Oncogene (2006) 25, 4207–4216 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Expression of GCIP in transgenic mice decreases susceptibility to chemical hepatocarcinogenesis

WMa1,2,3,4, X Xia5, LJ Stafford1,2,3,CYu1,2,3, F Wang1,2,3, G LeSage5 and M Liu1,2,3

1Alkek Institute of Biosciences and Technology, Texas A&M University System Health Science Center, Houston, TX, USA; 2Center for Cancer Biology and Nutrition, Texas A&M University System Health Science Center, Houston, TX, USA; 3Department of Molecular and Cellular Medicine, Texas A&M University System Health Science Center, Houston, TX, USA; 4The Graduate School of Biomedical Sciences, The University of Texas-Houston Health Science Center, Houston, TX, USA and 5Department of Internal Medicine, The University of Texas-Houston Health Science Center, Houston, TX, USA

Transcription factors with helix–loop–helix (HLH) motif Introduction play critical roles in controlling the expression of involved in lineage commitment, cell fate determination, Liver cancer is the third most common causes for cancer proliferation, and tumorigenesis. To examine whether death in the world. In 2005, a total of 15 420 estimated the newly identified HLH GCIP/CCNDBP1 deaths by liver/intrahepatic bile duct cancer are expected modulates cell fate determination and plays a role in in the US, including 10 330 male cases and 5090 female hepatocyte growth, proliferation, and hepatocarcino- cases (Jemal et al., 2005). Although lots of research has genesis, we generated transgenic mice with human GCIP been performed to study the mechanism of hepatocarci- driven by a liver-specific albumin promoter. We nogenesis, few of them are related to the early stage of demonstrated that in GCIP transgenic mice, the overall liver tumor development. Helix–loop–helix (HLH) liver growth and regeneration occurred normally after are key regulators of cell growth and differ- liver injury induced by carbon tetrachloride (CCl4). In the entiation in embryonic and adult tissues. They have diethylnitrosamine (DEN)-induced mouse hepatocarcino- been demonstrated to play critical roles in regulation of genesis, we demonstrated that overexpression of GCIP in gene expression, cell cycle control, cell lineage commit- mouse liver suppressed DEN-induced hepatocarcinogen- ment and numerous developmental processes (Blackwell esis at an early stage of tumor development. The number et al., 1990; Kreider et al., 1992; Zebedee and Hara, of hepatic adenomas at 24 weeks was significantly lower 2001). From the yeast to humans, over 240 HLH or not detected in GCIP transgenic male mice compared proteins have been identified to date (Massari and to the control mice under the same treatment. Although Murre, 2000). Based upon the presence or absence of GCIP has little inhibition on the number of hepatic tumors DNA-binding domain and other additional functional at later stages (40 weeks), hepatocellular tumors in GCIP domains, HLH proteins can be classified into five major transgenic mice are smaller and well-differentiated classes, basic HLH (bHLH) proteins, basic HLH compared to the poorly differentiated tumors in wild- Per-AhR-Arnt-Sim (bHLH-PAS) proteins, basic HLH type mice. Furthermore, we demonstrate that GCIP func- leucine zipper (bHLH-LZ) proteins, dominant-negative tions as a transcriptional suppressor, regulates the HLH (dnHLH) proteins, and HLH leucine zipper expression of , and inhibits anchorage-indepen- (HLH-LZ) proteins. In multicellular organisms, bHLH dent cell growth and colony formation in HepG2 cells, proteins are required for many important developmental suggesting a significant role of GCIP in tumor initiation processes, including neurogenesis (Lee et al., 1995; Ma and development. et al., 1996; Sun et al., 2001), myogenesis (Nabeshima Oncogene (2006) 25, 4207–4216. doi:10.1038/sj.onc.1209450; et al., 1993; Molkentin et al., 1995; Spicer et al., 1996), published online 27February 2006 hematopoiesis (Kreider et al., 1992; Zhao and Aplan, 1999; Barndt et al., 2000), and pancreatic development Keywords: GCIP; CCNDBP1; DIP1/HHM; transgenic (Jensen et al., 2000; Pin et al., 2001). On the basis of mouse; hepatocarcinogenesis; diethylnitrosamine; tissue distribution and dimerization capabilities, bHLH HLH-Zipper proteins protein can be further divided into two subgroups. One subgroup also known as the E proteins and includes E12, E47, E2-5, E2-2, HEB, and Daughterless. These proteins are ubiquitously expressed in many tissues and capable of forming either homo- or heterodimers Correspondence: Dr M Liu, Alkek Institute of Biosciences and (Zhuang et al., 1992; Shirakata and Paterson, 1995). Technology, Texas A&M University System Health Science Center, The DNA-binding specificity of E proteins is limited to 2121 W Holcombe Blvd, Houston, TX 77030, USA. E-mail: [email protected] the E-box site, which have the consensus sequence Received 28 September 2005; revised 27December 2005; accepted 27 CANNTG (Cordle et al., 1991; Voronova and Lee, December 2005; published online 27February 2006 1994). The other subgroup bHLH proteins show a GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4208 tissue-restricted pattern of expression and regulate liver injury induced by carbon tetrachloride (CCl4). tissue-specific cell growth and differentiation. Examples In the mouse model of hepatocarcinogenesis induced by include the myogenic bHLH family of proteins (MyoD, DEN, we observed that overexpression of GCIP in Myogenin, Myf-5, MRF4), the cardiogenic family of transgenic mice suppressed the number of chemical- proteins (eHand and dHand), the