Sublethal heat treatment promotes epithelial-mesenchymal transition and enhances the malignant potential of hepatocellular carcinoma


  • Shuhei Yoshida,

    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Miroslaw Kornek,

    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    2. Department of Medicine II, Saarland University Medical Center, Homburg, Germany
    Search for more papers by this author
  • Naoki Ikenaga,

    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Moritz Schmelzle,

    1. Transplantation Institute, Department of Medicine and Surgery, Beth Israel Deaconess Medical Center, Boston, MA
    2. Department of Surgery and Translational Centre for Regenerative Medicine, University of Leipzig, Germany
    Search for more papers by this author
  • Ryota Masuzaki,

    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    2. Transplantation Institute, Department of Medicine and Surgery, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Eva Csizmadia,

    1. Transplantation Institute, Department of Medicine and Surgery, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Yan Wu,

    1. Transplantation Institute, Department of Medicine and Surgery, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Simon C. Robson,

    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    2. Transplantation Institute, Department of Medicine and Surgery, Beth Israel Deaconess Medical Center, Boston, MA
    Search for more papers by this author
  • Detlef Schuppan

    Corresponding author
    1. Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Boston, MA
    2. Molecular and Translational Medicine, Dept. Medicine I, University of Mainz Medical School, Mainz, Germany
    • Address reprint requests to: Detlef Schuppan, M.D., Ph.D., Division of Gastroenterology and Liver Center, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, Boston, MA 02115. E-mail:; fax: 617-667-2767.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (grant nos.: 1 R21 DK075857-01A2 and U19 AI066313-05; to D.S.) and the German Research Foundation (DFG; grant no.: KO4103/1-1; to M.K.). M.S. also acknowledges grant support from the American Liver Foundation, the DFG (SCHM 2661/1-1 and 2661/1-2), and the German Federal Ministry of Education and Research (BMBF; PtJ-Bio, 0315883).

  • Some aspects of this study were presented at an oral sessions of Digestive Disease Week 2011 (Abstract no. 896) and at poster sessions of the American Association for the Study of Liver Diseases 2012 (Abstract no. 887) and 2013 (Abstract no. 1804).


Radiofrequency ablation (RFA) is a potentially curative therapy for hepatocellular carcinoma (HCC). However, incomplete RFA can induce accelerated invasive growth at the periphery. The mechanisms underlying the RFA-induced tumor promotion remain largely unexplored. Three human HCC cell lines were exposed to 45°C-55°C for 10 minutes, simulating the marginal zone of RFA treatment. At 5-12 days post-treatment cell proliferation, parameters of epithelial-mesenchymal transition (EMT), and activation of mitogen-activated protein kinases were analyzed. Livers from patients with viral hepatitis without and with HCC (n = 114) were examined to confirm the relevance of altered kinase patterns. In vivo tumorigenic potential of heat-treated versus untreated HCC cells was studied in nude mice. Heating to 55°C killed all HCC cells, whereas 65%-85% of cells survived 48°C-50°C, developing spindle-like morphology and expressing CD133, cytokeratin (CK)7, CK19, procollagen-α1(I), and Snail at day 5 after heat exposure, which returned to baseline at day 12. Heat-exposed HCC cells showed enhanced proliferation and prominent activation of p46-Shc (Src homology and collagen) and downstream extracellular signal-related kinase (Erk)1/2. In patients, Shc expression correlated with malignant potential and overall survival. Blocking Erk1/2 reduced proliferation and EMT-like changes of heat-treated HCC cells. Implantation of heat-exposed HEPG2 cells into nude mice induced significantly larger, more aggressive tumors than untreated cells. Conclusions: Sublethal heat treatment skews HCC cells toward EMT and transforms them to a progenitor-like, highly proliferative cellular phenotype in vitro and in vivo, which is driven significantly by p46Shc-Erk1/2. Suboptimal RFA accelerates HCC growth and spread by transiently inducing an EMT-like, more aggressive cellular phenotype. (Hepatology 2013;58:1667–1680)


7-aminoactinomycin D






bovine serum albumin


carboxyfluorescein succinimidyl ester


chronic hepatitis C




type I procollagen


cycle threshold


division indices


Dulbecco's modified Eagle's medium


epidermal growth factor receptor


epithelial-mesenchymal transition


extracellular signal-related kinase


estimated tumor weight


fluorescein-activated cell sorting


fetal bovine serum


flow cytometry


fibroblast growth factor receptor


hepatocellular carcinoma


heat shock protein


insulin-like growth factor 1 receptor


c-Jun N-terminal kinase


labeling indices


mitogen-activated protein kinase


overall survival


phosphate-buffered saline


platelet-derived growth factor receptor beta


phosphorylated Erk1/2


protein induced by vitamin K absence/antagonist-II


phosphorylated stress-activated protein kinase


quantitative reverse-transcriptase polymerase chain reaction


radiofrequency ablation




standard error of the mean


Src homology and collagen


tumor node metastasis


vascular endothelial growth factor.

