Hypoxia-driven metabolic reprogramming of adipocytes fuels cancer cell proliferation

Abstract

Objective: Obesity increases the risk of certain cancers, especially tumours that reside close to adipose tissue (breast and ovarian metastasis in the omentum). The obesogenic and tumour micro-environment share a common pathogenic feature, oxygen deprivation (hypoxia). Here we test how hypoxia changes the metabolome of adipocytes to assist cancer cell growth.

Methods: Human and mouse breast and ovarian cancer cell lines were co-cultured with human and mouse adipocytes respectively under normoxia or hypoxia. Proliferation and lipid uptake in cancer cells were measured by commercial assays. Metabolite changes under normoxia or hypoxia were measured in the media of human adipocytes by targeted LC/MS.

Results: Hypoxic cancer-conditioned media increased lipolysis in both human and mouse adipocytes. This led to increased transfer of lipids to cancer cells and consequent increased proliferation under hypoxia. These effects were dependent on HIF1α expression in adipocytes, as mouse adipocytes lacking HIF1α showed blunted responses under hypoxic conditions. Targeted metabolomics of the human Simpson-Golabi-Behmel syndrome (SGBS) adipocytes media revealed that culture with hypoxic-conditioned media from non-malignant mammary epithelial cells (MCF10A) can alter the adipocyte metabolome and drive proliferation of the non-malignant cells.

Conclusion: Here, we show that hypoxia in the adipose-tumour microenvironment is the driving force of the lipid uptake in both mammary and ovarian cancer cells. Hypoxia can modify the adipocyte metabolome towards accelerated lipolysis, glucose deprivation and reduced ketosis. These metabolic shifts in adipocytes could assist both mammary epithelial and cancer cells to bypass the inhibitory effects of hypoxia on proliferation and thrive.

Keywords: adipocytes; cancer cells; hypoxia; lipids; metabolites; obesity.

Figure 1

Hypoxic conditioned cancer cell media increases adipocyte lipolysis. (A) Glycerol release from SGBS adipocytes cultured in conditioned media (CM) from breast (MDA-MB-231, red) and ovarian (SKOV3, pink) cancer cells respectively under normoxia (21% O2) or hypoxia (0.5% O2), n=3/group (mean from 3 independent experiments). (B) Glycerol release from primary C57BL/6J mouse subcutaneous (SAT) and visceral (VAT) adipocytes cultured with CM from MET-1 (red) or ID8 (pink) cancer cells respectively under normoxia (21% O2) or hypoxia (0.5% O2), n=3/group (biological replicates). Non esterified fatty acids (NEFA) (C, D) and glycerol (E, F) release from human obese subcutaneous (SAT) and visceral (VAT) adipocytes cultured in conditioned media from breast (MDA-MB-231, red; BC) and ovarian (SKOV3, pink; OvCa) cancer cells respectively under normoxia (21% O2) or hypoxia (0.5% O2), n=4/group (biological replicates). Data are mean+/- SEM. Significance by ANOVA *p<0.05, **p<0.01, ***p<0.001. Note: connecting lines show significance within groups in the same O2 tension. If only “*” is used, this indicates significance between different O2 tensions. Abbreviations ad=adipocytes, BC=breast cancer cells, OvCa=ovarian cancer cells. ad+BC (or OvCa) 21%= both cells in 21% O2; ad+BC (or OvCa) 0.5%= both cells in 0.5% O2.

