Na/K-ATPase Y260 Phosphorylation–mediated Src Regulation in Control of Aerobic Glycolysis and Tumor Growth

We report here the identification of α1 Na/K-ATPase as a major regulator of the proto-oncogene Src kinase and the role of this regulation in control of Warburg effect and tumor growth. Specifically, we discovered Y260 in α1 Na/K-ATPase as a Src-specific phosphorylation and binding site and that Y260 phosphorylation is required for Src-mediated signal transduction in response to a number of stimuli including EGF. As such, it enables a dynamic control of aerobic glycolysis. However, such regulation appears to be lost or attenuated in human cancers as the expression of Na/K-ATPase α1 was significantly decreased in prostate, breast and kidney cancers, and further reduced in corresponding metastatic lesions in patient samples. Consistently, knockdown of α1 Na/K-ATPase led to a further increase in lactate production and the growth of tumor xenograft. These findings suggest that α1 Na/K-ATPase works as a tumor suppressor and that a loss of Na/K-ATPase-mediated Src regulation may lead to Warburg phenotype in cancer.

Y260 phosphorylation represents a general feature of Src signaling. If α1 Na/K-ATPase represents an important general regulator of Src, we would expect an increase in Y260 phosphorylation when cells are stimulated not only by ouabain but also other receptor ligands. We therefore tested Y260 phosphorylation in three different signaling pathways where Src is necessary for signal transduction. First, ouabain increased Y260 phosphorylation in both time and concentration-dependent manner in LLC-PK1 cells (Fig. 2d,e). To ensure that this effect was Y260-specific, Y10 phosphorylation was also probed. Consistent with a previous study 20 , ouabain showed no effect on Y10 phosphorylation (Fig. 2e). Second, Y260 phosphorylation was stimulated by EGF. EGF, like ouabain, produced a time-and dose-dependent stimulation of Y260 phosphorylation of Na/K-ATPase as well as Src at Y418 (Fig. 2f,g). Furthermore, we plated cells on to fibronectin-coated plates and analyzed the stimulation of Y260 phosphorylation in integrin signaling pathway 21 . An increase in Y260 phosphorylation was noted and correlated well with the activation of Src (Fig. 2h). These data indicate that Y260 phosphorylation represents a general feature of Src regulation, not only relevant to receptor function of Na/K-ATPase but also to receptor tyrosine kinases and integrin signaling.
Y260 is required for Src-mediated signaling. To investigate the role of Y260 in Src-mediated signal transduction, we generated a stable cell line that expresses a loss of function (Y260A) mutant rat α1 Na/K-ATPase. An α1 Na/K-ATPase knockdown cell line, PY-17 was used for transfection 7 . A normal rat α1 Na/K-ATPase-rescued cell line, AAC-19, was used as a control. Control experiments (Supplementary Fig. 2A) show that the clone 21 and AAC-19 expressed comparable amount of rat α1. 3 H-ouabain binding assays indicate that the expression of endogenous pig α1 Na/K-ATPase in Y260A mutant cells only amounted to about 1% of the total Na/K-ATPase ( Supplementary Fig. 2D) 13,22 . Functionally, the expressed Y260A mutant was expressed in the plasma membrane, fully capable of rescuing the expression of β1 subunit and forming a functional Na/K-ATPase exhibiting comparable ouabain-sensitive ATPase activity as in AAC-19 cells ( Supplementary Fig. 2B-E).
Next, we measured the effects of Y260A mutation in two different signaling pathways where Src plays an important role. First, Y260A mutant and AAC-19 cells were treated with different concentrations of ouabain, and subjected to Western blot measurements of Src, ERK, and Akt activities. Ouabain stimulated Src, ERK and Akt phosphorylation in AAC-19 cells. These stimulations were abolished by the Y260A mutation ( Fig. 3a-c). To confirm that the observed effect was due to the inhibition of Na/K-ATPase/Src interaction, we conducted immunoprecipitation analysis. Y260A mutation resulted in a significant decrease (~60%) in the binding of α1 Na/K-ATPase to Src as compared to AAC-19 cells (Fig. 3d). Control experiments show that another clone of Y260A mutant cells, clone 24, like clone 21, also failed to respond to ouabain stimulation ( Supplementary  Fig. 3A,B).