Radiofrequency ablation (RFA) is accepted as a potentially curative therapy for the early stages of primary hepatocellular carcinoma (HCC).[1] RFA induces tumor necrosis with low complication rates and is superior to percutaneous ethanol injection in tumor ablation.[2] However, suboptimal RFA treatment for HCC has been reported as a risk factor of early diffuse recurrence.[3] Large tumor size is a major risk factor of local recurrence because of poorly defined margins.[4] Such recurrent HCC appears to behave more aggressively than before RFA[5-8] and significantly reduces overall survival (OS) of HCC patients.[9] Phenotypic and functional alterations of HCC cells subjected to heat treatment have not been studied.

Epithelial-mesenchymal transition (EMT) is thought to be a critical factor in progression of cancer and dictating metastasis. Several oncogenic pathways (such as those of growth and transcription factors, integrins, Wnt/β-catenin, and Notch) can induce EMT.[10] In particular, the Ras/extracellular signal-related kinase (Erk)1/2 pathway has been shown to activate two EMT-related transcription factors, namely, Snail and Slug.[11] A recent clinical study has shown a correlation of Snail transcript levels with capsular and portal invasion of HCC.[12] Other inducers of EMT in HCC are TWIST1 and CHD1L.[13, 14] Moreover, expression of type I procollagen (COL1A1) is a useful marker of transition to a mesenchymal phenotype.[15]

Src homology and collagen (Shc) is a central SH2-containing cytoplasmic adaptor protein, which directly binds to tyrosine kinase receptors, such as epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor beta (PDGFRβ), insulin-like growth factor 1 receptor (IGF-1R), and fibroblast growth factor receptor (FGFR). Notably, Shc is a central player in malignant transformation.[16-20] Mitogenic, transforming, and proinvasive signal transduction from Shc to mitogen-activated protein kinases (MAPKs) by Grb2-Sos has been well studied,[21-23] with p46-Shc and p52-Shc being central upstream regulators of MAPK activation, whereas the alternatively spliced p66-Shc isoform appears to promote apoptosis.[24, 25] However, the role of Shc isoforms in liver diseases is still poorly understood. Although we previously reported that p46-Shc phosphorylation is a hallmark of hepatocarcinogenesis and liver regeneration in rats,[26, 27] the role of Shc in human HCC has not been studied yet.

Here, we demonstrate that sublethal heat treatment of HCC cells, as might occur in marginal zones of RFA therapy, endows these cells with a higher proliferative and carcinogenic potential in vitro and in vivo. These properties are linked to EMT-like changes and appear driven by p46-Shc and Erk1/2 activation.

Materials and Methods

Cell Cultures and Heat Treatment

Adherent monolayers of HEPG2, HuH7, and HEP3B hepatoma were grown to 70% confluence, trypsinized, washed in Dulbecco's modified Eagle's medium (DMEM), collected in 1.5-mL Boilproof Microtubes (#MAX-815; Phenix Research Products, Candler, NC) in 1 mL of medium (5 × 105 cells), and immediately exposed to heat shock using a digital dry bath incubator (ISOTEMP 145D; Thermo Fisher Scientific, Waltham, MA) at temperature settings of 37˚C, 45˚C, 50˚C, and 55˚C for 10 minutes. Cells were then seeded into 75-cm2 cell-culture flasks in 15 mL of DMEM with 10% fetal bovine serum (FBS) and maintained at 37˚C. DMEM was exchanged three times per day until day 3 after heating to remove debris and dead cells. At day 3 or 5 after heating, surviving HCC cells were subcultured into new 75-cm2 flasks after adjustment of cell numbers to 1 × 106.

Flow Cytometry for Cell Proliferation and Apoptosis

At day 3 after heating, cells were trypsinized, washed with 0.1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS), and resuspended to 1 × 106 cells/mL in 0.1% BSA/PBS.[28] An equal volume of 10 µM of carboxyfluorescein succinimidyl ester (CFSE; catalog no. C34554; CellTrace CFSE Cell Proliferation kit; Invitrogen, Carlsbad, CA) in 0.1% BSA/PBS was added to the cell suspension and incubated in the dark at 37°C for 10 minutes, followed by washing with 1% BSA/PBS and seeding into flat-bottomed six-well culture plates (catalog no. 353046; Falcon; BD Biosciences, San Jose, CA). After 48 hours (72 hours for HEP3B), cells were trypsinized, washed with PBS, and analyzed for CFSE staining by flow cytometry (FCM). For general FCM, cells (1 x 105) were incubated with appropriate antibody for 30 minutes on ice, washed with 0.1% BSA/PBS, incubated with 0.1 µg/mL of propidium iodide[28] for 5 minutes, and analyzed on a FACSCalibur flow cytometer (Becton Dickinson, San Diego, CA). FlowJo software (TreeStar, San Carlos, CA) was used to analyze all proliferation data.

