Article Text

Download PDFPDF

Original article
PTEN antagonises Tcl1/hnRNPK-mediated G6PD pre-mRNA splicing which contributes to hepatocarcinogenesis
  1. Xuehui Hong1,
  2. Ruipeng Song1,
  3. Huiwen Song2,
  4. Tongsen Zheng1,
  5. Jiabei Wang1,
  6. Yingjian Liang1,
  7. Shuyi Qi3,
  8. Zhaoyang Lu1,
  9. Xuan Song1,
  10. Hongchi Jiang1,
  11. Lianxin Liu1,
  12. Zhiyong Zhang4
  1. 1Key Laboratory of Hepatosplenic Surgery, Ministry of Education, Department of General Surgery, The First Affiliated Hospital of Harbin Medical University, Harbin, China
  2. 2Department of Cardiovascular Medicine and Department of Emergence, Baoan Hospital, Southern Medical University, ShenZhen, China
  3. 3Department of Gerontology, The First Affiliated Hospital of Harbin Medical University, Harbin, China
  4. 4Department of Surgery, Robert-Wood-Johnson Medical School University Hospital, University of Medicine & Dentistry of New Jersey, New Brunswick, New Jersey, USA
  1. Correspondence to Dr Zhiyong Zhang, Department of Surgery, Robert-Wood-Johnson Medical School University Hospital, Rutgers University, 125 Paterson Street, New Brunswick, NJ 08901, USA; zhiyongzhng3{at}gmail.com Dr Lianxin Liu, Key Laboratory of Hepatosplenic Surgery, Ministry of Education, Department of General Surgery, the First Affiliated Hospital of Harbin Medical University, #23 Youzheng Street, Harbin, Heilongjiang Province, 150001, China; liulx{at}ems.hrbmu.edu.cn Dr Hongchi Jiang, Key Laboratory of Hepatosplenic Surgery, Ministry of Education, Department of General Surgery, the First Affiliated Hospital of Harbin Medical University, #23 Youzheng Street, Harbin, Heilongjiang Province, 150001, China; jianghongchi01{at}163.com

Abstract

Background Mounting epidemiological evidence supports a role for phosphatase and tensin homologue (PTEN)-T cell leukaemia 1 (Tcl1) signalling deregulation in hepatocarcinogenesis.

Objective To determine the molecular and biochemical mechanisms by which the PTEN/Tcl1 axis regulates the pentose phosphate pathway (PPP) in hepatocellular carcinoma (HCC).

Methods We compared levels of PTEN and glucose-6-phosphate dehydrogenase (G6PD) mRNA in human HCC and healthy liver tissue. We measured PPP flux, glucose consumption, lactate production, nicotinamide adenine dinucleotide phosphate (NADPH) levels and lipid accumulation. We investigated the PTEN/Tcl1 axis using molecular biology, biochemistry and mass spectrometry analysis. We assessed proliferation, apoptosis and senescence in cultured cells, and tumour formation in mice.

Results We showed that PTEN inhibited the PPP pathway in human liver tumours. Through the PPP, PTEN suppressed glucose consumption and biosynthesis. Mechanistically, the PTEN protein bound to G6PD, the first and rate-limiting enzyme of the PPP and prevented the formation of the active G6PD dimer. Tcl1, a coactivator for Akt, reversed the effects of PTEN on biosynthesis. Tcl1 promoted G6PD activity and also increased G6PD pre-mRNA splicing and protein expression in a heterogeneous nuclear ribonucleoprotein (hnRNPK)-dependent manner. PTEN also formed a complex with hnRNPK, which inhibited G6PD pre-mRNA splicing. Moreover, PTEN inactivated Tcl1 via glycogen synthase kinase-3β (GSK3β)-mediated phosphorylation. Importantly, Tcl1 knockdown enhanced the sensitivity of HCC to sorafenib, whereas G6PD knockdown inhibited hepatocarcinogenesis.

Conclusions These results establish the counteraction between PTEN and Tcl1 as a key mechanism that regulates the PPP and suggest that targeting the PTEN/Tcl1/hnRNPK/G6PD axis could open up possibilities for therapeutic intervention and improve the prognosis of patients with HCC.

  • LIVER METABOLISM
  • HEPATOCELLULAR CARCINOMA
  • MOLECULAR MECHANISMS
  • MOLECULAR BIOLOGY
  • GLUCOSE METABOLISM
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known about this subject?

  • Phosphatase and tensin homologue (PTEN)-signalling is deregulated in hepatocarcinogenesis. Additionally, PTEN inhibits glycolysis in cancer.

  • Tcl1 is a coactivator for Akt and its expression is increased in liver cancer stem cells.

  • Heterogeneous nuclear ribonucleoprotein (hnRNPK) and glucose-6-phosphate dehydrogenase (G6PD) expression are increased in hepatocellular carcinoma (HCC). Additionally, hnRNPK is involved in G6PD pre-mRNA splicing.

What are the new findings?

  • PTEN and Tcl1 oppositely regulate the pentose phosphate pathway (PPP) and biosynthesis in human liver tumours.

  • Both PTEN and Tcl1 can bind to hnRNPK, but oppositely regulate G6PD pre-mRNA splicing. Furthermore, PTEN interacts with G6PD and inhibits its activity by lowering the formation of a dimeric G6PD holoenzyme. However, Tcl1 promotes G6PD activity.

  • PTEN inactivates Tcl1 via GSK3β-mediated phosphorylation, which disrupts interaction between Tcl1 and hnRNPK.