Second, we assessed the role of Y260 in the EGFR signaling. Y260A mutant clone 21 and AAC-19 cells were exposed to different concentrations of EGF and measured for EGFR phosphorylation. As depicted in Fig. 3e, EGF stimulated Src kinase activity in AAC-19, but not in Y260A mutant cells. In accordance, EGF stimulated the phosphorylation of EGFR at Y845, a known Src-phosphorylation site 23 , in AAC-19, but not in Y260A mutant cells. Interestingly, we observed that the basal EGFR Y845 phosphorylation was significantly increased in Y260A mutant cells, indicating an important role of Y260-mediated Src interaction in the regulation of basal Src and Src-effector activity.
To corroborate this finding, we first measured the basal protein tyrosine phosphorylation. As shown in Supplementary Fig. 2F, Y260A cells expressed more tyrosine-phosphorylated proteins. To test whether this effect was Src-mediated, we measured Src phosphorylation at Y418 and Y529. As shown in Supplementary Fig. 2F, Src activity (pY418) was increased almost 2 folds in Y260A cells but there was no change in Y529 phosphorylation. Consistently, basal ERK activities were increased in Y260A cells in comparison to those in AAC-19 cells (Fig. 3b), which was sensitive to PP2, a Src inhibitor ( Supplementary Fig. 2G).
Y260A mutation leads to metabolic switch. The above studies indicate that Y260 phosphorylation is a major regulatory mechanism of Src-mediated signal transduction in the plasma membrane in response to a variety of stimuli. To assess the general significance of this newly discovered signaling mechanism, we probed a potential role of Y260 in the regulation of cellular metabolism. In the routine culture of Y260A mutant cell lines (both clone 21 and 24), we noticed faster acidification of culture medium than that of control AAC-19 cells. Measurement of medium confirmed that Y260A mutant cells produced 80% more lactate than that of AAC-19 cells (Fig. 4a). This suggests that disruption of Y260-mediated Src interaction may cause a metabolic switch from mitochondrial oxidative phosphorylation to aerobic glycolysis. To further test this working hypothesis, we first measured cellular ATP content in response to the inhibition of glycolysis by 2-deoxyglucose (2-DG). Compared to AAC-19 cells, Y260A cells were much more sensitive to 2-DG (Fig. 4b). Second, we measured the effect of glucose depletion on cell growth. Y260A mutant cells failed to grow in the absence of glucose whereas AAC-19 cells grew. In fact, the number of AAC-19 cells more than doubled during this time (Fig. 4c).
Then we determined extracellular acidification rate (ECAR) in both AAC-19 and Y260A cells in the presence of different inhibitors 24 . Cells were first cultured in glucose-free basal medium. The addition of glucose to cells induced a large increase in ECAR, indicating an increase in aerobic glycolysis (Fig. 4d). To measure the maximal capacity of glycolysis, this was followed by the addition of oligomycin that inhibited mitochondrial ATP production, resulting in a further increase in ECAR in AAC-19, but not in Y260A cells. Moreover, the addition of glycolysis inhibitor 2-DG completely reversed the increase in ECAR. Data analyses indicated a diminished glycolytic reserve in Y260A mutant cells and an increase in aerobic glycolysis (Fig. 4e). When oxygen consumption rate was measured, no major defects in mitochondrial function (e.g. respiratory control ratio and coupling efficiency) were noted ( Supplementary Fig. 4B,C) except that the reserve capacity was reduced in Y260A mutant cells (Fig. 4f). Because oncogene activation is known to change cellular metabolic phenotype 25-33 , we tested whether this effect was dependent on Src dysregulation by α1 Na/K-ATPase in Y260A cells. As depicted in Fig. 4g and h, Src inhibition by treatment of Y260A cells with PP2, a Src inhibitor, was sufficient to reduce both lactate production ( Fig. 4g) and glycolysis rate (Fig. 4h). To further investigate the metabolic adaptation, we compared the gene expression profiles of Y260A mutant cells and control AAC-19 cells by RNAseq analysis. Several important genes involved in the glycolytic metabolism were significantly upregulated in Y260A mutant cells ( Supplementary Fig. 4D). Notably, hexokinase 2 isoform (HK2), pyruvate dehydrogenase kinase (PDK) and lactate dehydrogenase (LDHA) were significantly increased, which has been also observed in cancer cells that underwent metabolic switch 1 . We also observed a significant increase in the expression of glucose transporter GLUT4 and a few amino acid transporters in Y260A, providing further support to the notion that these mutant cells display a metabolic adaptation similar to cancer cells. To further confirm that these effects were mediated by the dysregulation of Src kinase, we treated Y260A cells with PP2 to see if it could reverse the change in metabolic gene expression. As shown in Supplementary Fig. 4E, PP2 treatment restored the mRNA level of most of these upregulated genes, as measured by qPCR.
