Erythropoietin receptor regulates tumor mitochondrial biogenesis through iNOS and pAKT

Abstract

Erythropoietin receptor (EPOR) is widely expressed in healthy and malignant tissues. In certain malignancies, EPOR stimulates tumor growth. In healthy tissues, EPOR controls processes other than erythropoiesis, including mitochondrial metabolism. We hypothesized that EPOR also controls the mitochondrial metabolism in cancer cells. To test this hypothesis, we generated EPOR-knockdown cancer cells to grow tumor xenografts in mice and analyzed tumor cellular respiration via high-resolution respirometry. Furthermore, we analyzed cellular respiratory control, mitochondrial content, and regulators of mitochondrial biogenesis in vivo and in vitro in different cancer cell lines. Our results show that EPOR controls tumor growth and mitochondrial biogenesis in tumors by controlling the levels of both, pAKT and inducible NO synthase (iNOS). Furthermore, we observed that the expression of EPOR is associated with the expression of the mitochondrial marker VDAC1 in tissue arrays of lung cancer patients, suggesting that EPOR indeed helps to regulate mitochondrial biogenesis in tumors of cancer patients. Thus, our data imply that EPOR not only stimulates tumor growth but also regulates tumor metabolism and is a target for direct intervention against progression.

Keywords: OXPHOS; VDAC1; erythropoietin receptor; mitochondrial biogenesis; nitric oxide (NO); respirometry; tumor metabolism.

Figure 1

Knockdown of EPOR impairs tumor growth of A549 lung cancer xenografts in Foxn1nu mice. A549 control cells (shSCR1, purple and shSCR2, red) or A549 EPOR knockdown cells (shEPOR1, green and shEPOR2, blue) were subcutaneously injected (3 x 10⁶ cells in 100 µl PBS/Matrigel) into Foxn1nu mice. Panel (A) shows a representative western blot image of EPOR (63 kDA) and β-actin (44 kDA) protein expression in shSCR and shEPOR A549 tumors (left panel). Western blotting images were analyzed by MCID Analysis 7.0 and shown is relative EPOR protein expression of shSCR (purple shSCR1 tumors, red shSCR2 tumors) and shEPOR (green shEPOR1 tumors, blue shEPOR2) tumors normalized to β-actin (n=6) (right panel). Panel (B) shows the tumor growth curves (left panel), tumor size 28 days after tumor cell implantation (middle panel), and tumor size 56 days after tumor cell implantation (right panel) for shSCR A549 and shEPOR tumors (n=8). Please note: The middle panel has a logarithmic scale, and in the right panel, no data for shSCR1 tumors are shown because the experiment was already terminated 28 days after tumor cell implantation. Data are presented either as scattered blots with mean, as mean, or as a box plot with min to max whiskers. A Student’s t-test (black symbols), a Kruskal Wallis test with Dunn’s multiple comparison test (grey letters), or a one-way ANOVA with Bonferroni post hoc test (black letters) was performed (**p<0.01); letters a and b indicate groups that statistically (p<0.05) differ from each other.

Figure 2

Knockdown of EPOR reduces cellular respiration of human A549 lung cancer xenografts in Foxn1nu mice. Biopsies of human A549 tumors that either express EPOR (shSCR1/2) or not (shEPOR1/2) were isolated from Foxn1nu mice and mass-specific respiration was immediately measured by high-resolution respirometry. Panel (A) shows the mass-specific respiration per unit weight of freshly isolated tumor biopsies of shSCR and shEPOR A549 tumors (n=6-7). L_N, respiration in the absence of adenylates; P_ETF, capacity for fatty acid β-oxidation; P_C1, submaximal state 3 respiration through complex I; P, maximal state 3 respiration - oxidative phosphorylation capacity; ETS, electron transport system capacity; P_C2, submaximal state 3 respiration through complex II. Relative mRNA expression of human and murine genes was analyzed by qPCR in A549 tumors: Shown are (B) mRNA levels of human superoxide dismutase 1 (SOD1), human superoxide dismutase 2 (SOD2), human catalase (CAT), human glutathione peroxidase 3 (GPX3) and human glutathione peroxidase 4 (GPX4) normalized to human b-Actin (ACTB) mRNA levels as well as (C) mRNA levels of murine superoxide dismutase 1 (Sod1), murine superoxide dismutase 2 (Sod2), murine catalase (Cat), murine glutathione peroxidase 3 (Gpx3) and (murine glutathione peroxidase 4 (Gpx4) normalized to murine b-Actin (Actb) mRNA levels of shSCR and shEPOR A549 tumors (n=16). Data are shown as means and standard deviations (A) or as scattered blots with mean and individual data distribution (B, C) of each control or EPOR-knockdown clone (control: shSCR1 purple, shSCR2 red; EPO- knockdown: shEPOR1 green and shEPOR2 blue tumor samples). The graphs in panel (B) are on a logarithmic scale. Data were analyzed by a Student’s t-test (black stars) or by a Mann-Whitney test (grey stars). ***p<0.001; **p<0.01.

