Introduction
Epithelial-mesenchymal transition (EMT) is an essential process for tumor invasion and metastasis (1–5). EMT is the conversion of epithelial cells into more invasive mesenchymal cells and characterized by a loss of cell-cell contact and apical-basal polarity through repression of the expression of E-cadherin and other epithelial markers and activation of the expression of mesenchymal markers (such as vimentin and fibronectin). TGF-β and Wnt are known to induce EMT during cancer development and progression (1–3,5–8). The transcription factor Snail is implicated in the repression of E-cadherin expression in response to TGF-β and Wnt. We previously showed that Wnt/Snail signaling induces the Warburg effect (also termed as glycolytic switch) (9). The Warburg effect is that cancer cells mainly use glycolysis for ATP production instead of mitochondrial oxidative phosphorylation, even in the presence of oxygen. The glycolytic switch increases the availability of biosynthetic precursors for nucleotides, lipids and amino acids needed for tumor cell proliferation (10–14). Mitochondrial dysfunction is closely linked to the induction of glycolytic switch (15–18). In cancer cells, inhibition of the glycolytic switch results in growth failure; thus, molecules implicated in glycolytic switch are regarded as potential target for cancer therapies.
Dlx-2 is one of the human distal-less (Dlx) gene family proteins that play an important role(s) in the embryonic development (19,20). Increased Dlx-2 expression is observed in a number of tumor tissues, suggesting an essential role of Dlx-2 in carcinogenesis (21–23). Recently, Dlx-2 was shown to be induced by TGF-β and involved in the shift of the TGF-β tumor suppressive activity to its tumor promoting activity (23). Inhibition of Dlx-2 expression has been shown to impair the metastasis ability of B16 melanoma cells (23). In addition, we previously showed that Dlx-2 is induced by reactive oxygen species (ROS) and is implicated in metabolic stress-induced necrosis (21). ROS contribute to cancer development and progression (24,25). However, the precise mechanism of Dlx-2 in tumor progression remains to be elucidated.
In this study, we show that Dlx-2 is implicated in TGF-β- and Wnt-induced EMT and glycolytic switch via Snail induction. We also show that TGF-β/Wnt suppressed mitochondrial respiration through inhibiting cytochrome c oxidase (COX), the terminal enzyme of the mitochondrial respiratory chain, by Dlx-2/Snail cascade. COXVIc appeared to be a common target of Dlx-2, Snail, TGF-β and Wnt. Taken together, our results show that Dlx-2 plays an important role in TGF-β/Wnt-induced EMT, glycolytic switch and mitochondrial repression and COX inhibition.
Materials and methods
Cell culture
MCF-7, Madin Darby Canine Kidney (MDCK) and L cells were obtained from the American Type Culture Collection (ATCC; authenticated by short tandem repeat profiling). Wnt3a-secreting L cells and HCT116 cells were provided by Dr D.S. Min and Dr Y.J. Kim, respectively (Pusan National University, Pusan, Korea). The cell lines were passaged two times per week and low-passage cultures (passages 5–25) were used for the experiments. The cells were routinely tested negative for mycoplasma using the Mycoplasma PCR Detection kit (iNtRON Biotechnology). MCF-7 and MDCK cells were cultured in Eagle’s minimal essential medium (EMEM; Hyclone, Logan, UT, USA); and L cells and Wnt3a secreting L cells in Dulbecco’s modified Eagle’s medium (DMEM; Hyclone); HCT116 cells were cultured RPMI supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS, Hyclone) and 1% penicillin-streptomycin (PS, Hyclone) in a 37°C humidified incubator with 5% CO2. Recombinant TGF-β (R&D Systems, MN, USA) was applied to cells at a concentration of 10 ng/ml.
Transfection and short hairpin RNA (shRNA) interference
The expression vectors pCAGGS-Dlx-2 (provided by Dr John L.R. Rubenstein, University of California at San Francisco) and pCR3.1-Snail-Flg (provided by J.I. Yook, Yonsei University, Korea) were transfected into MCF-7 cells using jetPEI (Polyplus transfection). pSUPER vectors for shRNA against control, Dlx-2, Snail, Smad2, Smad3, Smad4, β-catenin, TCF4, Axin1, Axin2 and COXVIc (abbreviations; shCon, shDlx-2, etc.) were produced and transfected as described previously (9).