neurogenic family of induced hepatocarcinogenesis at the early stage of tumor proteins (Neurogenin, NeuroD, Mash-1, Mash-2, Hes, development. At the late stage, hepatocellular tumors NSCL), and the haemopoietic family of bHLH proteins in GCIP transgenic mice are smaller and well- (SCL/TAL-1, Lyl-1, and ABF-1) (Murre et al., 1989; differentiated compared to the poorly differentiated Lassar et al., 1991; Hu et al., 1992; Kee et al., 2000). tumors in wild-type mice. We further demonstrated that With few exceptions, they are incapable of forming GCIP functions as a general transcriptional suppressor. homodimers and preferentially heterodimerize with the Overexpression of GCIP inhibited anchorage-indepen- E proteins. The heterodimers can bind both canonical dent cell growth and colony formation whereas down- (CANNTG) and noncanonical (CACNAG) E-box sites regulation of GCIP promotes colony formation in (Gaubatz et al., 1994). Dominant-negative HLH pro- HepG2 cells. These data suggest that GCIP may teins, including Id proteins (Id1, Id2, Id3, Id4) and emc, function as an important negative regulator in tumor are small proteins lack a basic region. Owing to a lack initiation and progression but does not alter the of the basic region, the Id proteins appear to function proliferation of normal hepatocytes following liver as dominant-negative regulators of bHLH proteins injury. These findings suggest GCIP maybe an attractive through the formation of ‘nonfunctional’ Id/bHLH target for prevention of hepatocellular carcinoma. heterodimers, thus quench the activity of the E proteins, inhibit cell differentiation and regulate tumorigenesis (Benezra et al., 1990; Sun et al., 1991; Peppelenbosch et al., 1995; Yokota and Mori, 2002; Perk et al., 2005). Results Mice deficient in the Id family of proteins have profound defects in neurogenesis, angiogenesis, vasculari- Generation of transgenic mice overexpressing GCIP under zation of tumor xenografts, impaired immune responses, the control of the albumin promoter and tumorigenesis (Lyden et al., 1999; Rivera et al., The full-length human GCIP was cloned from a human 2000; Rivera and Murre, 2001; Ruzinova and Benezra, bone marrow cDNA library (clontech) (Xia et al., 2000). 2003; Sikder et al., 2003). Human GCIP cDNA fragment (1.1 kb) containing the GCIP (CCNDBP1) is a HLH leucine zipper (HLH- entire coding region was inserted into the EcoRI site LZ) proteins without basic DNA-binding region, which of an albumin promoter-driven expression vector was identified and characterized in our lab and others to (Figure 1a) and a chicken insulator is inserted at interact with cyclinD1 (DIP1) and as a human homo- 30-end of the transgene to ensure the expression of logue of MAID protein (HHM) (Hwang et al., 1997; target gene. Two GCIP transgenic lines were developed Terai et al., 2000; Xia et al., 2000; Yao et al., 2000). and utilized for characterization. Transcription of GCIP gene is localized on 15q15, a region human GCIP was identified by Northern blot analysis frequently deleted or undergoes loss of heterozygosity of liver mRNA using a 32P-labeled human GCIP cDNA (LOH) in different tumors, including colorectum, probe. Expected transcripts of the human GCIP breast, lung, and bladder tumor (Natrajan et al., transgene were found only in GCIP transgenic mice 2003). Previous data demonstrate that GCIP is ex- but not in control wild-type mice (Figure 1b), whereas pressed specifically in foci of the 2-acetylaminofluorene/ transcript of the mouse GCIP were found in both partial hepatectomy (AAF/PH) model, which is used to transgenic mice lines and control wild-type mice. The monitor the differentiation of rat hepatic stem cells into expression of GCIP mRNA in transgenic mice was hepatocytes, suggesting GCIP may be involved in the approximately two to five folds higher than that of the hepatic stem cell development and differentiation endogenous murine GCIP gene. RT–PCR with primers processes (Terai et al., 2000). GCIP was also showed specific to human GCIP also showed the presence of to regulate the transcription of hepatocyte nuclear human GCIP mRNA in the liver of both Tg1 and Tg2 factor 4 (HNF4), which has E-box sequences in its GCIP transgenic mice, with higher levels of expression promoters, thus suggesting that GCIP might be involved in Tg1 GCIP transgenic animals, which is consistent in the liver growth and differentiation (Terai et al., 2000; with our results obtained from Northern blot analysis Takami et al., 2005). High-level expression of GCIP was (Figure 1b and c). Furthermore, we examined the found from the very early stages of hepatocarcino- expression of GCIP protein in transgenic mice using genesis, suggesting that GCIP might be involved in specific anti-GCIP antibody in Western blot analysis. As hepatocarcinogenesis (Takami et al., 2005). shown in Figure 1d, GCIP protein was at about two- to In this study, we have generated GCIP transgenic fourfold higher in transgenic mice than the endogenous mice in the liver and characterized the effects of GCIP in mouse GCIP protein (Figure 1d). liver regeneration after injury and in hepatocarcino- All GCIP transgenic mice appear to develop and genesis induced by diethylnitrosamine (DEN). Our data reproduce normally. The GCIP transgenic mice showed demonstrate that in GCIP transgenic mice, the no significant difference in liver weight relative to total overall liver growth, cell proliferation, and regeneration body weight in comparison to age-matched controls occurred normally compared to the wild-type mice after in first 12 months (data not shown). Examination of