Figure 3

Knockdown of EPOR reduces mitochondrial content of human A549 lung cancer xenografts in Foxn1nu mice. Biopsies of human A549 tumors expressing either EPOR (shSCR1/2) or not (shEPOR1/2) were isolated from Foxn1nu mice, and mitochondria-specific respiration was measured by high-resolution respirometry. Panel (A) shows mitochondria-specific respiration normalized to cytochrome c oxidase (COX) activity in freshly isolated shSCR and shEPOR A549 tumor biopsies (n=6-7). L_N, respiration in the absence of adenylates; P_ETF, capacity for fatty acid β-oxidation; P_C1, submaximal state 3 respiration through complex I; P, maximal state 3 respiration - oxidative phosphorylation capacity; ETS, electron transport system capacity; P_C2, submaximal state 3 respiration through complex II. Panel (B) shows a representative western blot image of specific subunits from complexes of the oxidative phosphorylation (OXPHOS) from control tumors (shSCR) and EPOR-knockdown tumors (shEPOR) by using an anti-total OXPHOS antibody cocktail: Complex V: 55 kDa (ATP5A, ATP synthase mitochondrial F1 complex alpha 1); Complex III: 48 kDa (UQCRC2, cytochrome b-c1 complex subunit 2); Complex II: 30 kDa (SDHB, succinate dehydrogenase [ubiquinone] iron-sulfur subunit); Complex I: 20 kDa (NDUFB8, NADH dehydrogenase [ubiquinone] 1 beta subcomplex subunit 8). Complex IV: 17 kDa was visualized using an anti-cytochrome c oxidase antibody. VDAC1: 31 kDa (voltage-dependent anion-selective channel 1) was used as a mitochondrial marker independent of OXPHOS complexes and β-actin 44 kDa was used as a loading control. The band intensity of proteins after western blotting was quantified using MCID Analysis 7.0 and normalized either to β-actin to estimate expression levels per cell, or to VDAC1 to normalize the protein levels to mitochondrial content. Relative protein expression levels of VDAC1, complex I, complex II, complex (III), COX-IV (complex IV), and complex V are shown for control (shSCR) and EPOR-knockdown (shEPOR) tumors (n=4-6). (C) Mitochondrial content was determined by the ratio of human (left panel) or murine (middle and right panel) MT-ND1 (mitochondrially encoded NADH dehydrogenase 1) mitochondrial DNA to human or murine β2M (β-2microglobulin) genomic DNA, respectively, which were quantified by qPCR from DNA extracts of A549 control (shSCR) and EPOR-knockdown (shEPOR) tumors or the liver (right panel) (n=12). (D) Likewise, mitochondrial content of in vitro cultured shSCR and shEPOR clones was determined by the ratio of human MT-ND1 to human β2M genomic DNA. Data are presented as (A) mean and standard deviation or as scattered blot with mean and individual data distribution for each clone (shSCR1 purple, shSCR2 red, shEPOR1 green, and shEPOR2 blue tumor samples). Data were analyzed by a Student’s t-test (black p-values; stars) or by a Mann-Whitney test (grey stars). ***p<0.001; **p<0.01; *p<0.05.

