Potential conflict of interest: Dr. Mills consults for and received grants from Astra Zeneca. He consults for and owns stock in Catena Pharmaceuticals. He also consults for Critical Outcome Technologies, Daiichi, Targeted Molecular Diagnotics, Foundation Medicine, Han AlBio Korea, Komen Foundation, Novartis, Symphogen, and Tau Therapeutics. He owns stock in PTV Ventures and Spindle Top Ventures. He received grants from Celgene, CeMines, Exelixis/Sanofi, GlaxoSmithKline, Roche, and Wyether/Pfizer/Puma.
This research is supported, in part, by the 2011 and 2012 cycle of MD Anderson Sister Institute Network Fund (to J-S.L.), 5U54 CA112970-08 (to G.B.M.), 5P01CA099031-07 (to G.B.M.), P30 CA016672 (to G.B.M.), and CA016672 (MD Anderson Cancer Center Support Grant) from the National Institutes of Health.
Metabolic changes are common features of many cancer cells and are frequently associated with the clinical outcome of patients with various cancers, including hepatocellular carcinoma (HCC). Thus, aberrant metabolic pathways in cancer cells are attractive targets for cancer therapy. However, our understanding of cancer-specific regulatory mechanisms of cell metabolism is still very limited. We found that Tat-activating regulatory DNA-binding protein (TARDBP) is a novel regulator of glycolysis in HCC cells. TARDBP regulates expression of the platelet isoform of phosphofructokinase (PFKP), the rate-limiting enzyme of glycolysis that catalyzes the irreversible conversion of fructose-6-phosphate to fructose-1,6-bisphosphate. Silencing of TARDBP expression in multiple HCC cell lines leads to impaired glucose metabolism and inhibition of in vitro and in vivo growth of HCC cells. Notably, the microRNA 520 (miR-520) family is an intermediate regulator of TARDBP-mediated regulation of glycolysis. Mechanistically, TARDBP suppressed expression of the miR-520 family, which, in turn, inhibited expression of PFKP. We further showed that expression of TARDBP is significantly associated with the overall survival of patients with HCC. Conclusion: Our study provides new mechanistic insights into the regulation of glycolysis in HCC cells and reveals TARDBP as a potential therapeutic target for HCC. (HEPATOLOGY 2013;)
TARDBP was identified first as a transcription factor that binds to the human immunodeficiency virus transactivation response region1 and later as an RNA-binding protein linked to neurodegenerative diseases, such as frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS).2-4TARDBP is one of the frequently mutated genes in sporadic and familial ALS, as well as in patients with FTLD, providing evidence of a direct link between TARDBP abnormalities and neurodegeneration.4 Although roles of TARDBP have been extensively studied in the motor neuron linking to FTLD and ALS, recent reports suggested that TARDBP might play important roles in cellular metabolisms, including glucose metabolism and lipid metabolism.5, 6 In addition, recent studies also suggested functional roles of TARDBP in human cancer.7-9TARDBP expression is significantly altered in leukemia, and TARDBP is significantly associated with susceptibility to Ewing's sarcoma.8, 9 However, although the link of TARDBP in human diseases has been confirmed by numerous reports, it is not clear how TARDBP contributes to diseases because very little is known about the molecular functions of TARDBP, except for its roles in RNA metabolism.10
Most cancer cells, including hepatocellular carcinoma (HCC) cells, have very high demand for cellular metabolism to meet the need for new building blocks and energy required for cell growth.11-13 In particular, oncogenic transformation of cells is frequently associated with an increase in glycolytic flux, mainly caused by increased expression of glycolysis-regulating genes. MYC and HIF1A are the best-known transcriptional regulators controlling expression of glycolysis genes, such as LDHA, HK2, PDK1, and GLUT1, whose expression levels are highly elevated in cancer cells.14, 15 However, because glycolysis is highly facilitated in cancer cells, more transcriptional regulators that actively promote glycolysis are expected to be involved.
In this study, we demonstrated that expression of TARDBP is significantly elevated in HCC and that it regulates the expression of PFKP, the rate-limiting enzyme for glycolysis, through negative regulation of microRNA 520s (miR-520s). Thus, our study provides evidence that TARDBP is a novel transcriptional regulator of glycolysis in cancer and a potential therapeutic target for treatment of patients with HCC.