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4209

Figure 1 Generation and characterization of the GCIP-transgenic mice. (a) Schematic representation of GCIP transgene construct. The 1.1 kb cDNA of human GCIP was ligated downstream of albumin promoter for overexpression in the hepatocytes. A chicken insulator is inserted at 30-end of the transgene to ensure the expression of target gene. The pronucleus of FVB mice fertilized eggs were injected with GCIP transgene construct and eggs were transferred to the oviduct of pseudopregnant female mice. (b) Northern blot analyses of RNAs from adult liver tissues of GCIP transgenic and wild-type (WT) mice were performed with the 32P-labeled human GCIP cDNA probe and mouse GCIP cDNA probe, respectively. GAPDH cDNA probe was used as a control to equal loading of total RNA. (c) RT–PCR with primers specific to the human GCIP and murine GCIP were preformed to show the presence of human GCIP mRNA in liver of both Tg1 and Tg2 of GCIP-transgenic mice, with higher levels of expression in Tg1- GCIP transgenic animals. RT–PCR with GAPDH-specific primer was used as control. (d) Western blot analysis of GCIP protein in the liver of transgenic mice and WT mice with anti-GCIP rabbit polyclonal antibody. Western blot analysis of b-actin was served as a loading control.

multiple livers from Tg1 and Tg2 lines of transgenic GCIP mice and wild-type mice revealed no significantly difference in liver structure and hepatocyte morphology.

Normal liver regeneration and hepatocyte proliferation in GCIP transgenic mice after acute CCl4 treatment Figure 2 Expression of GCIP in transgenic mice does not affect The liver has a remarkable capacity to regenerate after histology, liver injury, and cell proliferation in livers following injury. To determine the effect of GCIP overexpression intraperitoneal injection of CCl4. (a) Histological analysis of wild-type (WT) and GCIP-transgenic mice (Tg) after acute CCl4 treatment, on hepatocyte proliferation and regenerative capability, respectively. Liver tissues were fixed, paraffin-embedded, then stained we induced liver injury and regeneration on 7-week-old by hematoxylin and eosin (H&E) as described in Materials and male wild-type and GCIP transgenic mice by performing methods. (b) Serum enzyme levels (ALT) show similar liver damage a single intraperitoneal injection of the liver toxin CCl4, responses for WT and GCIP-transgenic mice. Plasma ALT activity was which is known to cause hepatic injury by forming measured using the GP-Transaminase kit as described. No significant difference between GCIP transgenic and WT mice at different time trichloromethyl free radical and altering the hepatocyte points was observed (Po0.01) after acute CCl4 treatment. (c)Kinetics membrane permeability (Recknagel et al., 1989). of DNA synthesis and cell proliferation in GCIP transgenic and WT Histologically, GCIP transgenic mice and wild-type mice. Cellular proliferation was measured by BrdU incorporation as mice showed similar extent of hepatocellular injury at the percentage of the number of BrdU positive hepatocytes in total hepatocytes by four or five visual fields per animal. All results are based each time point (from 38 to 72 h) following acute on analysis of 3–5 animals per time point and a minimum of 3000 cells CCl4 exposure (Figure 2a), demonstrating little per animal was counted. No significant difference was detected in or no difference in liver regeneration and heptocyte hepatocyte DNA synthesis between GCIP transgenic and WT mice at proliferation. different time points after acute CCl4 treatment.

