Epigenetic combination therapy as a tumor-selective treatment approach for hepatocellular carcinoma




Innovative epigenetic therapeutics comprise histone deacetylase inhibitors (HDAC-I) and demethylating agents (DA). It was recently found that HDAC-I compounds exhibit profound therapeutic activities against hepatocellular carcinoma (HCC). A comprehensive preclinical investigation was performed on the potential of a combined HDAC-I/DA epigenetic regimen for the highly chemotherapy-resistant HCC entity.


Human HCC-derived cell lines or primary human hepatocytes (PHH) were treated with HDAC-I compound suberoylanilide hydroxamic acid (SAHA) or DA compound 5-aza-2′-deoxycytidine (5-aza-dC) or both and examined for cellular damage, proliferation, histone acetylation pattern, and DNA methylation. In vivo activities were investigated in a xenograft hepatoma model.


Monotherapeutic application of SAHA or 5-aza-dC was found to induce substantial antiproliferative effects in HCC-derived cells, strongly enhanced by combined SAHA and 5-aza-dC treatment. PHH from different human donors did not exhibit any relevant cellular damage even when applying high doses of the combination regimen, whereas HCC-derived cell lines showed a dose-dependent damage. In vivo testing demonstrated a statistical significant inhibition of hepatoma cell growth for the combined treatment regime.


Because the combined HDAC-I/DA epigenetic approach was found to produce significant antitumor effects in HCC model systems and did not impair cellular integrity of untransformed hepatocytes, this combination therapy is now considered for further investigation in clinical trials. Cancer 2007;109:2132–41. © 2007 American Cancer Society.

The possibility to modulate epigenetic alterations of tumor cells, such as the acetylation pattern of DNA-associated histone proteins by histone deacetylase inhibitors (HDAC-I) or the methylation of cellular DNA by substances with demethylating activity (DA),1–3 opens up fascinating new treatment options, in particular for those tumors that exhibit an inherent resistance to cytostatic agents, as for hepatocellular carcinoma (HCC).4, 5

In human-derived HCC tumor tissues, compared with nonmalignant liver tissue, aberrant DNA methylation patterns have been described for several genes such as SOCS-1,6–8 p16INK4A,9, 10 cyclin D2,6 APC,6, 7 E-cadherin and p15INK4B,7 GADD 45,11 HAI-2/PB,12 or ASCL.13 To our knowledge, the exact contribution of these methylation patterns to the development and maintenance of hepatoma cells has not been determined yet in detail. Notably, atypical methylation patterns have also been found in precancerous lesions of the liver6 and have been correlated to the extent and severity of liver fibrosis.14

5-azacytidine (Vidaza; Pharmion, Boulder, Co) and 5-aza-2′-deoxycytidine (5-aza-dC; decitabine), constitute prototype DNA methylation inhibitor compounds that can be incorporated into the cellular DNA.15 Both substances, 5-azacytidine and 5-aza-dC, have been approved by the U.S. Food and Drug Administration (FDA) for the treatment of myelodysplastic syndrome in 2004 and 2006, respectively. Once incorporated, they are recognized by DNA methyltransferases (DNMTs) during replication, thereby leading to a widespread genomic hypomethylation.16, 17 The results of a recent Phase I trial demonstrated that a continuous infusion of 5-aza-dC is well tolerated in patients with solid tumors, paving the way for future clinical trials employing DA.18

First-generation HDAC-Is, such as the short chain fatty acid MS-275, the cyclic peptide depsipeptide (FK-228), or suberoylanilide hydroxamic acid (SAHA), are currently being tested in Phase I/II clinical trials in hematologic malignancies and some solid tumors, but not yet in patients harboring HCC.3, 19–21 Several early studies reported that a remodeling of the chromatin structure by HDAC-I might be a promising concept to treat HCC.22–24 Our own results addressing HCC-derived cells show that valproic acid (VPA) exhibits profound HDAC-I activity,25, 26 and that a new compound called ITF2357 selectively inhibits proliferation and induces apoptosis of hepatoma cells.27 Furthermore, we found an HDAC-I induced preferential killing of tumor cells by natural killer cells28 and a sensitization to the apoptosis-inducing ligand TRAIL.27 Thus, HDAC-I seems to be an attractive new treatment option for HCC.