Immunoblotting and quantitative real-time PCR (qRT-PCR)
Immunoblotting and qRT-PCR were performed as described previously (9,26). Immunoblotting was performed with the following antibodies: Dlx-2 (Millipore, Billerca, MA, USA); Snail (Abgent, San Diego, CA, USA); E-cadherin, vimentin, COXVIIc and COX19 (Santa Cruz, CA, USA); COXVIc, COXVIIa, and COXVIIb (Mitoscience, Eugene, OR, USA); SCO2 and COX18 (Abcam, Cambridge, MA, UK); α-tubulin (Biogenex, CA, USA). Total mRNA was isolated from cells by using the TRIzol (Invitrogen, Carlsbad, CA, USA) according to the supplier’s instructions. Transcript levels were assessed with qRT-PCR with primers for Dlx-2, Snail, E-cadherin and β-actin. Values are normalized to β-actin.
Immunofluorescence (IF) microscopy
MCF-7 cells were fixed for 10 min in 3.7% formaldehyde in PBS, permeabilized in PBS containing 0.2% Triton X-100 for 30 min, and blocked with 2% BSA in 0.1% PBST for 3 h. For E-cadherin staining, cells were incubated with mouse anti-E-cadherin (Santa Cruz) antibody for overnight at 4°C and immunostained with Alexa-Fluoro-488-labeled anti-mouse secondary antibody (Molecular Probes, NY, USA). Hoechst 33342 (Molecular Probes) was used to stain cell nuclei.
Chromatin immunoprecipitation (ChIP) assay
ChIP assays were conducted using a ChIP Assay kit (Millipore). IgG, anti-Dlx-2 or anti-Snail (Santa Cruz) was used to immunoprecipitate DNA-containing complexes. ChIP-enriched DNA was analyzed by PCR using primers complementary to the promoter regions.
Assays for mitochondrial respiration, COX activity, glucose (Glc) consumption, lactate (Lac) production and ATP production
Mitochondrial respiration and COX activity were measured as described previously (9,27). For mitochondrial respiration assay, exponentially growing cells (1.5×106) were washed with TD buffer (137 mM NaCl, 5 mM KCl, 0.7 mM Na2HPO4, 25 mM Tris-HCl, pH 7.4), and were collected and resuspended in complete medium without phenol red. The cells (5×105) were transferred to the Mitocell chamber equipped with a Clark oxygen electrode (782 Oxygen Meter, Strathkelvin Instruments, Glasgow, UK). Oxygen consumption rates were measured after adding 30 μM DNP to obtain maximum respiration rate and its specificity for mitochondrial respiration was confirmed by adding 5 mM KCN (29). COX activity was determined by measuring the KCN-sensitive COX-dependent O2 consumption rate by adding 3 mM TMPD in the presence of 30 μM DNP and 20 μM antimycin A. Glc, Lac and intracellular ATP levels were determined using a glucose oxidation assay kit (Sigma, MO, USA), a colorimetric and fluorescence-based lactate assay kit (BioVision, CA, USA), and an ATP Bioluminescence Assay kit (Roche, Switzerland), respectively. The level of ATP produced by aerobic respiration and glycolysis was determined by measuring Lac production and oxygen consumption (9,28).
Human tumor samples
All human tissues from patients #70331 (infiltrating ductal carcinoma), #69965 (invasive ductal carcinoma), #69941 (metaplastic carcinoma) and #70168 (pleomorphic lobular carcinoma) with breast cancer and normal matched tissue pairs from the same individuals were provided by the National Biobank of Korea, PNUH in compliance with all the regulations related to biomedical research with human samples, including informed consent of the patients for the use of their samples. We performed qRT-PCR and immunoblotting with cancer tissues. TRIzol extraction of total RNA and subsequent extraction of protein was carried out essentially according to the manufacturer’s specifications (Invitrogen Corp.). To a 50–100 mg tissue segment, 1 ml of TRIzol was added, and the sample was homogenized with 2–3 min homogenisation with a tissue lyser (Qiagen, Hilden, Germany) at 30 Hz.