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4210 Alanine transaminase (ALT), is used as a specific GCIP transgenic males versus 38.5% of wild-type males maker for hepatocyte injury (Wroblewski, 1959). To developed liver adenomas by 24 weeks (Figure 3a–d, determine the potential role of GCIP in hepatocyte Table 1). However, after40 weeks, 85.7% of GCIP injury and regeneration, we examined whether over- transgenic male mice developed liver tumors while 100% expression of GCIP affects ALT level in transgenic of the wild-type male mice developed liver tumor after GCIP mice and in wild-type control mice. As shown in DEN-treatment (Figure 3e–h, Table 1). The tumors in Figure 2b, in CCl4-induced liver injury, ALT levels GCIP transgenic mice at 40 weeks are smaller and well- increased and reached peak at 24 h in both wild-type and differentiated (Figure 3g and h) compared to those GCIP transgenic mice. There was no significant found in wild-type mice (Figure 3e and f). These data difference at ALT levels between wild-type and GCIP were consistent with the microfoci development at the transgenic mice up to 72 h post-CCl4 injection early stage in the wild-types of mice and suggested that (Figure 2b), indicating little effect of GCIP on the tumor development in the GCIP transgenic mice was injury responses of the liver induced by CCl4 injection. delayed. To further examine whether overexpression of GCIP in transgenic mice affect cell proliferation during injury responses, we examined BrdU incorporation in the GCIP suppresses the tumorigenesis in HepG2cells CCl -injured liver and evaluated hepatocyte DNA To examine the effect of GCIP in tumorigenesis, we 4 examined how GCIP regulated anchorage-independent synthesis and cell proliferation after CCl4 injection. As shown in Figure 2c, both GCIP transgenic and wild-type cell growth through soft agar assays. HepG2 cells were mice showed similar patterns for BrdU incorporation stably transfected with pCMV-Tag2B control vector, pCMV-Tag2B-GCIP, and pU6-GCIPsiRNA plasmid, with peak incorporation occurring at 48 h post-CCl4 injection, suggesting GCIP has little effect on DNA respectively. The transfected cells were analysed and compared in soft agar assays for tumorigenesis. As synthesis and cell proliferation after CCl4 injection in the transgenic liver. Together, these data support the shown in Figure 4, overexpression of GCIP inhibited idea that GCIP transgenic mice in the liver showed an HepG2 cell colony formation on soft agar (Figure 4a unaltered liver damage and hepatocyte proliferation and b), whereas downregulation of GCIP by specific after acute CCl treatment. siRNA significantly increased the colony formation 4 of HepG2 cells (Figure 4a and b), suggesting that GCIP can function as a potential suppressor gene GCIP overexpression suppressed diethylnitrosamine- for anchorage-independent cell growth and colony induced hepatocarcinogenesis in early stage of liver formation. tumor development To elucidate the role of GCIP in hepatocarcinogenesis, GCIP functions as a transcriptional repressor through DEN was administered to GCIP transgenic mice to TSA/NaB insensitive pathways induce liver tumorigenesis. As female mice are known to GCIP contains a putative HLH domain, an Asp/Glu- be resistant to hepatocarcinogenesis in experimental rich acidic domain, and a leucine-zip domain. The mouse models, including those employing chemical distinct domain structures suggest that GCIP could be a carcinogenesis, only male mice were used and analysed potential transcriptional regulator. To further determine in our experiments (Diwan and Meier, 1976; Moore the role of GCIP in general transcriptional regulation, et al., 1981; Vesselinovitch, 1987; Lee et al., 1998; we used L8G5-luc reporter, a luciferase reporter gene Nakatani et al., 2001). In this commonly used neonatal controlled by eight copies of the binding site for LexA DEN-induced hepatocarcinogenesis model, a single dose immediately adjacent to five copies of the binding site of 10 mg/kg DEN was intraperitoneally injected into for GAL4 (Hollenberg et al., 1995) (Figure 5a). LexA- 14-day-old wild-type and GCIP transgenic mice. The VP16 fusion coactivator was used as positive control to mice were killed 4, 6 or 10 months later. DEN-induced microfoci could be readily observed under light micro- scope in wild-type liver by 4 months. The most common foci were basophilic cell type, followed by clear cell type Table 1 Incidence of liver tumors in GCIP transgenic mice and wild-type mice treated with diethylnitrosamine (DEN) and the mixed type containing both types of cells. Hepatocyte dysplasia was apparent with increased Mouse incidence of liver tumor (weeks) nuclear/cytoplasm ratio, pleomorphism, vacuolization, Genotype 24 (weeks) 40 (weeks) and clear cell change. In general, while variations existed, there were no apparent differences in the WT 5/13 (38.5%) 8/8 (100%) histological changes between the wild-type and GCIP TG 1/13 (7.7%) 6/7 (85.7%) transgenic livers. However, there were significant The number of mice per group possessing at least one liver adenoma/ differences in the number of foci and the size of foci tumor, relative to the total number of mice in that group. The numbers between the GCIP transgenic mice and the wild-type in parentheses represent the percentage of liver tumor occurrence rate. mice. The number of hepatic adenomas at 24 weeks was Diethylnitrosamine was administered as a single intraperitoneal significantly lower or not detected in GCIP transgenic injection of 10 mg/kg body weight at 14 days of age. Incidence of male mice compared to the control mice under the same liver tumors in GCIP transgenic mice is significantly lower than that in wild-type mice at 24 weeks based on t-test (Po0.05). Abbreviations: treatment by 24 weeks (Figure 3a–d). Only 7.8% of WT, wild-type mice; TG, GCIP transgenic mice.

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4211

Figure 4 GCIP regulates tumorigenesis in HepG2 cells in colony formation assays. (a) Overexpression of GCIP suppresses ancho- rage-independent colony formation of HepG2 cells while down- regulation of GCIP by siRNA promotes colony formation. HepG2 cells were stably transfected with empty pCMV-Tag2B, pCMV- Tag2B-GCIP, pU6-GCIPsiRNA, plated in soft agar, and assayed for colony formation after 3 weeks. (b) The effect of GCIP on tumorigenesis was counted as the number of colonies formed on soft agar assays. Overexpression of GCIP inhibited HepG2 colony formation (column 2) while downregulation of GCIP by siRNA promoted colony formation (column 3).

