Loss of phosphatase and tensin homolog enhances cell invasion and migration through aKT/Sp-1 transcription factor/matrix metalloproteinase 2 activation in hepatocellular carcinoma and has clinicopathologic significance

Authors

  • Karen Man-Fong Sze,

    1. State Key Laboratory for Liver Research, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    2. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    Search for more papers by this author
    • *These authors contributed equally to this work.

  • Kris Lai-Ting Wong,

    1. State Key Laboratory for Liver Research, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    2. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    Search for more papers by this author
    • *These authors contributed equally to this work.

  • Glanice Kin-Yan Chu,

    1. State Key Laboratory for Liver Research, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    2. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    Search for more papers by this author
  • Joyce Man-Fong Lee,

    1. State Key Laboratory for Liver Research, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    2. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    Search for more papers by this author
  • Tai-On Yau,

    1. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    Search for more papers by this author
  • Irene Oi-Lin Ng

    Corresponding author
    1. State Key Laboratory for Liver Research, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    2. Department of Pathology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong
    • Room 127B, University Pathology Building, Department of Pathology, The University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong
    Search for more papers by this author
    • Fax: (852)-2872-5197


  • Potential conflict of interest: Nothing to report.

Abstract

Phosphatase and tensin homolog (PTEN) is frequently inactivated in cancers and is associated with advanced stages of cancers or metastasis. However, the molecular mechanism of PTEN in hepatocellular carcinoma (HCC) metastasis is unclear. In this study, we found frequent (47.5%, n = 40) protein underexpression of PTEN in human HCCs compared with their corresponding nontumorous livers. Significantly, PTEN underexpression was associated with larger tumor size (P = 0.021), tumor microsatellite formation (P = 0.027), and shorter overall survival of patients (P = 0.035). Using different cell models, we observed that PTEN-knockdown HCC cells and PTEN-knockout mouse embryonic fibroblasts (MEFs) had enhanced cell migratory and invasive abilities. In addition to activation of AKT, there was up-regulation of the Sp1 transcription factor (SP1) and matrix metalloproteinase 2 (MMP2), as well as MMP2 activation in PTEN-knockdown HCC cells and PTEN−/− MEFs. With dual luciferase reporter assay, exogenous expression of SP1 in HCC cells led to enhanced MMP2 promoter activity by up to 74%, whereas deletion of the putative SP1 binding site on the MMP2 promoter led to reduced promoter activity by up to 65%. Using chromatin immunoprecipitation assay, we documented increased binding of SP1 to the MMP2 promoter in PTEN-knockdown HCC cells. Overexpression of SP1 and MMP2 was significantly but negatively associated with PTEN underexpression in human HCCs. Conclusion: Our results show that PTEN was underexpressed in HCCs, and this underexpression was associated with more aggressive biological behavior and poorer patient survival. We have provided the first evidence that MMP2 up-regulation upon PTEN loss is SP1-dependent. Our findings indicate that PTEN plays a significant role in down-regulating HCC cell invasion via the AKT/SP1/MMP2 pathway. (HEPATOLOGY 2011;)

Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide and the second most common fatal cancer in Southeast Asia.1 HCC has a poor prognosis with high mortality. The majority of patients present late with advanced HCC,2 thereby limiting potentially curative therapeutic options. Cancer metastases, both intrahepatic and extrahepatic, are major factors for the mortality of HCC patients. Nonetheless, the molecular mechanisms underlying HCC metastasis remain largely unclear.

Phosphatase and tensin homolog (PTEN) is one of the most frequently mutated tumor suppressors, only second to p53. It is constitutively expressed, whereas p53 is a stress-responsive tumor suppressor. PTEN is an upstream negative regulator of the survival phosphoinositide 3-kinase (PI3K)/AKT cascade; activation of this signaling is frequently observed in multiple cancers due to loss of PTEN. AKT is a central regulator of various cellular processes—including cell survival, proliferation, growth, angiogenesis, and metabolism—via phosphorylation of various substrates.3 Hence, loss of PTEN gatekeeper function plays a pivotal role in promoting carcinogenesis. Clinically,4, 5 inactivation of PTEN is often associated with advanced stages of disease and metastasis in various cancers.3, 6 Although loss of PTEN in human cancers has been documented, the exact roles of PTEN in HCC have not been fully elucidated. Understanding the causative molecular mechanisms of cancer metastasis is important because it may open up new, targeted therapeutic interventions.