Molecular links between the modification of histone proteins and the methylation of DNA have led to the idea of combining both principles as a new epigenetic combination strategy in cancer; however, the exact molecular mechanisms of an enhancement of the antitumoral activity compared with each substance alone is not yet completely understood.2, 29–33 In the meantime, early clinical studies mainly addressing toxicity issues could demonstrate the feasibility of such an approach.34, 35

In the current study, we investigated the effects of the DA substance 5-aza-dC alone or in combination with the HDAC-I compound SAHA (both drugs being currently employed in Phase I/II clinical trials) on human hepatoma cell lines as well as on primary human hepatocytes (PHH). To preclinically explore this new HCC treatment strategy, both single and combination regimens were also evaluated in a human hepatoma xenograft model.


Reagents and Cells

5-aza-2′-deoxycytidine (5-aza-dC; Sigma-Aldrich, Taufkirchen, Germany) and suberoylanilide hydroxamic acid (SAHA; Italfarmaco, Milan, Italy) were both dissolved at a concentration of 10 mM. PHH were kindly provided by T.S. Weiss according to Thasler et al.36 and cultured as described.27, 37 Hep3B and HuH7 cells were maintained in Dulebecco modified Eagle medium (DMEM) (10% fetal calf serum [FCS]), and HepG2 in MEM and DMEM (3:1; 10%FCS). Media and supplements were from Life Technologies (Rockville, MD).

Sulforhodamin B Cytotoxicity Assay

Hep3B, HepG2, and HuH7 cells were seeded in 12-well plates (2 × 104 cells/well). One day later, medium was added containing 5-aza-dC, SAHA, or both; treatment was performed for 5 days, and the medium was changed once after 48 hours. Growth inhibition was evaluated by sulforhodamin B (SRB) assay38; data plotted represent the mean of optical density measurements related to untreated cells (100% cellular protein content). In an analysis of variance (ANOVA) with factors 5-aza-dC, SAHA, cell line, and all interactions, we tested contrasts between 5-aza-dC monotreatment and combinations with SAHA for each cell line and the system of contrasts (JMP IN 5.1 software; SAS Institute Inc., Cary, NC).

Methylation-Specific Polymerase Chain Reaction

Semiconfluent HepG2 or HuH7 cells were treated with 5-aza-dC, SAHA, or both substances and 48 hours later an extraction of genomic DNA (Qiagen, Hilden, Germany, DNeasy protocol) was performed. The bisulfite modification of genomic DNA was performed as described with some minor modifications.39 In brief, genomic DNA (1 μg) was denatured in 50 μL by NaOH (0.2 M final concentration) for 20 minutes at 42°C. Next, 30 μL of 10 mM hydroquinone (Sigma Chemical Company, St. Louis, Mo) and 520 μL of 3 M sodium bisulfite (Sigma) (pH 5.0) were added and the sample was incubated for 18 hours at 60°C. DNA was purified (Qiagen PCR Purification Kit) and eluted into 250 μL of water; 125 μL of NaOH (1 M) then were added and incubated for 15 minutes at 37°C, followed by ethanol precipitation. A methylation-specific polymerase chain reaction (PCR) was performed using previously described primers for the genes SOCS-17 and p16INK4A.39 Each PCR product was analyzed on a 2% agarose gel.

HDAC Inhibitor Screening

Histone deacetylase inhibition activity was determined in triplicate according to the Cayman Chemical HDAC Activity/Inhibitor Screening Assay Kit protocol (Cayman Chemical by Biozol, Eching, Germany).