Ab, antibody; ALS, amyotrophic lateral sclerosis; ATP, adenosine triphosphate; AUC, area under the curve; ChIP, chromatin IP; CI, confidence interval; DOPC, dioleoyl phosphatidylcholine; FTLD, frontotemporal lobar degeneration; HCC, hepatocellular carcinoma; IP, immunoprecipitation; IV, intravenously; miR-520, microRNA 520; miRNAs, microRNAs; mRNA, messenger RNA; OS, overall survival; PFK, phosphofructokinase; PFKP, the platelet isoform of PFK; qRT-PCR, quantitative real-time polymerase chain reaction; ROC, receiver operating characteristic; shRNA, short hairpin RNA; siRNA, small interfering RNA; TARDBP, Tat-activating regulatory DNA-binding protein; UTR, untranslated region.
Materials and Methods
Additional Materials and Methods information can be found in the Supporting Materials.
Short Hairpin RNA, Small Interfering RNA, and MicroRNA Mimics.
shTARDBP (#38-TRCN0000016038 and #40-TRCN0000016040) and shControl (SHC002) clones were purchased from Sigma-Aldrich (St. Louis, MO). Cell lines of interest were then infected with virus particles from the 293 cells. Virus-infected cells were selected for 3 days with 2 μg/mL of puromycin. siLuc, siTARDBP (SMARTpool),16 and microRNA (miRNA) mimics were purchased from Dharmacon, Inc. (Lafayette, CO). Cells were transfected with indicated small interfering (siRNA) or mimic miRNA using Oligofectamine (Invitrogen, Carlsbad, CA) for 72 hours and used for assays.
Detection of Glucose Uptake, Lactate Levels, and Cellular Adenosine Triphosphate Levels.
Glucose uptake and lactate levels were measured using a Multiparameter Bioanalytical System (#YSI 7100; YSI Life Sciences, Inc., Yellow Springs, OH). After SK-Hep1 cells were transduced with short hairpin RNA (shRNA) viruses, the media from cultured cells were used for measuring glucose or lactate levels. Amounts of glucose taken up by cancer cells were measured by subtraction of remaining glucose level in cultured media from total glucose level in uncultured media. Cellular adenosine triphosphate (ATP) level was determined using an ATP Colorimetric Assay Kit (#K354-100; BioVision, Inc., Milpitas, CA) and normalized to cell number.
Female athymic nude mice (NCr-nu) were purchased from the NCI-Frederick Cancer Research and Development Center (Frederick, MD) and maintained as previously described.17 Tumor implantation, siRNA incorporation into dioleoyl phosphatidylcholine (DOPC) nanoliposomes, and delivery in vivo were carried out as previously described.17, 18 For TARDBP silencing experiments, mice were randomized into one of the following treatment groups (n = 10 per group): control siRNA-DOPC (150 μg/kg intravenously [IV] twice-weekly) and TARDBP siRNA-DOPC (150 μg/kg IV twice-weekly). The control siRNA sequence used was UUCUCCGAACGUGUCACGU [dT][dT], and TARDBP siRNA sequences were the same as those used for the cell-line experiments. Human HCC cells (SK-Hep1) were subcutaneously injected into mice (1.0 × 106 cells/animal) on day 0, and siRNA treatment was started on day 7. Five weeks later, mice were euthanized and subjected to necropsy, and tumors were harvested.
TARDBP Expression Is Elevated in HCC.
Because recent studies suggested a potential link of TARDBP to cancer, we first examined the expression level of TARDBP in human liver tissues. Expression of TARDBP was significantly higher in tumors than in normal liver tissues surrounding the tumors (P = 1.0 × 10−14 by Student t test, Fig. 1A), indicating potential roles of TARDBP in HCC. Consistent with the gene-expression data from patient tissues, expression of TARDBP was detected in all HCC cell lines examined (Fig. 1B).
We next depleted expression of TARDBP with specific siRNAs to TARDBP to test whether TARDBP plays significant roles in the growth of HCC cells. Silencing of TARDBP expression with specific siRNAs significantly attenuated growth of SK-Hep1 and HUH7 cells (Fig. 1C and Supporting Fig. 1A), strongly suggesting that TARDBP is necessary for growth and survival of HCC cells. Consistent with cell growth assay, colony formation was also significantly reduced upon depletion of TARDBP with specific siRNAs (Fig. 1D). Similar levels of growth inhibition upon silencing of TARDBP expression were observed in additional HCC cells (SNU-449 and Hep3B) (Supporting Fig. 1A,B).