Figure 2

HIF1αKO adipocytes show blunt cancer-induced lipolytic response in hypoxia and do not assist cancer cell proliferation. (A, B) Glycerol release from adipose-specific HIF1αKO (KOad, blue bars) or control littermate (WTad, green bars) subcutaneous or visceral fat cells cultured in conditioned media from breast MET-1 or ovarian ID8 cancer cells in normoxia (Nx; 21% O₂) or hypoxia (Hx;0.5% O₂, 24h). In vitro proliferation of MET-1 (C) and ID8 (D) cancer cells alone (black) or cultured in adipocyte CM from KO (blue line) or WT (green line) adipocytes for 3 days in hypoxia (0.5% O₂). Values are mean+/-SEM from 3 independent experiments. * p<0.05, ** p<0.01, ***p<0.001. (C, D), the ACM was prepared in 0.5% O_2. In the proliferation assay, for MET-1,ACM was prepared from WT or KO SAT adipocytes. For ID-8, the ACM was from VAT adipocytes. p<0.05, ** p<0.01, ***p<0.001. CCM= cancer cell conditioned media; ACM=adipocyte conditioned media.

Figure 3

HIF-1α protein levels in adipocytes cultured with hypoxic or normoxic breast cancer MET-1 conditioned media. (A) Conditioned media was prepared from MET-1 cells cultured in normoxia (Nx, 21% O₂) or hypoxia (Hx, 0.5% O₂), and added to mouse subcutaneous adipocytes cultured in Hx or Nx. Western blot for HIF-1α protein in MET-1 cells (B) and adipocytes (C) in Nx and Hx. Abbreviations: adipo Nx or Hx=adipocytes cultured in Nx or Hx, CCM Hx = cancer conditioned media prepared in 0.5% O₂; Nx CCM= cancer conditioned media prepared in 21% O₂.

Figure 4

HIF1αKO adipocytes limit lipid transfer to cancer cells in hypoxia. (A) mouse primary subcutaneous adipocytes from control littermates (WT) or adipose-specific HIF1αKO (KO) mice were pre-labelled with a fluorescent lipid (fluoro-lipid) analogue. Fluoro-lipid adipocytes were then co-cultured with MET-1 cancer cells for 24h in hypoxia (0.5%) or normoxia (21% O2). Fluro-lipid transfer to MET-1 cells was evaluated by confocal imaging. Note the loss of fluoro-lipid in adipocytes cultured in hypoxia. (B) Confocal microscopy images (magnification x40) of fluoro-lipid (green) accumulation in MET-1 (nuclear co-staining, blue) co-cultured with fluoro-labelled HIF1αKO or WT adipocytes. Insert image at x100 magnification indicating that lipids are taken up by MET-1. Fluorescent quantification of % of lipid covered area corrected for the % of area of total nuclei. Scale bars=50 micrometers. Values presented as mean+/- SEM. n=4-5/group. Significance by 2-way ANOVA, * p<0.05, **p<0.01.

Figure 5

Human adipocyte metabolome is modulated in the presence of breast mammary cells under hypoxia. SGBS adipocytes were cultured with MCF-10A cells CM in normoxia (NX, 21% O2, white bars) or HX (0.5% O2, black bars) for 24h.Targeted LC/MS analysis in SGBS media (n=3/group). Representation of changes in energy metabolites, TCA cycle intermediates, pentose phosphate pathway (PPP) intermediates. Ordinate axes represent metabolite peak area. Values presented as mean+/- SEM. Significance by Students t-test *p<0.05, **p<0.01, ***p<0.001. Note: in Nx, adipocytes and CCM was prepared in Nx. In Hx, adipocytes and CCM in Hx.

Figure 6

Summary Schematic. (1) Fat expansion in obesity shares a common pathogenic feature with the cancer micro-environment, reduction of local oxygen levels. (2) Hypoxia limits the proliferation of cancer cells especially when nutrients are not available from neighbour fat cells. (3) When cancer cells reside close to fat cells (i.e. breast or ovarian omental metastasis), they induce lipid release from fat cells, hijack and use these lipids to by-pass the hypoxic-inhibitory effect on proliferation in a HIF-1α dependent manner. (4) Non-malignant mammary cells when reside in hypoxic environments can also facilitate metabolic shifts in fat cells (increased lactate, fumarate, lipids) and increase their proliferation status. This suggests that an obesogenic hypoxic microenvironment could be potentially a driver of cancer growth.