  • Tcl1 knockdown enhances the sensitivity of HCC to sorafenib, whereas G6PD knockdown inhibits hepatocarcinogenesis.

How might it impact on clinical practice in the foreseeable future?

  • G6PD and Tcl1 predict clinical outcomes of treatment for patients with HCC.

  • Our data suggest that the PTEN/Tcl1/hnRNPK/G6PD axis might serve as an effective new therapeutic target.

Introduction

Hepatocellular carcinoma (HCC) is the third most common cause of cancer-related death worldwide and its incidence is increasing. Improved understanding of the metabolism during hepatocarcinogenesis would help to optimise prevention, early diagnosis and treatment of HCC. Common characteristics of cancer cells are that they maintain high rates of glycolysis and convert a majority of glucose into lactic acid even in the presence of oxygen.1 Additionally, cancer cells have been previously reported to produce substantial alterations in several energy metabolism pathways, including glucose transport, oxidative phosphorylation and the pentose phosphate pathway (PPP).2

The PPP is important for both glucose catabolism and biosynthesis.3 During glycolysis, glycolytic intermediates can be diverted into PPP, which contributes to macromolecular biosynthesis by producing reduced nicotinamide adenine dinucleotide phosphate (NADPH). NADPH also functions as a crucial antioxidant, quenching the reactive oxygen species produced during rapid proliferation of cancer cells.4 Thus, tumour cells appear to coordinate glycolysis and anabolism to provide an overall metabolic advantage to cancer cell proliferation and disease development. However, the detailed mechanisms underlying this coordination remain largely unknown.

Oncogenes and tumour suppressors have been linked to the regulation of glucose and energy metabolism. The dysregulation of glucose transport and energy metabolism pathways by oncogenes and tumour suppressors has been proposed as an important biomarker for cancer detection and as a valuable target for the development of new anticancer treatments.5 In many cancers, including HCC, phosphatase and tensin homologue (PTEN) tumour suppressor activity is decreased compared with that in the normal tissues. Recent studies indicated that PTEN also has a role in modulating metabolism, including glycolysis and oxidative phosphorylation.6 However, the mechanism by which PTEN regulates the PPP and biosynthesis in HCC is less well understood.

Additionally, it has been shown that the proto-oncogene, T-cell leukaemia 1 (Tcl1) expression can mediate chemoresistance in HCC stem cells.7 Besides acting as a positive modulator of Akt activity, Tcl1, which was initially identified in leukaemia, can also enhance NF-κB activity.8 In chronic lymphocytic leukaemia, higher Tcl1 protein levels correlate with more aggressive clinical features. Yet the involvement of Tcl1 in the PPP and biosynthesis in HCC is unknown.

In this study, we sought to elucidate the role of the PTEN/Tcl1 axis in the PPP and anabolic biosynthesis, as well as the mechanism by which the PTEN/Tcl1 axis regulates glucose-6-phosphate dehydrogenase (G6PD) pre-mRNA splicing and activity, which contribute to liver cancer progression.

Materials and methods

Site-directed mutagenesis of Tcl1

The Tcl1 point mutations were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, California, USA), as described previously.9 The following primers were used for mutant constructs: S86A (S1A) forward, 5′-CTGATGGACGATACCGAGCCTCAGACTCCAGTTTCTG-3′; and reverse, 5′-CAGCTGGAGTCTGAGGCTCGGTATCGTCCATCAG-3′. S86/90A (S2A) forward, 5′-CGATACCGAGCCTCAGACTCCGCTTTCTGGCGC-3′; and reverse, 5′-GCGCCAGAAAGCGGAGTCTGAGGCTCGGTATCG-3′. S86E (S1E) forward, 5′-CTGATGGACGATACCGAGAGTCAGACTCCAGTTTCTG-3′; and reverse, 5′-CAGCTGGAGTCTGAGGCTCGGTATCGTCCATCAG-3′.

All point mutants were sequenced to check for the accuracy of the mutation.

Animal experiments

All mice were obtained from the laboratory animal centre of the Chinese Academy of Sciences, Shanghai. The experimental protocol was reviewed and approved by the committee on the use of live animals in teaching and research of Harbin Medical University, Harbin, China (SYSK 2010-012). Tumour size was measured twice weekly using a digital caliper. Tumour volume was calculated using the formula: length × (width)2×0.52. Male BALB/c (5–6 weeks old) mice (n=10/group) were inoculated subcutaneously in the flank with 2×106 Huh7 or Huh7-sh-G6PD cells suspended in phosphate-buffered saline (PBS).

The orthotopic HCC mouse model was established as previously described.10 Huh-7-luciferase-transfected (Luci-Huh-7) cells transfected with Huh7-control, Huh7-sh-G6PD, Huh-sh-heterogeneous nuclear ribonucleoprotein (hnRNPK), Huh-sh-Tcl1, Huh-sh-Tcl1/hnRNPK, wild-type Tcl1 or Tcl1-S1E mutant were harvested from subconfluent cultures and washed once in serum-free medium and resuspended in PBS. Suspensions consisting of single cells only, with >90% viability, were used for the injections. Animals were anaesthetised with 2.5% isoflurane/air mixture and injected intraperitoneally with 40 mg/mL D-luciferin potassium salt dissolved in PBS, according to the manufacturer's protocol. After 8–12 min, the animals were placed in the machine laterally and a digital image was taken through the acquisition of pseudocolour images, which stand for the dimensional distribution of detectable photons from active luciferase generated spontaneously by the animals. After 1 week of orthotopic implantation, mice (n=10/group) were given sorafenib once daily (orally, 10 mg/kg on days 1–5 of each week). PTEN+/+ and PTEN−/− mice (wild-type and PTENdel/del mice, c; 129S4-PTEN m1Hwu/J) were from the Jackson Laboratory.