Western Blot Assays

Western blot assays were performed as described40 employing rabbit anti-acetyl-histone H3 antibody (Ac-H3), 1:20,000 (Biomol, Hamburg, Germany), and anti-vinculin antibody, 1:5000 (Sigma-Aldrich). Acetyl-Histone H3 levels were estimated by performing a densitometric analysis; values are shown relative to untreated controls.

Cytotoxicity Assay for Human Hepatoma Cell Lines and PHH

PHH were cultured for 3 days in 12-well plates (4 × 105 cells/well). Hep3B, HepG2, and HuH7 cells were seeded in 12-well plates (1 × 105 cells/well). The next day, 5-aza-dC, SAHA, or a combination of both, were added. After 48 hours, culture media were collected and analyzed using commercial procedures for activity of lactate dehydrogenase (LDH) (Sigma TOX7 In Vitro Toxicology Assay Kit; Sigma-Aldrich) or aspartate aminotransferase (AST).41

Animal Treatment Protocol

Five-week-old female BALB/cOlaHsd-Foxn1nu mice (Harlan-Winkelmann, Borchen, Germany), housed under pathogen-free conditions, received an inoculation of 1 × 107 HuH7 hepatoma cells into the right flank. When palpable tumors became detectable, animals were divided randomly to 4 groups: intraperitoneal injection of vehicle only (control group), 5-aza-dC (1 mg/kg), SAHA (50 mg/kg), or a combination of 5-aza-dC (1 mg/kg) and SAHA (50 mg/kg). Tumor volumes were estimated according to the formula V = L × W2 (in which L indicates length and W indicates width) at the start of the treatment regimen (designated Day 1) and then from Day 3 on daily. Animals were sacrificed by carbon dioxide asphyxiation either when the tumor size reached 20 × 20 mm or at the end of treatment (Day 12). Individual 3-parametric logistic curves were fitted to cubic roots of tumor volumes in a hierarchical nonlinear mixed model with log-normal rates and days of half-maximal effect using WinBUGS 1.4. The Dunnett test was performed on individual posterior means at Days 0 and 10. All animal experiments were performed in agreement with the laws of the German government concerning the conduct of animal experimentation. The protocol was approved by the local ethics committee for animal experimentation.

Histologic Analysis

Completely removed tumor specimens from the xenograft mouse investigation with adjacent liver tissue were fixed in 4% buffered formalin for 24 hours and embedded in paraffin. Serial sections were performed and stained with hematoxylin and eosin. The histologic analysis was performed by a pathologist in a blinded fashion.


Antiproliferative Activity of DNA Methylation Inhibitor 5-Aza-dC in Human Hepatoma-Derived Cell Lines

To investigate the antiproliferative capacity of 5-aza-dC in the context of HCC 3 well-characterized human hepatoma cell lines (Hep3B, HepG2, and HuH7) were incubated with different concentrations of 5-aza-dC (5–50 μM); after 5 days of continuing treatment an SRB cytotoxicity assay was performed. The viability of all hepatoma cell lines tested was found to be significantly reduced at all concentrations employed, being most prominent for HuH7, followed by HepG2 and Hep3B (Fig. 1). Dose-dependent viability was fitted by hyperbolic curves decreasing to asymptotes of 50%, 59%, and 62%, respectively. Remarkably, the observed antiproliferative effect did not increase substantially when concentrations higher than 5 μM of 5-aza-dC were used (Fig. 1). Compared with untreated control cells, viability for 5 μM 5-aza-dC-treated HuH7, HepG2, and Hep3B cells was calculated to be 60%, 70%, and 77%, respectively. We chose the lowest investigated dose of 5-aza-dC of 5 μM as a standard dose for all further experiments.

Figure 1.