For analysis of apoptosis, 1 × 105 cells/mL were trypsinized and collected 24 h after heat treatment using fluorescein-activated cell sorting (FACS) buffer (1 × PBS, 2% FBS, and 0.1% sodium azide), fixed in 2% paraformaldehyde in FACS buffer for 15 minutes, washed, and stained with 10 µL of Annexin V/fluorescein isothiocyanate or 20 µL of 7-aminoactinomycin D (7-AAD). For cytokeratin (CK)7/19 staining, cells were permeabilized with 0.1% Triton X-100 for 15 minutes at 4°C. FCM data were analyzed by FlowJo software (TreeStar).

Patient Samples

Liver biopsy specimens were obtained from 50 patients with chronic hepatitis C (CHC) and without detectable HCC before starting antiviral therapy. Patients' characteristics are given in Table 1A. F3-F4 of histological stage was denoted as “advanced fibrosis.” Another 60 biopsies and 4 operative specimens were obtained from HCC tissues of patients with chronic hepatitis B (n = 6), C (n = 52), B and C coinfection (n = 2), or other causes (alcoholic: n = 1; primary biliary cirrhosis: n = 1; Wilson's disease: n = 1; unknown: n = 1; Table 1B). Four of the HCC samples and clinical data used in the current study were from a previous publication.[29] Tumor node metastasis (TNM) stage of HCC was determined according to the criteria of the International Union against Cancer and the American Joint Committee on Cancer,[30] and histological grading was performed according to the criteria of an International Working Party.[31] Tissues had been frozen for western blotting immediately at −80°C or formalin-fixed. Informed consent was obtained from each patient before study participation.

Table 1. Clinical Features of Patients Without and With HCC
  1. a

    Abbreviations: SD, standard deviation; HBV, hepatitis B virus; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated.

  2. b

    n = 50.

  3. c

    n = 64.

A. Patients Without HCCa 
Age, years, mean ± SD (range)62.7 ± 7.4 (47-74)
Viral infection, HCV(+)50
Histological background, fibrosis 
B. Patients With HCCb 
Age, years, mean ± SD (range)61.8 ± 7.6 (45-78)
Gender, male/female34/30
Viral infection 
Histological background, fibrosis 
Maximum size of tremor, mm 
Histological grade 
TNM stage 
Serum AFP, ng,mL 
Serum AFP-L3, % 

Malignant Potential of Heat-Treated HEPG2 Cells In Vivo

Female 8-week-old Balb/c nude (nu/nu) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Animal studies were approved by the Institutional Animal Care and Use Committee of the Beth Israel Deaconess Medical Center (Boston, MA).

HEPG2 cells were maintained at 37˚C for 3 days in vitro after a 10-minute exposure to 37˚C (controls), 45˚C, 48˚C, or 50˚C, followed by harvest with trypsin/ethylenediaminetetraacetic acid and resuspension in 50% growth-factor–reduced Matrigel (BD Biosciences) in PBS to a final cell count of 2.5 × 107 cells/mL. A volume of 0.2 mL of the cell suspensions was injected subcutaneously (SC) in the right flank of each mouse (6 mice per group), as described before.[32] Estimated tumor weight (ETW) was calculated every 2 days after injection using the following formula: ETW (mg) = Length (mm) × (Width (mm))2/2).[32] Fifteen days after tumor cell injection, animals were sacrificed and tumors were harvested and immediately stored in −80˚C for further analysis.

Statistical Analysis

Data are expressed as means ± standard error of the mean (SEM). Statistical analyses were performed using Microsoft Excel (Microsoft Corp., Redmond, WA) and GraphPad Prism (version 5.00; GraphPad Software Inc., San Diego, CA). Multiple comparisons were performed by one-way analysis of variance. Two planned comparisons were performed to each of the control groups using Dunnett's post-test. The OS curve of patients with positive Shc-labeling indices (LI; %) was plotted using Kaplan-Meier's method, and differences were analyzed statistically by the log-rank test. Differences among selected experimental groups with P values <0.05 were considered significant. Correlation coefficients were calculated by GraphPad Prism (version 5.00; GraphPad Software Inc.). A two-tailed P value was selected and confidence intervals were set to 95%. R > 0.5 was considered a strong correlation.


Survival of HCC Cells After Heat Treatment

As revealed by AnnexinV-7/AAD double staining, when exposed to 45˚C for 10 minutes, more than 96% of HEPG2 and 91.6% HuH7 cells had survived after 24 hours without signs of apoptosis, whereas survival was decreased to 65.6% and 87.6%, respectively, at 50˚C, and all cells died after exposure to 55˚C (Fig. 1A).

