Quantitative evaluation of RASSF1A methylation in the non-lesional, regenerative and neoplastic liver

Background Epigenetic changes during ageing and their relationship with cancer are under the focus of intense research. RASSF1A and NORE1A are novel genes acting in concert in the proapoptotic pathway of the RAS signalling. While NORE1A has not been previously investigated in the human liver, recent reports have suggested that RASSF1A is frequently epigenetically methylated not only in HCC but also in the cirrhotic liver. Methods To address whether epigenetic changes take place in connection to age and/or to the underlying disease, we investigated RASSF1A and NORE1A gene promoter methylation by conventional methylation specific PCR and Real-Time MSP in a series of hepatitic and non-hepatitic livers harboring regenerative/hyperplastic (cirrhosis/focal nodular hyperplasia), dysplastic (large regenerative, low and high grade dysplastic nodules) and neoplastic (hepatocellular adenoma and carcinoma) growths. Results In the hepatitic liver (chronic hepatitic/cirrhosis, hepatocellular nodules and HCC) we found widespread RASSF1A gene promoter methylation with a methylation index that increased from regenerative conditions (cirrhosis) to hepatocellular nodules (p < 0.01) to HCC (p < 0.001). In the non-hepatitic liver a consistent pattern of gene methylation was also found in both lesional (focal nodular hyperplasia and hepatocellular adenoma) and non-lesional tissue. Specifically, hepatocellular adenomas (HA) showed a methylation index significantly higher than that detected in focal nodular hyperplasia (FNH) (p < 0.01) and in non-lesional tissue (p < 0.001). In non-lesional liver also the methylation index gradually increased by ageing (p = 0.002), suggesting a progressive spreading of methylated cells over time. As opposed to RASSF1A gene promoter methylation, NORE1A gene was never found epigenetically alterated in both hepatitic and non-hepatitic liver. Conclusion We have shown that in non-lesional, regenerative and neoplastic liver the RASSF1A gene is increasingly methylated, that this condition takes place as an age-related phenomenon and that the early setting and spreading over time of an epigenetically methylated hepatocyte subpopulation, might be related to liver tumorigenesis.

In an autopsy series, Waki et al recently reported [20], that RASSF1A gene promoter is largely methylated in the human liver and that this epigenetic change may be related to age.
Given that age-related methylation is a process that affect non-lesional cells and precedes tumor formation [21], the early setting of a hepatocyte subpopulation with RASSF1A methylation, might be a potential and long-lasting for oncogenic injuries.
We designed a study aimed to ascertain the frequency of RASSF1A methylation in non-lesional, cirrhotic, hyperplastic, dysplastic and neoplastic liver. Moreover we performed a quantification study, by Real-Time PCR, to document whether the gene is methylated in relation to age and/or to the underlying disease. Given that the functional proapoptotic role of RASSF1A as tumor suppressor, has been recently associated to the homologue gene NORE1A [22][23][24] we also evaluated the methylation status of the latter gene in the same non-lesional and pathological conditions.

Tissue samples and DNA extraction
Consecutive cases from 38 hepatitic/cirrhotic patients with and without hepatocellular carcinoma (HCC) and/ or hepatocellular nodules (HN), from 13 patients with focal nodular hyperplasia (FNH) or hepatocellular adenoma (HA) and 5 human cell lines (hepatoma cell lines: Plc/Prf/5, Huh6, Huh7 and HepG2, and Chang, a normal hepatocyte cell line) were collected for study. Patient's characteristics (age, sex, etiology of hepatitis, type of lesions) are reported in Table 1. The series included 26 HCCs, 17 high-grade dysplastic nodules (HGDN), 28 non high-grade dysplastic nodules (large regenerative/lowgrade dysplastic nodules), 5 hepatocellular adenomas (HA) and 8 focal nodular hyperplasias (FNH). The study was conducted in accordance with the guidelines of the Ethics Committee and Internal Scientific Board of our institution (Istituto Clinico Humanitas, Rozzano, Milan, Italy).
After identification of the different lesions on the hematoxylin-eosin stained slides, individual lesions were carefully microdissected from 20-µm-thick paraffin sections. In all cases hepatocytes represented the main cellular population (≥ 80% of overall cellularity). In selected cases (Table 1) of FNH and HA, different samples were taken from the same lesional and non-lesional liver in order to assess individual inter-sample methylation pattern and levels heterogeneity. The non-lesional liver samples were taken far from the lesions.
Collected materials were dewaxed by washing in xylene, and rinsed in ethanol. The dried tissues were digested using proteinase K and subjected to classical DNA extraction using phenol/chloroform/isoamylalcohol and ethanol precipitation. After precipitation, DNA was resuspended in water and quantified by spectrophotometric analysis (260 nm).