For details of additional experimental materials and procedures, please see the online supplementary materials and methods.

Results

Inverse correlation between PTEN and G6PD expression in patients with HCC

As a first step to probe the relationship between PTEN and biosynthesis in HCC, we conducted a G6PD expression analysis based on quantitative reverse-transcription PCR (qPCR) data derived from human HCC cohorts (see online supplementary table S1). We focused on G6PD because G6PD catalyses the first and rate-limiting step of the PPP pathway, which produces NADPH to fuel macromolecular biosynthesis (see online supplementary figure S1). Dysregulation of PTEN and G6PD were evident and well correlated with the pathological grade in overt HCC: G6PD levels were upregulated, particularly in advanced HCC cases (p<0.001, figure 1B); and PTEN levels were downregulated in early and advanced HCC (p<0.001, figure 1A). These data suggested an inverse correlation between PTEN and G6PD in patients with HCC, which was further strengthened by Pearson analysis (Pearson r=−0.9411, p<0.0001, figure 1C) and encouraged us to investigate whether PTEN regulates the PPP pathway.

Figure 1

The inverse relationship between phosphatase and tensin homologue (PTEN) and glucose-6-phosphate dehydrogenase (G6PD) expression levels in patients with hepatocellular carcinoma (HCC). (A and B) Dot plot of the expression of PTEN and G6PD in patients with HCC by real-time PCR. The results are expressed relative to glyceraldehyde-3-phosphate dehydrogenase. (C) Scatterplot depicts relative PTEN and G6PD mRNA levels by qRT-PCR in tumours of the same cohort of patients with HCC as in (A and B). Pearson correlation coefficient was used to measure of association. Statistical significance was concluded at ***p<0.001.

PTEN deficiency promotes PPP flux, glucose consumption and lactate production

To investigate whether PTEN regulates the PPP, we compared the oxidative PPP flux in PTEN+/+ and PTEN−/− stable HCC and HCT116 cells. Cells were cultured in medium containing [2-13C]glucose and the glucose metabolites were measured. As shown in figure 2A, the absence of PTEN resulted in a significant increase (∼60%) in oxidative PPP flux, indicating that PTEN suppressed the PPP. The deficiency of PTEN concomitantly led to a strong enhancement in glucose consumption (figure 2B). Inhibition of G6PD using either small interfering RNA (siRNA) or 6-aminonicotinamide (6-AN)), a G6PD inhibitor, reduced glucose consumption of PTEN depleted cells to levels similar to that of the PTEN +/+ cells, which seems to be unaffected by the absence of PTEN, (figure 2B and online supplementary figure S2A). These results showed that PTEN deficiency promoted glucose consumption mainly through an increased PPP flux. Additionally, the deficiency of PTEN also increased lactate production (figure 2C and see online supplementary figure S2B). However, inhibition of G6PD in these cells increased lactate production, regardless of PTEN status. Therefore, glucose flux through the PPP may itself lower lactate production. The inhibition of lactate production may be related to the ability of PTEN to decrease glycolysis or increase oxidative phosphorylation.6

Figure 2

Phosphatase and tensin homologue (PTEN) inhibits the increases in pentose phosphate pathway (PPP) flux, glucose consumption and lactate production, nicotinamide adenine dinucleotide phosphate (NADPH) levels and lipid accumulation. (A) PTEN+/+ and PTEN−/− hepatocellular carcinoma (HCC) and HCT116 cells were cultured in medium containing [2-13C]glucose. Oxidative PPP flux (top) was measured based on the rate of glucose consumption and the ratio of 13C incorporated into carbon 2 (indicating glycolysis) and carbon 3 (indicating PPP) of lactate by NMR spectroscopy. Protein expression (bottom) was analysed. (B) PTEN+/+ and PTEN−/− HCC and HCT116 cells were treated with glucose-6-phosphate dehydrogenase (G6PD) small interfering RNA (siRNA) and control siRNA. Top panel: glucose consumption. Bottom panel: the expression of PTEN, G6PD and actin. (C) Lactate levels and NADPH levels (D) in cells from B. (E) PTEN+/+ and PTEN−/− HCC and HCT116 cells treated with or without 6-aminonicotinamide (6-AN) were cultured in the presence of insulin, isobutylmethylxanthine, dexamethasone and rosiglitazone. Lipid contents were examined by Oil Red O staining and quantified by absorbance at 500 nm. All data are means±SD (n=3 for each panel). Statistical significance was concluded at *p<0.05, **p<0.01, ***p<0.001 and # represents no statistical significance.

PTEN modulates NADPH levels and lipid accumulation

NADPH is mainly generated in the oxidative branch of the PPP; therefore, employment of the PPP is an interesting option for NADPH-dependent processes. PTEN deficiency led to a marked increase of NADPH levels in HCC and HCT116 cells (figure 2D). Treatment with G6PD siRNA minimised the difference in NADPH levels between PTEN+/+ and PTEN−/− cells. To validate the cell culture findings in vivo, we compared the NADPH levels in liver and spleen tissues from PTEN−/− and PTEN+/+ mice. The liver tissues from PTEN−/− mice exhibited substantially increased NADPH levels, compared with those in the corresponding tissues from PTEN+/+ mice (see online supplementary figure S2C). However, in spleen tissue, the activity of G6PD was very low. In contrast to PTEN knockdown, overexpression of PTEN led to a significant decrease in NADPH levels (data not shown).

