Glucose oxidation modulates anoikis and tumor metastasis

Abstract

Cancer cells exhibit altered glucose metabolism characterized by a preference for aerobic glycolysis or the Warburg effect, and the cells resist matrix detachment-induced apoptosis, which is called anoikis, a barrier to metastasis. It remains largely unclear whether tumor metabolism influences anoikis and metastasis. Here we show that when detached from the matrix, untransformed mammary epithelial cells undergo metabolic reprogramming by markedly upregulating pyruvate dehydrogenase (PDH) kinase 4 (PDK4) through estrogen-related receptor gamma (ERRγ), thereby inhibiting PDH and attenuating the flux of glycolytic carbon into mitochondrial oxidation. To decipher the significance of this metabolic response, we found that depletion of PDK4 or activation of PDH increased mitochondrial respiration and oxidative stress in suspended cells, resulting in heightened anoikis. Conversely, overexpression of PDKs prolonged survival of cells in suspension. Therefore, decreased glucose oxidation following cell detachment confers anoikis resistance. Unlike untransformed cells, most cancer cells demonstrate reduced glucose oxidation even under attached conditions, and thus they inherently possess a survival advantage when suspended. Normalization of glucose metabolism by stimulating PDH in cancer cells restores their susceptibility to anoikis and impairs their metastatic potential. These results suggest that the Warburg effect, more specifically, diminished glucose oxidation, promotes anoikis resistance and metastasis and that PDKs are potential targets for antimetastasis therapy.

Figures

Fig 1

Matrix detachment upregulates PDK4 in mammary epithelial cells. (A) Quantitative RT-PCR measurement of individual PDKs in HMEC under attached (Att) and suspended (Susp) conditions. (B) Time course of PDK4 induction in HMEC under suspension. (C) Quantitative RT-PCR analysis of expression of PDKs in MCF10A cells. (D) Induction of PDK4 in suspended MCF10A cells at the indicated time points. (E) Quantitative RT-PCR analysis of absolute levels of PDKs in attached and suspended MCF10A cells. cDNA plasmids of each PDK were used as references. (F) Northern blotting of PDK4 in attached and suspended MCF10A cells. (G) Immunoblotting of PDKs, LDHA, and tubulin in attached and suspended MCF10A cells.

Fig 2

Depletion of PDK4 in MCF10A cells enhances anoikis. MCF10A cells were transduced with a retroviral vector (EV) or two independent shRNAs targeting PDK4 (shPDK4) and cultured under attached (Att) and suspended (Susp) conditions for 24 h (unless otherwise indicated). Error bars represent standard deviations. (A) Northern blotting of PDK4 depletion in MCF10A cells under attached (A) and suspended (S) conditions. 18S rRNA was used as a loading control. (B) Quantitative determination of PDK4 knockdown efficiency in MCF10A cells under attached and suspended conditions. The RNA level of PDK4 in suspended MCF10A cells was set as 100. (C) Trypan blue exclusion assay of cell viability in control and PDK4-depleted MCF10A cells under attached and suspended (for 48 h) conditions. (D) Measurement of caspase 3/7 activity in control and PDK4-depleted MCF10A cells. (E) Fluorescence-activated cell sorting analysis of annexin V/7-AAD staining in control and PDK4-depleted MCF10A cells. The x axes show annexin V staining, and y axes show 7-AAD staining. (F) Statistics of total apoptotic cells based on the annexin V/7-AAD analysis.

Fig 3

Overexpression of PDK1 or PDK4 blunts anoikis. MCF10A cells were transduced with control (EV) or PDK1- or PDK4-expressing lentiviruses and subjected to anoikis analysis (24 h in suspension). (A and B) Quantitative RT-PCR (A) and Western blotting assay (B) results to verify overexpression of PDK1 in attached MCF10A cells. (C) Quantitative RT-PCR validation of PDK4 overexpression in attached MCF10A cells. (D) PDH activities in PDK1- and PDK4-overexpressing MCF10A cells under attached conditions. PDH activity in control cells was set as 100. (E) Trypan blue cell viability assay of control or PDK1- or PDK4-overexpressing cells under attached and suspended conditions. (F) Caspase 3/7 activity assay in control or PDK1- or PDK4-overexpressing cells under attached and suspended conditions.

Fig 4

Activation of PDH enhances anoikis in MCF10A cells. MCF10A cells were transduced with a lentiviral vector (EV) or lentivirus expressing a constitutively active form of PDHE1α with a carboxyl-terminal Flag tag (PDHE1a). Error bars represent standard deviations (n = 3). (A) Measurement of PDH activity in MCF10A cells under attached (Att) and suspended (Susp) conditions. (B) Immunoblotting of exogenous PDHE1α expression with anti-Flag antibodies. (C) Measurement of PDH activity in control and PDHE1α-expressing cells cultured under attached and suspended conditions. (D) Trypan blue exclusion assay for cell viability in control and PDHE1α-expressing MCF10A cells under attached and suspended conditions. (E) Caspase 3/7 activity assay in control and PDHE1α-expressing MCF10A cells under attached and suspended conditions.

