Tumorigenesis is a complex process involving genetic, epigenetic, and metabolic alterations. Gestational arsenic exposure has been shown to increase hepatic tumors in adult male offspring of C3H mice, which spontaneously develop hepatic tumors often harboring activating Ha-ras mutation. We explored tumor-promoting changes by gestational arsenic exposure with a focus on Ha-ras mutation and gene expression changes. The results of this study demonstrated that gestational arsenic exposure particularly increased hepatic tumors with a C61A Ha-ras mutation. Real-time PCR analyses on the adult normal livers showed that two genes (Creld2, Slc25a30), whose expression are induced by endoplasmic reticulum stress and cellular oxidative stress, respectively, were significantly upregulated and two genes (Fabp4, Ell3), whose products are involved in lipid efflux and apoptosis, respectively, were significantly downregulated more than twofold by gestational arsenic exposure compared with control mice. The expression changes in the four genes were shown to be late-onset events and to some extent to be associated with corresponding histone modifications, and not with DNA methylation changes. The gene expression changes suggested alterations in lipid metabolism and associated oxidative stress augmentation. Consistently, expression of an oxidative-stress-inducible gene heme oxygenase-1 (HO-1) was upregulated in the livers of the arsenic group. We also found increased expression of retrotransposon L1 mRNA in the tumor-bearing livers of the arsenic group in comparison with control mice. These results suggested that gestational arsenic exposure induces tumor-augmenting changes, including oxidative stress and L1 activation, in a late-onset manner, which would particularly promote tumorigenic expansion of cells with a C61A Ha-ras mutation.
Naturally occurring inorganic arsenic has been identified as a contaminant of drinking well water and as a cause of serious health problems in many areas in the world (Hughes et al., 2011). The major manifestations are early skin lesions and late development of cancer in various organs, including the skin, lung, liver, and urinary bladder (Hughes et al., 2011). Recent epidemiological studies have also indicated that gestational arsenic exposure is associated with increased cancers in adulthood (Smith et al., 2006; Yuan et al., 2010). Consistent with these findings, Waalkes et al. (2003) have shown that gestational exposure of C3H mice, which tend to spontaneously develop tumors in adulthood (Köhle et al., 2008), to inorganic arsenite increases tumors in multiple organs (Tokar et al., 2011).
Multiple stages of hepatic carcinogenesis consist of “initia tion” when altered cell clones originate from stem cells or mature hepatocytes, “promotion” when some of these clones are selected and expanded, and “progression” when the transformed cells acquire the ability to metastasize (Köhle et al., 2008). Mutation of the ras oncogene family (Ha-, Ki-, and N-ras) at codon 12, 13, or 61 is frequently found in several types of cancers in humans and mice and is believed to be among the early events which occur in the initiation stage. It has been reported that 9–60% of the spontaneous hepatic tumors of C3H mice harbor a Ha-ras mutation, predominantly at codon 61 (Maronpot et al., 1995). The ras mutation keeps the protein in the active form which plays a role in various cancer-driving processes (Maronpot et al., 1995; Pylayeva-Gupta et al., 2011).
Previous studies in cell lines, animal models, and humans have reported that arsenic acts as a tumor promoter through genetic, epigenetic, and metabolic alterations (Rossman and Klein, 2011; Salnikow and Zhitkovich, 2008; Wanibuchi et al., 2004). As a result of genetic alterations, tumors exhibit three types of instabilities: chromosomal instability, microsatellite (DNA repeat) instability, and point mutation instability (Preston et al., 2010). Arsenic has been found to induce chromosomal instability, including deletion mutations and chromosome aberrations, but its ability to induce point mutations is believed to be weak (NRC, 1999). The effects of arsenic on the instability of DNA repeat sequences, such as long interspersed nucleotide element-1 (L1) retrotransposon, have not been elucidated. Among epigenetic modifications, such as DNA methylation and histone modification, arsenic is reported to predominantly affect DNA methylation in vitro and in vivo (Reichard and Puga, 2010; Ren et al., 2011). Administration of water containing a 42.5 or 85 ppm concentration of sodium arsenite to pregnant mice from day 8 to 18 of gestation has been found to increase tumor incidence and multiplicity in the livers of their male offspring in a dose-dependent manner when examined at 74 weeks of age. Hypomethylation of the promoter region of estrogen receptor α (ERα) and upregulation of ERα expression were found in the normal tissue of tumor-bearing livers in the gestationally arsenic-exposed group compared to the normal tissues of control mice (Waalkes et al., 2004). However, the causal relationship between alteration of DNA methylation and arsenic-induced tumor formation or tumor-promoting activity has yet to be verified.