Y260 phosphorylation is reduced in many cancer cell lines. In view of the important role of metabolic switch in cancer biology, the above findings led us to hypothesize that cancer cells may lose the capacity of α1 Na/K-ATPase-mediated Src regulation. To test this hypothesis, we compared the pY260 level in two different panels of cancer cell lines, which should be an indicator of Na/K-ATPase/Src interaction. As shown in Fig. 5a, Y260 phosphorylation was greatly reduced in both prostate and breast cancer cell lines in comparison to the corresponding control cell lines. To verify this, we immunoprecipitated Src kinase in selected cell lines and compared the co-precipitated α1 Na/K-ATPase level between cancer and control cell lines. We detected a significant decrease in Na/K-ATPase/Src interaction by co-immunoprecipitation ( Supplementary Fig. 5A).
Knocking down of α1 Na/K-ATPase increases aerobic glycolysis in DU145 cells and promotes the growth of tumor xenograft. To test whether α1 Na/K-ATPase-mediated Src interaction is important for control of aerobic glycolysis and tumor growth, we knocked down α1 Na/K-ATPase expression in DU145 prostate cancer cells using siRNA and generated stable cell lines A4-7, A4-3 and a control vector-transfected DU-P1 cell line. The DU-P1 cells expressed similar amount of α1 Na/K-ATPase like parental DU145 cells whereas A4-7 cells showed about 50% down-regulation ( Supplementary Fig. 5B). Functional analyses indicate that knockdown of α1 Na/K-ATPase caused a significant decrease in Na/K-ATPase/Src interaction (Fig. 5b). A significant increase in Src and its effector FAK (Focal Adhesion Kinase) activity was noted in A4-7 cells (Fig. 5c). Moreover, EGF was able to stimulate Src activation and Y845 phosphorylation of EGFR in control DU-P1 cells but not in A4-7 cells (Supplementary Fig. 5C). Like Y260A mutant cells, basal EGFR Y845 phosphorylation was increased in A4-7 cells. Moreover, A4-7 cells expressed higher amount of Myc, which is indicative of their more aggressive and proliferative nature (Fig. 5c) 34 . In accordance, we detected a further increase in lactate production in A4-7 cells, which was sensitive to Src inhibition by PP2 (Figs 5d and Supplementary Fig. 5E). To compare the tumor forming ability, we tested the cell proliferation rate of A4-7 cells against DU-P1 cells. As expected the cell proliferation rate of A4-7 was significantly higher than that of the control, and reversed by PP2 (Fig. 5e). Finally, when control DU-P1 and A4-7 cells were implanted into NOD/SCID mice, we observed a close to four-fold increase in tumor size of A4-7 vs DU-P1 (Fig. 5f).
Expression of α1 Na/K-ATPase is reduced in several human cancers, especially the metastatic lesions. To assess the clinical relevance of our new findings, we measured α1 protein expression in three different types of human cancer where the expression of α1 Na/K-ATPase in the normal epithelium is high. First, we analyzed 66 prostate carcinoma, 12 bone metastatic and 23 normal tissue samples. The expression of α1 Na/K-ATPase was significantly reduced (Fig. 6a) in prostate carcinoma (n = 66) vs control (n = 23). This was confirmed by paired analyses (normal vs carcinoma, n = 15). Most importantly, there were no detectable α1 Na/K-ATPase signals in 11 out of 12 bone metastatic samples. Second, we compared 10 normal breast tissue and 62 ductal carcinoma samples, and corresponding metastatic samples in the ovary and lymph node (Fig. 6b). The expression of α1 Na/K-ATPase was significantly decreased in the primary tumor and further reduced in metastatic samples. Third, a significant decrease in α1 Na/K-ATPase expression was also found in renal cell carcinoma (n = 61) in comparison with the control (n = 15) (Fig. 6c). Interestingly, a decrease in adrenal gland metastasis of renal carcinoma appeared to be less severe than that in bone lesions of prostate cancer. However, α1 Na/K-ATPase was highly expressed in normal kidney tubules (15 out of 15 scored 3 in normal tissue). This level of expression was detected in only about 10% of renal cell carcinoma (6 out of 61) and further reduced to 0% in metastatic lesions.
