The role of estrogen in modulating morphological and physiological features of liver became evident in early 1970s when a possible correlation between occurrence of hepatic neoplasms and use of oral contraceptives was suggested[26,27]. In the reproductive system, estrogen is known to act by binding to specific cytoplasmic and nuclear receptors. Hence, search began to identify such a receptor in the liver. In 1978, Duffy and Duffy first reported the presence of ER in normal human liver[28]. Subsequently, the presence of ERs in human HCC was demonstrated by Molteni et al[29] followed by Friedman et al[30] and Iqbal et al[31]. Since then, a number of studies have been reported addressing the expression of ERs in normal as well as neoplastic liver tissues. The early studies used indirect methods of receptor detection based on ligand binding assays. Table 1 gives the details of these studies[13,15,17,30-37]. These assays were quantitative and measured the amount of receptor in the samples as well as its affinity for the ligand in terms of dissociation constants. The percentage positivity for ER expression varied significantly among different studies. These variations may be attributed to the differences in sample size, methodologies, ethnicity of the population studied, stage of the disease and underlying etiologies. Earlier studies by Friedman et al[30] and Iqbal et al[31] showed that ER content is similar in HCC and normal liver. In contrast, later studies consistently showed that the expression of ER is decreased in HCC tissue specimens as compared to normal liver tissue specimens or the non-tumor part of the liver[17,32,38]. However, Eagon et al[33] documented elevated levels of cytosolic ER in 3 of the 9 tumors as compared to non-cancerous tissues. Nuclear ER expression was found to be suppressed in all HCC samples as compared to normal samples[33]. The major drawback of these studies was the use of binding assays for detection that do not provide any information on the subtype of ER, i.e. ERα and ERβ as known today. It is important to study the relative expression of both isoforms of ER since ERα and ERβ are known to have overlapping but quite distinct functions. There are few reports on direct detection of ER using specific antibodies. Table 1 gives the details of these studies[32,39-47]. However, all these studies have employed either immunohistochemistry (IHC) or enzyme-immuno assays (EIA) using antibodies specific for only ERα isoform. The ERβ isoform was identified later in 1996[48] and information on the expression of ERβ protein in HCC is lacking though few studies at mRNA levels have been documented.

Since the molecular characterization and cloning of ERα in the mid 1980s[49,50], attempts have been made to determine the expression of ER in liver tumors at mRNA level. Table 2 gives the details of these studies[18,19,40,45,51,52]. In situ hybridization using ER specific oligonucleotide sequence probe revealed that 11 out of 15 HCC tissue samples expressed ER mRNA[40]. Interestingly, the same samples were found to be negative for ER protein by IHC, suggesting the use of more sensitive methods and more specific antibodies for detecting ER at protein level. Subsequently, the mRNA expression of ER in HCC tissues was studied in different populations by a highly sensitive method of reverse transcriptase-polymerase chain reaction (RT-PCR). Villa et al[51] were the first to demonstrate the presence of wild type ERα in peritumoral and tumoral tissue of HCC patients using this technique. The use of RT-PCR further enabled the authors to detect a splice variant of ERα lacking exon 5 in the hormone binding domain[18,51]. A similar splice variant has been described in breast cancer tissues to be associated with tumor pathogenesis[53]. The significance of the variant ERα (vERα) in pathology, prognosis and treatment of HCC has also been studied. The presence of vER receptor is able to influence the natural history of patients with HCC by regulating tumor growth as well as patient survival. The presence of the liver vERα transcript in the tumor has been described to be the strongest negative predictor of survival in operable HCC patients[54,55]. Furthermore, the presence of vERα was found to correlate with a higher clinical aggressiveness of the tumor in comparison with the tumors characterized by wild-type ERα transcript. These tumors were responsive to megestrol and unresponsive to anti estrogen tamoxifen. High rates of vERα expression have been shown to be present in men at high risk of HCC development[56,57]. In patients with chronic hepatitis and cirrhosis, the expression of vERα has been associated with higher oxidative stress-induced DNA damage and c-myc mRNA expression, a factor indicating increased genomic instability, augmented cytoproliferation and carcinogenesis[5].