Figure 1.

Sublethal heat experiment generates HCC cells with high survival and proliferation capacity. (A) Three HCC cell lines were exposed to heat for 10 minutes, followed by maintenance at 37°C for 24 hours and FCM after double labeling for Annexin V and 7-AAD. When exposed to 45°C, more than 96% of HEPG2 and 91.6% of HuH7 cells had survived after 24 hours without signs of apoptosis, whereas survival was reduced to 65.6% and 87.6%, respectively, at 50°C, and all cells died at 55°C. (B) HCC cells were labeled with CFSE, and DI, as determined at day 5 or 6, were significantly higher in cells exposed to 50°C, compared to cells kept at 37°C. (C) Significant increase of transcripts related to cell proliferation in HCC cells after exposure to 50°C, as compared to 37°C, in all three cell lines at day 5 after heat treatment. Ki-67 transcripts, were elevated in HuH7 and HEP3B, as well as CyclinD1 transcripts in HEP3B cells after exposure to 45°C. Most of these expression patterns returned to baseline at day 12 after heat exposure. Error bars: ± SEM of three or more independent experiments. *P < 0.05; **P < 0.005; ***P < 0.0005.

Enhanced Proliferation of HCC Cells After Heat Treatment

Division indices (DI) of three HCC cell lines were determined after CFSE labeling, followed by FACS analysis. In all cell lines that were exposed to 50˚C, DI at day 5 (or 6) was significantly higher than in cells kept at 37˚C (Fig. 1B). This was paralleled by a significant increase of proliferation-related transcripts (Ki-67 and CyclinD1), with Ki-67 transcripts being already elevated in two cell lines after exposure to 45˚C (Fig. 1C).

Morphological and Phenotypic Changes After Heat Treatment

Distinct morphological changes, such as appearance of spindle-like cells (Fig. 2A), were only observed for HEPG2 cells exposed to 50˚C on day 5.

Figure 2.

Heat treatment induces CK7 and 19 and EMT-like morphological changes in HCC cells. (A) Spindle cell-like morphology in HEPG2 cells at day 5 after exposure to 50°C (arrows), as well as reversal to epitheloid morphology at day 12. (B) After permeabilization, HEPG2 cells were stained for CK7 and 19 at day 5 or 12 after heat treatment, followed by FCM. Both CK7 and 19 were significantly increased in heat-treated cells at day 5. (C) Data were confirmed for CK19 by quantitative western blotting. Increased CK19 returned to baseline at day 12 of heat treatment. Error bars: ± SEM of three or more independent experiments. *P < 0.05; **P < 0.005.

Intracellular staining, followed by FCM, demonstrated that the cholangiocyte markers, CK7 and CK19, were increased in HEPG2 cells at day 5 after exposure to 50˚C, whereas low baseline expression was observed in cells exposed to 37˚C or 45˚C (Fig. 2B). This was confirmed by western blotting for CK19 (Fig. 2C) and immunohistology (data not shown).

Lineage-Specific Transcript Profiles in HCC Cells After Heating

Cell phenotype-related transcript levels were analyzed in the three HCC cell lines by using quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR; Fig. 3). CK7 and 19 transcripts were dramatically elevated on day 5 day after exposure to 50˚C, compared to no or minor changes after treatment at lower temperatures, but this increase was transient and levels returned to or near baseline on day 12. Transcript levels of the putative stem cell and progenitor marker, CD133, showed similar kinetics, whereas the hepatocyte differentiation marker, albumin (ALB) was significantly reduced on day 5, to rise baseline on day 12.

Figure 3.

Heat-exposed HCC cells shift from a hepatocyte-like to a cholangiocytic/progenitor-like phenotype. Phenotype-related transcript levels (CK7, CK19, ALB, and CD133) were measured by qRT-PCR. All three HCC cell lines showed a significant increase of CK7, CK19, and CD133 transcripts after exposure to 50°C, as compared to 37°C, whereas ALB expression was suppressed. Changes reverted to baseline at day 12. Cycle numbers (Ct values) are shown above bars in each graph. Error bars: ± SEM of at least three independent experiments. *P < 0.05; **P < 0.005; ***P < 0.0005.

Expression of four central EMT markers (Snail, TWIST1, CHD1L, and COL1A1) was increased in all three HCC cell lines 5 days after treatment at 50˚C (Supporting Fig. 1). Most of these increases were significant or highly significant, especially for COL1A1. The transcript level (cycle threshold [Ct] value) for COL1A1 almost resembled that found in the well-established, activated human hepatic stellate cell line, LX-2[33] (Supporting Fig. 2). Tissue inhibitor of matrix metalloproteinase, another EMT-related marker, showed a similar trend (Supporting Fig. 3). Of note, EMT-related transcript levels returned to baseline on day 12 post–heat treatment. EMT-like changes, enhanced invasiveness, and migration of HEPG2 cells were confirmed by a 5- to 8-fold increased protein expression of Snail at day 5, but also at day 12 after heat treatment, and a significantly enhanced level of in vitro HEPG2 and HuH7 invasions at day 5 (Fig. 4A,B).