In order to delineate the effects of gestational arsenic exposure on the complex tumorigenic process, which should involve multiple factors, in the present study we investigated differences between the livers of control and gestationally arsenic-exposed mice in the experimental model of gestational arsenic exposure reported by Waalkes et al. (2003, 2004). We analyzed Ha-ras mutations in the tumors found in the livers of adult mice, and analyzed gene expression changes in normal adult livers in an attempt to find arsenic-induced changes which lead to the promotion of tumor formation. Involvement of epigenetic regulation was assessed in regard to the genes whose expression was found to be affected by gestational arsenic exposure. Because the results of the gene expression analysis suggested that lipid accumulation and a resulting increase in oxidative stress had occurred in the livers of the gestationally arsenic-exposed group, we measured the lipid contents and expression of heme oxygenase 1 (HO-1), the representative oxidative stress–inducible gene. We also assessed expression of the L1 retrotransposon in the adult livers as another possible genotoxic factor. The results of this study suggest that gestational arsenic exposure increases hepatic tumors which harbor a C61A Ha-ras mutation in adulthood and that late-onset events, including an increase in oxidative stress and L1 activation, are involved in the tumor increase.
MATERIALS AND METHODS
Animal treatment. Pregnant C3H/HeN mice were purchased from Clea Japan (Tokyo) or Japan SLC Inc. (Shizuoka). They were given drinking water or water containing 85 ppm sodium arsenite (Sigma, St Louis, MO) ad libitum from day 8 to 18 of gestation. Diet (CA-1, Clea Japan) was autoclaved for sterilization and given ad libitum. Throughout the experiment, the animals were maintained in a controlled environment at a temperature of 24±1°C and humidity of 50±10%, and under a 12/12h light/dark cycle. The mice were handled in a humane manner in accordance with the National Institute for Environmental Studies (NIES) guidelines for animal experiments.
Histological analysis. Twelve animals in each experimental group were used for histological examination. The animals were sacrificed by exsanguination from the inferior vena cava under overdose of pentobarbital. The livers were immersed in a phosphate-buffered 4% paraformaldehyde solution (pH 7.3) overnight. The liver tissue slices from the four main lobes, in addition to the tumors detected under a stereoscopic microscope, were embedded in paraffin, and 4-µm sections were stained with hematoxylin and eosin. The histology of the liver neoplasms was classified as hepatocellular adenoma and hepatocellular carcinoma according to the criteria by Harada et al. (1999). Briefly, hepatocellular adenoma is characterized by a well-circumscribed lesion composed of well-differentiated hepatocytes, and hepatocellular carcinoma is characterized by abnormal growth pattern and cytological and nuclear atypia.
Ha-ras mutation. Ha-ras mutation was analyzed according to the protocol provided by Dr Yukihito Ishizaka (National Center for Global Health and Medicine). The liver tissue was lysed in lysis buffer (50mM Tris-HCl (pH 8.0), 0.1M NaCl, 20mM EDTA, 1% SDS, 0.3mg/ml proteinase K). Genomic DNA was purified with phenol chloroform mixture, precipitated with ethanol, dried, and resuspended in buffer containing 10mM Tris-HCl and 1mM EDTA, pH 8.0. The DNA regions containing the mutation sites in Ha-ras gene were amplified by PCR using primer pairs with the sequences 5´-ggttctgtggattctctggtc-3´ and 5´-aaggacttggtgttgttgatg-3´ for codon 61 and 5´-cattggcaggtggggcaggag-3´ and 5´-taacacctacaagacctgggc-3´ for codons 12 and 13. The PCR products were electrophoresed and purified using Wizard DNA Clean-up System (Promega). The modified PCR products from each sample were cycle sequenced with the primers for Ha-ras amplification and a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and analyzed using an Applied Biosystems 3730 DNA analyzer (Applied Biosystems).