The matrix metalloproteinase (MMP) superfamily consists of metalloproteinases that function to degrade extracellular matrix, an essential process prior to cancer cell invasion. Venous invasion is a major problem associated with poor prognosis, and increasing numbers of studies investigating the regulation of MMPs contributing to cancer metastasis have emerged. In a report on radiation enhancement of cell invasion, MMP9 expression was up-regulated via PI3K/AKT/nuclear factor κB cascade in HCC cells.7 Moreover, hepatitis B virus X protein could induce expression of MMP2 and MMP9 gelatinases and promote HCC invasion through extracellular signal-regulated kinases and PI3K/AKT signaling transduction.8, 9 Taken together, because activated AKT signaling pathway leading to cancer metastasis is well documented, PTEN might also be involved in HCC metastasis.

In the present study, we addressed the clinical significance of PTEN in human HCCs and its functional implications and molecular mechanisms in HCC development and invasion. We found that PTEN was frequently underexpressed in human HCCs, and its underexpression was closely associated with more aggressive tumor behavior in terms of larger tumor size, tumor microsatellite formation, and shorter overall survival of patients. With knockdown of PTEN in HCC cells and using PTEN knockout mouse embryonic fibroblasts (MEFs), we have provided the first evidence that loss of PTEN contributed to HCC invasion by activating MMP2 via an Sp1 transcription factor (SP1)-dependent pathway. These results suggest an important role of PTEN in suppressing HCC invasion as well as the potential of targeting PTEN and AKT/SP1/MMP2 activation as chemotherapeutic targets for treatment of HCC.

Abbreviations

ChIP, chromatin immunoprecipitation; HCC, hepatocellular carcinoma; HPRT, hypoxanthine-guanine phosphoribosyltransferase; MEF, mouse embryonic fibroblast; MMP, matrix metalloproteinase; mRNA, messenger RNA; p-AKT, phosphorylated AKT; PCR, polymerase chain reaction; PI3K, phosphoinositide 3-kinase; PTEN, phosphatase and tensin homolog; SDS, sodium dodecyl sulfate; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; shRNA, short hairpin RNA; SP1, Sp1 transcription factor.

Patients and Methods

Patients and Samples.

Human HCC samples and their corresponding nontumorous liver samples from 40 Chinese patients (31 men, 9 women; age range, 34-74 years) who had surgical resection at Queen Mary Hospital, the University of Hong Kong, from 1992 to 2000, were randomly selected for study. All specimens were collected at the time of surgical resection, snap-frozen in liquid nitrogen, and kept at −80°C.

Cell Lines.

The human HCC cell lines BEL-7402 and SMMC-7721 were obtained from the Shanghai Institute of Cell Biology, Chinese Academy of Sciences. HepG2, PLC/PRF/5, and Huh7 cells were purchased from American Type Culture Collection. H2P and H2M cells (a gift from X. Y. Guan, The University of Hong Kong) were derived from an HCC patient with intrahepatic metastasis; H2P cells were isolated from the primary cancer, and H2M cells were isolated from its occlusive tumor venous thrombus.10 The HCC cell lines H2P, H2M, HepG2, PLC/PRF/5, Huh7, BEL-7402, and SMMC-7721 were maintained in Dulbecco's modified Eagle's medium with high glucose (Gibco-BRL, Grand Island, NY) supplemented with 10% fetal bovine serum.

Cell Transfection and Establishment of Stable Knockdown Cells.

Small interfering PTEN, small interfering SP1 duplexes, and short hairpin RNA (shRNA) PTEN in pRNATin-H1.4/Retro expression vector incorporated with sequence 5′-GGCGCUAU GUGUAUUAUUA-3′ were purchased from Dharmacon (Lafayette, CO) and GenScript USA Inc. (Piscataway, NJ), respectively. Expression vectors, small interfering RNA duplexes, and shRNA were transfected into BEL-7402 and SMMC-7721 using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. For establishing shRNA stably expressing cell lines, transfected cells were kept under 400 μg/mL hygromycin B selection for 14 days.

Preparation of MEFs.

The PTEN+/− knockout mouse line was a gift from T. W. Mak of the University of Toronto. Pregnant mice were sacrificed at 9.5 days postcoitus. Embryos were dissected from the uterus, and extraembryonic membranes and viscera were subsequently removed. The sliced embryos were soaked in 0.25% trypsin–ethylene diamine tetraacetic acid for 30 minutes. Cell suspensions were allowed to pass through a strainer and were then seeded on culture dishes.

Extraction of RNA and Quantitative Real-Time Reverse-Transcription Polymerase Chain Reaction.

Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's protocol and reverse-transcribed to generate complementary DNA using GeneAmp Gold RNA PCR Reagent Kits (Applied Biosystems, Foster City, CA). Probes for target human genes, MMP2, and endogenous controls (hypoxanthine-guanine phosphoribosyltransferase [HPRT]), were purchased from Applied Biosystems (TaqMan system). The thermal profile was 95°C for 10 minutes followed by 95°C for 15 seconds and 60°C for 1 minute for 40 cycles of amplification. To measure the amount of mouse MMP2 messenger RNA (mRNA) in MEFs, the primer set for MMP2 (forward 5′-CCCCTATC TACACCTACACCAAGAAC-3′ and reverse 5′-CATT CCAGGAGTCTGCGATGAGC-3′) and β-actin (forward 5′-GTGGGCCGCCCTAGGCACCAG-3′ and reverse 5′-CTCTTTGATGTCACGCACGATTTC-3′) was employed in SYBR green quantitative real-time reverse-transcription polymerase chain reaction (PCR) measurement. β-Actin was used as an endogenous control in this measurement.

Protein Preparation and Western Blot Analysis.

Total cellular protein was extracted by way of cell lysis in ice-cold radio immunoprecipitation assay buffer (50 mM Tris-HCl [pH 7.4], 1% Triton X-100, 1% sodium-deoxycholate, 0.1% sodium dodecyl sulfate [SDS], 150 mM NaCl, and 1 mM ethylene diamine tetraacetic acid), supplemented with 1 mM Na3VO4, 1 mM NaF, and 1× complete protease inhibitors (Roche Molecular Biochemicals, Indianapolis, IN), and separated by way of SDS–polyacrylamide gel electrophoresis (SDS-PAGE) for western blotting analysis. Commercially available primary antibodies against PTEN, phosphorylated AKTser473, total AKT (Cell signaling biotechnology, Danvers, MA), SP1 (Santa Cruz Biotechnology, Santa Cruz, CA), α-tubulin (Sigma, St Louis, MO), and β-actin (Sigma) were used for protein analysis.

Transwell Migration Assay.

A total of 2 × 105 cells suspended in 100 μL serum-free medium were seeded in the top chamber of the transwell (Techno Plastic Products, Trasadingen, Switzerland), and full serum medium was added at the bottom of the well. Cells were allowed to move across the pores and adhere on the bottom membrane of the transwell. Cells were then fixed with 75% methanol and stained with crystal violet for 1 hour. Five randomized fields were captured in each transwell under the microscope and counted. To rule out the effects of different cell proliferation rates that might alter the results, cells were treated with 10 μg/mL of mitomycin C before the assay was performed.

Matrigel Invasion Assay.

Matrigel basement membrane matrix (BD Biosciences) was diluted four-fold with serum-free medium and coated onto the membrane of Transwell filters. Cells were seeded as stated above on top of the Matrigel. Cells capable of invading secreted enzymes to digest the components of the Matrigel were allowed to move across and adhere onto the bottom membrane of the transwell. Cells were fixed and counted as stated in the Transwell migration assay. To rule out effects of different cell proliferation rates that might alter the results, cells were treated with 10 μg/mL of mitomycin C before the assay was performed.

Gelatin Zymography.

Conditioned medium of each sample was concentrated 10-fold using centrifugal filter devices (Millipore, Billerica, MA), mixed with equal portion of 2× sample buffer, then separated by way of SDS-PAGE with addition of 0.1% gelatin. Gels were incubated with 1× Zymogram renaturing buffer (2.7% [wt/vol] Triton X-100), followed by 1× Zymogram developing buffer (50 mM Tris-HCl [pH 7.4], 0.2 M NaCl, 5 mM CaCl2, and 1 mM ZnCl) at room temperature for 30 minutes. After overnight incubation at 37°C, gels were stained with R-250 Coomassie blue for 1 hour and washed with destaining solution. Enzyme activity was visualized as clear bands.

Dual Luciferase Reporter Assay.

Cells were cotransfected with either wild-type MMP2 promoter or MMP2 promoter with mutation of the putative SP1 binding motif (−1951 to +74 nucleotides derived from transactional start site) in pGL3-Basic and PGK-Renilla luciferase constructs. Cells were harvested 24 hours after transfection, and the luciferase activity driven by the MMP2 promoter under different cellular levels of exogenous SP1 protein was determined using the Dual Luciferase Assay System (Promega, Madison, WI). Twenty milliliters of lysate in triplicates were then transferred to a 96-well plate, and the luciferase activity was measured with a luminometer with sequential addition of luciferase assay reagent II and Stop & Glo reagent. PGK-Renilla luciferase was included for internal normalization. The experiment was performed at least three times independently.

