LINE-1 Hypomethylation in a Choline-Deficiency-Induced Liver Cancer in Rats: Dependence on Feeding Period

Chronic feeding of methyl-donor (methionine, choline, folic acid, and vitamin B12) deficient diet induces hepatocellular carcinoma formation in rats. Previous studies have shown that promoter CpG islands in various cancer-related genes are aberrantly methylated in this model. Moreover, the global genome in methyl-donor-deficient diet fed rats contains a lesser amount of 5-methylcytosine than control livers. It is speculated that more than 90% of all 5-methylcytosines lie within the CpG islands of the transposons, including the long/short interspersed nucleotide elements (LINE and SINE). It is considered that the 5-methylcytosines in LINE-1 limit the ability of retrotransposons to be activated and transcribed; therefore, the extent of hypomethylation of LINE-1 could be a surrogate marker for aberrant methylation in other tumor-related genes as well as genome instability. Additionally, LINE-1 methylation status has been shown to be a good indicator of genome-wide methylation. In this study, we determined cytosine methylation status in the LINE-1 repetitive sequences of rats fed a choline-deficient (CD) diet for various durations and compared these with rats fed a choline-sufficient (CS) diet. The methylation status of LINE-1 was assessed by the combined bisulfite restriction analysis (COBRA) method, where the amount of bisulfite-modified and RsaI-cleaved DNA was quantified using gel electrophoresis. Progressive hypomethylation was observed in LINE-1 of CD livers as a function of feeding time; that is, the amount of cytosine in total cytosine (methylated and unmethylated) increased from 11.1% (1 week) to 19.3% (56 weeks), whereas in the control CS livers, it increased from 9.2% to 12.9%. Hypomethylation in tumor tissues was slightly higher (6%) than the nontumorous surrounding tissue. The present result also indicates that age is a factor influencing the extent of cytosine methylation.


INTRODUCTION
When rats are chronically fed a diet devoid of a methyldonorsource (choline, methionine, folic acid,and vitamin B 12 ), they spontaneouslydevelop hepatocellular carcinomas (HCCs) [1][2][3][4][5]. This is a unique carcinogenesis model in which no known carcinogen is involved. After the initial discovery [6] of this model, many questions were raised concerning its validity including the possibility of carcinogenic contaminants in the diet. Later, it was definitively demonstrated that diets lacking in methionine and choline and containing no detectable level of carcinogens acted as a complete carcinogen [7,8]. In spite of extensive phenomenological studies, the mechanism by which dietary methyl-donor deficiency causes HCC formation is not understood, but it is suggested that various concurrent carcinogenic pathways may be involved [4]. Oxidative stress appears to play a major role in this model [4] because there is a significant increase in oxidized DNA (8-hydroxy-deoxyguanosine) levels from day 1 of feeding [4] and because antioxidant cofeeding inhibits cancer formation [9]. Nevertheless, no direct connection between oxidative stress and carcinogenesis has been elucidated.
Genome-wide demethylation of 5-methylcytosine has been regarded as a common epigenetic event in malignancies and may play a crucial role in carcinogenesis. In the rat methyl-donor deficiency models, promoter CpG islands in several cancer-related genes are known to be aberrantly methylated, as noted by changes in DNA 5-methylcytosine content [10]. Methyl donors including choline and methionine are required for S-adenosyl methionine (SAM) biosynthesis [11], and SAM is the substrate for DNA cytosine methyltransferase, the enzyme responsible for maintaining 48  DNA cytosine methylation. Therefore, it is possible to speculate that the dysregulation of Dnmt activity is a cause of the genome-wide decrease of 5-methylcytosine and aberrant methylation of specific genes. However, the selective nature of the presence of aberrantly methylated genes suggests that other factors are also involved. More than 90% of all 5-methylcytosines lie within the CpG islands in the transposons, including long/short interspersed nucleotide elements (LINE and SINE). The presence of 5-methylcytosine in LINE-1 is considered to limit the ability of retrotransposons to be activated and transcribed; therefore, loss of 5-methylcytosine in LINE-1 could result in an increase in retrotransposon activity, leading to propagation of aberrant methylation to other genes [12,13]. Genome-wide instability inevitably results from hypomethylation. LINE-1 methylation status has also been shown to be a good indicator of genome-wide methylation [14,15]. In humans, LINE-1 hypomethylation was demonstrated in patients having various cancers [16][17][18]. In the present study, using the combined bisulfite restriction analysis (COBRA) method [14,19], we evaluated the amount of cytosine/5-methylcytosine in the LINE-1 repetitive sequence in rats fed a choline-deficient (CD) diet for various times.