Using RT-PCR, in addition to ERα, the expression of the lately identified isoform of ER, i.e. ERβ, has also been studied in HCC patients. Iavarone et al[19] report that both ERβ and ERα wild type receptors either alone or together with vER are co-expressed more frequently in patients with chronic liver disease than in those with HCC. However, both ERs are similarly expressed in tumoral and extratumoral tissues of HCC patients[19]. In this study, HBV-related tumors either expressed wild type ERα and ERβ or expressed variant ER and ER delta 5 more often than HCV-related tumor, and HBV-related tumors showed a tendency towards loss of ERβ expression as the disease progressed from chronic inflammatory liver disease to HCC[19].

Breast cancer studies suggest that ERα:ERβ expression ratio changes during carcinogenesis and is believed to play a role in tumor development[58]. Recently, Wang et al[52] studied the expression of ERα and ERβ in HCC tissues of Korean population using RT-PCR, and assessed 32 tumoral and peritumoral tissues from HCC patients with underlying chronic HBV or HCV infection and observed that wild type and variant ERα are expressed in all the samples. However, the expression of vERα is stronger in tumor than in peritumor tissues. Interestingly, ERβ was found to be significantly over-expressed in HCV-infected HCC tissues as compared to HBV-infected HCC tissues. The differences in ER expression in HCV-infected HCC tissues compared to HBV-infected HCC tissues suggest different pathogenetic mechanisms. Overall from these studies there appears to be a change in co-expression pattern of ERα and ERβ from cirrhosis to HCC development in both HBV- and HCV- related tumors. Thus, these studies provide further evidence in support of importance of wild type and variant ERs in HCC, suggesting detailed investigations in this area.

Like estrogens, androgens have also been reported to play an important role in liver carcinogenesis. Iqbal et al[31] showed the presence of androgen receptors (AR) in HCC in 1983. However, it was not until 1985 that normal human liver was believed to express AR. In 1985, Nagasue and colleagues[14] demonstrated the presence of AR in normal human liver as well as in tumor and non-tumor parts of HCC tissues. Since then, several reports showing protein and mRNA expression of AR in liver have been published. Table 3[13,15,17,28,30-33,35,40,41,52,59-64] and Table 4[14-17,31,33-36,39,41,47,52,65-71] give the details of these studies. In general, AR is found to be over expressed in liver tumor compared to the adjacent normal tissue[15,17,33,36,38]. However, like ER expression studies, majority of the early studies employed indirect binding assays to detect AR in the liver tissues. More precise quantitative methods of direct detection of AR protein are needed. Reports using antibody-based detection of AR in liver tissues are very sparse.

The mRNA levels of AR have been assessed in the non-cancerous and HCC tissues primarily by RT-PCR (Tables 3 and 4). In 1994, Negro and colleagues[16] developed a non-radioisotopic in situ hybridization assay specific for human AR mRNA and found that 73% of HCC tissues could express variable amount of AR mRNA. However, normal hepatocytes were stained weakly in 42% of the non-neoplastic tissues. Though initial binding studies demonstrated higher AR levels in tumor tissues than in respective peri-tumoral part[15,17,33,36,38], more recent observations based on mRNA expression do not reveal any significant differences in tumor and peri-tumor tissues. Tavian et al[71] found higher AR mRNA levels in tumor than in the corresponding peri-tumoral tissue in a relatively small percentage of HCC samples, suggesting that AR mRNA levels are associated with the histological tumor differentiation showing a lower AR expression in poorly-differentiated HCC than in well-differentiated tumors. In contrast, AR levels in Korean population with HCC do not show differences between tumor and peri-tumor tissues using RT-PCR[52]. Due to these conflicting reports on AR expression in HCC, there is a need for detailed investigation of AR mRNA as well as protein levels using more sensitive and accurate detecting quantitative methods.

Several attempts have been made to establish the role of estrogen and its receptors in hepatocarcinogenesis using animal models. Rat is the most extensively used model to study liver carcinogenesis. Rat hepatocytes are known to express ERs. ERα is the predominant isoform expressed in rat hepatocytes while cholangiocytes express both ERα and ERβ[80]. However, Inoue et al[81] showed that the levels of ERβ are higher than those of ER in cultured rat hepatocytes[81]. Hepatic stellate cells from rats appear to contain mainly ERβ[82]. Due to the lack of information about the existence of various ER isoforms, in most of the earlier studies, hepatic stellate cells did not differentiate ER into ERα and ERβ. In addition, majority of the in vivo studies have been conducted in animal models of chemical carcinogenesis. Diethylnitrosamine (DEN) is the most commonly used carcinogen in rat and mouse models of HCC. The pathogenesis of HCC in DEN-induced carcinogenesis in animal models differs from that in humans and therefore may not be directly comparable to human HCC[83]. Nevertheless, the histology and genetic signatures are similar to human HCC and a striking gender disparity with male predominance is also observed in these animal models as seen in humans[84]. In addition to DEN, acetylaminofluorene (AAF), di(2-ethylhexyl) phthalate (DEHP), peroxisome proliferator, arsenic and carbon tetrachloride have been used to induce HCC in various animal models[85-90].