Chromatin Immunoprecipitation Assay.

A total of 3 × 106 cells were seeded 1 day before harvest, and chromatin immunoprecipitation (ChIP) assay was performed. Cells were fixed with 1% formaldehyde for 10 minutes, and the reaction was neutralized by adding glycine to a final concentration of 125 mM in the mixture. Formaldehyde cross-linked cells were collected by way of centrifugation, resuspended in membrane lysis buffer (5 mM KOH [pH 8.0], 85 mM KCL, 0.5% Nonidet P40, 0.5% SDS, and 1× complete protease inhibitors), and incubated in ice for 30 minutes. Cell nuclei were collected by way of centrifugation, and cross-linked DNA was followed by Micrococcal nuclease digestion for 20 minutes according to the manufacturer's protocol (New England Biolabs, Inc., Ipswich, MA). Digested DNA was released from nuclei by way of freeze-thaw cycles, and ChIP assay was performed according to the EZ-Chip assay kit (Millipore) protocol. The antibody against SP1 protein was used (Santa Cruz Biotechnology), and the primer set (forward 5′-ACTGAGGGTGGACGTAGAGG-3′ and reverse 5′-CAGATGTAGCCGGCTGGGCT-3′) covering the putative SP1 binding site on MMP2 promoter was employed for standard PCR measurement in the ChIP assay.

Immunohistochemistry.

Immunohistochemistry was performed on formalin-fixed, paraffin-embedded sections as described,11 using rabbit polyclonal antibody against SP1 (Santa Cruz Biotechnology) and MMP2 (Abcam plc, Cambridge, MA) at 1:150 and 1:1,000 dilution, respectively.

Clinicopathologic Correlation, Survival, and Statistical Analysis.

Clinicopathologic features of patients, including sex, tumor size, number of tumor nodules, cellular differentiation by Edmondson grading, presence of tumor encapsulation, tumor microsatellite formation, venous invasion without differentiation into portal or hepatic venules, direct liver invasion, background liver disease in the nontumorous liver tissues, and serum hepatitis B surface antigen status were analyzed using SPSS 17 (SPSS Inc, Chicago, IL). After resection, all patients were followed up monthly in the first year and quarterly thereafter. Actuarial survival was measured from the date of hepatic resection to the date of death or the last follow-up. The survival curves were assessed using the Kaplan-Meier method, and statistical differences between two groups was evaluated using a log-rank test. Categorical data were analyzed with the Fisher's exact test, whereas independent t tests were used for continuous data. P < 0.05 was considered significant.

Results

PTEN Was Underexpressed in Human HCCs and Associated with Aggressive Tumor Behavior and Poorer Patient Survival.

PTEN protein expression was examined by way of western blotting in 40 human HCCs. Nineteen (47.5%) HCCs exhibited underexpression (≥2-fold) of PTEN at the protein level compared with their corresponding nontumorous livers (Fig. 1A,B).

Figure 1.

PTEN protein was underexpressed in human HCCs. (A) PTEN protein expression in two representative HCC cases. The relative amount of PTEN protein was normalized with β-actin. The number below each case indicates the ratio of PTEN protein levels of tumor (T) and nontumorous (NT) liver tissues. PTEN was regarded as underexpressed when T/NT was ≤0.5; no change when the ratio was 0.5 < T/NT < 2; and overexpressed when the ratio was T/NT ≥2. (B) Summary of PTEN protein expression levels in HCC patients. (C) Correlation of PTEN underexpression with overall survival rates of patients. Underexpression of PTEN at protein level was significantly associated with poorer overall survival rates of patients (P = 0.035, log-rank test). (D) The protein expression of PTEN in five human HCC cell lines and in a pair including a primary HCC cell line and its metastatic cell line was analyzed. Reduced PTEN protein expression was observed in the metastatic H2M HCC cells compared with its primary H2P HCC cells.

Upon clinicopathologic correlation, underexpression of PTEN significantly correlated with larger tumor size (P = 0.021, Fisher's exact test) and presence of tumor microsatellites (P = 0.027, Fisher's exact test) (Table 1). Presence of tumor microsatellites is an established feature of intrahepatic metastasis of HCC. This finding showed that underexpression of PTEN was frequent in human HCCs and was related to tumor growth and tumor metastasis.