Animals and diets
Rats were treated strictly following the animal use protocol approved by the Institutional Animal Care and Use Committee in the Oklahoma Medical Research Foundation. Weaned male Fisher 344 rats were obtained from Charles River (Indianapolis, IN, USA) and divided into 2 groups each containing 3-5 animals and fed a choline sufficient (CS) or a choline deficient (CD) diet (Dyets Inc Bethlehem, IN, USA). Feeding periods were 1, 4, 24, and 56 weeks. In these experiments, protein in the diet was substituted by a defined amino acid diet (CSAA or CDAA diet), because the CDAA diet has been shown to considerably accelerate carcinogenesis without altering cancer pathology as compared with the conventional CD (Lombardi's CD diet) [20].

Sequence of LINE-1
The LINE-1 sequence of interest in this study is illustrated in Figure 1.The consensus sequence revealed that LINE-1 el-ements have a 5' untranslated region (UTR) with internal promoter activity, two open reading frames (ORFs), a 3'UTR that ends in an AATAAA polyadenylation signal, and a poly A tail [21,22].

Combined bisulfite restriction analysis
COBRA is a simple method of CpG methylation analysis which utilizes the cleaving ability of the restriction enzyme RsaI specifically at bisulfite-modified CpG sites [14]. Genomic DNA from rat liver tissues was extracted using a Quiagen's genomic DNA extraction kit (Qiagen, Valencia, CA, USA). Bisulfite modification of genomic DNA was performed as follows: 3 µg of DNA, digested with the restriction enzyme EcoRI, was incubatedwith 0.3 N NaOH in a volume of 20 µL for 15 minutes, and then combined with a 120 µL portion of 3. Subsequently, the PCR product was digested with 10 unit of RsaI at 37 • C, for 5 hours, isolated with ethanol-chloroform precipitation, and subjected to polyacrylamide gel electrophoresis. The gel was stained with ethidium bromide, and the band intensity in the fluorogram was analyzed with an imaging workstation (Nu-cleoTech Corp, Hayward, CA, USA). The data for densitometric analysis is presented as mean ± SE Tests for statistical significance were evaluated using Student t test.

RNA extraction and Semiquantitative RT-PCR
Total RNA was extracted from liver tissues of CS-or CD-diet fed rats with Qiagen RNeasy kit (Qiagen, Crawley, UK). After  RNA samples were treated with DNase I (Invitrogen, Carlsbad, CA, USA), RT-PCR was performed with SuperScript III One-Step RT-PCR System (Invitrogen, Carlsbad, CA, USA). Common GenBank accession number of ORF I and ORF II sequences in LINE-1 is DQ100473. PCR was performed under the following conditions: annealing temperatures were 60 • C for ORF I and 55 • C for ORF II, cycle number was 23, and primer sequences were ORF I forward; 5 -AAG AAA CAC CTC CCG TCA CA-3 , ORF I reverse; 5 -CCT CCT TAT GTT GGG CTT TAC C-3 , ORF II forward; 5 -CCC ACT CTC TCC CTA CTT A-3 , and ORF II reverse; 5 -TAT AGA GGA AGG CAA CTG AT-3 . The expression of the glyceraldehydes phosphate dehydrogenase gene (GAPDH) was used to normalize the transcript band intensity.

LINE-1 COBRA in artificially methylated DNA
To confirm the accuracy of COBRA, control genomic DNA obtained from rats fed a regular diet was enzymatically methylated with SssI methylase (Sigma) in the presence of the substrate S-adenosylmethionine, and its LINE-1 promoter CpG island methylation was determined with CO-BRA. The results indicate that there was 5.9-6.1% undigested DNA (Figure 2, 163 kb band in the lanes marked with Mcontrol +), suggesting that the amount of mutated DNA plus SssI methylation resistant cytosine is within this level.

LINE-1 COBRA in CD livers
LINE-1 methylation was analyzed with COBRA in CD and CS livers after various feeding periods, including 1, 4, 24, and 56 weeks. Figure 2 illustrates typical COBRA fluorograms obtained from livers of rats fed either CD or CS diet for 4 weeks. There was a clear tendency for the CD diet to promote hypomethylation in LINE-1 during the entire feeding period (1 week to 56 weeks). The amount of unmethylated cytosine in LINE-1 ranged from 9.2% at 1 week to 12.9% at 56 weeks in CS livers, while in the CD livers, it ranged from 11.1% at 1 week to 19.3% at 56 weeks ( Figure 3). There was a statistical significant difference between CS and CD livers at 4 and 56 weeks of feeding ( Figure 3).

LINE-1 hypomethylation in tumor and nontumor tissues
After 24 weeks of feeding the CD diet, rats began to have tumor nodules which were histologically identified as adenomas (19), and at 56 weeks, most of these tumor nodules had developed into HCC. COBRA of LINE-1 DNA for tumor and nontumor tissues showed that there is a tendency for the DNA in tumor tissues to be more hypomethylated than nontumor tissues, however, a statistically significant difference was obtained only in the 56-week fed animals (Figure 4).

LINE-1 transcript expression
The expression of LINE-1 (ORF1 and ORF2) gene transcript was assessed with semiquantitative RT-PCR.