Measurement of circularity
For circularity, microscopic images were analyzed with Axiovision LE software (Release 4.8 version). Circularity was measured with the Axiovision LE software Measure command that calculates object circularity using the formula circularity = 4π(area/perimeter2). Circularity value of 1.0 indicates a perfect circle. As the value approaches 0.0, it indicates an increasingly elongated polygon.
Statistical analysis
qRT-PCR and assays for mitochondrial respiration, COX activity, Glc consumption, Lac production and ATP production were performed at least in triplicate and most experiments were repeated more than twice. Data were analyzed by the Student’s t-test (unpaired, two-tailed) and results were expressed as mean ± SE. P<0.05 was considered statistically significant.
Results and Discussion
Dlx-2 induces EMT via Snail activation
We examined the effects of Dlx-2 overexpression in non-invasive breast cancer cell line MCF-7. Dlx-2 overexpression in MCF-7 cells induced the loss of cell polarity and the formation of elongated morphology with pseudopodia, which are a characteristic of mesenchymal cells, indicating that Dlx-2 may induce EMT (Fig. 1A). Spindle quantification also supported Dlx-2-induced EMT (Fig. 1A). The phenotypical change was accompanied by a decreased expression of an epithelial marker E-cadherin, as revealed by IF, qRT-PCR and immunoblotting (Fig. 1A–C). In addition, Dlx-2 increased the levels of a mesenchymal marker vimentin (Fig. 1C).
We examined whether Dlx-2 directly regulates E-cadherin expression. We found three putative Dlx-2 binding sites between −1,000 and +1,000 from the transcription start site (TSS) in the promoter of E-cadherin. However, Dlx-2 binding to the E-cadherin promoter was not detected in ChIP analysis (Fig. 1D). Thus, we thought that Dlx-2 may indirectly reduce E-cadherin by activating other E-cadherin repressor. Snail is a typical E-cadherin repressor for EMT (2). Therefore, we examined whether Snail is involved in Dlx-2-induced EMT. Dlx-2 overexpression prominently increased Snail protein (Fig. 1E) and mRNA levels (Fig. 1F). Three putative Dlx-2 binding sites were found in the Snail promoter (Fig. 1G). ChIP assay showed Dlx-2 binding to the Snail promoter (Fig. 1G), indicating that Dlx-2 regulates Snail expression. Note that Dlx-2 binding to the Snail promoters was detected only at an early time-point (3 h) after transfection (Fig. 1G). We examined whether Snail is involved in Dlx-2-induced EMT. shSnail (hereafter, shSnail) prevented Dlx-2-induced EMT and E-cadherin downregulation and vimentin upregulation (Fig. 1A–C), indicating that Dlx-2 induces EMT via Snail activation. We performed ChIP assay to examine Snail binding to the E-cadherin promoter. 4 E-boxes for Snail binding are found in human E-cadherin promoter; but Snail has been shown to bind to the E-boxes 1, 3 and 4 with the most strong binding activity for E-box 4 (6,29). Dlx-2 and Snail overexpression enhanced Snail binding to the E-box 4 of the E-cadherin promoter (Fig. 1H), indicating that Dlx-2-induced Snail can interact with E-box sites (including E-box 4) in the E-cadherin promoter to repress E-cadherin expression.
Dlx-2/Snail cascade is implicated in TGF-β- and Wnt3a-induced EMT
TGF-β and Wnt signaling pathways are known to induce EMT through Snail activation (2,3,8). Thus, we examined if the Dlx-2/Snail cascade is involved in TGF-β- and Wnt-induced EMT. TGF-β or Wnt3a-conditioned medium (CM, obtained from Wnt3a-secreting L cells) induced EMT and E-cadherin downregulation (Fig. 2A–F). TGF-β and Wnt3a also increased expression of Dlx-2 and Snail (Fig. 2B, C, E and F). Note that TGF-β increased Snail expression at both mRNA and protein levels, whereas Wnt induced Snail at the level of protein but not mRNA (Fig. 2B, C, E and F). In the Wnt signaling, Axin2 is one of the Wnt target genes and regulates EMT by acting as a chaperone for nuclear export of GSK3β that is the dominant kinase responsible for Snail protein turnover and activity, in human breast cancer cells, thereby increasing Snail protein stability in the nucleus (30). Thus, in Wnt signaling, Dlx-2 may be involved in GSK3β-mediated Snail protein turnover; although it remained to be elucidated. shDlx-2 suppressed Snail expression, whereas shSnail had no effect on Dlx-2 expression (Fig. 2B, C, E and F), indicating that Dlx-2 acts upstream of Snail. shDlx-2 or shSnail appeared to block TGF-β- and Wnt3a-induced EMT and E-cadherin downregulation (Fig. 2A–F).