NADPH is required for the lipid biosynthesis. To examine the effect of PTEN on lipid accumulation, we treated PTEN+/+ and PTEN−/− HCC and HCT116 cells with a combination of insulin, isobutylmethylxanthine, dexamethasone and rosiglitazone or with high glucose (25 mM) concentrations, which stimulated lipogenesis.11 ,12 We observed enhanced lipid levels in the PTEN−/− cells compared with PTEN+/+ HCC and HCT116 cells (figure 2E and online supplementary figure S2D). Such a huge difference in lipid accumulation, however, was eliminated upon 6-AN treatment or G6PD silencing in both PTEN+/+ and PTEN−/− cells (figure 2E and online supplementary figure S2E). Taken together, these results showed that PTEN inhibited NADPH production and lipid accumulation by decreasing the glucose flux through the PPP.

We also investigated the bioenergetic state of mitochondria in PTEN−/− and PTEN +/+ HCC and HCT116 cells: oxygen consumption rate and ATP production. As shown in online supplementary figure S3, a significant decline in oxygen consumption was seen in PTEN−/− cells, which was accompanied by reduced mitochondrial ATP production.

PTEN interacts with G6PD and decreases formation of the dimeric G6PD holoenzyme

To elucidate the mechanism by which PTEN regulates the PPP, we analysed the activity of G6PD. A deficiency of PTEN greatly increased G6PD activity in HCC cell lines while G6PD siRNA or 6-AN reduced this effect (figure 3A,B). Conversely, overexpression of PTEN in the PTEN-deficient HCC and HCT116 cells markedly decreased G6PD activity (see online supplementary figure S4).

Figure 3

Phosphatase and tensin homologue (PTEN) interacts with glucose-6-phosphate dehydrogenase (G6PD) and inhibits its activity by lowering the formation of dimeric G6PD holoenzyme. (A) G6PD activity in PTEN+/+ and PTEN−/− hepatocellular carcinoma (HCC) and HCT116 cells treated with or without 6-aminonicotinamide (6-AN). (B) G6PD activity in PTEN+/+ and PTEN−/− HCC and HCT116 cells with G6PD small interfering RNA (siRNA) or control siRNA (means±SD, n=3). (C) HepG2 cells were transfected with enhanced green fluorescent protein (eGFP)–G6PD alone or together with increasing amounts of Flag–PTEN. Lysates were immunoprecipitated with anti-Flag antibody. Immunoprecipitated proteins and input were analysed by Western blotting. (D) PTEN+/+ HCC and HCT116 cells lysates were incubated with anti-G6PD antibody or IgG. Immunoprecipitates (IP) and input were analysed by western blotting. (E) Extracts of PTEN+/+ and PTEN−/− HepG2 and Huh7 cells were treated with and without 5 mM disuccinimidyl suberate (DSS) and analysed by western blotting. The positions of various forms of G6PD are indicated. (F) HepG2 cells were transfected with Flag–G6PD, eGFP–G6PD and different amounts of haemagglutinin–PTEN. Cell lysates were incubated with anti-Flag antibody. Input and immunoprecipitates were analysed by western blotting. Results of one of three independent experiments is shown. Statistical significance was concluded at *p<0.05, **p<0.01, ***p<0.001 and # represents no statistical significance.

We next examined whether PTEN interacts with G6PD. Flag-tagged PTEN specifically interacted with enhanced green fluorescent protein (eGFP)–G6PD in vivo (figure 3C). Similarly, endogenous PTEN associated with endogenous G6PD (figure 3D).

We next examined how physical interaction between PTEN and G6PD is involved in PTEN-mediated inhibition of G6PD activity. Cellular G6PD is mostly in equilibrium with inactive monomer and active dimer. In HCC cells, the deficiency of PTEN caused a strong increase in G6PD dimers and a corresponding decrease in G6PD monomers (figure 3E). In a transfection assay, PTEN reduced the interaction of two differentially tagged G6PD proteins (Flag–G6PD and eGFP–G6PD) in a dose-dependent manner (figure 3F). These results suggested that PTEN may suppress the formation of the dimeric G6PD holoenzyme.

Tcl1 reverses the effects of PTEN on biosynthesis

Oncogenes and tumour suppressors have been linked to the regulation of glucose and energy metabolism. Next, we wanted to determine which oncogene can reverse the effects of PTEN on biosynthesis in HCC and HCT116 cells. Recently, it has been shown that Tcl1 family proteins augment Akt activity. Moreover, Tcl1 promotes Akt dimerisation. And it is well known that PTEN inactivates Akt. Accordingly, we hypothesised that Tcl1 might be one of the candidates which acts antagonistically to PTEN. First, from immunohistochemistry staining and western blots, we observed that Tcl1 is indeed expressed in HCC and HCC cell lines, but not in normal liver tissue (figure 4A). To clarify whether Tcl1 modulated the PPP, we knocked down Tcl1 in PTEN+/+ and PTEN−/− HCC and HCT116 cells. We found that Tcl1 knockdown in PTEN-deficient cells almost completely restored biosynthesis activity to levels similar to PTEN+/+ cells (figure 4B–E). Unexpectedly, we observed that overexpression of Tcl1 substantially increased G6PD protein levels and also G6PD mRNA in HCC and HCT116 cells (figure 4F). It is well shown that regulation of G6PD expression is unique because expression of G6PD is regulated by changes in the rate of pre-mRNA splicing.13 However, Tcl1 has not any known function in pre-mRNA splicing. We reasoned that Tcl1 might interact with other proteins which regulated G6PD pre-mRNA splicing.