Affymetrix GeneChip analysis. Affymetrix GeneChip analysis was performed as previously described (Nohara et al., 2008) with some modifications. Briefly, total RNA samples were prepared from pooled liver homogenates of four control mice and four arsenic-exposed mice. The total RNA was isolated using ISOGEN RNA isolation kit (Nippon Gene, Toyama) and purified using an RNeasy Mini Kit (Qiagen, Chatsworth, DA), creating two samples: one sample for the control and one for the arsenic group. Double-stranded cDNA was synthesized from 100ng of liver total RNA and was used to prepare biotin-labeled aRNA by in vitro transcription. About 15 µg sample of the biotin-labeled aRNA was fragmented and hybridized to a Mouse Genome 430 2.0 array (Affymetrix). After being hybridized for 16h, the array was washed, stained, and scanned. Data were analyzed with Affymetrix GCOS 1.4 software and GeneSpring GX 10.0 (Silicon Genetics, Redwood City, CA). This procedure was carried out twice with different total RNA sample sets and genes whose expression was consistently increased or decreased by twofold or more in the exposed groups compared with the control groups were selected.
cDNA preparation and real-time PCR. Total RNA of individual livers was prepared with an RNeasy Mini Kit. DNase treatment was repeated twice to completely remove contaminating genomic DNA which interferes with L1 RNA measurement. The purity of RNA was checked by confirming no amplification of target DNA occurs before reverse transcription. Reverse transcription reaction was performed by TaKaRa RNA PCR Kit (AMV) Ver.3.0 (TaKaRa, Shiga, Japan). Quantitative real-time PCR analysis was performed on LightCycler 480 instrument (Roche Diagnostics, Basel, Switzerland) as described previously (Nohara et al., 2006). The primer sequences and annealing temperatures used for real-time PCR are shown in Supplementary table 1. Primer sets ORF1a and ORF2a (Muotri et al., 2010) were used for real-time PCR of L1 ORF1 and ORF2, respectively.
Chromatin immunoprecipitation assay. Chromatin immunoprecipitation (ChIP) assay for chromatin modification was carried out as previously described (Suzuki and Nohara, 2007) with some modifications. Briefly, liver tissue was minced and treated with 1% formaldehyde to cross-link protein-DNA complexes and then sonicated to give a final DNA size range from 200 to 800bp. Immunoprecipitation was carried out by incubating with anti-acetyl-histone H3 (06-599, Upstate), anti-acetyl-histone H4 (06-866, Upstate), anti-dimethyl-histone H3 (Lys9) (ab1220, Abcam), anti-trimethyl-histone H3 (Lys4) (ab1012, Abcam), or anti-rabbit-control IgG (ab46540, Abcam) antibody, followed by incubation with protein A-agarose. After protein-DNA complexes were recovered from the protein A-agarose beads, the formaldehyde cross-linking was reversed and protein was hydrolyzed with proteinase K. DNA fragments were purified by QIAquick PCR Purification Kit (Qiagen) and subjected to real-time PCR. The primers to detect histone modifications were designed near the transcription start site (TSS) (Roh et al., 2005; Vakoc, et al., 2006) and the sequences of primers were listed in Supplementary table 1.
Bisulfite sequencing analysis. DNA methylation status was examined for the CpG islands in the proximal promoter region of Creld2, Slc25a30, and ERα and the CpG-rich region downstream of transcription start site of Ell3 by bisulfite sequencing. Genomic DNA (2 µg), prepared as described above, were digested with EcoRI and incubated with freshly prepared 0.3M NaOH for 15min. To this solution was added sodium metabisulfite, pH 5.0, and hydroquinone to give final concentrations of 2.0M and 0.5mM, respectively. The mixture was incubated for 16h at 50°C in the dark. The samples were desalted using the Wizard DNA Clean-up System (Promega), and the bisulfite reaction was terminated by adding NaOH to give a final concentration of 0.3M and incubating for 15min at 37°C. The DNA was then precipitated with ethanol, dried, and resuspended in buffer containing 10mM Tris-HCl and 1mM EDTA, pH 8.0. The modified DNA was amplified with the primers shown in Supplementary table 1. PCR products were cloned into pUC18 vector (TaKaRa Bio). Clones from each sample were cycle sequenced with M13RV primers and a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and analyzed with 3730 DNA analyzer (Applied Biosystems).