Use of oral contraceptives and synthetic estrogens in women is reported to be a major risk factor for the development of hepatocellular adenoma, a benign liver tumor with malignant potential[26,27]. Shimomura et al[91] studied the role of ethinyl estradiol (EE) in inducing HCC in female rats, following EE treatment for 12 mo, 8% of rats developed HCC, revealing that EE causes mutations in hepatocytes leading to DNA adduct formation and induces HCC development in affected cells. The initial events in HCC, i.e. DNA adduct formation by EE, appear to be carried out in an ER independent manner since tamoxifen, a known selective estrogen receptor modulator (SERM), inhibited ER expression and suppressed HCC, but did not affect DNA adduct formation. Exogenous estrogens have also been shown to promote hepatocarcinogenesis induced by other agents[92,93]. Campen et al[92] documented that administration of 17-alpha ethinylestradiol in ovariectomized rats promotes DEN-induced carcinogenesis in a dose dependent manner. Further, it was reported that synthetic female hormones act synergistically with ethanol to increase HCC incidence[93]. Alcohol could affect HCC development due to EE by promoting changes in ER kinetics and expression as well as in DNA adduct formation.

Liver is the major site of estrogen metabolism[94]. Alterations in sex hormone metabolism are also considered a critical factor determining the significance of sex hormones in the process of liver carcinogenesis. Eagon et al[86,87] reported that the activity of male estrogen-metabolizing enzyme, estrogen 2-hydroxylase and male specific estrogen sequestering protein is reduced in liver, which explains the raised serum estradiol levels but the decreased hepatic activity of cytosolic and nuclear ER observed in DEHP-induced HCC in male Fischer 344 rats. The expression of cytochrome P450 enzymes that play an important role in estrogen metabolism have also been shown to be affected during hepatocarcinogenesis. Waalkes and colleagues[88] have described the feminized pattern of P450 genes in male mice with HCC induced by exposure to arsenic in utero. The expression of female dominant CYP2A4 and CYP2B9 is increased whereas levels of male dominant CYP7B1 gene gets reduced in arsenic treated animals. Recent findings suggest that cytochrome P450 (CYP) is regulated by estrogen itself through the involvement of estrogen receptors[94]. Nonetheless, elucidation of the exact mechanism of regulation of CYP isoforms by estrogen in liver needs further investigation.

In contrast with earlier animal studies that support estrogens in promoting and inducing carcinogenesis, recent studies highlighted the protective role of estrogens in HCC development. Shimizu et al[8] showed that estrogen can suppress chemical hepatocarcinogenesis induced by dimethylnitrosamine (DEN)-2-acetylaminofluorene (AAF) in partial hepatectomy (PH) model of hepatocarcinogenesis. In addition, estrogen has been shown to prevent the progression of liver disease to HCC. Estradiol treatment could reduce hepatic steatosis and restore the impairment in mitochondrial and peroxisomal fatty acid β-oxidation in aromatase-deficient mice which lack intrinsic ability to produce estrogen[95]. Furthermore, estradiol treatment was also shown to result in a dose dependent suppression of hepatic fibrosis in hepatic fibrosis models of male rats[96,97]. The mechanism of protective action of estrogens against progression of chronic liver disease has been recently reviewed by Shimizu and Ito[7].

Recently, using a mouse model of DEN-induced hepatic carcinogenesis, Naugler et al[83] described a molecular mechanism explaining the lower HCC susceptibility in females and the anti-inflammatory role of estrogen in preventing HCC development. The authors investigated the relationship between HCC development and gender dependent expression of interleukin-6 (IL-6). IL-6 is a proinflammatory cytokine that plays an important role in chronic hepatitis, the prerequisite for progression to cirrhosis and HCC. The serum IL-6 levels were higher in male mice than in female mice after administration of DEN, leading to a higher rate of liver cell proliferation in male mice. This effect can be further mediated by ERα, suggesting that ERβ plays a little role in modulating the expression of IL-6.