Figure 4

Tcl1 reverses the effects of phosphatase and tensin homologue (PTEN) on biosynthesis. (A) Representative immunohistochemistry (IHC) staining of Tcl1 in normal human liver tissue and hepatocellular carcinoma (HCC), ×200. Western blotting analysis showed a significant increase of Tcl1 in 10 randomly selected HCC cases (cohort 1, n=110) and human HCC cell lines, N, surrounding non-tumour liver tissue; T, tumour tissue. (B) PTEN+/+ and PTEN−/− HCC and HCT116 cells were treated with Tcl1 small interfering RNA (siRNA) or control siRNA. Glucose consumption was analysed as described above. (C) Lactate levels and nicotinamide adenine dinucleotide phosphate (NADPH) levels (D) in cells from (B). (E) Glucose-6-phosphate dehydrogenase (G6PD) activity in HCC and HCT116 cells stably expressing PTEN small hairpin RNA (shRNA) or control shRNA transfected with Tcl1 siRNA or control siRNA. Tcl1 overexpression in HCC and HCT116 cells increases G6PD protein levels (F, lower panel) and mRNA level (upper panel, F). Data are means±SD (n=3). Statistical significance was concluded at *p<0.05, **p<0.01, ***p<0.001 and # represents no statistical significance.

Tcl1 interacts with hnRNPK and enhances G6PD pre-mRNA splicing

To identify Tcl1-interacting proteins, we used a Tcl1 cDNA with a haemagglutinin (HA) tag (HA-Tcl1). HepG2 and Huh7 cells were transfected with HA-Tcl1 and treated with dithiols, which cross-link and fix protein complexes in vivo. Tcl1-HA protein, together with candidate protein partners, was isolated and identified by mass spectrometry. Among the identified candidate partners (see online supplementary table S2), we focused on hnRNPK because only hnRNPK has a known role in the regulation of G6PD pre-mRNA splicing.14

The potential interaction between Tcl1 and hnRNPK was investigated. HCC and HCT116 cells were stably transfected with a Flag-hnRNPK construct and a recombinant Adeno-Tcl1. Lysates were immunoprecipitated with hnRNPK antibody. As shown in figure 5A, Tcl1 interacts with hnRNPK (figure 5A). A similar observation was reproducible in lysates immunoprecipitated with anti-Tcl1 and probed with both hnRNPK and Tcl1 antibodies (figure 5B). A glutathione S-transferase (GST) pull-down assay was also used to test direct interaction between recombinant preparations of hnRNPK and GST-Tcl1 in vitro. As shown in figure 5C, Tcl1 directly interacted with hnRNPK in vitro.

Figure 5

Tcl1 interacts with heterogeneous nuclear ribonucleoprotein (hnRNPK) and promotes glucose-6-phosphate dehydrogenase (G6PD) expression. (A) Hepatocellular carcinoma (HCC) and HCT116 cells were transfected with Flag-hnRNPK and then infected with Ad-TCL1. Lysates were immunoprecipitated with anti-hnRNPK and analysed by western blotting. (B) Lysates in (A) were immunoprecipitated (IP) with anti-Tcl1 and analysed by western blotting. (C) HepG2 cells were cotransfected with glutathione S-transferase (GST)-Tcl1 and Flag-hnRNPK. Lysates were GST-pulled down and immunoblotted (IB) with anti-Tcl1 and anti-hnRNPK. (D) The structure of the G6PD and oligonucleotide sequence used in the pull-down experiments. The sequence from nt 35 to 79 of exon 12 is shown diagrammatically. Oligonucleotide 50–72 nt was used in pull-down assays. (E) 50–72 nt or the non-specific (NS) oligonucleotide was covalently linked to adipic acid beads and incubated with nuclear extracts (150 μg of protein) from HepG2 cells overexpressing Tcl1. The proteins were eluted and analysed using the indicated antibodies. Data are representative of three independent experiments.

Next, we asked how interaction between Tcl1 and hnRNPK affected G6PD pre-mRNA splicing. It has been demonstrated that a cis-acting RNA element localised to nucleotides (nt) 43–72 of exon 12 in the G6PD mRNA was involved in the regulation of the G6PD pre-mRNA splicing (figure 5D). HnRNPK can bind to this cis-acting RNA element and inhibited G6PD pre-mRNA splicing.14 We designed one RNA oligonucleotide (nt 50–72) of exon 12, bound to adipic acid beads and mixed with hepatocyte nuclear extract protein to pull down hnRNPK. The more hnRNPK binds to exon 12, the less G6PD pre-mRNA splices. Based on this principle, we tested whether overexpression of Tcl1 regulates the binding of hnRNPK to G6PD exon 12. We found that little or no protein bound either to the beads alone, or to the non-specific oligonucleotide (NS, figure 5E), whereas Tcl1 decreased the binding of hnRNPK to nt 50–72 by 88±7%, for three separate pull-down assays. However, the amount of hnRNPK protein in the total nuclear extracts (figure 5E, input) did not vary with Tcl1 overexpression. Thus, Tcl1 restrains the binding of hnRNPK to G6PD.

PTEN binds to hnRNPK and inhibits G6PD pre-mRNA splicing

Previous analysis indicated that endogenous hnRNPC and PTEN interacted and co-localised within the nucleus, suggesting a potential role for PTEN in RNA regulation.15 We wondered whether PTEN also bound to hnRNPK. HCC and HCT116 cells were transfected with a Flag-hnRNPK construct and a recombinant Flag-PTEN. Lysates were immunoprecipitated with hnRNPK antibody. As shown in figure 6A, PTEN interacted with hnRNPK (figure 6A). The same complexes were detected when lysates were immunoprecipitated with anti-PTEN (figure 6B). GST pull-down assay was also used to test direct interaction between recombinant preparations of hnRNPK and GST-PTEN in vitro. We found that PTEN directly binds to hnRNPK (figure 6C).