Lipid contents. Total lipids were extracted from liver tissue according to Folch et al. (1957) and the amounts of triglycerides and total cholesterol were measured using CholesTest TG and CholesTest CHO (Sekisui Medical, Tokyo), respectively. Measurement was done by Skylight Biotech Inc. (Akita, Japan).
Statistical analysis. Differences among the control and arsenic-treated groups were analyzed by Mann–Whitney’s U-test. The statistical analysis was performed by using StatView statistical software (version 5.0, SAS Institute, Cary, NC). A p value less than 0.05 was considered significant.
Increased Hepatic Tumorigenesis as a Result of Gestational Arsenic Exposure
We examined 85 male offspring in a control group and 85 male offspring in a gestationally arsenic-exposed group for macroscopic hepatic tumors when the offspring had reached 74–84 weeks of age (Fig. 1). The mice in the control group and gestationally arsenic-exposed group were dissected and examined in an age-matched manner. Gestational arsenic exposure was followed by a tendency toward a higher number of mice having hepatic tumors and a greater multiplicity and size of the tumors than in the control group (Fig. 1A). Pathological examination of 12 mice in each group showed that adenoma was the predominant histological type of tumor, particularly in the gestationally arsenic-exposed group (Figs. 1B and 1C).
On the other hand, the incidence and multiplicity of hepatic tumors in the female offspring were not increased by gestational arsenic exposure because 3 of 25 female offspring in the control group and 2 of 25 female offspring in the arsenic group had developed hepatic tumors at 82 weeks of age.
Increased Hepatic Tumors Bearing a Ha-ras Mutation as a Result of Gestational Arsenic Exposure
We detected three different types of Ha-ras mutations at codon 61, that is, C61A mutations, A61T mutations, and A61G mutations, with C61A mutations predominating, in the hepatic tumor tissue of both the male control mice and the male gestationally arsenic-exposed mice (Figs. 2A and 2B), consistent with previous data (Maronpot et al., 1995). No mutations at codon 12 or 13 were detected in the hepatic tumor tissue of the 12 male mice examined (data not shown). No Ha-ras mutations were detected in the normal tissues in the livers (data not shown), indicating that the mutation at codon 61 was somatically acquired. The mutation analysis revealed that gestational arsenic exposure particularly increased the percentage of tumors containing a CAA → AAA Ha-ras mutation at codon 61 (C61A; Fig. 2B). The percentages of tumor-bearing livers with or without Ha-ras mutation in the total livers were calculated using the data shown in Fig. 1A and Fig. 2B, which demonstrated that gestational arsenic exposure rather reduced the percentage of livers having tumors without Ha-ras mutation and particularly increased the percentage of livers having tumors containing a C61A Ha-ras mutation (Fig. 2C).
Late-onset Changes in Gene Expression in Non-tumor-bearing Normal Livers After Gestational Arsenic Exposure
To identify factors which promote the later steps of hepatic tumorigenesis in the male mice gestationally exposed to arsenic, we searched for genes whose expression was changed in the as yet non-tumor-bearing livers from the arsenic-exposed group compared with those from the control group at 74 weeks of age by using Affymetrix GeneChips. A total RNA sample was prepared for the control (from a pool of four mice) and analyzed against a total RNA sample for the arsenic group (from a pool of four mice). Two independent experiments with different total RNA sample sets suggested that expression of 19 genes was consistently altered more than twofold in the arsenic-exposed group in comparison with the control group (Supplementary table 2). Expression of the 11 genes which had certain signal intensities (average signal intensity > 200 in control or gestationally arsenic-exposed group) was verified by real-time PCR. Expression of only two genes, Creld2 (cysteine-rich with EGF-like domains 2) and Slc25a30 (solute carrier family 25, member 30, or KMCP1), was shown to be significantly upregulated in the gestationally arsenic-exposed group, and two genes, Fabp4 (fatty acid binding protein 4, adipocyte) and Ell3 (elongation factor RNA polymerase II-like 3), to be significantly downregulated (Supplementary figure 1). The possible involvement of these genes in tumorigenic process is discussed below in the Discussion section. The expression of these four genes in the liver of female mice was unchanged at 82 weeks of age after gestational arsenic exposure (data not shown).