Transcriptional regulation could represent an important mechanism of changing α1 expression in human cancers. Database search reveals that the expression of α1 Na/K-ATPase mRNA was significantly reduced (p < 0.001) in renal clear cell carcinoma in the TCGA-KIRC database (n = 530). The mean log 2 values are 11 and 8.4 for the normal kidney and renal carcinoma, respectively. Most importantly, this expression pattern inversely correlated with patient survival rate, with lower expression being associated with a high mortality rate (Fig. 6e). A significant decrease in the expression of α1 Na/K-ATPase mRNA in prostate cancer was also detected. However, the difference was less than 0.3 Log 2 value. As such, there was no correlation between lowered mRNA expression and patient survival rate in prostate cancer ( Supplementary Fig. 6A,B).

Discussion
We report here the discovery of α1 Na/K-ATPase Y260 as a Src-specific phosphorylation and binding site, and an increase in Y260 phosphorylation as a general feature of Src-mediated signal transduction in response to the activation of membrane receptors. It appears that this dynamic regulation of Src by α1 Na/K-ATPase is significantly attenuated or lost in human cancer cells. This dysregulation increases the basal Src activity, resulting in the activation of Src effectors such as ERK, EGFR, Akt and FAK, all of which are implicated in cancer progression [35][36][37][38] . Furthermore, we provide evidence that the loss of Src regulation by α1 Na/K-ATPase causes a metabolic switch and promotes the formation and growth of tumor xenograft. A schematic diagram of α1 Na/K-ATPase-mediated Src regulation and the consequence of the loss of this regulation is given in Fig. 7. These and other important issues are further discussed.

Na/K-ATPase α1 isoform as an underappreciated albeit important regulator of Src family kinases.
Recent studies have indicated an important role of α1 Na/K-ATPase-mediated and Src-dependent pathways in control of embryonic development, renal salt handling and in disease progression where inflammation and ROS stress play an important role [39][40][41][42][43] . Interestingly, it appears that α1 Na/K-ATPase also interacts with other members of Src family kinases and that such interaction (e.g., Lyn) is a key to CD36-and CD40-mediated signal transduction [44][45][46] . The new findings presented here reaffirm the ability of α1 Na/K-ATPase to bind and regulate Src. Specifically Y260 and its phosphorylation by Src kinase appears to be very significant in many Src kinase mediated signaling pathways, since it participated in not just ouabain, but also EGF as well as integrin signaling. These data indicate that α1 Na/K-ATPase is an essential partner (regulator) of Src kinase in the plasma membrane, which is further supported by the loss of function Y260A mutant studies. Moreover, the Y260A mutant Na/K-ATPase is fully functional as an ion pump, but defective in binding and regulating Src. As such, it blocks not only ouabain but also EGF-induced Src signaling. Moreover, it also caused Src dysregulation as basal Src activity increased by almost 2 folds in Y260A cells. Concomitantly, we detected an increase in activity of ERK, which was sensitive to Src inhibitor, PP2 (Supplementary Fig. 2G). These findings are consistent with our prior observations that either knockdown of α1 Na/K-ATPase or expression of Src binding-α1 mutants increases basal Src and consequently ERK activity 13,16 . Thus, the new findings reaffirm our contention that the α1 Na/K-ATPase is a major regulator of the plasma membrane pool of Src. We further suggest that the loss of interaction between α1 Na/K-ATPase and Src will result in a dys-regulation of Src, which could lead to 1) the inability of ouabain-induced signaling and 2) increased basal Src activity. Although many proteins interact and regulate Src, to the best of our knowledge, α1 Na/K-ATPase appears to be the only one that can regulate Src simultaneously through its two domain-domain interactions [11][12][13] . This balanced and sequential regulation is essential for the dynamic nature of Src signaling 47 . Disruption of either interaction alters Src signaling regardless of whether basal Src activity is increased or decreased 13,15,22 , resulting in changes in cellular metabolism and growth.
The role of α1 Na/K-ATPase in the regulation of aerobic glycolysis and its implication in cancer biology. Although mutations in oncogenes like KRAS or PI3K genes have been implicated in the metabolic switch observed in cancer cells, how dysregulations of the proto-oncogene SRC affects tumor growth is not clear 1 . Src family kinases (SFK) are frequently hyper-activated in cancers but activating mutations or chromosomal rearrangements in Src are relatively rare in nature 2 . The precise molecular mechanism underlying the defective Src regulation in human cancers remains to be resolved.