Figure 6

Phosphatase and tensin homologue (PTEN) interacts with heterogeneous nuclear ribonucleoprotein (hnRNPK) and inhibits glucose-6-phosphate dehydrogenase (G6PD) expression. (A) Hepatocellular carcinoma (HCC) and HCT116 cells were transfected with Flag-hnRNPK and Ad-PTEN. Lysates were immunoprecipitated with anti-hnRNPK and analysed with anti-hnRNPK or anti-Tcl1. (B) Cell lysates in (A) were immunoprecipitated with anti-PTEN and analysed by western blotting. (C) HepG2 cells were cotransfected with glutathione S-transferase (GST)-PTEN and Flag-hnRNPK. Cell lysates were GST-pulled down and immunoblotted (IB) with anti-PTEN and anti-hnRNPK. (D) RNA were covalently linked to adipic acid beads and incubated with nuclear extracts (150 μg of protein) from HepG2 cells overexpressing PTEN. The proteins were eluted from the beads and 4 μL of eluted proteins were analysed by western analysis using the indicated antibodies. Data are representative of three independent experiments.

Next, we asked how interaction between PTEN and hnRNPK affected G6PD pre-mRNA splicing. Using the method described in figure 5E, we investigated whether overexpression of PTEN regulated the binding of hnRNPK to exon 12 of G6PD. As shown in (figure 6D), PTEN increased the binding of hnRNPK to nt 50–72 by 70.1% (n=3 pull-down assays). Thus, PTEN regulated the binding of hnRNPK to exon 12 of G6PD. Indeed, the pre-mRNA/mature-mRNA G6PD ratio is modified in PTEN−/− and PTEN+/+ cells (see online supplementary figure S5).

PTEN inactivates Tcl1 via GSK3β-mediated phosphorylation which disrupts interaction between Tcl1 and HnRNPK

Given that PTEN activated glycogen synthase kinase-3β (GSK3β) and Akt inhibits GSK3β activity via phosphorylation, we were curious whether Tcl1 was a substrate of GSK3β. We carefully checked the amino acid sequences of all three Tcl family members (Tcl1, Tcl1B and MTCP (mature T cell proliferation-1)), figure 7A). Based on theoretical algorithms (http://elm.eu.org/), we found that only the mapped Tcl1 sites (86-S/-S-D-S/S-90) perfectly resided within a consensus GSK3β phosphorylation sequence. Most substrates of GSK3β required pre-phosphorylation at a priming residue located at the fourth residue C-terminal to the site of GSK3β phosphorylation. The consensus sequence for GSK3β existed as S/T-X-X-X-S/T, in which the first S/T residue was the GSK3β target residue and the final S/T represented the priming site. The priming phosphorylation was generally executed by a separate kinase, such as c-Jun NH2-terminal kinase (JNK).16 Accordingly, in the theoretical analysis of the 86-S/-S-D-S/S-90 phosphorylation sites, S90 could serve as a priming site for GSK3β phosphorylation at S86.

Figure 7

Glycogen synthase kinase-3β (GSK3β) directly phosphorylates Tcl1, thereby inhibiting the interaction between Tcl1 and heterogeneous nuclear ribonucleoprotein (hnRNPK). (A) Sequence alignment of the human members of the TCL1 proto-oncogene family (upper panel). Five GSK3β's substrates and phosphorylation sites (lower left panel). Predicted Tcl1 site phosphorylated by GSK3 and its mutants (lower right panel). (B) The constitutively active GSK3β-S9A incubated with recombinant glutathione S-transferase (GST)-Tcl1, S1A and S2A mutants for 30 min at 37°C in the presence of [32P]γ-ATP. Proteins were separated by SDS-PAGE for autoradiography. (C) HepG2 cells were cotransfected with GST-Tcl1 or its S1A, S1E mutants and Flag-hnRNPK. Cell lysates were GST-pulled down and immunoblotted (IB) with anti-hnRNPK.

In order to examine whether Tcl1 represented a substrate of GSK3β, in vitro kinase assays were performed. As shown in figure 7B, a catalytic active GSK3β (GSK3-S9A) readily phosphorylated Tcl1, whereas it could not phosphorylate Tcl1-S1A and -S2A mutants (two non-phosphorylatable mutants). Furthermore, one phosphomimetic mutant, Tcl1-S1E, significantly restrained its interaction with hnRNPK. By contrast, the non-phosphorylatable Tcl1-S1A mutant did not interfere with its binding to hnRNPK (figure 7C).

Knockdown of G6PD induces HCC and colon cancer cellular senescence and inhibits tumour growth in Huh7 orthotopic tumour models

Next, we explored the functional significance of G6PD and Tcl1 knockdown in HCC and HCT116 cell lines, respectively. Two non-overlapping shRNAs significantly reduced the expression of G6PD and decreased cell proliferation (figure 8A), which was rescued by overexpression of G6PD (figure 8B).