In agreement with previous reports,1, 2 cell fractionation showed that TARDBP is predominantly localized in the nucleus of SK-Hep1 cells (Fig. 1E and Supporting Fig. 1C), suggesting that its biological roles in cancer cell growth might be mediated by its roles as a transcription factor or regulator of RNA processing.
TARDBP Regulates Glycolysis in HCC Cells.
To investigate downstream targets of TARDBP that could regulate cell growth, we carried out microarray experiments after depleting TARDBP in SK-Hep1 cells (Fig. 2A). As expected, silencing of TARDBP expression led to down-regulation of genes involved in cell growth (i.e., CDK6, RANBP1, and CENPE). Surprisingly, a large number of the down-regulated genes are directly involved in glucose transport and glycolysis (i.e., SLC2A1, PFKP, PFKFB4, PGK1, and ENO2), strongly suggesting potential roles of TARDBP in regulating glucose metabolism. Notably, expression of PFKP, among many glycolysis-related genes, was most significantly altered by TARDBP (P = 7.5 × 10−5 by Student t test; 4.7-fold). Expression of PFKP and other glycolysis-related genes was also significantly down-regulated by depleting TARDBP in two additional HCC cell lines (FOCUS and HUH7) when their expression was assessed by quantitative real-time polymerase chain reaction (qRT-PCR) (Fig. 2B,C). These results strongly suggested conserved and universal roles of TARDBP in glucose metabolism in HCC cells, particularly through regulation of PFKP.
Phosphofructokinase (PFK) is a key regulatory enzyme in glycolysis that catalyzes the irreversible conversion of fructose-6-phosphate to fructose-1,6-bisphosphate. Humans have three PFK isoforms: liver (PFKL); muscle (PFKM); and platelet (PFKP).19, 20 Thus, we examined whether TARDBP regulates expression of the other PFK isoforms in addition to PFKP. Results showed that protein expression of PFKL and PFKM was not altered by silencing of TARDBP expression, whereas all three siRNAs specific to TARDBP successfully down-regulated expression of PFKP in SK-Hep1 cells (Fig. 2D). Consistent with western blotting experiments, qRT-PCR experiments showed that TARDBP regulates expression of only PFKP in SK-Hep1 cells (Fig. 2E).
Because TARBDP regulated expression of many glycolysis genes, including PFKP, in multiple HCC cells, we determined the effect of depletion of TARDBP in metabolic response. Glucose uptake of SK-Hep1 cells was significantly reduced by silencing of TARDBP expression (Fig. 3A,B). Furthermore, silencing of TARDBP expression resulted in a decrease in lactate production and ATP levels, indicating a decrease of glycolysis (Fig. 3B). Thus, our findings strongly support the proposed roles of TARDBP in HCC cell growth through regulation of glucose and energy metabolism.
miR-520s Down-Regulates PFKP Expression.
We next attempted to determine the molecular mechanism of how TARDBP regulates PFKP expression. Given that the best-known function of TARDBP is RNA processing as an RNA-binding protein,21 we examined whether TARDBP directly interacts with the messenger RNA (mRNA) of PFKP. However, analysis of RNA immunoprecipitation (IP) data with anti-TARDBP antibody (Ab)21 failed to demonstrate interaction of TARDBP with PFKP mRNA (Supporting Fig. 2), suggesting that TARDBP likely regulates PFKP by other mechanisms. Because TARDBP positively regulates expression of PFKP and also functions as a transcription repressor,22 we hypothesized that PFKP could be negatively regulated by intermediate regulators that are, in turn, directly suppressed by TARDBP. Recent studies showed that TARDBP is involved in regulation of miRNAs,23, 24 suggesting that miRNAs might be good candidates for intermediaries between TARDBP and PFKP. To identify such intermediary regulators, we explored target miRNAs that can suppress PFKP based on sequence alignment (Fig. 4A). Sequence analysis with the starBase database25 revealed that 26 miRNAs contain direct binding sequences for the PFKP 3′ untranslated region (UTR) (Supporting Table 1). Interestingly, three all-independent prediction programs (target Scan, picTarm and miRanda) predicted the miR-520 and miR-302 family as major regulatory miRNAs for PFKP.26 Because previous studies showed that miR-520b and miR-520e can inhibit cancer cell growth,27-29 we next tested whether inhibition of cell growth by miR-520 is mediated by regulation of PFKP expression. When SK-Hep1 cells were treated with miR-520a-3p, miR-520b, and miR-520e (hereafter miR-520a/b/e), expression of PFKP was significantly down-regulated (Fig. 4B), suggesting that PFKP might be a direct target of miR-520a/b/e. However, expression of other glycolysis genes were not significantly altered by miR-520a/b/e (Supporting Fig. 3), suggesting that these miRNAs regulate glycolysis mainly through inhibition of PFKP. To determine whether miR-520a/b/e directly targets the 3′ UTR of PFKP mRNA, we assessed a luciferase reporter vector containing the 3′ UTR sequence of PFKP, including the predicted binding site for miR-520a/b/e in SK-Hep1 cells. Luciferase activity was significantly inhibited by the PFKP 3′ UTR sequence when only miR-520a/b/e were cotransfected (Fig. 4C). However, luciferase activity was not inhibited by a mutant 3′ UTR sequence (Fig. 4D,E), strongly demonstrating that miR-520a/b/e directly targets the 3′ UTR sequence of PFKP mRNA and inhibits the expression of PFKP.