RASSF1A and NORE1A methylation-specific PCR (MSP) Sodium bisulfite modification
We designed a set of primers targeted to amplify a 111 base pair sequence located in the upstream region of the RASSF1A promoter (ranging from -135 to -24 base pairs from the starting site of exon 1) ( Table 2). This sequence was selected because it showed the highest CpG island density of the whole promoter.
Methylation-specific PCR was carried-out as previously reported [18,25]. Genomic DNA was modified by treatment with sodium bisulfite, which converts all unmethylated cytosines to uracil, then to thymidine during the subsequent PCR step. One µg of DNA in a volume of 50 µl were denatured by adding sodiumhydroxide to a final concentration of 0.2 M and incubating for 10 min at 37°C. Thirty µl of 10 mM hydroquinone and 520 µl of 3 M sodium bisulfite (pH 5.0), both freshly prepared, were added to the denatured DNA. The sample was mixed gently, overlaid with enough mineral oil to cover the surface of the aqueous phase, and incubated at 55°C for 16-20 h. Under this conditions, complete selective conversion of unmethylated cytosine to uracil can be achieved [26]. The modified DNA was desalted through Wizard DNA Clean-  up System (Promega) and eluted in 50 µl of distillated water. Desulphonation was carried-out adding 5.5 µl of NaOH 3 M and incubating DNA for 5 min at room temperature. Modified DNA was precipited with sodium acetated 3 M, ethanol and 1 µl of glycogen as carrier; finally was resuspended in 50 µl of distillated water and DNA concentration was measured by spectrophotometric analysis. The DNA was stored at -20°C until use.

Non-Hepatitic Livers
The methylation status of RASSF1A and NORE1A was determined by the MSP method of Herman et al. with the following modifications, previously described [26]. Five microliters of modified DNA were amplified in a stage I PCR with a primer set that recognise the bisulfite-modified template but do not discriminate between methylated and unmethylated alleles (  Table 2. Annealing temperature for RASSF1A was 55°C to amplify methylated and unmethylated sequences, for a total of 25 cycles and annealing temperature for NORE1A was 58°C to amplify methylated and unmethylated sequences, for a total of 25 cycles. This nested PCRs are able to detect 1 methylated allele in 50,000 unmethylated alleles [26]. All PCRs were performed with positive methylation control, a human placental DNA treated in vitro with excess of SssI methyltransferase that generates a completely methylated DNA. Ten microliters of PCR products were run on a 2% agarose gel and visualized by staining with ethidium bromide. Real-time PCR is based on the continuous optical monitoring of a fluorogenic PCR [27]. PCR amplifications were performed using DyNAmo HS SyBR Green qPCR Kit (Finnzymes) according to the manufacturer's instructions. SYBR Green I is specific for double-stranded DNA and fluoresces when bound to the amplified double-stranded PCR product, thereby enabling direct quantification of amplified DNA without labeled probes [28].
Amplifications were carried out in 96-well plates in an Chromo 4 thermalcycler (MJ Research Inc.), with 0.2 µM primers specific to methylated or unmethylated bisulfitemodified template and 50 ng of modified DNA in a 20 µl final reaction volume. The primers used for RASSF1A realtime MSP were the same used for conventional MSP ( Table 2). Thermal cycling was initiated with a incubation step of 15 min at 95°C, followed by 45 cycles (95°C for 15 sec, 55°C for 20 sec, 72°C for 20 sec) with a single fluorescent reading taken at the end of each cycle. Each run was completed with a melting curve analysis to confirm the specificity of amplification and lack of primer dimers.
Threshold cycle values, C t , were determined by the Opticon2 software. A calibration curve was run in parallel with each analysis. Human cell lines, LoVo and HeLa, previously shown to be respectively RASSF1A emimethylated and unmethylated by conventional MSP, were used for constructing the calibration curve for the RASSF1A M and U real-time MSP (correlation coefficient were: r = 0.992 ± 0,007 for M and r = 0,997 ± 0,003 for U). To set-up the calibration curve, we prepared serial dilutions of LoVo and HeLa DNA, containing 10, 100, 500, 1000, 2500 and 5000 genome-equivalents, one genome-equivalent being the amount of a particular target sequence in a single reference cell (Figure 1). A conversion factor of 6.6 pg of DNA per diploid cell was used for expressing quantitative results in genome equivalents [29,30].
The methylation index (%) in a sample was calculated according to the following equation: where M and U are genome-equivalents of methylated and unmethylated RASSF1A sequences obtained by the interpolation of Ct values with M and U calibration curves.
A representative set of the original graphs (n = 12) from each group of lesions occurring in both hepatitic and non hepatitic liver is reported [see Additional file 1 and 2].