Figure 4

Knockdown of EPOR impairs iNOS expression and AKT phosphorylation in A549 lung cancer xenografts in Foxn1nu mice. Biopsies of human A549 tumors expressing EPOR (shSCR1/2) or not (shEPOR1/2) were isolated from Foxn1nu mice, and levels of key mitochondrial biogenesis, as well as nitric oxide synthesis genes and proteins, were quantified by qPCR and western blotting. (A) Shown are the human mRNA levels of mitochondrial biogenesis genes peroxisome proliferative activated receptor, gamma, coactivator 1α (PGC-1a), nuclear respiratory factor 1 (NRF1) and transcription factor A, mitochondrial (TFAM) quantified by qPCR and normalized to β-actin (ACTB) mRNA expression levels (n=12). Panel (B) shows the mRNA levels of nitric oxide synthase genes nNOS, iNOS, and eNOS from control (shSCR) and EPOR-knockdown (shEPOR) tumors quantified by qPCR and normalized to β-actin (ACTB) mRNA (n=6-12). Notably, iNOS mRNA was not detectable in five samples (two from clone shEPOR1 and three from shEPOR2), and the scale is logarithmic. Panel (C) shows representative western blot images of iNOS (130 kDa) from protein extracts of control tumors (shSCR) and EPOR-knockdown tumors (shEPOR) (n=3). β-actin (44 kDa) was used as a loading control. The band intensities of proteins after western blotting images were quantified using MCID Analysis 7.0 and normalized to β-actin. The relative protein expression levels of iNOS are shown for control (shSCR) and EPOR- knockdown (shEPOR) tumors (n=6). Furthermore, the plasma nitrate values of mice with shSCR and shEPOR tumors are shown (right panel). The dotted black line indicates the reference value for three tumor-free Foxn1nu mice (n=4-7). (D) Tumor sections of control (shSCR1) and EPOR-knockdown (shEPOR1) tumors were immunohistochemically stained for EPOR (brown) and iNOS (pink) and counterstained with hematoxylin (blue). Panel (E) shows a representative western blot image of phospho-AKT (60 kDa) and AKT (60 kDa) from protein extracts of control tumors (shSCR) and EPOR-knockdown tumors (shEPOR) (n=3). β-actin (44 kDa) was used as a loading control. The band intensity of proteins on western blotting images was quantified using MCID Analysis 7.0 and normalized to β-actin. Relative protein expression levels of pAKT and AKT are shown for control (shSCR) and EPOR-knockdown (shEPOR) tumors (n=6). Data are presented as scattered blots with the mean and individual data distribution of each clone (shSCR1 purple, shSCR2 red, shEPOR1 green, and shEPOR2 blue tumor samples). Data were analyzed by a Student’s t-test (black stars) or by a Mann-Whitney test (grey stars). ***p<0.001; *p<0.05.