TARDBP Regulates PFKP by Suppressing miR-520s.
Because expression of PFKP was down-modulated by miR-520s, we next tested whether miR-520a/b/e are regulated by TARDBP by measuring expression of these miRNAs by qRT-PCR after silencing of TARDBP in SK-Hep1 and SNU449. Results showed that expression of miR-520a/b/e was significantly increased when TARDBP was silenced (Fig. 5A), with expression of miR-520b strongly induced by silencing TARDBP in two HCC cell lines. TARDBP is a DNA-binding protein that binds to the GTGTGT sequence in target promoter regions.22 Thus, we examined whether TARDBP can directly bind to the miR-520b promoter. Indeed, a chromatin IP (ChIP) assay demonstrated that TARDBP directly bound to the miR-520b promoter region, specifically in GT-rich regions (Fig. 5B), suggesting that miR-520b is indeed a direct downstream target of TARDBP.
Our results strongly suggested that growth inhibition after depletion of TARDBP might be mediated by miR-520a/b/e. To test this hypothesis, we introduced miR-520a/b/e into two HCC cell lines and observed that cell growth was significantly reduced (Supporting Fig. 4). To further test whether growth inhibition is mediated by down-regulation of PFKP, we introduced exogenous myc-tagged PFKP after silencing TARDBP in SK-Hep1 cells. Because myc-tagged PFKP lacks a 3′ UTR sequence, its expression is not inhibited by miR-520s (Fig. 5C). Growth inhibition by silencing TARDBP was rescued by exogenous PFKP (Fig. 5D). Furthermore, glucose uptake, lactate production, and ATP level were significantly increased by exogenous PFKP (Fig. 5E). When induced miR-520b in TARDBP-silenced SK-Hep1 was inhibited by specific antisense miR-520b inhibitor, expression of PFKP was recovered (Supporting Fig. 5), demonstrating that regulation of PFKP by TARDBP is mediated through miR-520s. Taken together, our data suggested that PFKP is the main regulatory target for TARDBP-miR-520s-mediated regulation of cell growth. These observations agree very well with the finding of previous studies showing that miR-520b and miR-520e might function as negative regulators of cell proliferation.27, 30
Clinical Relevance of TARDBP, PFKP, and miR-520s Expression in Cancer.
Our data suggested functional roles of TARDBP and PFKP as positive regulators and miR-520s as a negative regulator of cell proliferation. This view is strongly supported by expression patterns of these genes in the NCI-60 cancer cell lines. Expression of both TARDBP and PFKP was very high in the vast majority of the 60 cancer cell lines (Supporting Fig. 6A). In contrast, miR-520a/b/e were barely expressed, strongly supporting its role as a negative regulator of cell growth. Expression of PFKP was the highest among PFK isoforms in NCI-60 cell lines (Supporting Fig. 6B), further supporting that cancer-specific expression of PFKP is regulated by miR-520a/b/e and TARDBP.
We next assessed the clinical relevance of TARDBP in HCC. Expression of TARDBP is significantly associated with prognosis when estimated by receiver operating characteristic (ROC) analysis. Areas under the curve (AUCs) of TARDBP expression over 3-year overall survival (OS) were 0.6 (95% confidence interval [CI]: 0.53-0.66; P = 0.007) (Fig. 6A). When patients were stratified according to expression level of TARDBP, patients with high TARDBP expression showed significantly shorter survival (P = 3.8 × 10−4; Fig. 6B). Association of TARDBP with prognosis is further supported by its significant correlation with the 65-gene risk score (r = 0.5; P = 2.2 × 10−16) (Fig. 6C) that was previously developed for prediction of recurrence.31 Significant positive correlation between expression of TARDBP and PFKP in HCC patients is also concordant with their roles as positive regulators for cell growth (Fig. 6D).