of the class I and class II HDACs, which are sensitive to the inhibition by TSA and NaB. Figure 3 Comparison of liver tumors arising in GCIP-transgenic and wild-type (WT) mice treated with DEN. (a–b) Histology of WT male mouse livers at 24-weeks after DEN-treatment shows GCIP inhibits the expression of cyclin D1 adenomas in five out of 13 WT mice. (c–d) No obvious adenoma To understand the potential mechanism of GCIP was observed in GCIP transgenic mice (only one small adenoma was found in GCIP transgenic mice). (e–f) Tumors of WT male inhibition of cell proliferation and tumorigenesis, we mice at 40 weeks’ old, showing poorly differentiated tumors and examined how GCIP affects the expression of cyclin D1, tumor necrosis (e) found in eight out of eight WT mice. (g–h) Liver a key molecule in cell proliferation and tumorigenesis. adenomas found in GCIP transgenic mice at 40 weeks, showing First, we investigated how GCIP regulates the transcrip- smaller and well-differentiated tumors in six out of seven GCIP- transgenic mice. Tumors were labeled by the cycle lines. tional activation of cyclin D1 by transfecting HepG2 cells with the pCMV-Tag2B vector, pCMV-Tag2B- GCIP plasmid, or pU6-GCIPsiRNA plasmids, together activated reporter to high levels of transcription with the À1745 cyclin D1 promoter-luciferase reporter. (Figure 5a). By fusing GCIP to the Gal4 DNA-binding This cylcin D1 reporter is the orginal fragment of cyclin domain, the Gal4-GCIP fusion protein was able to D1 50 sequence cloned from the PRAD1 break point efficiently reduce the transactivation of L8G5-luc (Motokura and Arnold, 1993). Cell extract were assayed reporter activated by LexA-VP16. This Gal4-GCIP- for luciferase or b-galactosidase activities 48 h later. As induced inhibition was not affected when cells were shown in Figure 5b, overexpression of GCIP can incubated with either of the two HDAC inhibitors, markedly inhibit cyclin D1 transcriptional activity while trichostatin A (TSA) and sodium butyrate (NaB) decreasing GCIP expression by specific siRNA in the (Figure 5a). Both TSA and NaB are well-known HDAC cells can alleviate the inhibitory effect of cyclin inhibitors and were able to strongly promote transcrip- D1 transcription (Figure 5b). Based on the inhibitory tional activation at indicated concentration. In the effect of GCIP on the transcriptional activity of cyclin presence of TSA or NaB, cotransfection of Gal4-GCIP D1 promoter in the cell, we further examined whether still strongly inhibited VP16-induced transcriptional GCIP could inhibit the expression of cyclin D1 protein activation in the L8G5-luciferase assays, suggesting that in the cells. Cells were transfected with pCMV-Tag2B GCIP induced transcriptional inhibition is independent vector, pCMV-Tag2B-GCIP, or pU6-GCIPsiRNA,

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4212

Figure 5 GCIP functions as a transcription repressor and regulates cyclin D1 expression. (a) Regulation of transcriptional activities by the expression of Gal4-GCIP. Cos-7cells were transiently cotransfected with L8G5-luc reporter, LexA-VP16, Gal4-GCIP, or Gal4- vector control as indicated. Gal4-GCIP can efficiently suppress L8G5-luc transcription activated by LexA-VP16. TSA or NaB has little or no effect on the GCIP-induced transcriptional repression. (b) GCIP inhibits transcriptional activation of À1745 cyclin D1 promoter. HepG2 cells were plated and transiently transfected with pCMV-Tag2B vector, pCMV-Tag2B-GCIP plasmid, or pU6-GCIPsiRNA, together with the À1745 cyclin D1-luciferase reporter. Overexpression of GCIP significantly inhibited the transcriptional activity of cyclinD1 promoter while decreasing GCIP expression can upregulate cyclin D1 transcription. (c) Inhibition of cyclin D1 protein expression by GCIP. HepG2 cells were plated and transiently transfected with pCMV-Tag2B vector, pCMV-Tag2B-GCIP, and pU6-GCIPsiRNA plasmid, respectively. Cell extracts were assayed for the expression of GCIP, cyclin D1, and b-actin with specific antibodies using Western blot analysis 72 h later. Results showed that overexpression of GCIP markedly inhibited cyclinD1 expression while decreasing GCIP expression by siRNA significantly increased the expression of cyclin D1 protein in the cells. (d) RT–PCR with primers specific to the murine cyclin D1 were preformed to show the expression of cyclin D1 mRNA in liver of both 24 weeks and 40 weeks old GCIP transgenic mice and wild-type (WT) mice treated with DEN, with lower levels of cylin D1 expression in 24 weeks old GCIP transgenic mice. RT–PCR with GAPDH-specific primer was used as control.

respectively. Total proteins were extracted and sepa- showed that lower levels of cylin D1 expression was rated by SDS-PAGE. Western blot analysis was present in 24 weeks’ old GCIP transgenic mice performed to examine the expression levels of the GCIP, compared with wild-type mice (Figure 5d). However, cyclin D1, and actin. As shown in Figure 5c, GCIP there is little difference between the expression levels of protein was decreased in the cells transfected with pU6- cyclin D1 in 40 weeks old GCIP transgenic and wild- GCIPsiRNA, confirming that pU6-GCIPsiRNA could type mice (Figure 5d), suggesting that cyclin D1 plays a efficiently decrease GCIP expression level (Figure 5c). key role in the effect of GCIP in tumor suppression. Down-regulation of GCIP greatly increased the expres- sion level of cyclinD1 protein in the cells transfected with pU6-GCIPsiRNA while overexpression of GCIP Discussion decreased the expression of cyclin D1 in cells transfected with pCMV-Tag2B-GCIP (Figure 5c), suggesting that In the previous study, our group reported that GCIP GCIP is able to inhibit tumorigenesis by regulating the was expressed highly in terminal differentiated tissues, expression and function of cyclin D1 in the cell. including heart, muscle, peripheral blood leukocytes, To further examine the in vivo effect of GCIP on and brain. Furthermore, in cells transfected with GCIP, cyclin D1 expression, RT–PCR analysis was performed phosphorylation of retinoblastoma (Rb) protein by to check the expression of cyclin D1 in liver of both 24 cyclin D-dependent protein kinase was reduced and weeks and 40 weeks GCIP transgenic mice and wild- E2F1-mediated transcription activity was inhibited, type mice that were treated with DEN. The results suggesting GCIP could play a role in cell proliferation