Expression of the above four genes was also examined in the livers of male mice at 6 weeks and 49 weeks of age (Fig. 3). No tumors was observed in their livers at these ages. Significant upregulation of expression of Creld2 and significant upregulation of expression of Slc25a30 were first detected at 49 weeks of age and at 74 weeks of age, respectively. Although expression of Fabp4 and Ell3 tended to be downregulated in the gestationally arsenic-exposed group at 6 weeks of age, significant suppression of Fabp4 and Ell3 was observed at 74 weeks of age and 49 weeks of age, respectively.
These results indicated that gestational arsenic exposure induces late-onset changes in the expression of a limited number of genes in the livers of adult male offspring.
Epigenetic Modification of the Genes Found to Exhibit Late-onset Changes in Expression After Gestational Arsenic Exposure
Because epigenetic modifications have been implicated in arsenic-induced changes in gene expression in many previous studies (Reichard and Puga, 2010; Ren et al., 2011; Rossman and Klein, 2011), we examined the four genes whose expression was found to be altered by gestational arsenic exposure in a late-onset manner (Creld2, Slc25a30, Fabp4, and Ell3) for the presence of epigenetic modification.
To examine the genes for histone modification, we measured the levels of H3 acetylation (H3Ac), H4 acetylation (H4Ac), H3K4 trimethylation (H3K4me3), and H3K9 dimethylation (H3K9me2) in the proximal promoter regions of the Creld2, Slc25a30, Fabp4, and Ell3 genes by ChIP assay. H3Ac and H4Ac are general markers for active genes, and H3K4me3 is a pivotal transcription-activating histone modification. H3K9me2 is the most common methylation state of H3K9 and the hallmark of heterochromatin where transcription is suppressed (Li et al., 2007). A clear correlation was observed between the suppression of Fabp4 expression and the significant increase in H3K9me2 induced by arsenic exposure (Fig. 4). The significant increase in H3K4me3 in the promoter of Slc25a30 induced by arsenic exposure paralleled the arsenic-induced upregulation of expression of this gene (Fig. 4).
DNA methylation in the CpG islands or CpG-rich region around the promoter was investigated by bisulfite sequencing as described in the Materials and Methods section. The results showed no marked difference in DNA methylation between the control mice and the mice that had been gestationally exposed to arsenic because almost all of the CpG islands of Creld2 and Slc25a30 were unmethylated in both groups (data not shown), and the CpG-rich region of Ell3 included methylated CpG at 3.4% in the control group and at 4.5% in the arsenic group, respectively (Supplementary figure 2). No CpG islands were found in the promoter region of Fabp4.
These results suggest that histone modification may to some extent be involved in the regulation of the late-onset arsenic-induced changes in expression of Fabp4 and Slc25a30. On the other hand, there was no evidence implicating DNA methylation in the arsenic-induced changes in gene expression.
Changes in Lipid Metabolism and HO-1 Expression in the Livers of Mice Gestationally Exposed to Arsenic
Creld2 is an endoplasmic reticulum (ER) stress-inducible gene (Oh-hashi et al., 2009), and ER stress is known to activate lipogenesis (Ozcan and Tabas, 2012). The adipocyte fatty acid–binding protein, Fabp4, plays a role in effluxing fatty acids from adipocytes (Smith et al., 2008), although its function in the liver remains to be clarified. Thus, the late-onset upregulation of Creld2 and downregulation of Fabp4 which were found to have occurred in the gestationally arsenic-exposed livers suggested lipid had accumulated in the livers as a result of increased lipogenesis and suppressed lipid efflux.
We therefore measured the triglyceride and the cholesterol contents of the livers of adult mice in both groups. The results showed increased triglyceride content in the arsenic-exposed group in comparison with the control group, but because of the large deviation in the control group, the difference was not statistically significant (Fig. 5A). The total cholesterol content of the livers in the control group and gestationally arsenic-exposed group was similar (Fig. 5A). To further assess the changes in lipid metabolism, we measured the expression of sterol regulatory element-binding protein 1 (Srebp1) and one of the target genes, glycerol-3-phosphate acyltransferase-1 (Gpat-1), in the livers of control mice and arsenic-exposed mice. Srebps are central transcription factors which control intracellular levels of cholesterol and fatty acids through feedback and feed-forward regulation at the transcriptional, translational, and posttranslational levels (Raghow et al., 2011). Real-time PCR analysis revealed that the expression of Gpat-1 (Wendel et al., 2010) was suppressed in the livers of the arsenic group in comparison with those of control group, whereas the expression of Srebp1 was not different in the two groups (Fig. 5B). These results suggest that Srebp1 activity is suppressed to reduce lipid metabolism in consequence of sensing lipid accumulation.