Therapeutic Efficacy of PFKP in a Mouse Model.
The critical roles of TARDBP and its downstream targets, the miR-520 family, in cell growth and the significant correlation of TARDBP with patient survival strongly suggested that TARDBP and its downstream targets would be potential therapeutic targets for cancer treatment. To test this, we carried out a mouse xenograft experiment with SK-Hep1 cells and siRNA specific to TARDBP. Compared to treatment with control siRNA, treatment with siTARDBP resulted in a significant reduction in tumor weight (Fig. 7A), recapitulating the effects of silencing TARDBP in vitro. Efficient silencing of TARDBP by siRNA was confirmed by immunostaining of TARDBP and its downstream target, PFKP, and further validated by qRT-PCR (Fig. 7B,C and Supporting Fig. 7). As expected, cell proliferation, as examined by Ki67 immunostaining, was significantly decreased in tumors treated with siTARDBP (Fig. 7B). In addition, lactate and ATP levels were also significantly decreased (Fig. 7C) and expression of miR-520b and miR-520e (Fig. 7D) was significantly increased in siTARDBP-treated mice, compared to control. These results clearly demonstrate the importance of TARDBP in tumor growth and the potential of TARDBP as a therapeutic target.
In the current work, we have presented a mechanistic link from TARDBP to PFKP, the rate-limiting enzyme of glycolysis, and we also have provided evidence suggesting that this pathway is associated with poor prognosis of HCC. A notable finding was the identification of the miR-520 family as an intermediary regulator of this pathway.
Although TARDBP was originally identified as a transcription repressor binding to the human immunodeficiency virus transactivation response region,1 downstream targets and molecular mechanisms related to its transcription repressor activity have not been properly explored. Our data demonstrated that TARDBP functions as a transcription repressor and its molecular activity as a transcription repressor plays key roles in the regulation of glycolysis in cancer cells. TARDBP strongly represses expression of the miR-520 family as evidenced by a significant increase in their intracellular levels upon TARDBP depletion. This notion was strengthen by a ChIP assay demonstrating that TARDBP directly bound to GT-rich regions in the miR-520b promoter. In addition to regulation of miR-520s at the transcriptional level, TARDBP may regulate miR-520s post-transcriptionally because the involvement of TARDBP in miRNA processing has been observed in other systems.23, 24 Our data are in good agreement with previous studies demonstrating repression of the spermatid-specific gene, SP-10, expression by TARDBP.22, 32 Thus, our study rediscovered a previously recognized molecular function of TARDBP and updated molecular mechanisms and biological roles of TARDBP, especially in cellular metabolism with the use of miRNAs as intermediary regulators.
There is growing evidence supporting that miRNAs play important roles in regulation of cellular metabolism.33, 34 Recent studies revealed that miRNAs, such as miR-124, miR-137, miR-340, miR-143, and miR-155, regulate glycolysis by directly targeting the 3′ UTR region of HK2 and PKM2.35-37 With prediction analysis based on sequence, we discovered that miR-520a/b/e are major regulators of glycolysis by directly targeting the 3′ UTR of PFKP mRNA. We further demonstrated that TARDBP-mediated suppression of miR-520a/b/e is important for the growth and survival of HCC cells.
Importantly, analyses of gene-expression patterns from multiple cancer lineages provide evidences supporting our findings related to molecular functions and mechanisms of TARDBP-mediated PFKP regulation. Expression of TARDBP is significantly higher in tumors than normal tissues. We also have found an inverse correlation of expression patterns among TARDBP, PFKP, and miR-520s. Expression of TARDBP and PFKP is significantly high in the vast majority of cancer cell lines, whereas expression of miR-520s is very low (Supporting Fig. 4A). This is in good agreement with mechanism postulating that TARDBP suppresses the expression of miR-520s that directly inhibit PFKP to maintain increased glycolysis in cancer cells (Fig. 7E).
In conclusion, we found that novel roles of TARDBP are linked to glycolysis by PFKP in HCC. Deregulation of cellular metabolism is one of the hallmarks of cancer cells,38 and altered components of the metabolic pathway represent attractive therapeutic targets.13, 39 Thus, the identification of the TARDBP/miR-520/PFKP axis regulating glycolysis and ATP production elicits a potential new approach to target the tumor-specific metabolic pathway.