Potential conflict of interest: Nothing to report.
The Glypican 3 (Gpc3) gene is expressed abundantly in the fetal liver, is inactive in the normal adult liver, and is frequently reactivated in hepatocellular carcinoma (HCC). This reactivation in HCC has led to considerable interest in Gpc3 as a diagnostic tumor marker and its possible role in tumorigenesis. Despite this interest, the basis for Gpc3 regulation is poorly understood. On the basis of the similarities between Gpc3 and alpha-fetoprotein expression in the liver, we reasoned that common factors might regulate these 2 genes. Here we identify zinc fingers and homeoboxes 2 (Zhx2) as a regulator of Gpc3. Mouse strain–specific differences in adult liver Gpc3 messenger RNA levels and transgenic mouse studies indicate that Zhx2 represses Gpc3 expression in the adult liver. We also demonstrate that Gpc3 is activated in the regenerating liver following a carbon tetrachloride treatment and that the level of Gpc3 induction is controlled by alpha-fetoprotein regulator 2 (Afr2). Conclusion: We show that Zhx2 acts as a repressor of Gpc3 in the adult liver, and this raises the interesting possibility that Zhx2 might also be involved in Gpc3 reactivation in HCC. We also show that Gpc3 is activated in the regenerating liver in an Afr2-dependent manner. Zhx2 and Afr2 represent the first known regulators of Gpc3. (HEPATOLOGY 2007.)
Glypicans are a small family of glycosyl phosphatidylinositol–anchored heparin sulfate proteoglycans that are found in species ranging from Drosophila to vertebrates.1 This family of proteins is associated with cell growth, development, and responses to various growth factors.2, 3 Six glypicans exist in vertebrates, several of which are globally expressed, whereas others exhibit more restricted developmental and tissue-specific expression.1, 3, 4 Of the different glypicans, the X-linked Gpc3 has been the focus of considerable attention because of its association with certain diseases. Loss of functional GPC3 is associated with Simpson-Golabi-Behmel syndrome in humans, which is characterized by general overgrowth with additional clinical manifestations.5 Changes in Gpc3 expression are also associated with certain tumors. For example, reduced Gpc3 expression is often found in breast and ovarian tumors and mesotheliomas.6–8 In contrast, Gpc3 expression is frequently increased in hepatocellular carcinoma (HCC).9 In this respect, Gpc3 behaves as an oncofetal gene, in that it is expressed in the fetal liver, normally silent in the adult liver, and reactivated in cancer. In fact, Gpc3 may be reactivated in HCC as frequently as alpha-fetoprotein (AFP), which has been used extensively as a marker of this cancer, and may be a more reliable early diagnostic marker than AFP.10, 11
Despite considerable interest in Gpc3 activation in HCC, factors involved in regulating Gpc3 in the liver have not yet been identified. Because the Gpc3 gene is expressed abundantly in the fetal liver and HCC and is silent in the normal adult liver, it seems reasonable that it may be regulated similarly to other genes with the same pattern of expression, such as AFP and H19. AFP encodes a major serum transport protein in the developing mammalian fetus and is expressed abundantly in the fetal liver and extraembryonic yolk sac.12–14H19 encodes an untranslated messenger RNA (mRNA) of unknown function and is also highly expressed in the yolk sac and fetal liver and in additional endoderm and mesoderm tissues.15 Both genes are repressed in the liver after birth. However, this silencing is reversible as both genes can be activated in the regenerating liver and in HCC.12, 15, 16