Antiproliferative effect of 5-aza-dC. Human-derived hepatoma cell lines Hep3B, HepG2, and HuH7 were treated with 5-aza-dC at various concentrations for 5 days. Cell viability was assayed in an sulforhodamin B (SRB) cytotoxicity assay. A 100% viability represents the optical density of untreated cells. Shown are results as a mean of 3 independent experiments and single standard deviations. Lines are hyperbolic decays to asymptotes with 95% confidence bands (dashed lines).

Enhanced Antiproliferative Activity Achieved by a Combination of a DNA Methylation Inhibitor and an HDAC Inhibitor

Previously we could show that compounds with HDAC-I activity inhibit proliferation and induce apoptosis of human hepatoma cells.41 To test for possible synergistic effects of HDAC-I compounds (such as SAHA) with DNA methylation inhibitor compounds (such as 5-aza-dC), HepG2, Hep3B, and HuH7 cells were either incubated with SAHA, 5-aza-dC, or a combination of both. After 5 days of treatment target cell viability was determined by an SRB cytotoxicity assay. Notably, the combination treatment employing both epigenetic modulators, SAHA and 5-aza-dC, induced the most prominent reduction of viability in all cell lines tested (Fig. 2, fourth bars [filled] from the left). Compared with the viability score of a treatment with 5-aza-dC only (Fig. 2, second bars [filled] from the left), the combined treatment approach led to a significant further reduction of the viability score in all 3 hepatoma cell lines (Fig. 2) (P = .023 for HuH7, P < .001 for Hep3B, P < .001 for HepG2). Combination treatment with SAHA and 5-aza-dC demonstrated a cell line-independent inhibition of viability (P < .0001).

Figure 2.

Combinatorial treatment of human-derived hepatoma cells with 2 different epigenetic therapy approaches. Hep3B, HepG2, and HuH7 cells were treated either with 5 μM of 5-aza-dC alone (second bars [filled] from the left) or 2 μM of suberoylanilide hydroxamic acid (SAHA) (third bars [open] from the left) or with a combination of both (fourth bars [filled] from the left) for 5 days. Sample cell viability was assessed by sulforhodamin B (SRB) cytotoxicity assays relative to the viability of untreated control cells (first bars [open] from the left). Shown are the mean of 3 independent experiments and 95% confidence interval. Significant reduction of the viability score was tested with analysis of variance (ANOVA).

Maintenance of Epigenetic Modulation Capacities in a Combined Treatment Approach

Next we investigated a possible interference of 5-aza-dC with SAHA and vice versa by choosing 2 well-described model genes, known to be inactivated in human hepatoma cells by methylation, namely, SOCS-18 and p16INK4A.9, 10 Suppressor of cytokine signaling (SOCS-1) is known to be unmethylated in HuH7 cells but its inactivation via methylation has been attributed to an important role in the development of HCC.8 p16INK4A is 1 of the most important tumor suppressor genes. It is well examined that p16INK4A is methylated in HuH7 and faint methylated in HepG2 cells and that the methylation contributes to the process of hepatic carcinogenesis.10 In the absence of both compounds (baseline control) SOCS-1 displayed a strong methylation pattern in HepG2 cells (Fig. 3A, upper panel, Lanes 1 and 2) and a very faint methylation pattern in HuH7 cells (Fig. 3B, upper panel, Lanes 1 and 2), whereas p16INK4A was methylated in HuH7 cells (Fig. 3B, lower panel, Lanes 1 and 2), but not in HepG2 cells (Fig. 3A, lower panel, Lanes 1 and 2). These findings agree with the current literature. Neither treatment of HepG2 nor HuH7 hepatoma cells with the HDAC inhibitor SAHA alone caused any substantial changes in the methylation patterns of SOCS-1 or p16INK4A (Fig. 3A and B, Lanes 5 and 6). As expected, treatment with 5-aza-dC induced a partial demethylation of SOCS-1 in HepG2 (Fig. 3A, upper panel, Lanes 3 and 4) and of p16INK4A in HuH7 cells (Fig. 3B, lower panel, Lanes 3 and 4). Furthermore, the faint methylation pattern of SOCS-1 in HuH7 cells was completely resolved in the course of treatment with 5-aza-dC (Fig. 3B, upper panel, Lane 4). Importantly, combined treatment of HepG2 or HuH7 hepatoma cells with 5-aza-dC plus SAHA did not result in a change of the demethylating activity induced by 5-aza-dC alone (Fig. 3, Lanes 7 and 8 vs. Lanes 3 and 4). These results demonstrate that SAHA does not substantially interfere with the overall demethylating activity induced by 5-aza-dC.