Table 1. Summary of the Findings of Clinicopathologic Correlation
Clinicopathologic FeaturesPTEN ExpressionP Value
UnderexpressionNo Underexpression
Sex  1.00
 Men1516 
 Women45 
Tumor size  <0.05
 <5 cm18 
 ≥5 cm1813 
Cellular differentiation (Edmondson grading)  1.000
 I-II89 
 III-IV1112 
Encapsulation  0.333
 Absent1412 
 Present59 
Tumor microsatellite formation  <0.05
 Absent413 
 Present159 
No. of tumor nodules  0.689
 11518 
 ≥243 
Direct liver invasion  0.532
 Absent1110 
 Present710 
Venous invasion  0.349
 Absent610 
 Present1311 
Background liver disease  0.067
 Normal and chronic hepatitis138 
 Cirrhosis613 

The overall survival rates of the 40 patients were 55.6%, 27.8%, and 16.7% months at 1, 3, and 5 years, respectively. Patients whose tumors had PTEN underexpression had significantly shorter overall survival rates compared with those whose tumors had no PTEN underexpression (median, 15.2 and 63.2 months, respectively; P = 0.035, log-rank test) (Fig. 1C).

Knockdown of PTEN Enhanced Cell Migration and Invasion in Human HCCs In Vitro.

From our clinicopathologic correlation findings, PTEN underexpression was associated with feature of intrahepatic metastasis. Interestingly, we observed a marked reduction of PTEN protein level in a higher metastatic potential HCC cell (H2M) compared with the cell line (H2P) derived from the primary HCC of the same patient and has a lower metastatic potential HCC cell (Fig. 1D).10 To assess the effect of PTEN on HCC cell migration, we knocked down the expression of PTEN by shRNA in the HCC cells lines BEL-7402 and SMMC-7721, which have relatively high levels of PTEN (Fig. 1D). Reduced PTEN expression was confirmed by western blotting; there was 35%-51% reduction of endogenous PTEN protein level in either BEL-7402 or SMMC-7721 cells (Fig. 2A). Using Transwell migration and Matrigel invasion assays (Fig. 2B,C and Supporting Information Fig. 1), we observed that knockdown of PTEN in HCC cells significantly enhanced cell migration and invasion, respectively, in both BEL-7402 and SMMC-7721 cells (P = 0.011 and 0.020, respectively, for cell migration assay and P = 0.004 and 0.012, respectively, for cell invasion assay, Student t test) (Fig. 2B,C and Supporting Fig. 1).

Figure 2.

Knockdown of PTEN in human HCC cell up-regulated cell migratory and invasive abilities in human HCC cells. (A) Underexpression of PTEN protein in stable PTEN-knockdown BEL-7402 cells. (B) Cell migration and (C) cell invasion assays. There was significant increase in the cell migratory and invasive abilities in PTEN-knockdown BEL-7402 cells compared with vector control cells (P = 0.011 and 0.004 respectively, Student t test).

PTEN−/− MEF Exhibited Increased Cell Migration and Invasion.

To further delineate the effects of complete PTEN loss on cell migration and invasion, PTEN−/− MEFs were established and the absence of PTEN protein expression was confirmed with western blotting (Fig. 3A). Similar to the PTEN-knockdown HCC cells, the PTEN−/− MEFs showed a significantly enhanced ability to migrate and invade, as assessed by the Transwell migration and Matrigel cell invasion assays, respectively (P < 0.001 for both, Student t test) (Fig. 3B).

Figure 3.

Functional analysis on PTEN+/+ and PTEN−/− MEFs. (A) Protein expression of PTEN in both PTEN+/+ and PTEN−/− MEFs. Significant up-regulation of p-AKT in PTEN-null MEFs was observed. (B) Cell migration and invasion analysis of PTEN+/+ and PTEN−/− MEFs using Transwell assay and Matrigel invasion assay. PTEN-knockout MEFs showed a significantly enhanced cell migration and invasion in vitro (P < 0.001 for both, Student t test).

Up-regulation and Activation of MMP2 in PTEN-Knockdown HCC Cells and PTEN−/− MEFs.

We then investigated whether the induced cell invasion by loss of PTEN, as shown in the cell invasion assay, involved degradation of the Matrigel. We evaluated the role of MMP2, a gelatinase responsible for degradation of collagen IV, in the enhanced cell invasion mediated by loss of PTEN. The mRNA levels of MMP2 in the PTEN-knockdown stable clones of SMMC-7721 and BEL-7402 were analyzed by quantitative PCR. MMP2 transcription was up-regulated by 3.5-fold and two-fold in the PTEN-knockdown BEL-7402 and SMMC-7721 cells, respectively (P = 0.002 and 0.006, respectively, Student t test) (Fig. 4A). Consistently, there was 1.8-fold up-regulation of MMP2 mRNA in the PTEN−/− MEF, compared with the wild-type MEFs (P = 0.041, Student t test) (Fig. 4B).