Insight into postnatal AFP and H19 repression came from mouse studies that revealed incomplete AFP repression in the livers of BALB/cJ mice.13, 17 This is in contrast to other mouse strains in which AFP is fully silenced in the liver at birth. Persistent AFP expression in the BALB/cJ adult liver, in which the AFP mRNA levels are about 10–20 times higher than those of other strains, is a recessive trait due to an unlinked gene called alpha-fetoprotein regulator 1 (Afr1). The BALB/cJ phenotype is consistent with Afr1 functioning as a repressor of AFP expression; a loss of this repressor in BALB/cJ mice leads to incomplete silencing of its target, AFP. The mouse H19 gene was originally cloned on the basis of a molecular genetics screen for Afr1 targets.18 We used positional cloning to identify Afr1 as the mouse zinc fingers and homeoboxes 2 (Zhx2) gene.19 Zhx2 is a member of a small family of vertebrate transcription factors that includes Zhx1 and Zhx3 all of which are predicted to contain two zinc fingers and four homeodomains and appear to function as transcriptional repressors.20 Reduced Zhx2 expression in BALB/cJ mice, due to the insertion of an endogenous retroviral element into the BALB/cJ Zhx2 allele (which we call the Zhx2Afr1 allele, in contrast to the wild-type Zhx2+ allele), leads to incomplete repression of the target genes AFP and H19 in the adult liver of this mouse strain.19 More recently, it has been shown that the human ZHX2 gene is often silenced in HCC; this silencing correlates with ZHX2 promoter hypermethylation.21 Taken together, these data indicate that reduced Zhx2 levels result in persistent AFP/H19 expression in the adult BALB/cJ liver and might be involved in AFP/H19 reactivation in liver tumors.
On the basis of similarities between AFP, H19, and Gpc3 expression during liver development and in liver cancer, we examined whether Gpc3 is also a target of Zhx2. Our data indicate that Gpc3 expression persists in the adult liver of BALB/cJ mice and is repressed in this strain in the presence of a Zhx2 transgene, indicating that Zhx2 controls postnatal Gpc3 mRNA levels. We also show that Gpc3 is reactivated in the regenerating liver and that this reactivation is controlled by another regulator called alpha-fetoprotein regulator 2 (Afr2). This identifies the first two factors involved in Gpc3 regulation. Finally, we show that Gpc3 is expressed in embryonal carcinoma F9 cells that are differentiated in vitro. However, Gpc3 activation in these cells is different from what is seen for AFP and H19.
AFP, alpha-fetoprotein; Afr1, alpha-fetoprotein regulator 1; Afr2, alpha-fetoprotein regulator 2; Alb, albumin; CCl4, carbon tetrachloride; cDNA, complementary DNA; e18, embryonic day 18; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; Gpc3, glypican 3; H, heart; HCC, hepatocellular carcinoma; K, kidney; L, liver; Lu, lung; M, DNA marker lane; MO, mineral oil; mRNA, messenger RNA; p1, postnatal day 1; p7, postnatal day 7; p28, postnatal day 28; PCR, polymerase chain reaction; PE, parietal endoderm; RA, retinoic acid; RT-PCR, reverse-transcription polymerase chain reaction; Sp, spleen; TTR, transthyretin; VE, visceral endoderm; Zhx2, zinc fingers and homeoboxes 2.
Materials and Methods
Mouse embryonal carcinoma F9 cells were passaged and differentiated as described.22 Differentiated cells were harvested at designated times, washed, pelleted, and frozen at −80°C.
All mouse procedures were approved by the University of Kentucky Institutional Animal Care and Use Committee following guidelines established by the National Institutes of Health. C3H/HeJ, BALB/cJ, and C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The TTR-Zhx2 transgenic mice (where TTR represents transthyretin) were generated by the University of Kentucky Transgenic Mouse Facility and described previously.19 The animals were killed by CO2 inhalation followed by cervical dislocation. Tissues were quickly removed, snap-frozen in liquid nitrogen, and stored at −80°C until RNA was prepared. A portion of the tissue (∼100 mg) was used to prepare RNA. To obtain fetal tissues, mated females were identified by the presence of a vaginal plug and killed 18 days later (embryonic day 18; e18). Livers were removed from the fetuses, snap-frozen in liquid N2, and stored at −80°C until RNA was prepared. For the analysis of Zhx2 regulation of target genes, RNA was prepared from tissues that were removed from adult BALB/cJ mice, C3H/HeJ mice, or mice in which the TTR-Zhx2 transgene was crossed into BALB/cJ mice. For liver regeneration and Afr2 analysis, C3H/HeJ (Afr2a) and C57BL/6 (Afr2b) mice (2–3 months old) were given intraperitoneal injections of 50 μL of mineral oil (MO) containing 10% (vol/vol) carbon tetrachloride (CCl4); control animals were given intraperitoneal injections of 50 μL of MO alone.23 Three days later, the animals were killed, and their livers were removed for RNA preparation and analysis.