Figure 3.

Maintenance of 5-aza-dC inherent demethylating activity in the presence of the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA). A methylation-specific polymerase chain reaction (PCR) was employed to investigate the methylation status of 2 model genes, SOCS-1 and p16INK4A, in (A) human hepatoma cell lines HepG2 and (B) HuH7. The cells were treated either with 5-aza-dC (5 μM) or SAHA (2 μM) or with a combination of both (5-aza-dC + SAHA) for 48 hours. Bands found to be visible in “U” lanes represent unmethylated DNA products (175 basepairs [bp] for SOCS-1; 151 bp for p16INK4A) generated by primers being specific for the unmethylation state of the target DNA. Bands found to be visible in “M” lanes represent methylated DNA products (160 bp for SOCS-1; 150 bp for p16INK4A) generated by primers specific for the methylation state of the target DNA.

Next we looked for a relevant influence of the DNA methylation inhibitor 5-aza-dC on the HDAC-I activity exerted by SAHA with a fluorescence-based in vitro assay using standardized nuclear extracts of HeLa cells (Fig. 4A). As a result, the evident inhibition of HDAC enzyme activity of >70% did not appear to differ if treated with SAHA alone or with SAHA plus 5-aza-dC (Fig. 4A, Bars 3 and 4). In contrast, the treatment of nuclear extracts with 5-aza-dC alone showed no inhibition of HDAC enzyme activity (Fig. 4A, Bar 2). In addition to these hepatoma cell-free experiments, Western blot analysis of treated Hep3B cells and PHH were performed. Hep3B cell extracts demonstrated that H3 histone proteins were hyperacetylated when treated with SAHA alone (Fig. 4B, Lane 3) or SAHA plus 5-aza-dC (Fig. 4B, Lane 4), but not when treated with 5-aza-dC alone (Fig. 4B, Lane 2) or mock-treated (Fig. 4B, Lane 1). Notably, cell extracts of PHHs that had been treated simultaneously either with 5-aza-dC or SAHA alone or a combination of both did not show any relevant changes in the H3 histone protein acetylation pattern (Fig. 4C); this finding is in keeping with our previous observation of a differential response pattern of malignant and nonmalignant liver cells with respect to HDAC-I treatment.41 Taken together, these experiments show that SAHA and 5-aza-dC both maintained specific epigenetic modulation capacities in a combined treatment approach.

Figure 4.

Maintenance of suberoylanilide hydroxamic acid (SAHA)-mediated inhibitory activity against histone deacetylase inhibitor (HDAC) in hepatoma cells in the presence of 5-aza-dC. (A) HDAC activity was determined in nuclear extracts of HeLa cells by a fluorescence-based assay kit. The percentage of HDAC inhibition in the presence of 5-aza-dC (5 μM), SAHA (2 μM), or a combination of both substances (5-aza-dC, SAHA) was calculated relative to the acetylation activity of nuclear HeLa cell extracts in the absence of any HDAC-I compound (control). Shown are the mean of 2 independent experiments and single standard deviations. (B and C) Western blot analysis of H3 histone acetylation in (B) Hep3B human hepatoma cells and (C) primary human hepatocytes (PHH). The cells were treated with 5-aza-dC (5 μM), SAHA (2 μM), or a combination of both substances (5-aza-dC, SAHA) as indicated for 24 hours. Western blot analysis was performed using cellular lysates; equal protein loading was verified by vinculin staining. Ac-H3 levels were estimated by performing a densitometric analysis; values are shown relative to untreated controls.