TGF-β induces EMT through activation of Smad signaling pathways (8). Smad2/3/4 shRNA suppressed TGF-β-induced EMT and E-cadherin downregulation (Fig. 2G) as well as Dlx-2 and Snail expression (Fig. 2G and H). Wnt induces EMT via canonical pathways, which includes β-catenin, TCF4 and Axin1/2. shRNA for β-catenin, TCF4 and Axin1/2 suppressed Wnt3a-induced EMT/E-cadherin downregulation (Fig. 2I) as well as Dlx-2 (but not Snail) expression (Fig. 2I and J). These results supported that the Dlx-2/Snail cascade is implicated in TGF-β- and Wnt3a-induced EMT.
We further examined the effects of shDlx-2 and shSnail on the EMT in HCT116 and MDCK cells. shDlx-2 and shSnail suppressed Wnt3a-induced EMT and E-cadherin downregulation in HCT116 cells (Fig. 2K and L). Similar inhibitory effects of shDlx-2 and shSnail on TGF-β-induced EMT were observed in MDCK cells. shDlx-2 and shSnail prevented TGF-β-induced EMT and E-cadherin downregulation (Fig. 2M and N).
Dlx-2/Snail signaling is involved in TGF-β- and Wnt3a-induced glycolytic switch and mitochondrial repression
Wnt3a/Snail cascade has been shown to induce glycolytic switch and mitochondrial repression (9). Therefore, we examined whether Dlx-2 is involved in the Snail-induced glycolytic switch. Dlx-2 overexpression significantly increased Glc consumption and Lac production (Fig. 3A). In addition, Dlx-2 overexpression decreased O2 consumption (Fig. 3A). ATP levels were similar in both control and Dlx-2 transfected cells (data not shown). By measuring oxygen consumption and Lac production, we estimated the relative contributions of glycolysis and aerobic respiration in total ATP production. Dlx-2 increased the ratio of ATP produced by glycolysis versus ATP produced by aerobic respiration (Fig. 3A), indicating that Dlx-2 induces glycolytic switch. Dlx-2-induced glycolytic switch and mitochondrial repression were prevented by shSnail (Fig. 3A), indicating that Dlx-2 induces glycolytic switch/mitochondrial repression via Snail.
Then, we examined if TGF-β and Wnt3a induce glycolytic switch/mitochondrial repression via the Dlx-2/Snail cascade. TGF-β and Wnt3a induced Glc consumption and Lac production (Fig. 3B and C). In addition, TGF-β and Wnt3a reduced O2 consumption (Fig. 3B and C). Although total ATP concentrations remained the same in all cells, TGF-β and Wnt3a increased the ratio of ATP produced by glycolysis versus ATP produced by aerobic respiration (Fig. 3B and C), indicating that TGF-β and Wnt3a induce glycolytic switch. In addition, shDlx-2 or shSnail decreased TGF-β- and Wnt3a-induced increase of Glc consumption and Lac production and impairment of O2 consumption (Fig. 3B and C), indicating that the Dlx-2-Snail axis is involved in TGF-β- and Wnt3a-induced glycolytic switch and mitochondrial repression.
Dlx-2/Snail signaling is involved in TGF-β/Wnt-induced COX inhibition
Changes in the activity of COX, the terminal enzyme of the mitochondrial respiratory chain, are closely associated with decreased mitochondrial respiratory activity. Therefore, we examined the effects of Dlx-2 on COX activity. Dlx-2 overexpression reduced COX enzymatic activity (Fig. 3D). Dlx-2-induced COX inhibition was prevented by shSnail (Fig. 3D), indicating that Dlx-2 induces COX inhibition via Snail activation.
We also found that TGF-β and Wnt3a reduce COX activity (Fig. 3E and F). We examined if Dlx-2-Snail axis is implicated in TGF-β- and Wnt3a-induced COX inhibition. shDlx-2 or shSnail decreased TGF-β- and Wnt3a-induced impairment of COX activity (Fig. 3E and F), indicating that the Dlx-2-Snail axis is involved in TGF-β- and Wnt3a-induced mitochondrial repression.