RNA was prepared from frozen tissues and pelleted F9 cells with Trizol (Invitrogen) according to the manufacturer's instructions. One microgram of RNA was reverse-transcribed to complementary DNA (cDNA) with Omniscript reverse transcriptase (Qiagen). Standard PCR was performed with Taq polymerase (Qiagen) or Thermo-Start PCR Master Mix (ABgene). PCR products were resolved on 2% agarose gels and visualized with ethidium bromide staining, and digital images were quantitated with ImageQuant 5.0 (Molecular Dynamics). Real-Time PCR reactions were prepared with iQ SYBR Green Supermix (Bio-Rad) and were amplified in a Bio-Rad MyiQ single-color real-time PCR detection system. Oligonucleotides were obtained from Integrated DNA Technologies (Coralville, IA); sequence information and PCR conditions are available upon request. In all cases, primers were from different exons so that amplicons from cDNA and contaminating genomic DNA would be of different lengths.
Gpc3 Expression is Silenced in the Perinatal Liver.
The mouse AFP and H19 genes are expressed abundantly in the fetal liver, repressed at birth, and reactivated in the adult liver during regeneration and in liver cancer. The incomplete repression of these two genes in the adult liver of BALB/cJ mice led us to identify Zhx2 as a regulator of postnatal AFP and H19 mRNA levels. We predicted that other Zhx2 targets would show patterns of expression that are similar to those of AFP and H19, that is, repressed at birth and reactivated in liver cancer. One gene that shows this pattern is Gpc3. To test whether Gpc3 is a target of Zhx2, we examined first whether Gpc3 is silenced after birth in wild-type (Zhx2+) mice (Fig. 1). RT-PCR was performed with RNA samples from e18, p1, p7, and p28. We found that the Gpc3 mRNA levels declined roughly 40-fold between p1 and p28, similarly to those of AFP (Fig. 1). Although the extent of AFP repression was more dramatic than that of Gpc3, the two genes showed similar temporal patterns of repression in the liver after birth. Albumin mRNA levels increased roughly 2-fold during this perinatal period, and this was consistent with previous studies showing continued albumin expression in the fetal and adult livers.
Adult Liver Gpc3 Levels Are Controlled by Zhx2.
Having demonstrated postnatal Gpc3 repression, we then determined whether Gpc3 was a target of Zhx2 control. To test this, we compared Gpc3 mRNA levels in the adult livers of BALB/cJ and C3H/HeJ mice. The hypomorphic mutation in the BALB/cJ Zhx2Afr1 allele leads to elevated hepatic AFP and H19 mRNA levels in comparison with the levels of these two genes in the presence of the wild-type Zhx2+ allele found in C3H/HeJ mice. RNA was prepared from adult livers and analyzed by RT-PCR (Fig. 2A). As expected, the Zhx2 mRNA levels were substantially higher in C3H/HeJ livers than in BALB/cJ livers. In contrast, the AFP levels were high in adult BALB/cJ livers, and this is consistent with the notion that Zhx2 acts to repress AFP in the postnatal liver. The Gpc3 levels were also elevated in the livers of BALB/cJ mice in comparison with those of C3H/HeJ mice, similarly to the levels of AFP. The difference in the steady-state Gpc3 mRNA levels between C3H/HeJ and BALB/cJ livers indicates that this gene is also a target of Zhx2.