Lack of HDAC-I/DA Combination Regimen Toxicity in PHHs

To investigate unwanted side effects potentially limiting the transfer of our innovative HDAC-I/DA treatment approach into the clinic, we first incubated PHH cells for 48 hours only with 5-aza-dC (employing increasing concentrations of up to 50 μM), subsequently measured the cellular integrity by an LDH release assay and cellular damage by an AST release assay in the cellular supernatant (Fig. 5A), and fitted a straight line. As a result, we did not find any relevant increases of the LDH (slope 0.0017; 95% confidence interval [95% CI], −0.108 to 0.112) or AST (slope 0.092; 95% CI, −0.250 to 0.435) activities of PHH, showing that 5-aza-dC, even in high concentrations, is well tolerated in this cell type.

Figure 5.

Hepatoma cell-selective toxicity of 5-aza-dC as a single treatment approach or in a combinatorial treatment setting together with suberoylanilide hydroxamic acid (SAHA). (A) Primary human hepatocytes (PHHs) were treated for 48 hours with various concentrations of 5-aza-dC; subsequently, lactate dehydrogenase (LDH) and aspartate aminotransferase (AST) release into the supernatant fluid compared with untreated controls was determined. (B) PHH were treated for 48 hours with various concentrations of SAHA alone, or with a combination of 5-aza-dC (5 μM) and SAHA; subsequently, relative LDH and AST released into the supernatant fluid was compared with untreated controls (left side and central diagram). The cell viability of PHH was assayed in a sulforhodamin B (SRB) cytotoxicity assay; 100% viability represents the optical density of untreated cells (right side). The results of the LDH release are the mean of 3 independent experiments with PHH from 3 different donors contrary to 1 representative donor (AST and SRB). All experiments with PHH were included as a positive control for cell death the incubation with a strong HDAC inhibitor alone (40 μM Trichostatin A; TSA). (C) Hep3B, HepG2, and HuH7 cells were either treated for 48 hours with 5-aza-dC (5 μM), various concentrations of SAHA alone, or with a combination of 5-aza-dC and SAHA. LDH release of hepatoma cells into the culture media was determined and found to be related to untreated control cultures. The results of the LDH release assay are shown as a mean of 3 independent experiments.

Next the toxicity of our combined HDAC-I/DA treatment approach was investigated by incubating PHH cells with 5 μM 5-aza-dC plus increasing concentrations of SAHA (up to 4 μM). Interestingly, SAHA treatment alone did not lead to a relevant increase in LDH or AST release in a linear regression fit (Fig. 5B, gray lines, left side and central diagram). These changes of the enzyme activities for each curve in Figure 5B fit a flat linear regression curve with no significant changes of the slopes from zero. Furthermore, an SRB assay using increasing concentrations of SAHA in the presence or absence of 5-aza-dC did not show any relevant reduction in viability (Fig. 5B, right side). In contrast to these observations in nonmalignant PHHs, all 3 hepatoma cell lines (Hep3B, HepG2, and HuH7) demonstrated an increase of LDH release when increasing amounts of SAHA were applied (Fig. 5C, black lines). This effect was found to be most prominent for HepG2 and HuH7 cells.

In summary, whereas PHH cells were found to tolerate treatment with 5-aza-dC alone or in combination with SAHA quite well, human-derived hepatoma cell lines showed substantial cellular damage undergoing our epigenetic treatment approach.

In Vivo Activity of a Combined Epigenetic Treatment

Finally, our epigenetic combination regimen was tested in a xenograft hepatoma model. For this purpose, nude mice harboring subcutaneously implanted HuH7 xenografts were treated intraperitoneally with either 1) 5-aza-dC, 2) SAHA, 3) a combination of both, or 4) placebo (control group) (Fig. 6).