DISCUSSION
Chronic CD diet is hepatocarcinogenic in male rats and global hypomethylation has been shown to exist from early feeding times [10,23]. In many cancers, global hypomethylation as well as hypo-or hypermethylation in specific genes are widely accepted epigenetic changes [24]; however, which gene or DNA region responsible for aberrant methylation, especially in the methyl-deficient diet models, is not clear.
Using a rat CD model, we applied the COBRA method to analyze LINE-1 methylation. COBRA requires small amount of DNA samples and was previously employed to analyze human LINE-1 hypomethylation [18]. We showed that the LINE-1 promoter was hypomethylated in the livers of rats fed a CD diet from as early as 4-weeks' feeding (14.4%) as compared to CS livers of the same feeding period (9.1%), and that hypomethylation increased as a function of feeding period up to 19.3% at 56 weeks ( Figure 3). It is not unexpected that there was a significant increase in unmethylated cytosine (or a decrease in 5-methylcytosine) in LINE-1 of control CS livers as a function of feeding period (12.9% at 56 weeks) because it has been shown that aging is a major cause of genome-wide hypomethylation in mice and rats [25,26]. For example, using Southern blotting, Mays-Hoopes et al estimated that there is an 8% decrease in LINE-1's 5-methylcyotsine content in the livers of 27-month old mice [26]. Furthermore, it is possible that the age-dependent increase of mutation in LINE-1 could provide false signals of hypomethylation in this assay. In humans, LINE-1 hypomethylation was detected only in HCC but not in nontumor liver cirrhosis [16]. Also, a recent report has indicated that in some cancers, such as lymphoma, renal cell carcinoma, and papillary carcinoma of the thyroid, the LINE-1 hypomethylation level is not significantly different from normal tissues [18]. The authors suggested that human cancers may be classified into two groups, a low (0-3.4%) LINE-1 hypomethylation group and a moderately high (6.8-9.5%) group [18]. We speculate that HCC in this model may mimic the situation in the low group. The same study also showed a linear correlation of methylation levels analyzed by COBRA with levels determined by using semiquantitative conventional Southern blotting hybridization analysis [18].
Florl et al reported that in human urothelial cancer, there were coordinate changes of LINE-1 and HERV-K DNA methylation, suggesting that hypomethylation affects a variety of retroelements to similar extents [12]. Thus, LINE-1 hypomethylation is thought to be one of the important surrogate markers of global hypomethylation [14,15]. In the present study, LINE-1 hypomethylation in CD livers increased in a time-dependent fashion from 11.1% at 1 week to 19.3% at 56 weeks. The cytosine (demethylated 5methylcytosine) level in LINE-1 in this experiment was much  lower than that of the global genome as determined by using an HpaII/MspI-based cytosine extension assay [23], which resulted in cytosine content increasing from 46% at 9 weeks up to 54% at 36 weeks. However, another study using HPLC analysis indicated that genome-wide unmethylated cytosine increased from 6% at 8 weeks to 11-14% at 22 weeks [10]. The cause of discrepancies in the three methods is unknown.
LINE-1 promoter methylation is thought to play an important role in transcriptional activation of retrotransposons [27]. Active retrotransposition can cause the movement of LINE-1 to anywhere within one chromosome and, as such, could disrupt tumor suppressor genes and/or activate oncogenes [28,29]. Indeed, human colon cancer has been shown to be associated with APC retrotransposon activity [28]. In addition, aberrant LINE-1 methylation may cause specific gene modification as well as genomic instability [13,30]. In rat methyl-deficiency models, CpG island aberrant methylation has been seen in several oncogenes and tumorsuppressor genes, such as alpha-fetoprotein (AFP) [31], c-Ha-ras [32], c-Ki-ras [32], c-fos [32], c-myc [33], Dnmt [34], glutathione S-transferase pi (GSTP) [35], p16 [36], and protein tyrosine phosphatase receptor O gene (PTPRO) [37]. Although LINE-1 hypomethylation should be followed by the increase in LINE-1 transcripts, our RT-PCR assessment did not show the increasing tendency. Previously, the similar discrepancy was reported in human liver cancer tissues [38].
In conclusion, the degree of hypomethylation in promoter CpG islands in LINE-1 repetitive sequences in the livers of rats fed a CD diet progresses as a function of feeding period. The level of LINE-1 hypomethylation was similar to that found in previous HPLC analysis for genome-wide hypomethylation. These results suggest that genome-wide hypomethylation occurs because of the choline deficiency diet and that LINE-1 methylation status is a good indicator of such. Moreover, because LINE-1 hypomethylation can activate its retrotransposon activity, it may be the root cause of aberrant methylation in several cancer-related genes in this model. This notion is yet to be proven.

ACKNOWLEDGMENT
Support from the National Cancer Institute, National Institutes of Health (R01 CA82506) is gratefully acknowledged.