Lipid accumulation in the liver often leads to increased sensitivity of the liver to oxidative stress (Rolo et al., 2012). In order to investigate the involvement of oxidative stress in arsenic-induced tumor promotion, we measured the expression of HO-1, a gene which is representative of the genes whose expression is induced by oxidative stress. Expression of HO-1 was found to be higher in the tumor-free livers of the gestationally arsenic-exposed group than in the tumor-free livers of the control mice (Fig. 6). Expression of HO-1 was also higher in the nontumor tissue of tumor-bearing livers from the arsenic-exposed group than in the nontumor tissue of tumor-bearing livers from the control mice. The tumor tissue from both the control group and the arsenic-exposed group showed higher levels of HO-1 expression (Fig. 6). No upregulation of HO-1 expression as a result of gestational arsenic exposure was detected in the tumor-free livers of 6- or 49-weeks-old male mice in comparison with the group (data not shown), and upregulation of HO-1 expression was found to be a late-onset event.
The above results suggest that gestational arsenic exposure induces late-onset changes in gene expression leading to lipid metabolism change and augmented oxidative stress, which promote tumorigenesis.
Increased Retrotransposon L1 Activity in the Livers of Mice Gestationally Exposed to Arsenic
Whereas most transposable elements have been rendered inactive by mutation, some L1 retrotransposons still have transposable activity (Beck et al., 2011). L1 expression and the subsequent retrotransposition can act mutagenically, for example, by directly disrupting genes by insertion into the genome or by causing double-stranded DNA breaks as a result of the endonuclease activity of its open reading frame (ORF) 2 product (Beck et al., 2011). Although L1 has been believed to be predominantly expressed in the germ line, evidence of its expression in cancer cells, brain tissues, and various other somatic tissues has been accumulating (Beck et al., 2011; Belancio, et al., 2010).
In the present study, we measured L1 ORF1 and ORF2 RNA in the normal tissue and tumor tissue of the livers of control adult mice and gestationally arsenic-exposed adult mice. Hepatic tissue from the gestationally arsenic-exposed group, including both normal tissue from tumor-free and tumor-bearing livers and tumor tissue, showed higher L1 expression than hepatic tissue from the control group (Fig. 7). In particular, the amount of L1 RNA in the normal tissues of tumor-bearing livers was significantly higher in the gestationally arsenic-exposed group than in the control group. The level of L1 RNA expression in the non-tumor-bearing livers of the 6- and 49-weeks-old male mice in the control group and gestationally arsenic-exposed group was similar (data not shown). Thus, an increase in L1 expression was also a late-onset event.
These results suggest the possibility of tardive L1 activation in the liver as a result of gestational arsenic exposure and promotion of greater genetic instability.
In the present study, we confirmed that gestational arsenic exposure increases the incidence and multiplicity of tumors in the livers of male C3H mice as previously reported (Waalkes et al., 2003). Another feature of the tumors in the arsenic group was the larger size of the tumors than in the control group (Fig. 1). Furthermore, we found that gestational arsenic exposure particularly increased hepatic tumors containing the C61A Ha-ras mutation (Fig. 2). Mutation at codon 61, including C61A, A61T, and A61T mutations, results in an amino acid substitution and impairs the guanosine triphosphate (GTP) hydrolysis reaction, thereby maintaining Ras in the active GTP-binding form (Pylayeva-Gupta et al., 2011). As a result, the Ha-ras mutation results in incessant activation of Ras and a variety of downstream signaling pathways, including the Raf-ERK and PI3 kinase pathways, which drive tumorigenesis by increasing proliferation, suppressing apoptosis, and altering metabolism, and it also activates apoptosis, depending on the status of the ras effectors (Pylayeva-Gupta et al., 2011). In consistent with the notion that gestational arsenic exposure aggravates the inherent predisposition to disease of the animal (States et al. 2011), our results indicated that gestational arsenic exposure increases Ha-ras mutation, which is one of critical factors leading to hepatic tumorigenesis in C3H mice.