Figure 6.

Suppression of hepatoma-derived tumor growth in vivo by a combinatorial treatment approach with 5-aza-dC and suberoylanilide hydroxamic acid (SAHA). Growth of the HuH7 xenograft in nude mice treated with daily intraperitoneal injection of vehicle, 5-aza-dC (1 mg/kg), SAHA (50 mg/kg) alone, or 5-aza-dC in combination with SAHA. Data are shown as the mean tumor volumes of the surviving animals in each treatment group. Treatment started when palpable tumors were evident in the animals (designated Day 1). The mice were divided randomly into control and experimental groups. Each group comprised 4 to 5 animals when treatment started; all tumor growth measure points are depicted in the diagram. The mean tumor growth of the combination treatment (5-aza-dC, SAHA) at the end of the investigation period was found to differ statistically significantly compared with the vehicle group (P = .023; Dunnett test) but the growth of single treatment groups did not.

As a result, animals in the control group (injection of vehicle only) showed large tumors at treatment Day 12 (Fig. 6, vehicle line, depicted in red); at this point the animals were sacrificed because tumor volumes reached the maximum size admitted by the treatment protocol. Animals in both single treatment groups with either 5-aza-dC (Fig. 6, blue line) or SAHA (Fig. 6, black line) or the combination treatment group (Fig. 6, green line) displayed reduced tumor volumes in comparison to the control group. At the end of the investigation period the maximum tumor volumes and single standard deviations calculated for 1) vehicle-treated animals (control animals) were 3214 mm3 ± 676; 2) 5-aza-dC-treated animals, 1529 mm3 ± 448; 3) SAHA-treated animals, 1531 mm3 ± 426; and 4) the combinatorial treatment group, 739 mm3 ± 316. Notably, the comparison of the mean tumor growth of the combinatorial treatment group was statistically significant different from the untreated control group (P = .023, Dunnett test), thereby supporting the data of our prior in vitro experiments and demonstrating an enhanced potential of our dual therapy approach also under in vivo conditions. A histologic analysis of mice liver tissues did not reveal relevant histologic changes in the combination treatment compared with untreated control animals (Fig. 7), further supporting our in vitro data employing PHH cultures that tolerated the combination treatment well.

Figure 7.

Histologic analysis of liver and tumor tissue. Liver and tumor specimens from daily intraperitoneal-injected animals treated with either vehicle alone (control; Panels A and C) or with the combination of 50 mg/kg of suberoylanilide hydroxamic acid (SAHA) and 1 mg/kg of 5-aza-dC (Panels B and D) were fixed, embedded in paraffin, and stained with hematoxylin and eosin. (A) Liver tissue with regular microarchitecture and no obvious cell damage. (B) Liver tissue with regular microarchitecture: only single hepatocytes show microvesicular intracytoplasmatic lipid vacuoles. (C) Tumor tissue with small areas of mass necrosis and apoptotic bodies. (D) Tumor tissue with confluent areas of mass necrosis and apoptotic bodies (magnification ×200).


The discovery that epigenetic alterations contribute substantially to both development and maintenance of human malignancies defines the chromatin structure and DNA methylation status as legitimate sets of targets for the development of new cancer therapy approaches.21 This evidence has led to the first FDA approval of an HDAC inhibitor for the treatment of cancer in 2006 (ie, SAHA for use in cutaneous T cell lymphoma), and the approval of 2 demethylating substances, 5-azacytidine (Vidaza) and 5-aza-dC (decitabine) for the treatment of myelodysplastic syndrome (MDS). 5-aza-dC was further found to have activity in a broad range of hematologic disorders such as acute or chronic myelogenous leukemia; however, to our knowledge, the majority of clinical data have been gathered for the treatment of MDS, with a recent Phase III clinical trial including 170 patients demonstrating clinical efficacy by documented durable responses, including 9% complete responses, and improved time to the development of acute myelogenous leukemia or death.42