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4213 and differentiation (Xia et al., 2000). Previous data from on the recruitment of hepatic oval cells. Hepatic oval Thorgeirsson’s group had shown that GCIP/HHM is cells are able to proliferate and differentiate into expressed specifically in the foci of the 2-acetylamino- hepatocytes. Oval cell proliferation is commonly seen fluorene/partial hepatectomy (AAF/PH) model and is in the preneoplastic stages of liver carcinogens and is likely to be involved in hepatic stem cell (hepatic oval believed to be the source of hepatocarcinogenesis after cell) proliferation and differentiation (Terai et al., 2000). the exposure to hepatocarcinogens (Knight et al., 2000; GCIP/HHM was also demonstrated to regulate the Roskams et al., 2003; Theise et al., 2003; Lemmer et al., transcriptional activation of hepatocyte nuclear factor 4 2004). GCIP is expressed specifically in the foci of the (HNF4) in reporter assays, suggesting a role of GCIP/ 2-acetylaminofluorene/partial hepatectomy (AAF/PH) HHM in liver growth and differentiation (Terai et al., model, indicating that it is more likely to be involved in 2000). Mouse GCIP, which is named as Maid (Maternal hepatic oval cell proliferation and differentiation but not Id-like molecule), was isolated from a subtraction hepatocyte proliferation (Terai et al., 2000). The cDNA library enriched for maternal transcripts that involvement of GCIP in hepatic oval cell proliferation are still present in the mouse 2 cell stage embryo (Hwang and differentiation may partly explain why there exists et al., 1997). However, Maid is just one of the shorter no difference in liver injury and hepatocyte prolifera- splicing isoform of the mouse GCIP, the full-length tion between GCIP transgenic and wild-type mice after mouse GCIP isoform has extra 47amino acids at its CCl4-treatment, while hepatic carcinogenesis after amino terminus and shares an overall amino acid DEN-treatment was attenuated in the GCIP transgenic sequence identity of 79.2% homology with GCIP. mice compared to wild-type mice. Interesting, human GCIP also has multiple different It is known that early hepatocellular carcinoma splicing isoform. Totally 14 splice patterns (SP) from the (HCC) is usually a well-differentiated carcinoma and Alternative Splicing Database (ASD) are shown for then gradually changes into a moderately differentiated human GCIP. Five of them were valuated at high and poorly differentiated phenotype during tumor confidence level. Alternative splicing is an important progression (Sugihara et al., 1992). GCIP expression mechanism to generate protein diversity and to deter- was also detected in human liver tumor samples with mine the binding properties, intracellular localization, immunostaining by specific anti GCIP antibody. GCIP enzymatic activity, protein stability and posttransla- protein was stained in 23 of 32 adenomatous hyperplasia tional modifications of proteins in a tissue-specific or (AH) samples (72%), 19 of 28 well-differentiated HCC development-specific way (Sierralta and Mendoza, 2004; samples (68%), and 9 of 18 poorly moderately Cooper, 2005; Lee and Wang, 2005; Stamm et al., 2005). differentiated HCC samples (50%). Expression of GCIP Splice site mutations of tumor suppressor genes often was high in the adenomatous hyperplasia (AH) and create truncated proteins as classical nonsense muta- well-differentiated HCC samples but relatively low in tions. Over the last two decades, many studies have the poorly differentiated HCC samples, suggesting that reported cancer-specific alternative splicing in the the decrease of GCIP might be a prerequisite event for absence of genomic mutations (Venables, 2004). The further tumor progression (Takami et al., 2005). In our ubiquitous expression of GCIP and multiple different present study, we demonstrated that overexpression of splicing isoforms of GCIP suggest that GCIP may have GCIP in mouse liver decreased susceptibility to DEN multiple functions and involve in multiple different induced hepatocarcinogenesis in transgenic mice in the processes. early stage of tumorigenesis. After given a single In this study, we have generated the GCIP transgenic intraperitoneal injection of DEN, 38.5% of wild-type mice by subcloning the human GCIP into a vector males developed liver tumors by 24 weeks, however, containing the albumin promoter to evaluate the in vivo only 7.8% of transgenic males developed liver tumors at function of GCIP in liver proliferation, differentiation, the same stage (Table 1). These results raised the and tumorigenesis. No abnormal development was possibility that GCIP may have a tumor suppressor detected in GCIP transgenic mice both before and after function in the early stage of the hepatocarcinogenesis. birth. Cell proliferation during liver regeneration in Although GCIP has little effect on the number of liver response to CCl4 induced liver injury was normal in tumors in the later stage of tumor development (86% in GCIP transgenic mice compared with the wild-type GCIP transgenic mice versus 100% in WT mice), tumors mice. Although GCIP may affect E2F-1 and cyclinD1 in GCIP transgenic mice are smaller and well-differ- activities in our in vitro assays, data obtained from the entiated compared to the tumors found in wild-type GCIP transgenic mice are similar to the overexpression mice, suggesting an important inhibitory role of GCIP of E2F-1 and cyclin D1 in mice liver (Conner et al., in liver tumor initiation and progression processes. 2000; Ledda-Columbano et al., 2002), in which no In our investigation of the potential mechanisms of obvious changes was observed in hepatic regenerative GCIP in tumorigenesis and transcription regulation, we growth and proliferation after partial hepatectomy found that GCIP could function as a transcription (Conner et al., 2000) and mitogenic stimuli (Ledda- repressor independent of the class I and class II HDAC Columbano et al., 2002), respectively. proteins. The D-type cyclins (cyclins D1, D2, and D3) Liver regeneration following liver injury is known to are key molecules in cell cycle progression, cell depend on two processes. The first one is through proliferation and tumorigenesis. The cyclin D proteins hepatocyte division and usually happens after acute liver complex with cdk4 and cdk6 and thereby regulate injury. If this process is impaired, liver repair depends transition from the G1 phase into the S phase by