Dlx-2/Snail signaling is implicated in TGF-β/Wnt-induced downregulation of multiple COX subunits and assembly factors
Eukaryotic COX is composed of 13 different subunits and its assembly is regulated by a sequential action of several nucleus-encoded assembly factors. We examined the effects of Dlx-2 and Snail on the gene expression of COX subunits and assembly factors using real-time PCR (Table IV). Dlx-2 downregulated the expression of COXVIc and COX19 (Table IV and Fig. 4A–C). Snail has been shown to decrease mRNA levels of COXVIc, COXVIIa and COXVIIc (9). Note that Snail-mediated COXVIIa and COXVIIc repression was not observed in Dlx-2 expressing cells. shSnail suppressed Dlx-2-induced reduction in the levels of COXVIc, but not COX19 (Fig. 4A and C), suggesting that Snail is implicated in Dlx-2-mediated COXVIc gene repression, but not in COX19 gene repression. Dlx-2 overexpression enhanced Snail binding to the COXVIc promoter, but not COX19 promoter (Fig. 4D), confirming that COXVIc, but not COX19, is regulated by a Snail-dependent mechanism. As expected, Dlx-2 also did not bind to the COX19 promoter (Fig. 4E).
We examined the effects of TGF-β and Wnt on the gene expression of COX subunits and assembly factors using real-time PCR. TGF-β decreased mRNA levels of COXVIc, COXVIIa and COX11 (Table IV and Fig. 4F). Snail-mediated COXVIIc repression was not observed in TGF-β-treated cells by unknown mechanism. shSnail suppressed TGF-β-induced reduction in the levels of COXVIc and COXVIIa, but not COX11 (Fig. 4F). shDlx-2 suppressed TGF-β-induced reduction in the levels of COXVIc (Fig. 4G). In case of Wnt3a, it decreased mRNA levels of COXVIc, COXVIIa and COXVIIc (Fig. 4H) (9). shSnail suppressed Wnt3a-induced reduction in the levels of COXVIc, COXVIIa and COXVIIc (Fig. 4H). shDlx-2 also suppressed Wnt3a-induced reduction in the levels of COXVIc (Fig. 4I).
COXVIc was a common target of TGF-β, Wnt, Dlx-2 and Snail. Because TGF-β- and Wnt-induced COX inhibition was suppressed by shDlx-2, COXVIc levels seem to be more important. Thus, TGF-β- and Wnt-induced COX inhibition is thought to be mediated by COXVIc inhibition by the Dlx-2/Snail-mediated pathway.
We examined the effects of shCOXVIc on mitochondrial respiration and COX activity. Without affecting the cell morphology (Fig. 4J), shCOXVIc inhibited mitochondrial respiration and COX activity (Fig. 4K).
The expression of Dlx-2, Snail and COXVIc in human tumors
To further examine the physiological relevance of Dlx-2/Snail/COXVIc cascade, we analyzed human tumor samples. We examined the expression of Dlx-2, Snail and COXVIc by qRT-PCR using RNAs extracted from paired biopsy of breast cancer and the corresponding normal tissues. Dlx-2 and Snail expression were higher and COXVIc expression was lower irrespective of the stage in breast cancer tissues compared with matched normal tissues (Fig. 4L). We also examined the expression of Dlx-2 and Snail protein using immunoblotting. Dlx-2 and Snail expression were higher in breast cancer tissues than in matched non-tumorigenic tissues (Fig. 4L). These results further support an important role of Dlx-2 and Snail in tumor development.
In this study, we show novel functions of Dlx-2 that contribute to tumor development and progression; to induce EMT and glycolytic switch. Dlx-2 induced EMT and glycolytic switch via Snail activation. The Dlx-2/Snail cascade was involved in TGF-β/Wnt-induced EMT and glycolytic switch. Furthermore, we found that TGF-β/Wnt suppressed COX in a Dlx-2/Snail-dependent manner. COXVIc downregulation appeared to play an important role in TGF-β/Wnt-induced COX inhibition. Taken together, our findings suggest that Dlx-2 plays an important role in TGF-β- and Wnt-induced tumor progression and aggressiveness.