Figure 4.

Enhanced transcriptional level and proteolytic cleavage activity of MMP2 in PTEN-deficient cells. (A) Quantitative analysis of MMP2 mRNA expression in both BEL-7402 and SMMC-7721 PTEN-knockdown clones. HPRT was used as a normalization control (P = 0.002 and 0.006, respectively, Student t test). (B) Quantitative real-time PCR analysis of MMP2 mRNA expression in PTEN knockout MEF cells. β-actin was used for normalization control (P = 0.041, Student t test). (C) Measurement of MMP2 proteolytic cleavage activity in PTEN-knockout MEFs using gelatin zymogram. A clear activated MMP2 was present in PTEN−/− MEF but not in PTEN+/+ MEF.

It is known that the activated form of MMP2 (62 kDa) is produced by enzymatic cleavage of the pro-MMP2 (72 kDa) upon digestion by plasminogen, such as urokinase plasminogen activator.12 Because activated MMP2 digests gelatin in the polyacrylamide gel and produces a digested halo area at the corresponding molecular weight of the MMP2 in gelatin zymography, we performed gelatin zymography and documented activation of MMP2 in PTEN−/− MEFs. A 62-kDa, enzymatically cleaved product of MMP2 was observed in the PTEN−/− MEFs but not in the PTEN+/+ MEFs (Fig. 4C), indicating the presence of the activated form of MMP2 in the PTEN−/− MEFs.

Activation of AKT and Up-regulation of SP1 upon PTEN Loss.

Consistent with the notion that PTEN suppresses AKT phosphorylation, we confirmed an up-regulation of p-AKTSer473 protein level in the PTEN−/− MEF, whereas the total AKT protein level remained unchanged (Fig. 3A).

It has been reported that the SP1 transcription factor is one of the key components regulating the MMP2 promoter activation13 and that up-regulation of SP1 transcriptional activity occurs through phosphorylated AKT (p-AKT) activation in human cancers.14 We observed elevated protein levels of SP1 in the PTEN-knockdown BEL-7402 and SMMC-7721 HCC cells and PTEN−/− MEFs (Fig. 5A).

Figure 5.

SP1 is involved in MMP2 transcriptional activation in PTEN-deficient cells. (A) SP1 was overexpressed in both PTEN-knockdown BEL-7402 and SMMC-7721 cells and in PTEN knockout MEFs on western blotting analysis. (B) Dual luciferase reporter assay. Upon exogenous expression of SP1, the promoter activity of wild-type MMP2 was increased by 50% compared with SP1-nontransfected BEL-7402 and SMMC-7721 cells (P = 0.016 and < 0.001, respectively, Student t test). However, the promoter activity of MMP2 was significantly reduced when the two putative SP1 binding sites (98 to 63 nucleotides upstream of transcriptional start site) on the MMP2 promoter were deleted in both cell lines. (P = 0.006 and P < 0.001, respectively, Student t test).

SP1 Enhanced MMP2 Transcription.

Next, we investigated the role of SP1 as an intermediate molecular target linking loss of PTEN and MMP2 activation in HCC cells. We evaluated the activity of the MMP2 promoter using Dual luciferase reporter assay with or without exogenous expression of SP1. Exogenous expression of SP1 protein in both BEL-7402 and SMMC-7721 cells enhanced the wild-type MMP2 promoter activity (P = 0.016 and P < 0.001, respectively, Student t test) (Fig. 5B). When the putative SP1 binding site (located at 98-63 nucleotides upstream of the transcriptional start site) was deleted, there was a significant reduction of the MMP2 promoter activity compared with the wild-type MMP2 promoter, in BEL-7402 and SMMC-7721 cells (P = 0.006 and P < 0.001, respectively, Student t test) (Fig. 5B). The results suggest that SP1 regulates MMP2 transcription in human HCC.

Moreover, transient depletion of SP1 resulted in significantly reduced MMP2 mRNA level in both PTEN-knockdown BEL7402 and SMMC-7721 cells (Fig. 6A). Furthermore, with ChIP assay, we demonstrated an enrichment of SP1 bound on the MMP2 promoter in PTEN-knockdown BEL-7402 cells compared with the vector control cells (Fig. 6B). Taken together, our data suggest that, in the PTEN-knockdown HCC cells and PTEN−/− MEF, loss of PTEN activates AKT and up-regulates SP1, which in turn up-regulates MMP2, leading to increased cell invasion.