Figure 4.

and cell migration in hepatoma cells increase at day 5 after heat treatment and are associated with higher Snail expression. (A) Western blotting of Snail extracted from HEPG2 cells heat-pretreated at 45°C, 48°C, or 50°C or left untreated (37°C). Loading was normalized to β-actin, and band intensity was quantified by densitometry. Snail is undetectable in untreated HEPG2 cells, but is clearly up-regulated after treatment at 48°C and 50°C. NIH-3T3 cell lysate was used as a positive control (“pos” in the band and “+” in the graph). (B) Tumor invasion assays with HEPG2 and HuH7 cells. Hepatoma cell lines were heat-pretreated at 50°C or left untreated. Invading cells increased significantly after heat treatment. Erk1/2 inhibition clearly reduced the invasion of heat-treated HEPG2 cells. Error bars: ± SEM of at least three independent experiments. *P < 0.05.

Shc Expression and Phosphorylation Are Increased in Heat-Exposed HCC Cells

Preheating (50˚C) preheating increased Shc transcript levels 5.0-, 2.8-, and 5.1-fold at day 5 in HEPG2, HuH7, and HEP3B, respectively (Fig. 5A). On the protein level, this was paralleled by enhanced expression of p46-Shc and phosphorylated Erk1/2 (p-Erk1/2; Fig. 5B). The other two major phosphorylated MAPKs (phosphorylated stress-activated protein kinase/c-Jun N-terminal kinase [p-SAPK/JNK] and p38 MAPK) only increased insignificantly after heat treatment (Supporting Fig.4). Phosphorylation levels returned to baseline at day 12 after heat treatment. Notably, expression of heat shock protein (HSP)27, 70, and 90 was significantly increased at day 5 post–heat treatment temperature dependently and also reverted to baseline levels at day 12 (Supporting Fig. 5).

Figure 5.

Phosphorylated Shc and Erk1/2 are up-regulated in heat-treated hepatoma cells and correlate with malignant potential of human HCC. (A) Increased Shc transcript levels in three HCC lines after heat treatment at 50°C, as determined by qRT-PCR. Changes reverted to baseline at day 12. Cycle numbers (Ct values) are shown above bars in each graph. (B) Shc, Erk1/2, and their phosphorylated forms (p-Shc and p-Erk1/2) in HEPG2 cells after heat treatment were determined by western blotting and densitometry. Up-regulation of p46Shc and p-p46Shc at day 5 after heat treatment at 45°C or 50°C, compared to untreated controls. Exposure to 50°C induced the highest p-p46Shc and p-Erk1/2 (p42/p44) levels, with an almost 40-fold increase in p44, whereas p-p52Shc was reduced. (C) Shc protein expression in normal liver (NL), cirrhosis resulting from CHC, well-differentiated (WD), moderately differentiated (MD), and poorly differentiated (PD) HCC, respectively. Magnification, ×200. (D) LI (in % of Shc-positive hepatocytes) were determined by counting of five high-power fields per section at ×200 magnification. (E) Kaplan-Meier curves of OS according to Shc expression (more than 65% positive staining versus less than 65% positive staining) in 84 patients with advanced fibrosis, which includes non-HCC (F3-F4) and HCC (HBV: n = 6; HCV: n = 54; HBV/HCV: n = 2; other: n = 4), showing a significantly shorter survival in patients with high Shc expression (P < 0.0316; log-rank test). (F) Quantitation of Shc and phosphorylated Shc (p-Shc) from livers of patients with chronic hepatitis without cirrhosis, cirrhosis, or HCC by western blotting and densitometry. Total p52Shc was significantly increased in HCC sections, as compared to non-HCC tissue, whereas total p46Shc was increased insignificantly. On the other hand, both phosphorylated p46Shc (p-p46Shc) and p-p52Shc were significantly up-regulated in HCC versus chronic hepatitis and normal liver. Error bars: ± SEM of at least three independent experiments. *P < 0.05; **P < 0.005; ***P < 0.0005. HBV, hepatitis B virus.

Shc Expression Predicts Survival in Human HCC

Liver specimens from 64 HCC patients, 20 patients with cirrhosis, and 30 subjects with CHC without cirrhosis were examined for Shc expression (Fig. 5C). Shc staining was absent in healthy liver, but dramatically increased in HCC tissue, whereas samples with CHC without cirrhosis showed an intermediate expression (P < 0.0005 for HCV cirrhosis versus HCC and for HCV without cirrhosis versus HCV cirrhosis; Fig. 5D).