In the present study, we investigated the potential of a combined epigenetic treatment of HCC employing a DNA demethylating substance in combination with an HDAC-I compound. This endeavor was stimulated by several observations that define a special role of epigenetic alterations in the HCC context: 1) the repression of growth-regulating genes in HCC-derived cell types by DNA methylation,8–10 2) antiproliferative and hepatoma cell-specific proapoptotic effects exerted by HDAC-I activity,22, 41 and 3) a proven interaction of DNA methyltransferases with cellular HDAC enzymes leading to the silencing of cellular regulatory genes.2, 33, 43

In an earlier study, we described a hepatoma cell-specific antiproliferative effect exerted by the 2 HDAC-I substances VPA and ITF2357.41 Moving forward from this, we now demonstrate also that a DNA methylation inhibitor compound, 5-aza-dC, induces a substantial inhibition of hepatoma cell proliferation. Interestingly, the combination of DNA methylation inhibition (5-aza-dC) together with HDAC inhibition (SAHA) led to an attractive synergistic therapeutic effect, suggesting that 2 dissimilar epigenetic mechanisms (DNA methylation inhibition, HDAC inhibition) do not substantially interfere on a molecular level. This assumption is supported by the determination of the HDAC activity in nuclear extracts or the acetylation status of DNA associated histone proteins and by the determination of the methylation status of 2 model genes, SOCS-1 and p16INK4A.

Interestingly, current concepts suggest 1) that either DNA methylation may be responsible for the “locking in” of defined genes into a silenced, hypoacetylated chromatin state33, 44 or 2) that histone modifications are even able to determine DNA methylation patterning.45 Taking these multiple interactions between these 2 different epigenetic modifications in cancer cells into account, combinatorial epigenetic modulation regimes seem to be an ideal target for future therapeutic strategies.

Recently, it was described that VPA not only exhibits an inherent HDAC-I activity, but also exerts a replication-independent DNA demethylating activity.46 This observation raised speculations that HDAC-I might generally lead to a replication-independent demethylation, too, but this dual molecular activity was not reported for other substances with HDAC-I activity as yet. In our experiments it is of interest that SAHA did not change the methylation status of SOCS-1 or p16INK4A, but 5-aza-dC did. Notably, these 2 methylation-silenced genes play a crucial role in the carcinogenesis of HCC.8, 10

Each new therapeutic option of HCC requires an in-depth analysis of unwanted toxicities to surrounding normal, nonmalignant hepatocytes. For the first time we have tested the application of an epigenetic combination regime on primary human-derived liver cells and did not find any toxicity in doses considered clinically relevant. This important preclinical result is supported by recent clinical studies in which the application of either 5-aza-dC or HDAC-I compounds did not provoke any dose-limiting liver toxicity.18, 47, 48 For substances with HDAC-I activity, World Health Organization grade 1 or 2 liver toxicities have been reported in trials applying MS-27549 or VPA only when given in combination with all-trans retinoic acid,50 but not for VPA alone51 or SAHA.52

Most notably, we could substantiate our in vitro experiments with the results of an animal experiment. The epigenetic combination regimen tested in a xenograft hepatoma model demonstrated the most prominent inhibition of tumor growth, whereas the single compound treatment equally reduced the tumor growth at a lower level. These findings are congruent with the results of the proliferation assays and support the effort for further investigations in clinical trials.

To our knowledge to date, clinical trials making use of a combination of a demethylating substance and an HDAC-I compound have only been reported in the context of a pharmacokinetic study,34, 35 but efficiencies of such combinatorial therapeutic approaches for solid tumors (including HCC) have not yet been investigated. Therefore, a combination of these epigenetic drugs for HCC seems to be a novel, logically founded step that should be evaluated further.43


We thank Andrea Schenk and Irina Smirnow for excellent technical assistance and Heike Heitmann for exceptional support in the animal facility.