Figure 6.

(A) Depletion of SP1 protein by siRNA against SP1 showed significant reduction of MMP2 mRNA in PTEN-knockdown BEL-7402 (P = 0.031) and SMMC-7721 (P = 0.043) cells. HPRT was used for normalization control. (B) ChIP assay. A marked increase of SP1 bound on the −186 nucleotides upstream of the transcriptional start site of MMP2 was seen in PTEN-knockdown BEL-7402 cells compared with the corresponding vector control in the ChIP assay. Absence of antibody added served as a negative control.

SP1 and MMP2 Proteins Were Up-regulated in PTEN-Underexpressed Human HCCs.

We further evaluated the possible association among the expression of PTEN, SP1, and MMP2 in human HCCs. Immunohistochemistry showed positive staining in the nuclei for SP1, whereas for MMP2, the staining was cytoplasmic (Fig. 7). Overexpression of SP1 and MMP2 was significantly but negatively associated with PTEN underexpression in human HCCs (P = 0.010 and 0.022, respectively, Fisher's exact test) (Fig. 7). There was, however, no association between SP1 and MMP2 expression using immunohistochemistry (P = 0.740). No association between the clinicopathologic features and SP1 and MMP2 overexpression was found.

Figure 7.

SP1 and MMP2 protein expression in human HCC samples. (A) A representative case of HCC showing PTEN protein underexpression in the tumor. The relative amount of PTEN protein was normalized with β-actin. (B) Immunohistochemical staining for SP1 and MMP2 on the tumor and nontumorous liver of the same case as in (A). There was marked increase in the nuclear staining of SP1 in the tumor cells. In addition, marked overexpression of MMP2 was also seen in the cytoplasm of the tumor cells compared with its corresponding nontumorous liver. (C) Overexpression of SP1 and MMP2, respectively, was significantly but negatively associated with PTEN underexpression in human HCCs.

Discussion

Despite the fact that PTEN has been extensively studied and is implicated in cell migration,15-19 its underlying molecular mechanisms in HCC progression and metastasis have not been clearly elucidated. Therefore, it is of strategic importance to characterize the functional consequence of PTEN underexpression and the deregulated downstream signaling. In line with our clinical findings, the metastatic cell line H2M had a lower PTEN protein expression compared with its primary HCC counterpart cell line H2P.10 To study the functional consequences of PTEN loss in HCC, two HCC cell lines, SMMC-7721 and BEL-7402, with relatively higher endogenous PTEN levels were used to establish the stable PTEN-knockdown clones. Because more than one stable knockdown clone was used, this likely eliminated clonal effect in the assays. We documented that knockdown of PTEN significantly promoted cell migration and invasion in vitro, suggesting its possible role in HCC metastasis. Similar results were obtained with both cell lines, which suggests that the enhanced cell migration and invasion associated with loss of PTEN expression was not cell line–specific. In addition, our PTEN-deficient MEFs also exhibited increased cell migration and invasion compared with wild-type MEFs. This indicates that the association of PTEN loss with metastasis is not cell type–specific. Overall, our in vitro results were consistent with our clinical observation that PTEN underexpression was significantly associated with more frequent tumor microsatellite formation in human HCCs. Tumor microsatellite formation in human HCCs is one of the histologic features of tumor metastasis. Furthermore, we observed that the PTEN−/− MEFs also possessed enhanced proliferation rate (Supporting Fig. 2), and this was consistent with our finding of p-AKT activation, thereby triggering the prosurvival pathways. The enhanced cell proliferation is in agreement with our finding of an association of PTEN underexpression with increased tumor size in human HCCs. To eliminate the additive effect or cell proliferation in contribution to the enhanced cell migration and invasion observed, mitomycin C, a drug that blocks cell proliferation, was added in these assays.