Statistical analysis
Differences in the degree of methylation among categories of lesions were analyzed by linear mixed models (random effect model) accounting for the within subject correlations. Statistical association between age and methylation index were analyzed by Pearson correlation. p < 0.05 was considered to be statistically significant. Statistical analysis were performed by using STATA 9 software. RASSF1A gene promoter methylation with the etiology of cirrhosis.
A gradual but significant increase in the methylation index of RASSF1A was seen from cirrhosis to HN to HCC ( Figure 3A). Cirrhotic tissue had a methylation index lower than that detected in HN (p < 0.01) and the latter a methylation index significantly lower than that detected in HCC (p < 0.001). HN were grouped together because no statistical difference was seen as to the methylation index among LGDN, HGDN and LRN (data not shown).
No correlation was found between RASSF1A methylation index with the etiology of cirrhosis.

Quantitative analysis for the prevalence of methylated alleles by Real Time MSP
Hepatocellular adenomas (HA) showed a methylation index significantly higher than that detected in focal nodular hyperplasia (FNH) (p < 0.01) and in non-lesional tissue (p < 0.001) ( Figure 3B). As shown in Table 3 a certain degree of inter-sample heterogeneity was seen in methylation index in individual cases of non-lesional liver, FNH and HA with multiple sampling. Specifically, a methylation index < 10% was detectable in 7/13 cases (15/34 samples) of non-lesional liver, 2/8 cases (4/22 samples) of FNH but never in HA. By contrast a methylation index > 50% was seen in 2/13 cases (2/34 samples) of nonlesional liver, 2/8 cases (3/22 samples) of FNH and 3/5 cases (7/13 samples) of HA. In non-lesional liver, a gradual increase was seen in the methylation index by ageing (p = 0.002) (Figure 4).

NORE1A gene promoter methylation in hepatitic and nonhepatitic liver
In all samples under study NORE1A gene was always found unmethylated by conventional MSP (Figure 2; lower panel).