The role of AR has also been studied in animal models of chemical induced carcinogenesis[87,98-100]. In DEN treated Wistar rats, a 20-fold increase in hepatic AR concentration was reported in females, suggesting that increased hepatic AR concentration is correlated with accelerated tumor development in these animals, in which male rats showed a slower tumor development with no change in AR concentrations[98]. Subsequent studies in the same model revealed that removal of ovary increases AR levels in the liver of female rats but testosterone treatment does not further enhance AR levels[99]. On the other hand, normal adult males with intact testis or testosterone treatment maintain high levels of AR but in castrated rats estrogen treatment reduces AR expression[99]. Animal studies demonstrated that the expression of both AR and ER increases during preneoplastic stages and that progression towards cancer development can suppress ER and maintain AR expression levels[94,100].

Interestingly, anti-androgen treatment has been shown to reduce AR levels in liver as well as the size and number of tumors in male Spargue Dawley rat model of hepatocarcinogenesis[101]. It has been shown that inhibition of AR positive HCC with anti-androgen cypertone acetate in male mice involves cell cycle arrest and to some extent induction of apoptosis due to increased synthesis of transforming growth factor-β 1 (TGF-β1)[102]. In another model of chemical-induced liver carcinogenesis, inhibition of androgens using 5-alpha reductase inhibitors significantly suppressed HCC development in rats[103]. Recent studies in a xenograft model of hepatocarcinogenesis in nude mice suggested that AR expression remains elevated until development of tumor and starts declining as the size of tumor increases[104]. It is therefore proposed that androgen therapy may be ineffective after establishment of the tumor. Nevertheless, for better understanding and rationale design of hormone-based therapies, it is mandatory to study the role of ERs and AR in animal models mimicking the natural course of disease progression to HCC development as in humans. Currently available HBV and HCV transgenic mice depicting features close to human HCC pathogenesis, appear to be promising models for future in vivo studies.

Estrogen action and the role of ERs in carcinogenesis have been well documented in mammary carcinoma and the studies have revealed the involvement of estrogens in key cellular processes such as apoptosis, cell cycle, proliferation, oxidative stress and inflammation. The progress in understanding the role of estrogen in regulating various cellular events in liver carcinogenesis has been rather slow. However, the research conducted over recent years provides key insights in this direction.

The classical mode of estrogen action is the genomic mechanism in which ERs function as ligand-activated transcription factors[105]. Activated ERs translocate to the nuclei and regulate the expression of specific target genes. These transcriptional regulations are achieved through interaction with estrogen responsive element (ERE) sequences located in the promoter region of the target gene[106]. However, one third of the genes regulated by ERs in humans do not contain ERE-like sequences[107]. ERs can also regulate the transcription of such genes without binding to DNA through protein-protein interactions with other transcription factors, such as AP-1 and Sp-1 in the nuclei[108]. In addition, this transcriptional control at alternate response elements is also facilitated by non-genomic actions of estrogen. The non-genomic functions of estrogen are initiated by membrane-localized ERs and are associated with activation of various signaling pathways especially protein kinases[109]. The functions of many transcription factors are regulated through protein kinase-mediated phosphorylation including CREB, NF-κB and AP-1 and these transcription factors may thus be targets for non-genomic actions of estrogens. This possible convergence of genomic and non-genomic actions at multiple response elements provides an extremely fine degree of control for the regulation of transcription for ERs (Figure 1)[105,110]. In the following section, we discuss the findings of recent in vitro studies highlighting the significance of ER in mediating genomic and non-genomic actions of estrogen in liver cells to modulate the expression of a number of genes involved in cellular processes central to carcinogenesis.