Figure 5

iNOS and AKT are required to mediate the EPOR effect on mitochondrial biogenesis. A549 shEPOR1 (green symbols) and shEPOR2 (blue symbols), LLC1 murine Lewis lung carcinoma cells (grey bars), and human MCF-7 breast cancer cells (white bars) were cultivated in vitro. (A) EPOR-knockdown shEPOR1 A549 cells were transfected with huEPOR or a control (mCherry) plasmid, and 72 h after transfection, mRNA and protein were isolated. Shown is a representative western blot image of human erythropoietin receptor (EPOR) (63 kDa), pAKT (60 kDa), and loading control β-actin (44 kDa) (left panel), as well as mRNA levels of erythropoietin receptor (EPOR), inducible nitric oxide synthase (iNOS), transcription factor A, mitochondrial (TFAM), cytochrome c oxidase subunit 4.2 (COX-IV), voltage-dependent anion-selective channel 1 (VDAC1), and superoxide dismutase 2 (SOD2) quantified by qPCR and normalized to β-actin (ACTB) mRNA (n=6). Further shown is mitochondria content (right panel) determined by the ratio of human MT-ND1 (mitochondrially encoded NADH dehydrogenase 1) mitochondrial DNA to β2M (β-2microglobulin) genomic DNA, which was quantified by qPCR from genomic DNA extracts. (B) EPOR-knockdown shEPOR1 and 2 A549 cells were incubated with lentiviral vectors to stably express inducible nitric oxide synthase (iNOS) or mCherry (control). Additionally, cells were transfected with a plasmid to overexpress constitutively active myr-AKT (24). The cells were incubated for 72 h, and mRNA was isolated. iNOS, TFAM, and VDAC1 mRNA levels were quantified using qPCR and normalized to ACTB. Furthermore, the mitochondria content (right lower panel) was determined by the ratio of human MT-ND1 (mitochondrially encoded NADH dehydrogenase 1) mitochondrial DNA to β2M (β-2microglobulin) genomic DNA, which was quantified by qPCR from genomic DNA extracts (n=6). (C) Shown are images of shEPOR1 (upper row) and shEPOR2 (middle and bottom rows) of A549 cells stably expressing iNOS or mCherry (control) and were transfected with a plasmid to myr-AKT or not. Cells were incubated with Mitotracker (green) and Hoechst (blue), and images were taken using a fluorescence microscope and quantified using ImageJ. Shown in the right panel is the Mitotracker signal normalized to the Hoechst signal (n=4-6). (D) Murine LLC1 (grey bars) and human MCF-7 cells (white bars) were incubated with 200 µM L-NAME and/or 5 µM API-1 for 72 h. Mitochondrial content was determined by the ratio of murine (LLC1) or human (MCF7) MT-ND1 (mitochondrially encoded NADH dehydrogenase 1) mitochondrial DNA to murine or human β2M (β-2microglobulin) genomic DNA, which was quantified by qPCR from genomic DNA extracts (n=3). (E) Further shown are RNA levels of iNOS, TFAM, COX-IV, and VDAC1 quantified by qPCR and normalized to ACTB from LLC1 and MCF-7 cells either treated in vitro for 72 h with 200 µM L-NAME + 5 µM API-1 (Inhib.) to simultaneously inhibit iNOS and AKT or not (Ctrl.) (n=3). Data are shown as scattered blots with mean and individual data distribution of each clone (shEPOR1 green and shEPOR2 blue) or as bars with scatter dot plots (LLC1 grey and MCF-7 white). Data were analyzed by a Student’s t-test (black stars), a Mann-Whitney test (grey stars), an one-way ANOVA with Bonferroni post hoc test (black stars), or a Kruskal Wallis test with Dunn’s multiple comparison test (grey stars) (***p<0.001; **<0.01; *p<0.05).

Figure 6

EPOR expression correlates with VDAC1 expression in human lung cancer biopsies. Human non-small lung cancer tissue arrays were immunohistochemically stained for EPOR, VDAC1, and iNOS. (A) Shown is a representative tumor core image stained for EPOR (brown), iNOS (red), and counterstained with hematoxylin (blue). (B) EPOR and iNOS expression (left panel), as well as EPOR and VDAC1 expression (right panel), were quantified by ImageJ and normalized to the total measured tumor core area. Shown are Pearson correlation analyses of normalized EPOR (x-axis) and iNOS and VDAC1, respectively (y-axis) expression levels (in %) from tumor core images of human lung adenocarcinoma (n=19). (C) Shown are three representative tumor core images stained for EPOR (brown), VDAC1 (red), and counterstained with hematoxylin (blue). (D) EPOR and VDAC1 expression (i.e., stained area) were quantified by ImageJ and normalized to the total measured tumor core area. Shown are Pearson correlation analyses of normalized EPOR (x-axis) and VDAC1 (y-axis) expression levels (in %) from all tumor core images (upper left panel; n=214) or tumor core images of human adenocarcinoma lung tumors (upper right panel; red; n=118), human squamous cell carcinoma lung tumors (lower left panel; orange; n=65) and human large cell carcinoma (lower right panel; blue; n=30). (E) Analyses of VDAC1 mRNA expression in lung adenocarcinoma patients using the lung cancer explorer (37). The first and second panels show VDAC1 levels in normal lungs and lung adenocarcinoma in the TCGA_LUAD_2016 study (1^st panel) (38) and from Takeuchi_2006 study (2^nd panel) (39). In panels 3-5, Kaplan-Meier survival curves show an association between overall survival of lung adenocarcinoma patients and the mRNA expression of VDAC1 in the TCGA_LUAD_2016 study (3^rd panel) (38), the Takeuchi_2006 study (4^th panel) (39), and the Schabath_2016 study (5^th panel) (40). The datasets were split into low and high VDAC1 expression by using the overall mean of VDAC1 expression and were analyzed with a log-rank test.