND of the average mean pixel density±s.d. of select proteins. (b) Representative immunohistochemistry (IHC) staining for CD31 (brown colour, some indicated with white arrows) from the indicated Py8119 tumours, haematoxylin (blue) being used for background nuclear staining. An isotype-negative control staining is included (negative control). Scale bar, 200 μm. (c) Quantification of IHC staining in b. Data represent the average area of CD31-positive staining over total area±s.d. At least three fields per section and three tumours per group were used in the analysis. (d) Data represent the average number of CD31⁺ cells as a percentage of total cells counted by flow cytometry±s.d. in E0771 from tumours of the indicated mice (n=5 all groups). (e,f) Average Vegfa mRNA expression relative to Ppia in Py8119 tumours (e) and E0771 tumours (f) from the indicated mice±s.d. (n=3 for each group). All studies represent results from two to three repeats. For all panels, the group means were compared by Student's t-test to determine significance.

Figure 5. IL-1β-induced Vegfa expression in adipocytes.

(a) Vegfa expression is increased in non-myeloid cells under obesity. CD11b⁺ and CD11b⁻ cells were purified by magnetic beads in tumours from ND or DIO mice. Data represent the mean fold change compared with ND±s.d. Expression is relative to Actb (n=3 for each group). (b,c) Indicated cells were left untreated (cont), treated with 100 ng ml⁻¹ rIL-1β or CM from non-treated or 100 ng ml⁻¹ rIL-1β-treated primary BMDM. Data represent the average mRNA expression of Vegfa relative to Ppia as fold change compared with cont±s.d. (d) Adipocytes were treated as indicated with 100 ng ml⁻¹ rIL-1β, 5 μM NFκB inhibitor (BMS345541) and 40 μM JNK inhibitor. Data represent the average mRNA expression of Vegfa relative to Ppia±s.d. (n=3 for all groups). (e) Primary mammary adipocytes were treated with 100 ng ml⁻¹ rIL-1β for the indicated time, and the indicated proteins were separated by SDS–PAGE followed by immunoblotting with the indicated antibodies. All studies represent results from two to three repeats. For all panels, the group means were compared by Student's t-test to determine significance.

Figure 6. Metformin inhibits tumour growth and angiogenesis in diet-induced obese mice.

Mice were treated as in Fig. 1a; except after 9 weeks one group of ND and one group of HFD mice were fed with 0.5% metformin water. (a,b) Average Py8119 tumour volume (a) or body weight (b) ±s.e.m. (n=5 ND+Met, n=10 HFD+Met, n=14 ND, n=24 HFD). Two-way ANOVA was used to determine significance. (c) Representative IHC staining for CD31 from the indicated mice. Scale bar, 200 μm. (d) Quantification of IHC staining in c. Data represent the average area of CD31⁺ staining over total area±s.d. At least three fields per section and three tumours per group were used in the analysis. (e) Average mRNA expression of Vegfa±s.d. in tumours from the indicated mice (n=3). (f) Adipocytes were treated with 100 ng ml⁻¹ rIL-1β and 500 μM metformin as indicated. Data represent the average Vegfa mRNA level relative to Ppia as a fold change compared with non-treated cells±s.d. (n=3). Tumour study was repeated in a different cohort of animals. All other studies represent results from two to three repeats. Group means were compared by Student's t-test to determine significance.

Figure 7. NLRC4 and CASP1 are associated with macrophage markers and poor outcome in human breast cancer.

(a,b) NLRC4 mRNA is elevated in normal breast tissues from obese individuals (a) and is positively correlated with BMI (b). Number of cases indicated. Data from GSE33256 GEO data set. Statistical significance was determined by Welch's t-test (a) and linear regression and F test (b). (c) The mean NLRC4 and NLRP3 expression ±95% confidence interval (CI) in PAM50 subtypes of breast and normal breast tissue from TCGA-invasive breast cancer data set. Number of cases indicated. (d) CASP1 mRNA is elevated in obese breast cancer patients with luminal B and basal-like breast cancers. Data are presented as the mean±95% CI. Data are from GSE20194 GEO data set. Statistical significance was determined by Welch's t-test. (e) Correlation of NLRC4 expression (left two panels) or NLRP3 expression (right two panels) with overall survival in months within all human breast cancer patients (all) or within the ER+ breast cancers (ER+), analysed using KM-Plot meta-data set for invasive breast cancer. Statistical significance was determined by log-rank test comparing low versus high groups with number of cases indicated. (f) NLRC4 expression is positively correlated with macrophage markers including CD163 and common markers for myeloid cells such as CD68 and CD33, and is also correlated with IL1B expression in human breast cancer from TCGA (n=960 analysed from cBioPortal). Spearman and Pearson r values are indicated.