Interestingly, Y260A mutant cells undergo a metabolic switch from mitochondrial oxidative phosphorylation to aerobic glycolysis (i.e., the increase in lactate production), a phenomenon commonly seen in oncogenic cells. The expression of several important glycolytic genes is upregulated in these cells, which further supports the phenotype. Consistently, PP2 treatment not only reduced the expression of these up-regulated genes involved in glycolysis, but also blocked the increase in lactate production (Fig. 4g,h). In short, these studies provide a molecular insight of how α1 Na/K-ATPase may work as a regulator of cellular metabolism through Src 4,5 .
We further suggest that this mechanism may provide a novel link connecting Na/K-ATPase α1 subunit downregulation in cancer to defective Src regulation and henceforth altered cellular metabolism. We demonstrate here that cells harboring Y260A mutant Na/K-ATPase have increased aerobic glycolysis and lactate production. Y260 phosphorylation was significantly reduced in human cancer cell lines, indicating a loss of Src regulation by α1 Na/K-ATPase. This finding is further substantiated by our tissue array data that show a significant decrease in α1 Na/K-ATPase in three different types of human cancers. Others have also reported a reduction in Na/K-ATPase α1 expression in lung, skin and testicular cancers [48][49][50][51] . Apparently, both transcriptional (e.g., kidney) and post-transcriptional mechanisms (e.g., prostate) are involved in the down-regulation of α1 Na/K-ATPase. The post-transcriptional regulation appears to involve increased endocytosis of α1 Na/K-ATPase 52 .
The importance of α1 Na/K-ATPase in cancer biology was further substantiated by two additional observations. First, the expression of α1 Na/K-ATPase is further reduced in metastatic samples from prostate, breast cancer, and renal cell carcinoma. Src activity is higher in metastatic lesions relative to primary tumor tissues 53,54 . It is plausible that the decrease in the expression of α1 Na/K-ATPase diminishes the regulation of Src, leading to an increase in basal Src activity. This is supported by the data that Src and its effectors such as EGFR and FAK activity were further increased in α1 knockdown A4-7 cells (Fig. 5c and Supplementary Fig. 5C). Moreover, A4-7 cells had increased lactate production and c-Myc expression (Fig. 5c,d), which is indicative of their more aggressive status. This is reaffirmed by an almost four-fold increase in tumor size when xenografted into mice (Fig. 5f). This observation is consistent with knockdown studies reported with other Src regulators like Csk 55 . Second, data analysis of renal clear cell carcinoma in the TCGA-KIRC database (n = 530) reveals a significant decrease in the expression of α1 mRNA in patient samples. This decrease is inversely correlated with the survival of patients suffering from renal clear cell carcinoma.
In short, our new findings provide strong evidence that α1 Na/K-ATPase works like a tumor suppressor by regulating the cellular Src kinase. We suggest that this newly discovered signaling mechanism may be targeted for developing new cancer therapeutics, which is supported by our early studies 52 . Furthermore, we speculate that Y260 phosphorylation may be utilized as an indicator for Na/K-ATPase/Src interaction in cancers. This may be of particular importance in some cancer types where α1 Na/K-ATPase expression is reportedly high 56 . Needless to say, this hypothesis remains to be tested. Finally, it is important to note that although metabolic switch is well established to play a role in tumorigenesis, our data do not allow us to conclude that increases in tumorigenesis in A4-7 cells are due to increased glycolysis. Interestingly, we observed that this switch actually inhibits cell proliferation in Y260A cells (Supplementary Fig. 4A). This apparent discrepancy in cellular fate may be explained by the recent findings that suggests activation of oncogene in normal cells can induce them to enter a non-proliferative state called senescence, whereas oncogene activation in cancer cells can cause them to become hyper proliferative 57,58 . Immunohistochemical (IHC) Staining Analysis of Human Samples. Na/K-ATPase α1 IHC staining was performed by US Biomax (Rockville, MD) using human kidney, prostate and breast tissue microarray, and was carried out as described before 52 . As such, it is not possible to record the details of the patients 52 . Antibody for IHC-mouse monoclonal anti-Na/K-ATPase α1 antibody (Millipore). Nucleus was counterstained with hematoxylin. Two independent investigators examined the staining intensity and scored each slide three times as defined: 0, absent; 1, weak; 2, moderate; 3, strong. A mean score was recorded.