Telomerase activation has been implicated in hepatocarcinogenesis and the expression of human telomerase reverse transcriptase (hTERT) that encodes for the catalytic subunit of the multicomponent enzyme telomerase hTERT is the prerequisite for telomerase activation[111,112]. Several studies indicated that estrogen regulates transactivation of the hTERT gene by direct interaction of activated ER with an imperfect ERE sequence in the hTERT promoter[113]. Estrogen treatment has been shown to up-regulate the expression of hTERT mRNA and protein in three normal human hepatic cell lines (hc-cells, hNheps and WRL-68) expressing ERα to varying degrees[90]. Furthermore, estrogen exposure prevents shortening of telomeres and decreases the number of cells undergoing senescence, indicating that estradiol acts as a positive modulator of the hTERT gene in the liver[90]. However, the mechanism of ER-mediated transactivation of hTERT in the liver is not well understood. In contrast, in HepG2 cells, estrogen modulation of telomerase activity has been found to be regulated post-transcriptionally via the IP3/PKC pathway[114,115]. IP3 production has been shown to be up-regulated by estrogens in HepG2 cells[114]. Furthermore, estradiol-induced IP3/PKC-alpha production is dependent on either ERα or ERβ expression in both HepG2 and Hela cells[114]. It is hypothesized that membrane ER-mediated IP3/PKC-alpha pathway represents an alternative signaling pathway utilized by cells when low ER levels are unable to activate classic ER-mediated genomic mechanisms as in HepG2 cells[114].

A similar regulatory mechanism has been observed in case of estrogen modulation of expression of cyclin D1 gene in hepatoma cells. Cyclin D1, important for progression of cells through G1 phase of cell cycle, is a well defined target for estrogen action in mammary carcinoma[116,117], although no detectable estrogen responsive element like sequence in the cyclin D1 gene promoter has been reported in these cells[118]. The cyclin D1 mechanism identified in mammary carcinoma cells involves direct interaction of ERα and Sp1 or ERα and Ap-1[119]. Interestingly, Marino et al[23] demonstrated that in HepG2 cells, estrogen-induced activation of cyclin D1 transcription can occur independently of the transcriptional activity of ER. They further showed that the effect of 17-beta estradiol on HepG2 cells is mediated by activation of the MAPK/ERK pathway by membrane-localized ER that increases the expression of cyclin D1 gene through activation of AP-1 transcription factor[23], suggesting that non-genomic signaling pathways play an the pivotal role in estrogen-mediated regulation of gene expression at multiple response elements.

Besides, modulating the molecules involved in cell cycle control and cell proliferation, estrogen has also been shown to regulate the expression of genes crucial for apoptosis of hepatocytes and dysregulation of apoptosis in hepatic cells is reported to be a significant factor in accelerating hepatocarcinogenesis or tumor progression in HCC[120]. The Bcl-2 family of proteins regulates one of the key steps in the conserved apoptotic pathway. Among the members of this family, Bcl-2 and Bcl-xL act as inhibitors of apoptosis where as Bax and Bak promote apoptosis[121,122]. Ethinyl estradiol is known to increase the levels of Bcl-2 protein in cultured female rat hepatocytes[123]. Estradiol and idoxifene, two selective estrogen receptor modulators, are known to induce the expression of Bcl-2 protein in male rat liver tissues[124]. Omoya et al[9] and Inoue et al[81] also demonstrated that estradiol is able to stimulate the expression of Bcl-2 and Bcl-xL and to suppress Bad expression in oxidative stress-induced early apoptotic rat hepatocytes. Similar findings have been recently documented in response to estradiol treatment of Huh-7 human hepatoma cells describing a dose dependent increase in expression of Bcl-2 and Bcl-xL and a reduction in Bad levels[22]. No change was observed in expression of pro-apoptotic protein Bax. The regulation of Bcl-2 gene expression by estrogen in mammary carcinoma cells has been shown to be mediated indirectly through activation of Sp-1 transcription factor[125]. However, the precise mechanism of Bcl-2 transactivation in hepatocytes has not been clearly understood.

One of the most interesting mechanisms of transcriptional regulation at alternate response elements by estrogen is through inhibition of transcription factor NF-κB. Studies demonstrating a mutually antagonistic cross-talk between these families of transcription factors have been recently reviewed[126]. The ER has been shown to mediate opposition of NF-κB functions at various levels by inhibiting the activation of signaling pathways, preventing nuclear translocation, blocking DNA binding or inhibiting recruitment of co-activators for transcription[126]. Estrogen has been shown to bring about its anti-inflammatory and anti-oxidant effects on liver cells by suppressing the NF-κB activity as evident from the following studies. It was reported that 17 beta-estradiol-bound ERα interferes with cytokine-induced activation of a NF-κB reporter in HepG2 cells, suggesting that estrogen exerts its anti-inflammatory and protective effects on human liver cells[127]. Moreover, in an in vivo model, estrogen treatment has been shown to block the induction of hepatic expression of inflammatory vascular cell adhesion molecule-1 (VCAM-1), tumor necrosis factor-α (TNF-α), and regulate normal T-cell expression and secretion upon activation[128]. In a mouse model of DEN-induced HCC, ERα was suggested to suppress IL-6 production, a pro-inflammatory molecule, through the involvement of the NF-κB pathway[83]. Estrogen has also been reported to suppress oxidative stress-induced reactive oxygen species (ROS) generation, lipid peroxidation, activation of AP-1 and NF-κB as well as loss of Cu-Zn SOD activity in cultured rat hepatocytes[2,9].