We previously showed that the loss of Zhx2 in the livers of BALB/cJ mice could be complemented, as judged by AFP and H19 mRNA levels, by the overexpression of a Zhx2 transgene.19 In these mice, a Zhx2 cDNA was driven by a TTR enhancer/promoter expression cassette (TTR-Zhx2); this TTR cassette results in high liver-specific expression of linked transgenes.19, 24 As further confirmation that Zhx2 regulates Gpc3, we tested whether the Gpc3 levels were responsive to the TTR-Zhx2 transgene (Fig. 2B). To accomplish this, TTR-Zhx2 transgenic (C3H X BL/6) mice were crossed to BALB/cJ; the resulting TTR-Zhx2–positive F1 mice were backcrossed to BALB/cJ. The resulting F2 offspring were genotyped for the endogenous Zhx2 allele (Zhx2Afr1/+ or Zhx2Afr1/Afr1) and the TTR-Zhx2 transgene (TG+). In agreement with the previous data (Fig. 2A), the Gpc3 mRNA levels were lower in mice that had one wild-type Zhx2 allele (Zhx2Afr1/+, lane 3, Fig. 2B) than in those with two mutant Zhx2 alleles (Zhx2Afr1/Afr1, lane 1). Furthermore, the Gpc3 levels were low in TG+Zhx2Afr1/Afr1 mice (lane 2), demonstrating that transgene-derived Zhx2 could lead to reduced Gpc3 mRNA levels. In fact, the Gpc3 levels were lower in the transgenic mouse than in the wild-type mouse (compare lanes 2 and 3). This stronger repression of the target gene Gpc3 may be due to the higher levels of Zhx2 expression in transgenic mice in comparison with wild-type mice.
Zhx2 Repression of Gpc3 in BALB/cJ Mice is Liver-Specific.
Previous studies have shown that persistent expression of AFP and H19 in adult BALB/cJ mice is observed only in the liver. These 2 genes are expressed in other adult tissues (that is, AFP in the gut and H19 in the gut and muscle), but strain-specific differences in AFP and H19 mRNA levels are not seen in these other tissues. This is somewhat surprising because Zhx2 is ubiquitously expressed in adult mice. Because Gpc3 is expressed at low levels in other adult organs, we tested whether Gpc3 mRNA levels might show strain-specific differences in tissues other than the liver (Fig. 3). RT-PCR revealed that the Gpc3 levels were ∼4-fold higher in the livers of adult BALB/cJ mice when compared with those of adult C3H/HeJ livers (Fig. 3B). However, the Gpc3 levels were essentially the same in the hearts, lungs, spleens, and kidneys of these two mouse strains (Fig. 3B). Thus, the lack of complete Gpc3 repression in adult BALB/cJ mice is also restricted to the adult liver, as it is for AFP and H19.
Gpc3 is Regulated by Afr2.
Although AFP and H19 are silenced in the liver after birth and remain repressed in the normal adult liver, this silencing is reversible because both genes are transiently reactivated during liver regeneration. Strain-specific differences in the degree of AFP and H19 reactivation identified a second postnatal regulator of these genes called Afr2.13, 25 Mice with the Afr2a allele, found in most strains of mice, including C3H/HeJ, exhibit roughly 10-fold higher AFP and H19 mRNA levels during liver regeneration than mice containing the rare Afr2b allele (C57BL/6). These two alleles are codominant because intermediate AFP and H19 mRNA levels are seen in the regenerating livers of Afr2a/b heterozygous mice. We tested whether Gpc3 would also be activated during liver regeneration and regulated by Afr2, similarly to AFP and H19. Liver regeneration was initiated in C3H/HeJ (Afr2a) and C57BL/6 (Afr2b) mice by a single intraperitoneal injection of the hepatotoxin CCl4 in mineral oil (control mice received an injection of mineral oil alone). The livers were removed after 72 hours, a time in which AFP and H19 mRNA levels are highest during the regenerative period. Liver RNA was prepared, and the Gpc3 levels were analyzed with RT-PCR (Fig. 4). Both the AFP and β-actin mRNA levels were analyzed as controls. These data revealed that AFP was highly induced by CCl4 in C3H/HeJ mice (Fig. 4, compare lanes 8–10 with lanes 6 and 7) and that this induction was dramatically reduced in C57BL/6 mice (Fig. 4, compare lanes 3–5 with lanes 1 and 2). Gpc3 was also activated during regeneration, and like AFP, the CCl4 induction was greater in C3H/HeJ mice than in C57BL/6J mice. Thus, Gpc3 is also a target of Afr2-mediated regulation in the regenerating adult liver.