Materials
Cell culture and cell line generation. All cell lines were purchased from and maintained according to ATCC (Mannassa, VA). LLC-PK1 was serum starved before being used for signaling experiments. AAC-19, Y260A and DU145, A4-7 cell lines were cultured in the presence of 0.5% and 1% serum respectively for 24 hours before being used for the signaling experiments. LLC-PK1 cells were transfected (Lipofectamine 2000) with pEYFP-C1 vector containing α1 CD2, α2 CD2 or pEYFP-C1 empty vector. After verifying YFP expression visually, the cells were selected with 1 mg/ml G418 for one week. G418 resistant clones were selected and expanded. Cells were then cultured without G418 for at least three generations before being used for experiments. CD2N and CD2C cells were generated in similar manner. Y260A cells were generated by transfecting PY-17 cells with pRC/CMV-α1 AACm1 vector harboring mutation at Y260 and selected with ouabain (3 μM). Ouabain-resistant clones were isolated and expanded into stable cell lines. The cells were cultured for at least 3 generations without ouabain before being used for any experiment. DU-P1, A4-7 and A4-3 cell lines were generated by transfecting DU145 cells with a α1 Na/K-ATPase-specific siRNA containing vector and selecting with Puromycin as described above. All constructs were verified by DNA sequencing.
Cell growth assay and MTT assay. Cell growth assay and MTT assay were performed as previously described 52,59 . Biochemical measurement of ATP and lactate. ATP measurements were performed using CellTiter-Glo Luminescent Cell Viability Assay kit. 10,000 cells per well were cultured in 96-well culture plate. After treatment with 2-DG at indicated concentrations in serum-free DMEM for 45 minutes, assay reagents were reconstituted and added into culture plate. Luminescent counts of the reactants were determined from an opaque-walled 96-well plate with a microplate reader. Lactate measurement was done by colorimetric methods as described by Barker 60 . In PP2 study, culture medium containing serum and 5 μM PP2 was replaced in culture dish and collected after 4 hours for measurements.

Bioenergetics. Properties of cellular bioenergetics were characterized using Seahorse XFp Extracellular Flux
Analyzer by measuring oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) following the guideline provided by the manufacture. Prior to the start of Seahorse assay, optimal FCCP concentration for each cell line was determined by titration studies. Unless indicated otherwise, 10,000 cells per well were seeded with culture medium. Bicarbonate-free medium (Agilent Technologies) with different substrates (10 mM glucose, 2 mM glutamine and 1 mM pyruvate for Cell Mito Stress Test assays; 2 mM glutamine for Glycolysis Stress Test assays) was replaced one hour prior to the assay. Baseline OCR and ECAR rates were measured three times before various inhibitors, stimulants, substrates, or compounds were added through the drug delivery ports. In PP2 study, cells were pretreated with 5 μM PP2 for 4 h and then analyzed.

H ouabain binding assay and ATPase Activity Assay. H-Ouabain Binding Assay and ATPase Activity
Assay were performed as previously described 59 .
Cell-attachment induced integrin signaling. Cell-attachment induced integrin signaling were done as previously described 15 . RNAseq Analysis. RNAseq analysis was performed by BGI Tech, China. Briefly, RNA was extracted from cell lysates, treated with DNase 1 and magnetic Oligod(T) beads were used to isolate mRNA. mRNAs were then fragmented with fragmentation buffer and cDNAs were synthesized using them as templates. After agarose gel electrophoresis, suitable fragments were selected for PCR amplification as templates. During QC steps, Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System were used for quantification and qualification of the sample library. The library was sequenced using HiSeqTM 2000 sequencer. Bioinformatic analysis was performed by deep analysis of gene expression. RNA Extraction, cDNA Synthesis and Quantitative PCR. Total RNA was isolated with the QIAGEN RNeasy Mini Kit. The same amount of total RNA was used for synthesizing first-stand cDNA with the SuperScript III First-Stand Synthesis SuperMix for qRT-PCR (ThermoFisher). The cDNA from each sample was used as a template for the quantitative PCR (Syber green) with Roche LightCycler ® 480 Real-Time PCR System. All primers were synthesized by Integrated DNA Technologies (IDT). β-actin was used as internal control.
Statistical analysis. Data were recorded as mean +/− SEM (Standard Error of Mean). Student's T-test were used to measure differences between two individual groups and one-way analysis of variance (ANOVA) was used to measure differences between more than two groups. Two-way ANOVA was used to measure between more than two groups while each group contained more than one variables. One-way ANOVA (Bartlett's test) or paired T-test were used to measure differences in the tissue array data, where appropriate. Paired T-test followed by Wilcoxon signed rank test were used to analyze gene expression data from TCGA database. Survival analysis was measured by log-rank survival test. p value less than 0.05 was considered as significant.
Data availability. The authors confirm all relevant data and figures are included in the paper and/or its supplementary information files.