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4214 phosphorylation and inactivation of the retinoblastoma ALT analyses protein (pRb) tumor suppressor. Amplification and/or Blood was drawn from anesthetized mice through heart overexpression of the cyclin D1 (bcl1, PRAD1,and puncture and the serum was separated by centrifugation for CCND1) gene has been found in 11–13% of human 20 min. Blood plasma levels of ALT (aminotransferase) hepatomas (Nishida et al., 1994, 1997; Suzui et al., 2002) activity were immediately measured using the GP-Trans- aminase kit (No. 505-P, Sigma, St Louis, MO, USA) in per and other human cancers, such as in 30–40% of human time point after CCl injection. breast cancers (Kenny et al., 1999; Pestell et al., 1999). 4 Overexpression of cyclin D1 is sufficient to initiate hepatocarcinogenesis in transgenic mice (Deane et al., Hepatocyte DNA synthesis assays 2001, 2004). Thus, cyclin D1 can function as an Hepatocyte DNA synthesis was measured by bromodeoxy- oncogene in the liver and is a potential target for uridine (BrdU) incorporation. Mice were given intraperitone- hepatoma prevention and therapy. Our finding that ally injections of 20 mg/ml BrdU (Boehringer, Mannheim, Indianapolis, IN, USA) in a dose of 50 mg/g body weight 2 h GCIP suppresses the liver tumorigenesis by significantly before killing. The liver samples were collected, fixed and down-regulating the transcriptional activity and protein paraffin-embedded at per time point after CCl4 treatment as expression of cyclin D1 suggest a novel mechanism for described in the text. Sections, 5-mm, were cut for immuno- GCIP as a potential tumor suppressor gene in human histochemistry (IHC). An ABC staining kit was purchased cancers. Future studies are underway in our laboratory from Santa Cruz Biotechnology. Immunohistochemical stain- to help us understand the function of GCIP as a tumor ing follows the manufacturer’s protocol from Santa Cruz suppressor and a regulator in cell proliferation and Biotechnology, Santa Cruz, CA, USA. Briefly, anti-BrdU differentiation processes. primary antibody (Sigma, no. 2531) was used at 1:500 in block serum. BrdU incorporation in fixed liver sections was visualized with an anti-BrdU monoclonal antibody and an alkaline phosphatase-conjugated secondary antibody. Posi- Materials and methods tively staining hepatocytes nuclei and unstained cells were counted under a Nikon Digital camera Dxm1200 microscope, Generation of GCIP transgenic mice and BrdU incorporation was expressed as the percentage of 1.1 kb human GCIP cDNA containing the entire coding region the number of labeled hepatocytes in four or five visual fields for the full-length human GCIP was inserted into the EcoRI per animal (magnification  100). site of an albumin promoter-driven expression vector (Figure 1a). A chicken insulator is inserted at 30-end of transgene to ensure the expression of target gene. This Diethylnitrosamine-induced liver tumor and histological transgene was used to generate transgenic mice by injecting procedures GCIP transgene into the pronucleus of FVB mice fertilized At 14 days of age, mice were given a single intraperitoneal eggs, then eggs were transferred to the oviduct of pseud- injection of DEN (Sigma Chemical Co.) in a dose of 10 mg/kg opregnant female FVB mouse. Transgenic mice were identified of body weight. At 21 days of age, mice were separated by sex by Southern blot analysis of EcoRI digested tail genomic DNA and their transgenicity was determined. using a 32P-labeled 1.1 kb GCIP cDNA probe. To confirm At 24 and 40 weeks of age, animals were killed and analysed. Southern results, PCR analysis of mouse tail DNA was Representative sections from each lobe of the liver and visible performed using oligonucleotide primers designed from the tumors were fixed, sectioned, stained with hematoxylin and human GCIP gene. Transgenic mice and control mice used in eosin (H&E), and examined for the presence of adenomas. this study were propagated by breeding with GCIP hetero- Statistical analysis of tumor incidence was carried out using zygote transgenic males with wild-type FVB females. Mice the Student’s t-test. were maintained in 12-h light/12-h dark cycles with free access Liver tissues were fixed overnight in Histochoice Tissue to food and water. All animal studies were performed in Fixative MB (no. H120–4L, Amresco, Solon, OH, USA), accordance with the guidelines of the Institutional Animal Use dehydrated through a series of ethanol treatments, and and Care Committee of the Institute of Biosciences and embedded in paraffin according to standard procedure. Technology, Texas A&M University System Health Science Sections were prepared and stained with H&E. Center. Northern blot analyses GCIPsiRNA plasmid construction Total RNA was isolated from liver using TRIZOL reagent For stable expression of GCIP siRNAs within cells, the U6 (Invitrogen, Carlsbad, CA, USA) and separated on 1% promoter was cloned in front of GCIP gene-specific targeting agarose-formaldehyde gel. RNA was transferred by capillary sequence (19 nt sequences GACTCAATGAGGCAGCTGT blotting overnight onto Hybond þ membrane (Amersham- from GCIP cDNA separated by a 9 nt spacer from the reverse Pharmacia Biotech, Piscataway, NJ, USA) in 10  SSC. complement of the same sequence) and five thymidines (T5) as GCIP probe was labeled with [a-32P]dCTP (Amersham) using termination signal. the Random Primers DNA Labeling system (Invitrogen), and hybridized according to the manufacturer’s protocol. Experimental liver injury by administration of CCl4 Prehybridization and hybridization were done in 10 ml Rapid All mice were 7-week-old males and kept under 12-h light/12-h Hyb buffer (Amersham) at 681C for 1 h and overnight, dark lighting cycles at 231C. Three to five 7-week-old male respectively. The blot was washed two times in 2  SSC with mice were used in each experimental group. A single dose of 0.1% SDS at room temperature for 30 min, followed by two 2.0 ml/kg of body weight (2:5 v/v in mineral oil) was washes in 0.1  SSC with 0.1% SDS at 681C for 30 min. We administered by intraperitoneal injection for CCl4-induced stripped the GCIP probe with 0.1% SDS and reprobed the liver damage study. After the mice were weighed, anesthetized blot with b-actin probe using the same protocol. Blots were and exsanguinated, livers were excised for analysis. exposed 3 h to overnight on BioRad Phosphorimager screen.