The particular increase in tumors containing the C61A Ha-ras mutation may be attributed to oxidative-stress-induced G:C to T:A transversion. A G:C to T:A transversion has been reported to be induced following formation of 8-oxo-2′-deoxyguanosine (8-oxodG), a representative oxidative DNA damage, in vitro (Sibutani et al., 1991). Increased G:C to T:A transversion has also been reported in the livers of mice which are deficient in the antioxidant enzyme CuZn-superoxide dismutase, and such mice have a higher incidence of hepatic cancer (Busuttil et al., 2005). The increase in oxidative stress found in the present study at a later stage by gestational arsenic exposure might be involved in the increase in tumor cells containing a C61A Ha-ras mutation. The formation of reactive oxygen species and increase in oxidative stress have been suggested as an immediate effect of arsenic toxicity (Hughes et al., 2011). Another possible causation of the increase in tumor cells carrying a C61A Ha-ras mutation by gestational arsenic exposure might be the oxidative stress induced at the initiation stage in the fetal liver.
A recent study by Waalkes et al. (2008) reported that gestational arsenic exposure plus postnatal 12-O-tetradecanoyl phorbol-13-acetate (TPA) application to Tg.AC mice increased the incidence of squamous cell carcinomas in comparison with TPA application alone. The Tg.AC transgenic mouse carries an oncogenic v-Ha-ras transgene under control of the mouse ζ-globin promoter, which is transcriptionally silent until activated by tumor promoters, including TPA, and other stimuli (Humble et al., 2005). Thus, the results reported by Waalkes et al. (2008) and the results obtained in this study both consistently indicate that gestational arsenic exposure increases tumors bearing a Ha-ras mutation. On the other hand, the study by Waalkes et al. (2008) reported that gestational arsenic exposure increased v-Ha-ras expression in the fetal skin, implying that arsenic increased v-Ha-ras expression regulated by the mouse ζ-globin promoter in Tg.AC mice. Thus, the mechanism by which arsenic induced the increase in tumors in the two studies seems to have been different.
In the present study, we also attempted to identify factors which promote the growth of the altered cells in adulthood by analyzing changes in gene expression using microarrays and real-time PCR in normal adult livers in which no tumors had yet developed. The results showed that gestational arsenic exposure induced large late changes in the expression of a limited number of genes in the livers of adult mice (Fig. 3). The genes whose expression was upregulated by arsenic were Creld2 and Slc25a30, and those whose expression was downregulated were Fabp4 and Ell3. Because Creld2 is an ER stress-inducible gene (Oh-hashi et al., 2009), its upregulation in the liver suggested increased ER stress. ER stress may activate lipogenesis (Ozcan and Tabas, 2012), which is a key event that leads to hepatic steatosis, that is, intracellular triglyceride accumulation in the liver. Also, the downregulation of the gene encoding Fabp4, a molecule involved in fatty acid efflux (Smith et al., 2008), may lead to lipid accumulation in the liver. Consistently, lipid measurements showed a tendency toward accumulation of triglyceride in the livers of the arsenic group in our study (Fig. 5A). In addition, we detected suppressed expression of Gpat-1, one of the target genes of the lipid-level regulators, Srebps (Ye and DeBose-Boyd, 2011), which suggested suppressed Srebp activity due to accumulation of triglyceride (Fig. 5B).
Steatosis gives rise to increased oxidative stress (Powell et al., 2005; Rolo et al., 2012), and the oxidative stress associated with steatosis is believed to play significant roles in tumorigenesis by causing DNA damage and by altering signal transductions and inducing epigenetic alterations (Ziech et al., 2011). The upregulation of a member of mitochondrial transporter family Slc25a30, whose expression has been reported to be upregulated by cellular oxidative stress (Haguenauer et al., 2005), in the arsenic-exposed livers may also be a sign of increased oxidative stress. In accordance with the notion, we detected upregulation of the representative oxidative-stress-inducible gene HO-1 in the normal hepatic tissue of adult mice in arsenic group (Fig. 6), and that was also a late-onset event. These results suggest that gestational arsenic exposure induces lipid accumulation in the liver as a late-onset event which leads to increased oxidative stress and promotes tumorigenesis. How these factors function in relation to Ha-ras signaling has yet to be clarified. In addition, reduced efficiency of the antioxidant pathway in association with aging may be another factor involved in the tardive effects of gestational arsenic exposure (Powell et al., 2005). Although HO-1 is a well-known sensor of oxidative stress, it has also been demonstrated to play a role in tumorigenesis by stimulating proliferation and exerting anti-apoptotic activity (Was et al., 2010). Thus, upregulation of HO-1 may be one of the factors which are involved in the promotion of tumorigenesis. Downregulation of Ell3, which has been reported to induce cell-cycle arrest and apoptosis (Johnstone et al., 2001), was also detected by gestational arsenic exposure. The changes in gene expression found in the present study are promising clues to the elucidation of the process of the arsenic-induced tumor increase.