Discussion
We investigated the methylation profile of RASSF1A and NORE1A promoter genes, in non-lesional, cirrhotic, hyperplastic, dysplastic and neoplastic (benign and malignant) hepatocellular growths. These two genes are functionally thought to act in concert to mediate proapoptotic signal on the RAS pathway in vivo [31]. Given that RAS gene is not alterated in human hepatocarcinogenesis, we sought to analyze by MSP if these RAS effectors are epigenetically altered by promoter methylation in the above conditions. In keeping with several authors [13][14][15][16] we observed high frequency of RASSF1A methylation in HCC (100%) and associated cirrhosis (81.1%). In addition we observed a high RASSF1A methylation frequency (86.6%) in hepatocellular dysplastic nodules.
Interestingly, the RASSF1A methylation rate in cirrhosis (81-82%) and corresponding HCC (98-100%), as detected by MSP (qualitative assay), occured with a surprisingly overlapping frequency in 3 different geographic regions: Italy (mediterranean area, present work), North Europe [32] and China [14]. Given the different etiology of chronic liver disease in these regions, the above results might suggest that the very high rate of RASSF1A methylation in the cirrhosis-HCC sequence is a peculiar early molecular feature of hepatocarcinogenesis in the cirrhotic setting, unrelated to a specific etiology.
Lee et al. [17] reported 67% promoter methylation in HCC, 10% in dysplastic nodules and no methylation in non neoplastic hepatitic/cirrhotic liver. This discrepancy Non-lesional liver: methylation index of RASSF1A according to age Figure 4 Non-lesional liver: methylation index of RASSF1A according to age. might be due to the sensitivity of our method (nested MSP), relying on the analysis of a different set of CpGs. Indeed Lee et al. [17] designed a set of primers targeted to amplify a sequence of exon 1 of the gene and not of its promoter, as we did.
Opposite results to RASSF1A, were obtained by analyzing NORE1A promoter methylation, which was not investigated previously in the human liver. This gene, which is supposed to act in concert with RASSF1A as proapoptotic mediator of RAS signalling, was never found methylated in our series of non-lesional, cirrhotic, hyperplastic, dysplastic and neoplastic liver. Hesson et al. [24] reported that this gene was variably methylated in a number of different human tumors such as breast, pulmonary, kidney and colorectal. In the present study, the unmethylation of NORE1A in the liver might be related to opposite epigenetic alteration of RASSF1A. According to this hypothesis Hesson et al. [33] recently reported frequent RASSF1A, but not NORE1A, methylation in gliomas. Given that both genes act in concert, it can be speculated that the methylation of NORE1A could result biologically redundant.
Based on these results we performed a RASSF1A quantitative analysis of methylation by Real-Time PCR on nonlesional, cirrhotic, hyperplastic, neoplastic liver and for the first time on dysplastic hepatocellular nodules.
In hepatitic livers, our quantitative analysis showed that HCC, as compared to cirrhosis, had a significantly higher methylation index of RASSF1A. This result is in keeping with that of Zhong et al. [15] measuring the number of methylated hepatocytes in HCC and cirrhosis by PCR-RFLP. Interestingly, in our series, quantitation of RASSF1A methylation showed that hepatocellular nodules had a methylation index in between cirrhosis and HCC, in keeping with their putative HCC precursor role.
The methylation of RASSF1A gene promoter, already set in our cases of cirrhosis, also supports a tumor-independent mechanism of gene methylation. Waki et al [20] recently showed, in an autopsy study, that all patients over 40 yrs had the RASSF1A gene promoter methylated in the liver, suggesting an age-related mechanism of gene methylation.
To probe whether RASSF1A gene promoter was also methylated in connection to age we included in our series, nonlesional liver samples taken far from lesion in patients with FNH or HA.
In the non-lesional liver, one or more samples of 5 out of 9 patients under 40 yrs, showed a methylation index grater than 10%. Interestingly, methylation index in non-lesional liver showed a gradual increase with ageing, that started before 30 years [32]. These results suggest that agerelated methylation of RASSF1A gene promoter takes place early in a small subpopulation of cells of the human liver and that this fraction of hepatocytes is fated to increase by ageing. This observation resembles that reported by Nakagawa et al. [34] as to the methylation pattern of hMLH1 in the human colon, where methylated cells are distributed at multiple sites in the colon and methylation increased with age. Conceivably the fraction of hepatocytes already methylated with age at the RASSF1A gene promoter, might become, over time, more susceptible to oncogenic injuries. Because our results were obtained in a limited number of cases, the gradual increase of RASSF1A methylation index by age should be confirmed in larger series including younger subjects such as children and fetuses. As detected between HCC (a neoplastic growth) and cirrhosis (a regenerative growth), we also observed a statistically significant increase in the methylation index between hepatocellular adenoma (a neoplastic growth) and focal nodular hyperplasia (a hyperplastic growth). It has been suggested by Toyota et al. [21] that age-related methylation by potentially affecting genes that regulate both cell growth and differentiation could account in part for the hyperproliferative state that precedes tumor formation, thus linking age-related methylation to cancer-related methylation. The analysis of RASSF1A methylation index in the whole series of cases showed a gradual increase in the methylation index from non-lesional liver to regenerative/hyperplastic conditions (chronic liver disease and FNH), to preneoplastic lesions (HN) to overt tumors (HA and HCC). This result suggests that quantitative analysis of RASSF1A gene promoter methylation, rather than the detection of methylation bands "per se", might be clinically relevant. Quantitative results should also be compared to expression data (not available in the present series given the lack of specific antibodies and the unavailability of frozen material for mRNA analysis), to document the functional role of methylation.

Conclusion
In conclusion we have shown that in non-lesional, cirrhotic, hyperplastic, dysplastic and neoplastic liver growths the RASSF1A gene is methylated, that this condition takes place as an age-related phenomenon and that the early setting and spreading over time of an epigenetically alterated hepatocyte subpopulation, might be related to liver tumorigenesis.