Fig 5

Knockdown of PDK4 increases mitochondrial respiration and ROS levels in suspended MCF10A cells. Empty vector (EV) or PDK4 shRNA (shPDK4)-transduced MCF10A cells were cultured under attached (Att) and suspended (Susp) conditions. Error bars represent standard deviations (n = 3). (A) Measurement of PDH activity. (B) Oxygen consumption rate (OCR). (C) ATP levels. (D) Overall ROS levels in control and PDK4-depleted cells under attached and suspension (for 24 h) conditions. (E) Mitochondrial ROS detected by using MitoSOX Red in control and PDK4-depleted cells under suspension (for 24 h) conditions. (F and G) Cell viability (F) and caspase 3/7 activity (G) in cells untreated (-) or treated with 2.5 mM GSH or 100 μM LA under attached and suspension (for 48 h) conditions.

Fig 6

ERRγ activates PDK4 in response to matrix detachment. (A) Quantitative RT-PCR analysis of ERRγ induction in MCF10A cells by matrix detachment. (B) ChIP assay of Flag-ERRγ binding to the PDK4 promoter. Control and Flag-ERRγ stable cells were subjected to ChIP with IgG and anti-Flag antibodies. Bound DNA was analyzed by quantitative PCR. Vascular endothelial growth factor (VEGF) exon 3 served as a negative control. (C) Activation of a PDK4-Luc reporter by exogenous ERRγ in transiently transfected MCF10A cells (attached). (D) Depletion of ERRγ in MCF10A cells diminished induction of endogenous PDK4 following cell detachment. MCF10A cells were transduced with a retroviral empty vector (EV) or shRNA targeting ERRγ (shERRg). Cells were plated in suspension for 24 h, and total RNA was extracted and analyzed by quantitative RT-PCR. (E) Trypan blue exclusion assay for cell viability in control and ERRγ-depleted MCF10A cells under attached and suspended conditions. (F) Caspase 3/7 activity assay in control and ERRγ-depleted MCF10A cells under attached and suspended conditions. (G) Schematic of metabolic reprogramming by activation of the ERRγ-PDK4 pathway in untransformed mammary epithelial cells following matrix detachment.

Fig 7

PDK4 is highly expressed in SKMEL-5 cells and contributes to anoikis resistance. (A) Comparison of PDK4 levels in attached MCF10A and SKMEL-5 cells by quantitative RT-PCR. (B) Depletion of PDK4 in attached SKMEL-5 cells by shRNA. (C) Trypan blue analysis of control and PDK4-depleted SKMEL-5 cells under attached and suspended conditions.

Fig 8

Activation of PDH sensitizes MDA-MB-231 cells to anoikis. MDA-MB-231 cells were transduced with a control virus (EV) or virus expressing a constitutively active form of PDHE1α (as in Fig. 4) and subsequently subjected to matrix detachment. (A) Verification of exogenous PDHE1α expression by immunoblotting with anti-Flag antibodies. Tubulin was used as a loading control. (B) Measurement of PDH activity in control and PDHE1α-expressing cells under attached conditions. (C) Trypan blue exclusion assay for cell viability assay in control and PDH-activated MDA-MB-231 cells under attached and suspended conditions. (D) Caspase 3/7 activity assay in control and PDH-activated MDA-MB-231 cells under attached and suspended conditions. Error bars represent standard deviations (n = 3).

Fig 9

Depletion of PDK1 in MDA-MB-231 cancer cells increases glucose oxidation. (A) Quantitative RT-PCR measurement of absolute levels of each PDK in MDA-MB-231 cells under attached (Att) and suspended (Susp) conditions. (B) Relative induction of PDKs in MDA-MB-231 cells grown under attached and suspended conditions. (C) Immunoblotting analysis of PDK1 depletion in MDA-MB-231 cells. MDA-MB-231 cells were infected with retrovirus of either empty vector (EV) or two independent shRNAs targeting PDK1 (shPDK1-1 and shPDK1-2). Tubulin was used as a loading control. (D) Measurement of PDH activity in control and PDK1-depleted cells under attached (A) and suspended (S) conditions. (E) Intracellular lactate levels in attached control and PDK1-depleted MDA-MB-231 cells. (F) Measurement of oxygen consumption rate (OCR) in attached control and PDK1-depleted MDA-MB-231 cells. (G) Mitochondrial ROS in control and PDK1-depleted cells under suspension (for 24 h) conditions. Representative histograms for MitoSOX Red flow cytometry are shown. The x axis shows the fluorescent signal intensity, and the y axis shows cell numbers. (H) Quantification of mean fluorescent intensity of MitoSOX Red in control and PDK1-depleted suspension cells, measured by flow cytometry (the results shown in panel G).

Fig 10

Increased glucose oxidation in MDA-MB-231 cancer cells restores anoikis and decreases metastasis. (A) Trypan blue exclusion assay in control and PDK1-depleted MDA-MB-231 cells under attached (Att) and suspended (Susp) conditions (for 48 h). (B) Caspase 3/7 activity assay in control and PDK1-depleted cells under attached and suspended conditions. (C) Lung tumor nodules resulting from tail vein injection of equal numbers of control and PDK1-depleted cells followed by H&E histological analysis. (D) Statistical analysis of numbers of lung tumor nodules.