Figure 8

Knockdown of glucose-6-phosphate dehydrogenase (G6PD) induces hepatocellular carcinoma (HCC) and colon cancer cellular senescence and inhibits tumour growth in Huh7 orthotopic tumour models. (A) HCC and HCT116 cells were treated with 2 μg/mL doxycycline for 5 days, analysed by western blotting for G6PD and ascertained for the effects on proliferation. (B) The anti-proliferative effect of G6PD knockdown could be completely rescued by overexpression of mouse G6PD. Cells in (A) were either stained for β-galactosidase (β-GAL) activity or used in a chromogenic β-GAL assay for quantification (C). P21 expression was also determined by western blotting (C). Experiments were done in triplicate and were repeated at least three times. (D) Representative bioluminescence images corresponding to Huh7 orthotopic hepatic tumours (left panel). Volume of Huh7 orthotopic tumours was determined at different time points. Data points represent the mean±SD (middle panel). The ratio of liver weight/body weight (right panel). Statistical significance was concluded at *p<0.05, **p<0.01, ***p<0.001.

Furthermore, G6PD knockdown significantly increased the number of cells positive for SA-β-GAL (β-galactosidase) staining; the expression of p21, one classic marker for senescence, was upregulated in G6PD knockdown cells and restored to normal level when mouse G6PD was overexpressed (figure 8C). Therefore, suppression of G6PD induced cellular senescence in HCC and HCT116 cells.

We also tested whether G6PD knockdown would inhibit tumour growth in Huh7 orthotopic tumour models and mouse subcutaneous xenograft models (see online supplementary figure S6). As shown in figure 8D, knockdown of G6PD markedly retarded tumour growth. Our in vivo orthotopic tumour study reinforced our in vitro results and established the potential significance of G6PD in liver cancer progression.

Tcl1 knockdown enhances the sensitivity of HCC to sorafenib

Although multikinase inhibitor sorafenib can increase the survival of patients with advanced HCC, it is becoming apparent that combination treatments are critical to overcome the complex genomic aberrations in this disease. We found that HCC cell lines transfected with Tcl1 siRNA were more sensitive to doxorubicin (data not shown) or sorafenib, which inhibited proliferation and induced apoptosis (figure 9A,B). Furthermore, Huh7 cells transfected with one phosphomimetic Tcl1-S1E mutant showed higher sensitivity to sorafenib, and tumour growth was slowed in the Huh7 orthotopic tumour model compared with that in wild-type Tcl1 and Tcl1-S1A mutants (figure 9C,D). These observations indicated that combination treatments with sorafenib and Tcl1 siRNA or inactivation had the potential to improve therapeutic options for HCC.

Figure 9

Tcl1 knockdown enhances the sensitivity of hepatocellular carcinoma (HCC) to sorafenib. (A) and (B) HepG2 and Huh7 cells transfected with scramble small interfering RNA (siRNA) or Tcl1 siRNA were treated with 15 μM sorafenib for 24 h. Cells were stained with annexin V-FITC and propidium iodide for apoptotic assays. Mean±SD of three independent experiments. (C) Representative bioluminescence images corresponding to Huh7 orthotopic hepatic tumours (n=10 per group). (D) Volume of Huh7 orthotopic tumours was determined at different time points. Data points represent the mean±SD (left panel). The ratio of liver weight/body weight (right panel). (E) Proposed model depicting the antagonisation of phosphatase and tensin homologue (PTEN) to Tcl1-heterogeneous nuclear ribonucleoprotein (hnRNPK)– glucose-6-phosphate dehydrogenase (G6PD) axis that regulates the pentose phosphate pathway: (1) PTEN activates glycogen synthase kinase-3β (GSK3β) to phosphorylate Tcl1, thereby inhibiting the interaction between Tcl1 and hnRNPK; (2) PTEN interacts with hnRNPK and inhibits G6PD pre-mRNA cleavage; (3) PTEN interacts with G6PD and inhibits its activity (dimerisation). Statistical significance was concluded at *p<0.05, **p<0.01, ***p<0.001.

We also explored the functional significance of hnRNPK and found that hnRNPK knockdown inhibited proliferation, whereas hnRNPK overexpression rescued the effects of Tcl1 knockdown in HCC and HCT116 cells (see online supplementary figure S7).

G6PD and Tcl1 predict clinical outcomes of sorafenib therapy

To further determine the clinical relevance of G6PD and Tcl1 in HCC treatment, we analysed the survival follow-up information from an independent cohort of 65 subjects who received sorafenib treatment after liver cancer surgery (see online supplementary table S3). Lower expression of G6PD and Tcl1 were significantly associated with better progression-free survival and overall survival in these patients (figure 10A–D). These data strongly suggested important pathological roles for G6PD and Tcl1 in HCC.

Figure 10

Glucose-6-phosphate dehydrogenase (G6PD) and Tcl1 predict clinical outcomes of sorafenib therapy. Kaplan–Meier estimate of (A and C) overall survival and (B and D) progression-free survival in 65 patients with hepatocellular carcinoma (HCC) who received sorafenib treatment. Marks on graph lines represent censored samples. High Tcl1 and low Tcl1 refer to samples with high and low levels of Tcl1 expression, respectively; High G6PD and low G6PD refer to samples with high and low levels of G6PD expression, respectively. The p values were calculated using the log rank test.

Discussion

The findings shown here identify an important role for the PTEN/Tcl1/hnRNPK axis in regulating G6PD pre-mRNA splicing and activity in HCC. Through this regulation, PTEN exerts a powerful surveillance on the metabolic pathways that are critical for both the PPP and biosynthesis.