In addition to genomic and non-genomic actions of estrogen mediated by nuclear and membrane ER, mitochondria have also recently been identified as important targets of estrogen and ERs[110]. Early binding studies on sub-cellular fractions indicated that ER is present in rat liver mitochondria[129]. Both ERα and ERβ have been reported to be present in the mitochondria of human HepG2 cells[130-132]. The mitochondrial genome has been shown to contain sequences that have partial homology to the estrogen responsive elements[132-134]. Both ERα and ERβ bind to mitochondrial DNA and the binding can be increased by estradiol using mobility shift assays and surface plasmon resonance[135]. These results suggest that estradiol is directly involved in the regulation of mitochondrial DNA transcription (Figure 1). Regulation of apoptosis and oxidative metabolism by estrogens in mitochondria may be important in the normal liver and in the development of HCC.

Ethinyl estradiol treatment has been shown to elevate the expression levels of mitochondrial DNA-encoded cytochrome C oxidase subunit III (CO III) and ATP synthase 6 in vivo as well as in HepG2 cells[136]. This increased expression of mitochondrial transcripts is accompanied by increased mitochondrial superoxide production and respiratory chain activity that require cytochrome P450-mediated biotransformation of ethinyl estradiol and 17-beta estradiol to catechol metabolites[136,137]. In addition to CO III, the levels of CO I and CO II encoded by mitochondrial DNA have also been found to be elevated in ethinyl estradiol treated female rat hepatocytes. This effect is accompanied by increased mitochondrial superoxide production, high ATP levels and increased Bcl2 production, and is suggested to play a role in ethinyl estradiol-mediated inhibition of apoptosis[123]. In contrast, 17-beta estradiol and 17-beta estradiol like compounds, diethylstilbestrol (DES), tamoxifen and genistein, have been found to induce apoptotic effects in human hepatoma Hep3B cell line[138]. These compounds cause the leaking of cytochrome C from mitochondria and activation of caspase-3 in an ER dependent manner. In another study, the two isoforms, ERα and ERβ, showed their opposing actions on apoptosis in a poorly differentiated HCC cell line HA22T[139]. Over-expressed ERβ but not ERα induces the expression of caspase-8 and TNF-α in HA22T cells in response to estradiol treatment, indicating that the death receptor-mediated apoptotic pathway is activated[139].

Differential roles of ERα and ERβ have also been observed in non-genomic actions of estrogen in the liver[21,23]. There is indirect evidence that membrane ER may exist in human liver as the binding of gold tagged estrogen-BSA conjugate on the surface of clathrin-coated pits in HepG2 cells has been demonstrated by electron microscopic visualization[140]. The non-genomic mechanism of action of sex steroids on the plasma membrane involves the activation of protein kinase cascades (Figure 1). Two major cascades, protein kinase C, and mitogen-activated protein (MAP) kinase are active and important in carcinogenic liver cells. Protein kinase C cascade and its second messenger IP3 are important in cell proliferation and have been discussed in this review in context of transcriptional regulation of hTERT expression by estrogen. The mitogen-activated protein (MAP) kinase cascade is another pathway that is regulated by the action of sex steroids at the plasma membrane. This complex signaling cascade involves three major pathways: ERK, p38, and JNK[141]. In HepG2 cells, estradiol has been found to rapidly increase the phosphorylation of ERK[21,23]. Naringenin, an anti-estrogenic flavonone, induces the activation of p38 in ERα containing HepG2 cells or in ERβ containing human colon adenocarcinoma DLD-1 cells[142], suggesting that naringenin has an antiestrogenic effect only on the ERα expressing cells, whereas it mimicks the estradiol effects on ERβ expressing cells. The role of ERα and ERβ in the regulation of MAP kinase cascade has been further studied in cell lines expressing either ERα or ERβ[21]. It was found that estrogen-bound ERα can rapidly activate the ERK and AKT signal transduction pathways leading to cell cycle progression and inhibition of apoptosis, whereas estrogen-complexed ERβ can induce rapid phosphorylation of p38 leading the cells to the apoptotic cycle and cell death. These studies further support the functional antagonism between ERα and ERβ with respect to estrogen-induced cell proliferation and emphasize the need to study the independent and interactive role of both isoforms in hepatocarcinogenesis.