Next, we formed two groups with high and lower Shc-LIs (≥65% or <65%; n = 54 and n = 30, respectively) in patients with advanced fibrosis (without and with HCC). When comparing both groups by Kaplan-Meier's analysis, OS rate of patients with Shc-LI ≥65% was significantly lower than with Shc-LI <65% (P = 0.0316; Fig. 5E).

When Shc-LI (%) in these patients was compared with hematological parameters associated with hepatocarcinogenesis (alpha-fetaprotein [AFP]-L3%, AFP, and protein induced by vitamin K absence/antagonist-II [PIVKA-II]) and liver function (alanine aminotransferase, aspartate aminotransferase, total bilirubin, alkaline phosphate, gamma-glutamyl transpeptidase, ALB, and platelet count) in all samples (Supporting Table 3), a strong correlation was only found with AFP-L3 (%) (r = 0.5312; P < 0.0001). However, no strong correlation between Shc-LI and the other parameters was observed.

Comparison of Shc Expression and Its Tyrosine Phosphorylation Among Human Liver Specimen by Western Blotting

Expression of both phosphorylated Shc-variants, p46- and p52-Shc, was assessed by semiquantitative western blotting in homogenized lysates of human liver specimens (Fig. 5F). Although p52-Shc was strongly expressed in both cirrhosis and HCC specimens (p = 0.0374 and p = 0.0054, respectively), p46-Shc was detected only in HCC, whereas no p66-Shc could be detected in any samples. In all HCC samples, phosphorylated p46-Shc expression was much stronger than phosphorylated p52-Shc expression (P = 0.0313).

Blockade of Erk1/2 Signaling Blunts Proliferation and Reverses EMT-Like Changes of Heat-Treated HCC Cells

Five days after heat treatment (50˚C), HEPG2 cells were exposed to the Erk1/2 inhibitor, U0126, whereas HEPG2 cells kept at 37˚C served as controls. Notably, effective Erk1/2 inhibition, as evidenced by complete suppression of Erk1/2 phosphorylation (Fig. 6A), blunted enhanced proliferation (Fig. 6B) and essentially normalized (or reduced) all parameters related to EMT, except for significantly reduced, but still elevated, CK19 and COL1A1 (Figs. 4 and 6C,D).

Figure 6.

Selective inhibition of Erk1/2 activation blunts proliferation and EMT-like changes in HCC cells exposed to sublethal heat. (A) Effective inhibition of Erk1/2 kinase by U0126 (10 μM) was confirmed by western blotting. U0126 was added to untreated HEPG2 cells or 3 days after heat treatment at 50°C. Western blotting was performed after 48 hours. (B) U0126 significantly reduced the cell division index of CFSE-labeled HEPG2 cells after heat treatment (50°C; P < 0.0005), as well as Ki-67 and CyclinD1 transcript levels. (C) U0162 down-regulated EMT-related transcript levels of Snail, CHD1L, and COL1A1 in heat-exposed HEPG2 cells. (D) U0126 significantly suppressed cholangiocyte/progenitor cell-related transcript levels of CK19 and CD133 in heat-treated HEPG2 cells. Error bars: ± SEM of at least three independent experiments. *P < 0.05; **P < 0.005; ***P < 0.0005.

Heat-Exposed HCC Cells Induce More Aggressive Tumors In Vivo

To evaluate how far heat pretreatment may cause a more aggressive (EMT-like) tumor growth in vivo, we transplanted heat-exposed HEPG2 cells into nude mice. According to our in vitro profiling of both HEPG2 and HuH-7 cells, we expected the highest rate of proliferation and EMT-like changes between days 3 and 5 after heat treatment at 48˚C or 50˚C (Supporting Figs. 6 and 7). Therefore, 5 × 106 HEPG2 cells kept at 37˚C, or pretreated at 45˚C, 48˚C, or 50˚C, were SC implanted on day 3 after heating. Tumor formation and ETW were evaluated every 3 days, and at day 15 after implantation, all mice were sacrificed. ETW showed that HCC grew faster in the 48˚C and 50˚C groups than in the 37˚C group (Fig. 7A). No mice died before sacrifice, and absence of tumor growth was observed in 1 mouse each of the 37˚C and 45˚C groups (Supporting Table 4). Median tumor weight was 298, 202, 57.5, and 19.5 mg in the 50˚C, 48˚C, 45˚C, and 37˚C groups, respectively (Fig. 7B; Supporting Table 4). Western blotting in harvested tumors showed higher p-Erk/Erk (p42/p44) ratio in the 48˚C and 50˚C groups than in the 37˚C group (P < 0.05 and P < 0.05, respectively; Fig. 7C). However, no significant changes were detected in Shc and p-Shc expression among these four groups (data not shown). Ki-67 positivity was higher in the center than in the periphery of tumors (P < 0.05 for the 48˚C and 50˚C groups; Fig. 7D). Transcript levels of Ki67 and of the EMT markers, TWIST1 and COL1A1, were significantly elevated in the 50˚C group, compared to the 37˚C group (P < 0.05; Fig. 7E). Other EMT or stem-cell–related transcripts showed no significant difference and a trend to be increased at best. Using hematoxylin and eosin histology, there was no difference in necrosis, vacularization, or invasiveness of the tumors of the four implantation groups. Moreover, the amount of Snail protein between the four groups was comparable (Supporting Fig. 8). Similarly, the groups did not show significant differences in pancytokeratin, CK7, CK19, and apoptosis (terminal deoxynucleotidyl transferase dUTP nick end labeling staining; data not shown).