Gpc3 is Activated During F9 Cell Differentiation.
Our data indicate that Gpc3 is regulated similarly to AFP and H19 in the liver. AFP and H19 are also expressed at high levels in the yolk sac visceral endoderm (VE) but are not expressed in the parietal endoderm (PE). F9 embryonal carcinoma cells provide a tissue culture system to study this regulation. F9 cells (similar to the primitive endoderm) can be differentiated in vitro along a PE lineage or a VE lineage. Both AFP and H19 are activated when F9 cells are differentiated into VE, but they remain silent when these cells are differentiated into PE.26, 27 We therefore examined Gpc3 expression in differentiated F9 cells. F9 cells were treated with retinoic acid (RA), grown in a suspension (VE pathway) or grown as adherent monolayers in the presence of RA and dibutyryl cyclic adenosine monophosphate (PE pathway), and harvested at designated times. RNA was prepared and analyzed with RT-PCR (Fig. 5). In agreement with previous studies, AFP was activated in the VE pathway, with AFP mRNA levels first detected 5 days after the treatment, but it was not present in undifferentiated F9 cells and not activated in the PE pathway. In contrast, Gpc3 was expressed in undifferentiated F9 cells. After a transient decline at day 1 in both pathways, the Gpc3 levels increased in both pathways. Gpc3 is expressed in e18 mouse yolk sac (data not shown), although we do not know whether it is expressed in both parietal and visceral yolk sac tissues. Nonetheless, the pattern of Gpc3 expression in differentiating F9 cells is different than that of AFP.
There has been considerable interest in Gpc3 because it is commonly reactivated in HCC.9Gpc3 activation may be as frequent as that seen for AFP, which has been used historically as a marker of HCC.10, 11 Moreover, Gpc3 is reactivated more often than AFP in small dysplastic liver nodules and therefore may be a more valuable marker for early diagnosis.10, 11 Despite this clinical interest in Gpc3, its regulation in the liver and the basis for its reactivation in the adult liver are not well understood. The data presented here identify Zhx2 as the first factor that controls Gpc3 mRNA levels. This was shown by the persistent Gpc3 expression in the adult BALB/cJ liver and by the ability of a Zhx2 transgene to repress Gpc3 in a BALB/cJ background. It is intriguing that the three known targets of Zhx2 repression—AFP, H19, and Gpc3—are commonly reactivated in HCC. Although our data do not test whether Zhx2 is involved in Gpc3 reactivation in HCC, it has been shown that silencing of the human ZHX2 gene by DNA methylation is a common occurrence in HCC.21 It is possible that ZHX2 silencing in HCC would allow targets of this repressor to become re-expressed in tumors. A more recent study suggests that ZHX2 protein levels may in fact be increased in HCC,28 indicating that additional studies will be needed to resolve these conflicting data.