In comparison with ER, there is limited information about genomic and non-genomic functions of AR in the liver. Like ER, AR has also been shown to regulate gene transcription by binding to androgen responsive sequences (ARE)[143,144]. Yoon et al[145] demonstrated that androgen can directly regulate the expression of transformation growth factor-beta 1 (TGF-β1) through binding of AR to ARE in TGF-β1 promoter, suggesting that such activation might regulate the progression of HCC in both human and animal models[145]. Furthermore, AR has been shown to interact with a newly identified transcription factor, paternally expressed gene 10 (PEG 10) in hepatoma cell line[146]. PEG 10 has growth promoting properties and is implicated in hepatocarcinogenesis[147,148]. Dihydrotestosterone (DHT) promotes hepatoma formation in nude mice through PEG 10 activation. In addition, DHT treatment is shown to up-regulate hTERT expression in hepatoma cell lines in a PEG-10 dependent manner[146]. These studies indicate that PEG-10-mediated transactivation of target genes by AR has an essential role in hepatocarcinogenesis.

To the best of our knowledge, AR has not been detected in the liver mitochondria. The information about the membrane localization of AR in human liver cells is also lacking. However, AR has been reported to occur in the plasma membranes of male rat liver[149]. Androgens are also involved in the regulation of the MAP kinase signaling pathways as orchiectomy of H-ras 12V transgenic mice decreases phospho-MEK and phospho-ERK in liver tissues. In addition, orchiectomy reduces hepatotumorigenesis in male mice while ovariectomy increases phospho-MEK and phospho-ERK in liver tissue from female mice, but ovariectomy does not affect the incidence of tumorigenesis[150]. Detailed investigations are urgently needed to confirm the existence of non-genomic signaling actions of androgens in liver and the role of AR in mediating these functions.

Han et al[24] recently reported that HBV protein (HBx) interacts with ERα. HBx is a multifunctional protein involved in neoplastic transformation in cultured cells and can induce HCC in transgenic mice. HBx associates with both ERα and delta 5 deletion variant of ERα (vERα) and inhibits ERα transcriptional activity by recruiting histone deacetylase enzyme, HDAC-1[24]. HDAC-1 belongs to the family of enzymes involved in deacetylation of hyperacetylated histone tails, leading to compaction of chromatin and transcritptional repression[153]. Both HBx and vERα have additive effects on suppression of ERα transactivation[24].

ERα has also been shown to interact with nonstructural (NS) 5B protein of HCV[25]. NS5B is a RNA-dependent RNA polymerase, which plays a central role in viral genome replication[154]. HCV replication takes place in a replication complex consisting of viral RNA and non structural proteins including NS5B[155]. The replication complex forms on the surface of intracellular membranes including endoplasmic reticulum membrane and is associated with lipid rafts rich in caveolin 2 (CAV 2) on these membranes[156-158]. Using chemical biology approach, Watashi et al[25] demonstrated that ERα facilitates the interaction of NS5B with CAV 2 in lipid rafts and hence promotes the participation of NS5B in HCV replication complex. However, they did not find that ERβ affects HCV replication in the same study[25]. An important observation of the study is that tamoxifen inhibits ERα actions and suppresses HCV genome replication, further supporting the potential for anti-ER drugs in developing new anti-HCV strategies.

Like ER, HBx protein has also been shown to interact with AR. HBx functions as a positive transcriptional co-regulator to increase AR-mediated transcriptional activity. This transcription enhancement is increased in the presence of androgen in a concentration-responsive manner. However, HBx does not physically associate with ligand-bound AR in the nuclei, suggesting that HBx augments AR activity by increasing the phosphorylation of AR through HBx-mediated activation of the c-Src kinase signaling pathway[159]. In contrast, Zheng et al[160] demonstrated that HBx can physically bind to AR in the liver and alter the subcellular localization of AR both in the presence and absence of dihydrotestosterone (DHT). Further studies indicated that HBx can enhance the gene transactivation activity of AR by enhancing its DNA binding activity in a DHT-dependent manner.