Figure 7.

Heat-treated HCC cells induce larger, more aggressive tumors in nude mice. HEPG2 cells were pretreated at 45°C, 48°C, or 50°C for 10 minutes or were maintained at 37°C. After 3 days in culture, 5 × 106 cells were mixed with 50% Matrigel in 0.2 mL of PBS and injected SC in the right flank of nude mice (n = 6 mice per group). (A and B) HEPG2 exposed to 48°C and 50°C displayed faster HCC growth, as assessed by ETW and a significantly higher tumor weight at sacrifice (15 days after implantation), compared to tumors arising from untreated cells. *P < 0.05; **P < 0.005. (Individual data are shown in Supporting Table 4.) Mean tumor weight was highest in the 48°C group, and median weight was highest in the 50°C group. (C) Continuous up-regulation of p-Erk (p42/p44) in tumors generated with HEPG2 cells pretreated at 48°C and 50°C at harvest 15 days after implantation. (D) Numbers of Ki-67-positive HCC cells, as assessed in five high-power fields per tumor (magnification, ×400), were significantly increased in inner tumor mass and invasive front of the 48°C and 50°C versus the 37°C and 45°C groups. (E) Expression of EMT-related genes (qRT-PCR from HCC tissue). Only COL1A1 and Ki-67 expression demonstrated a significant difference (P = 0.0105 and 0.0214, respectively). Error bars: ± SEM of at least three independent experiments. *P < 0.05.


Local recurrences of HCC can progress rapidly after RFA,[5, 6], and cancer cells up-regulate CK19 (i.e., a feature of cholangiocarcinoma and hepatic progenitor cells).[34, 35] Recent studies also describe other progenitor cell biomarkers, such as CD133, that characterize HCC with enhanced malignant potential.[36, 37] Here, we demonstrate that hepatoma cells that were exposed to sublethal heat for 10 minutes adopted molecular and functional characteristics of hepatic progenitors (CK7, CK19, and CD133), coupled with increased proliferation, up-regulation of genes that are involved in EMT (TWIST1, Snail, COL1A1, and CHDL1) and an enhanced malignant potential in vivo. Moreover, the observed EMT and aggressiveness of HCC cells exposed to sublethal heat were dependent on activation of the MAPKs, Erk1/2 (and upstream Shc).

We describe and analyze, for the first time, that heat-exposed HCC cells undergo EMT-like changes, gain progenitor characteristics, and display a higher malignant potential. We note that activation of Erk1/2 has recently been identified as an important regulator of EMT in tight association with Snail.[38, 39] Three types of EMT are known: Type 1 describes the invasion of transitional cells into the mesenchyme during development; type 2 occurs when epithelia transform into myofibroblast-like cells during wound healing and repair; and type 3 is the adoption of mesenchymal properties by cancer cells that permit their infiltration and migration into the circulation to generate distant metastases.[40] Thus, Snail and TWIST1 can induce type 3 EMT in pancreatic and breast cancer cells,[41, 42] and TWIST1 protein has also been shown to directly trigger type 3 EMT and promote invasion by activation of Ras.[43] Taken together, our data indicate that sublethal heat exposure of HCC promotes type 3 EMT by induction of Snail, TWIST1, and other (functional) EMT markers and upstream p46Shc-Erk1/2 activation.