We also show that Gpc3 is reactivated in the regenerating liver after CCI4 intoxication and that this increase is governed by the regulator Afr2. This result again demonstrates a similarity between Gpc3 and AFP/H19 regulation. Although the product of Afr2 has not yet been identified,25, 29 this is a second factor that regulates Gpc3 expression. Gpc3 activation has also been shown in both the D-galactosamine and 2-acetylaminofluorene/partial hepatectomy models of liver regeneration.30 In both cases, Gpc3 and AFP activation has been observed specifically in hepatic progenitor cells. Because there is interest in using Gpc3 as a diagnostic marker for HCC, it will be important to evaluate whether Gpc3 levels increase during liver injury in humans.
Although Gpc3, AFP, and H19 show similar patterns of expression in the liver, we also observed differences in their expression in other tissues. Gpc3 was expressed in undifferentiated F9 cells and showed a transient repression and re-expression in F9 cells differentiated along both the VE and PE pathways. This is in contrast to AFP and H19, which are inactive in F9 cells and become active only when cells are differentiated along the VE pathway.15, 22, 31 Because F9 cells provide an in vitro model for yolk sac gene regulation,26, 27 these data suggest that Gpc3 is regulated differently than AFP and H19 in extraembryonic tissue. Interestingly, Gpc3, like AFP, is frequently activated in yolk sac tumors.12, 32, 33 We have also found that Gpc3 is expressed at moderate levels in the adult mouse heart, lungs, and kidneys and at low levels in the liver and spleen. This pattern of Gpc3 expression in adult mouse tissues is similar to what was seen in an analysis of human tissues.9 In general, Gpc3 expression is more widespread than that of AFP, which is essentially silent in adult tissues (with the exception of very low levels in the liver and gut), and H19, which continues to be expressed at moderate levels in adult muscle.15, 18
The presence of four predicted homeodomains in Zhx2 suggests that it functions as a DNA-binding protein, although the role of these domains has not been investigated and consensus DNA sequences bound by any Zhx proteins have not been identified.20, 34 Previous data from our laboratory showed that the 250–base pair AFP promoter was sufficient to confer Zhx2 control to a linked transgene.35 This indicates that the cis-acting site(s) required for AFP regulation resides in this 250–base pair region, but we have not identified the specific motif required for Zhx2 control. Furthermore, we and others have data indicating that Zhx2 also acts at the posttranscriptional level to regulate steady-state AFP and H19 mRNA levels36 (L.A.M., B.T.S., and M.L.P., unpublished data, 2007). It will be of interest to determine whether similar posttranscriptional regulation of Gpc3 also occurs.
The difference in AFP and H19 mRNA levels between BALB/cJ and other mouse strains is seen only in the adult liver; the AFP and H19 mRNA levels in the gut and muscle are the same in these different mouse strains.13, 18, 23 The adult Gpc3 mRNA levels are also the same in the hearts, lungs, spleens, and kidneys of BALB/cJ and C3H/HeJ mice. Because Zhx2 is ubiquitously expressed, the liver specificity of the BALB/cJ phenotype is somewhat unexpected. It is still not clear what mechanisms account for the liver-specific Zhx2Afr1 phenotype. It may be that Zhx2 interacts with other liver-specific factors to repress target genes at birth, that other Zhx proteins can compensate for the reduction of Zhx2 in organs other than the liver, or that Zhx2 levels in BALB/cJ livers are below a threshold that is needed for the repression of target genes (Zhx2 levels in wild-type livers are low versus those in other organs but still above this threshold).20
In summary, we have identified Zhx2 as a known regulator of Gpc3 in the adult liver. Further studies will be needed to determine whether Zhx2 is involved in Gpc3 reactivation in HCC. We also have shown that Gpc3 is reactivated in the regenerating liver after a CCI4 treatment and that this induction is controlled by Afr2. Gpc3 is the third known target of Zhx2, and it is likely that additional targets exist. Microarray analysis, ChlP-CHIP experiments, and studies in Zhx2 knock-out mice should help to identify these additional targets and elucidate further the function of Zhx2 in the control of hepatic gene expression in the healthy adult liver and in liver disease.
We thank Michelle Glenn and Michael Green for their technical assistance and members of our laboratories for helpful discussions.