To move across the Matrigel layer in the cell invasion assay, cells need to secrete certain enzymes to degrade the extracellular matrix. Indeed, the Matrigel coated onto the invasion chamber in the cell invasion assay comprise three main substances: 56% laminin, 31% collagen IV, and 8% entactin, all by weight. MMP2 and MMP9 gelatinases primarily act to destroy the major constituent collagen IV. Using gelatin zymography, we observed that when PTEN was absent or down-regulated, there was an increased proteolytic cleavage of MMP2, resulting in an activated form of MMP2. Furthermore, we also assessed the expression levels of MMP2 in the stable PTEN-knockdown clones of SMMC-7721, BEL-7402, and PTEN−/− MEFs. Endogenous MMP2 mRNA expression was markedly up-regulated in these cell lines. This finding suggests that, in our HCC knockdown cells and knockout MEF models, the enhanced cell invasion mediated by loss of PTEN involved MMP2 up-regulation. Our results were consistent with those from studies on murine cardiac fibroblasts cells.20 It has been reported that MMP9 is another factor playing important roles in cell invasiveness in HCC via the PI3K pathway.8 Surprisingly, in our study, MMP9 was not detected in gelatin zymography in both wild-type and PTEN knockout MEF cells, even when MM9 transcripts were abundantly expressed in both MEF cell lines (data not shown), suggesting that secretion of MMP9 might not be PTEN-dependent in the MEF model.

We further delineated the molecular pathway by which PTEN knockdown enhanced cell invasion. Previous reports have suggested that SP1 is one of the key regulators of the MMP2 promoter,13, 21, 22 and activation of AKT leads to phosphorylation of SP1, resulting in enhanced transcriptional activity of SP1.14, 23-25 Therefore, we speculated that SP1 might contribute to MMP2 activation in PTEN-deficient cells. Consistent results of enhanced SP1 endogenous protein expression and its binding affinity with the MMP2 promoter were observed in PTEN-knockdown BEL-7402 and SMMC-7721 cells. Furthermore, there was a significant but negative association of both SP1 and MMP2 protein expression by immunohistochemistry with PTEN underexpression in human HCCs. Thus, our data provide the first evidence that MMP2 up-regulation upon PTEN loss is SP1-dependent and suggest that the PTEN/AKT/SP1/MMP2 pathway plays an important role in regulating the cell invasive ability in HCC cells.

In this study, we documented that PTEN protein was frequently (47.5%) underexpressed in human HCCs. Its underexpression was significantly associated with larger tumor size and tumor microsatellite formation. Significantly, PTEN underexpression was associated with shorter overall survival of patients. Our findings are consistent with those of a number of previous studies showing underexpression of PTEN at both mRNA and protein levels in human HCCs.4, 5, 26-28 The significant association of PTEN underexpression with HCC progression, metastasis, and poorer prognosis in our study was in line with those from previous studies. As we aimed to focus on the relationship between PTEN and HCC invasion in this study, we did not examine the causes of underexpression. Indeed, PTEN is frequently lost or mutated in sporadic cancers and heritable diseases,3, 27, 29 and this may be attributed to chromosomal or allelic losses, mutations, or epigenetic silencing due to DNA methylation or histone deacetylation. Indeed, PTEN mutations span the entire gene and cause PTEN truncations; only a small fraction of the mutations is present at its phosphatase domain.29 However, among the 96 human HCC cases analyzed from a Taiwan group,6 only three samples carried PTEN mutations, whereas p53 mutations were observed in 26 samples. This suggests that mutation of PTEN may not be a major event in HCC development. Hence, other inactivation mechanisms of PTEN should be investigated.

Our clinicopathologic analysis has provided evidence that PTEN underexpression in human HCCs was associated with increased tumor size and thus was related to disease progression. Our PTEN-knockdown HCC cell models have simulated the loss of PTEN expression during disease progression. Accordingly, finding small molecules or drugs that could be delivered into tumor cells and up-regulate PTEN expression would be promising as novel therapeutic interventions. Previous studies have shown that activation of peroxisome proliferator-activated receptor γ was able to up-regulate PTEN in human macrophages and pancreatic cancer cells.30, 31 Furthermore, an antidiabetic drug, rosiglitazone, has been reported to serve as a selective ligand of peroxisome proliferator-activated receptor γ, which up-regulates PTEN by promoting its binding to PTEN promoter. Such PTEN activation may also be effective in reducing p-AKT and corresponding cell proliferation. Furthermore, it has later been demonstrated that the drug could inhibit cell migration in BEL-7402 HCC cells through up-regulation of PTEN expression.32 Hence, restoration of PTEN expression in HCC might be a potential new therapeutic approach.

In conclusion, we have documented frequent underexpression of PTEN in human HCCs, and PTEN underexpression was associated with a more aggressive biological behavior and shorter overall survival of patients. Our findings also demonstrated that PTEN played a significant role in down-regulating cell invasion via the AKT/SP1/MMP2 pathway.

Acknowledgements

We thank Dr. Tak W. Mak for providing the PTEN+/− knockout mouse line and Dr. D. Y. Jin for providing SP1 expression plasmid. We thank Dr. Terence K. W. Lee for critical reading of the manuscript.

Ancillary