Another novel finding of our study is the likely important upstream role of Shc in HCC progression in general and after sublethal heat treatment. p46-Shc and its phosphorylation are clearly enhanced after heat exposure, which is upstream of p-Erk1/2 activation, whereas p-SAPK/JNK and p38 MAPK remained unchanged. Shc functions as an adapter molecule of the EGFR and other tyrosine kinase receptors, such as PDGFRβ, IGF-1R, and FGFR, involved in oncogenic activation.[16-20] The signaling cascade induced by Shc activation, named the alternative pathway,[44] is thought to be a master regulator of tumor growth, differentiation, and development.[17, 45] p-Erk1/2 itself is a well-known regulator of cell proliferation, malignant transformation, and tumor progression.[46] Enhanced expression of Shc, especially activated p46-Shc, is a general phenomenon in hepatocarcinogenesis.[27] In this line, we showed that Shc expression strongly correlated with a serum marker of enhanced malignant potential (AFP-L3) and overall patient survival. Heat treatment (50°C) activated p46-Shc in HEPG2 cells and activated p-Erk1/2. Thus, these data suggest that p46-Shc expression, and its activation by phosphorylation, is a central switch for activation of Erk1/2, which then accelerates both malignant transformation and tumor progression in HCC after sublethal heat exposure.

In our previous study of spontaneous hepatocarcinogenesis in the Long-Evans Cinnamon rat, we showed that total activated Shc (p46- and p52-Shc) was highly increased in hepatoma cells, with a prominent activation of p46-Shc in HCC specimens.[27] We could also demonstrate that p46-Shc was strongly up-regulated in the early stage of liver regeneration in rats with 70% hepatectomy, supporting its role also as a primary inducer of hepatocyte regeneration.[26] At present, it is unclear why only phosphorylated p46Shc is up-regulated during proliferation.

A recent report that employed the hepatic implantation model of VX2 carcinoma (an anaplastic squamous cell carcinoma in rabbits) supports our findings, although conditions were less well defined and the data largely descriptive. The researchers carried out RFA with three temperature settings (55°C, 70°C, and 85°C) for 5 minutes.[47] Only the 55°C treatment significantly induced a more rapid progression of remnant tumor, with increased protein levels of proliferating cell nuclear antigen, matrix metalloproteinase 9, vascular endothelial growth factor (VEGF), hepatocyte growth factor, and interleukin-6 in tumor tissue. Another report showed that an aggressive and heat-resistant subclone of HepG2 cells expressed higher levels of hypoxia-inducible factor 1 alpha and VEGF-A in vitro and in nude mice in vivo, suggesting that heat treatment may also promote tumor angiogenesis.[48]

HSPs are known inhibitors of apoptotic cell death and inducers of proliferation and invasion/metastasis in gastrointestinal cancer cells.[49] We demonstrated that heat treatment at 48°C and 50°C significantly induced HSP27, HSP70, and HSP90 proteins at day 5. The elevated expression of these HSPs very likely contributes to the higher malignant potential of the heat-treated hepatoma cells, as was previously reported.[50]

Notably, we demonstrated that inhibition of Erk1/2, which is downstream of Shc, almost normalized Ki-67, CyclinD1, Snail, COL1A1, CK19, and CD133 expression and almost completely reverted the EMT- and progenitor-like phenotype of heat-exposed HCC cells.

Importantly, we could verify our in vitro data by implanting HEPG2 cells into nude mice with or without previous heat treatment. Here, cells with previous exposure to 48°C/50°C induced an impressive, higher in vivo tumor growth, which was accompanied by enhanced proliferation and up-regulation of some, but not all, markers of EMT.

The finding of only modest EMT-like changes in vivo at the day of tumor harvest (18 days after implantation of cells) can be explained by the kinetics of the observed EMT in vitro, because all the changes found in heat-exposed HCC cells at day 5 returned to baseline at day 12. Our data also indicate that the heat-induced EMT-like changes with activation of p46-Shc and Erk1/2 in HCC are reversible in vivo as well, and that invasion and metastasis of HCC may also transiently occur within a short time frame after RFA. The reversibility of the EMT phenotype is important because it largely rules out that heat treatment generates aggressive subclones, and thus potential artifacts, but rather shows that this mechanism is operative and of relevance for in vivo HCC in general. This knowledge may provide a rational basis for the short-term use of adjuvant antiproliferative therapy in RFA-treated HCC. Moreover, we show that heat exposure affects HCC cell lines to a different degree. Because this will likely apply to subclones of HCC cells in vivo, observation of single hepatoma cell clones may not be representative of overall tumor response.[48]

In conclusion, although RFA is currently an important, potentially curative therapy for HCC, inadequate or sublethal treatment might induce a higher malignant potential and cause recurrent HCC with a worse prognosis. Our data underline the need to apply sufficiently high temperatures and secure wide therapeutic margins, in an attempt to completely eradicate all residual HCC within the treated focal lesion. Our results further suggest that p46-Shc expression and its phosphorylation may be a strong predictor of malignant transformation, tumor invasion, and metastasis of HCC, and that its downstream effecter Erk1/2 is key to EMT-like changes that confer an enhanced malignant potential in insufficiently heat-treated HCC cells. Finally, Erk1/2 (or further upstream molecules) may be an attractive therapeutic target for a short-term adjuvant therapy to prevent HCC recurrence after RFA therapy.