Recent data have shed new light on the role of PTEN beyond tumour suppression and particularly in metabolism.17 A major problem for the immediate future is to dissect the mechanism by which PTEN affects metabolism. It is known that metabolism can be controlled by allosteric interactions, covalent modifications of enzymes and changes in their amounts.18 Notably, we demonstrated here that PTEN inhibits G6PD through direct interactions, indicating that PTEN may act as a catalyst to the induction of conformational changes in G6PD. PTEN-induced conformational conversion of G6PD may represent a previously unknown mechanism of metabolic regulation. Additionally, studies suggest that PTEN may function within higher molecular mass complexes. Vazquez et al,19 found that PTEN can be separated into two populations: a monomeric subpopulation and a higher molecular mass subpopulation which contains hnRNPC.15 Our results identified hnRNPK as another new protein within the PTEN complex. Further characterisation of other components within the PTEN complex will be instrumental in expanding our understanding of PTEN function. Given the importance of PTEN in suppressing the PPP, the prevalent inactivation of PTEN in HCC probably accelerates glucose consumption and directs glucose towards rapid production of macromolecules by means of an increase in the PPP flux. Therefore, PTEN inactivation contributes to the Warburg effect and also links it to the PPP and enhanced biosynthesis in HCC.

Although we indicated that Tcl1 reversed the effects of PTEN on biosynthesis in HCC, little is known about its role in HCC. That is partially because the physiological expression of Tcl1 is primarily restricted to early embryonic cells and B cells. However, in pathological conditions, Tcl1 is overexpressed in a variety of human diseases.20 Recent findings suggest that Tcl1 plays a critical role in HCC because the Oct4-Tcl1-Akt pathway affected liver cancer stem cell survival and drug sensitivity by the regulation of ABCG2.7 Our study also identified an important role for Tcl1 in regulating the sensitivity of HCC to sorafenib and doxorubicin. Moreover, based on our mass spectrometry data, it seems that Tcl1 also binds to other proteins, such as HSP90 and ATM, closely related to liver tumour cell survival, suggesting one bigger role for Tcl1 in HCC. Despite these interesting findings, the potential relationship between Tcl1 and these partners needs to be further investigated. Additionally, it is necessary to identify more Tcl1 partners in order to elucidate Tcl1 functions.

We and others have performed mass spectrometry-based proteomic screens to identify hnRNPK as a potential human Tcl1 substrates in liver and lung tumour.21 HnRNPs are a group of RNA-binding proteins, whose main function is to bind newly translated messenger RNA (mRNA), in particular pre-mRNA.22 We found that both PTEN and Tcl1 can interact with hnRNPK, but oppositely regulated G6PD pre-mRNA splicing. Given that other hnRNPs are also involved in G6PD pre-mRNA splicing,13 whether PTEN and Tcl1 interact with them remains to be determined. Overexpression or altered subcellular expression of hnRNPK in numerous tumours has been positively correlated with poor clinical outcome. A recent study described the usefulness of hnRNPK as a tumour biomarker for detecting early HCC.23 Thus, hnRNPK would be an interesting therapeutic target to regulate liver tumorigenesis and progression.

G6PD molecules form catalytically active dimers by binding NADPH. The active enzyme exists in a dimer–tetramer equilibrium. High pH and ionic strength shift the equilibrium towards the dimer, whereas low pH causes a shift towards the tetramer.24 Additionally, it has been reported that p53 prevents formation of the active dimer.25 Similarly, we confirmed that PTEN inactivates G6PD. Regulation of G6PD expression is solely controlled by changes in the rate of pre-mRNA splicing.14 This mechanism may be shared across other metabolic genes, even those regulated by transcriptional mechanisms.26 Regulating gene expression by changes in the rate of pre-mRNA splicing in addition to transcriptional regulation would result in a more rapid response to metabolic changes.26

To further confirm and extend these findings, especially given that Tcl1 is primarily expressed in leukaemia, we explored whether the PTEN/Tcl1/hnRNPK/G6PD axis also exists in leukaemia (data not shown) and colon cancer. We found that this is indeed the case. Hence, we propose that the PTEN/Tcl1/hnRNPK/G6PD axis may have a more general function beyond HCC.

There is no standard treatment for HCC. In recent years, treatment with sorafenib plus doxorubicin, compared with doxorubicin monotherapy, resulted in greater median time to progression, overall survival and progression-free survival. Therefore, the potential for developing new agents and strategies against HCC is great. Our data suggest the use of Tcl1 and G6PD as new biomarkers of the risk and prognosis of HCC to help in the selection of therapeutic modalities. We propose a strategy to target Tcl1 and G6PD as a potential adjuvant therapy in combination with other methods of HCC treatment.

In summary, the PTEN/Tcl1/hnRNPK/G6PD axis is involved in the key regulatory steps of the PPP and biosynthesis in HCC and is a valuable target for the development of new anticancer treatments.

References

View Abstract

Supplementary materials

Footnotes

  • XH, RS, HS, TZ and JW contributed equally.

  • Contributors The first five authors contributed equally to this paper. XH, RS, HS and TZ: study design, acquisition of data, analysis and interpretation; JW, YL, SQ, ZL and XS: acquisition of data, technical support. HJ and ZZ: drafting of the manuscript and material support. ZZ and LL: study design, drafting and critical revision of the manuscript and obtaining funding.

  • Funding This study was supported by the National Natural Science Foundation of China (81272705), Program For Innovative Research (in science and technology) in higher educational institutions of Heilongjiang Province (2009td06), Heilongjiang Province Science Fund for outstanding youths (JC200616), Foundation of Harbin Science and Technology Bureau for creative young talents (2010RFQXS069) and Foundation of Health Department of Heilongjiang Province (grant No. 2009-043). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Research ethics committee of the First Affiliated Hospital of Harbin Medical University.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.