Because previous studies have reported that arsenic modulates gene expression by altering DNA methylation (Reichard and Puga, 2010; Ren et al., 2011), we examined the involvement of DNA methylation in the changes in expression of the four genes (Creld, Slc25a30, Fabp4, and Ell3) whose expression was altered as late-onset events by arsenic as described above. However, as reported in the Results section, no evidence of involvement of DNA methylation in the changes in gene expression was found. Furthermore, although hypomethylation of the promoter region of ERα and corresponding upregulation of ERα expression were reported in the normal tissue of tumor-bearing livers of mice gestationally exposed to arsenic in comparison with the normal livers of control mice (Walkees et al., 2004), neither of these changes was detected in the present study (Supplementary figure 3). The results of the present study indicated that neither changes in DNA methylation nor changes in the expression of ERα is a prerequisite for the increase in hepatic tumorigenesis as a result of gestational arsenic exposure. On the other hand, the results of the present study indicated that H3K9me2 and H3K4me3 may be involved in the tardive transcriptional regulation of Fabp4 and Slc25a30, respectively (Fig. 4).
We also investigated L1 mRNA expression in the liver in the present study. The results of a recent study conducted on a DMBA/TPA-induced model of cutaneous tumorigenesis in L1 transgenic mice which enabled detection of L1 retrotransposition suggested that L1 retrotransposition occurred in the tumor tissue during tumor promotion induced by repeated TPA application (Okudaira et al., 2011). Consistent with the findings in that study, the results of the present study showed that L1 expression in the adult liver was increased by gestational arsenic exposure; however, the significant upregulation was detected in the normal tissue of tumor-bearing livers. The discrepancy between the regions where TPA or arsenic seems to increase L1 activity in the two studies may be attributable to the difference in detection methods (L1 retroposition and mRNA measurement) or the difference between the animal models used in the two studies. L1 mRNA expression is regulated by multiple factors, including tissue-specific transcription factors, DNA methylation in the promoter region, and premature polyadenylation (Belanco et al., 2006). Further study will be necessary to elucidate how gestational arsenic exposure affects L1 mRNA expression. Increased L1 transcription may be one of the tumor-promoting stimuli in arsenic-exposed livers.
In summary, the results of this study suggest that gestational arsenic exposure increases hepatic tumors particularly bearing a C61A Ha-ras mutation in adult male mice and that a late-onset increase in oxidative stress and L1 activity is involved in the tumor increase (Fig. 8). These findings would provide important clues for identifying the targets of tumorigenesis by gestational arsenic exposure.
National Institute for Environmental Studies (0710AG333, K.N., 1115AA082); Ministry of the Environment of Japan, R&D Project for Environmental Nanotechnology (K.N.); Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research (19590611, 23390166, K.N.); Banyu Foundation Research Grant (K.N.).
The authors wish to thank Dr Michael P. Waalkes for his helpful advice on the animal experiment and Dr Linda Birnbaum for her encouragement. They also wish to thank Dr Yukihito Ishizaka for his kind advice on the analysis of Ha-ras mutation and Dr Takashi Sugimura, Dr Toshikazu Ushijima, Dr Satoshi Yamashita, and Dr Hidenori Ojima (National Cancer Center Research Institute), Dr Katsuhiro Ogawa (Asahikawa Medical School), and Dr Hiroshi Nitta and Mr Yuji Koresawa (NIES) for their helpful discussions. The authors also wish to thank M. Matsumoto and S. Itaki for their excellent technical and secretarial assistance.