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4215 Cell culture and luciferase reporter assays the pCMV-Tag2B-GCIP plasmid, and the siRNA plasmid- The HepG2 cells were obtained from ATCC (Rockville, MD, specific for GCIP. Transfected cells will be selected with G418 USA) and maintained in a 371C incubator with 5% CO2 for 7days. After selection, cells will be analysed for colony humidified air in Dulbecco’s minimal modified Eagle’s medium growth by suspending in 0.3% agarose medium containing (DMEM) supplemented with 10% (v/v) fetal bovine serum. DMEM þ 5% FBS and layered onto a 2.5-ml bed of 0.6% At 48 h after transfection of the plasmids, the cells in a agarose in a 35-mm dish with grids. Plates will be incubated 3 24-well plate were lysed and harvested in 200 ml reporter lysis weeks, and the number of colonies >100 mm will be counted. buffer (Promega), and cell lysates were assayed for luciferase activities. The luciferase assay was carried out using luciferase Statistical analysis assay kit (Promega) and Packard Topcount Scintillation Counter. Extracts were also assayed for b-galactosidase Difference between wild-type and GCIP-transgenic animals in activity with the Galacto-Light Plust b-Galactosidase the liver tumor incidence was examined for statistical t Reporter Gene Assay Systems system (Tropix). Each extract significance using the -test. was assayed three times, and the mean relative light unit (RLU) was corrected by values obtained from an extract Acknowledgements prepared from nontransfected cells. The relative luciferase activity was calculated as RLU/b-galactosidase. This work was supported partially from NIH Grants (5R01HL064792 and 1R01CA106479). We would like to Soft agar assay thank members of the Liu laboratory and members of the Human liver HepG2 cells were cultured in DMEM containing Center for Cancer Biology and Nutrition for their comments 10% fetal calf serum and transfected by pCMV-Tag2B vector, and discussion.

References

Barndt RJ, Dai M, Zhuang Y. (2000). Mol Cell Biol 20: Lee GH, Ooasa T, Osanai M. (1998). Cancer Res 58: 6677–6685. 1665–1669. Benezra R, Davis RL, Lockshon D, Turner DL, Weintraub H. Lee JE, Hollenberg SM, Snider L, Turner DL, Lipnick N, (1990). Cell 61: 49–59. Weintraub H. (1995). Science 268: 836–844. Blackwell TK, Kretzner L, Blackwood EM, Eisenman RN, Lemmer ER, Vessey CJ, Gelderblom WC, Shephard EG, Van Weintraub H. (1990). Science 250: 1149–1151. Schalkwyk DJ, Van Wijk RA et al. (2004). Carcinogenesis Conner EA, Lemmer ER, Omori M, Wirth PJ, Factor VM, 25: 1257–1264. Thorgeirsson SS. (2000). Oncogene 19: 5054–5062. Lyden D, Young AZ, Zagzag D, Yan W, Gerald W, O’Reilly Cooper TA. (2005). Cell 120: 1–2. R et al. (1999). Nature 401: 670–677. Cordle SR, Henderson E, Masuoka H, Weil PA, Stein R. Ma Q, Kintner C, Anderson DJ. (1996). Cell 87: 43–52. (1991). Mol Cell Biol 11: 1734–1738. Massari ME, Murre C. (2000). Mol Cell Biol 20: 429–440. Deane NG, Lee H, Hamaamen J, Ruley A, Washington MK, Molkentin JD, Black BL, Martin JF, Olson EN. (1995). Cell LaFleur B et al. (2004). Cancer Res 64: 1315–1322. 83: 1125–1136. Deane NG, Parker MA, Aramandla R, Diehl L, Lee WJ, Moore MR, Drinkwater NR, Miller EC, Miller JA, Pitot HC. Washington MK et al. (2001). Cancer Res 61: 5389–5395. (1981). Cancer Res 41: 1585–1593. Diwan BA, Meier H. (1976). Cancer Lett 1: 249–253. Motokura T, Arnold A. (1993). Genes Cancer 7: Gaubatz S, Meichle A, Eilers M. (1994). Mol Cell Biol 14: 89–95. 3853–3862. Murre C, McCaw PS, Vaessin H, Caudy M, Jan LY, Jan YN Hollenberg SM, Sternglanz R, Cheng PF, Weintraub H. et al. (1989). Cell 58: 537–544. (1995). Mol Cell Biol 15: 3813–3822. Nabeshima Y, Hanaoka K, Hayasaka M, Esumi E, Li S, Hu JS, Olson EN, Kingston RE. (1992). Mol Cell Biol 12: Nonaka I. (1993). Nature 364: 532–535. 1031–1042. Nakatani T, Roy G, Fujimoto N, Asahara T, Ito A. (2001). Hwang SY, Oh B, Fuchtbauer A, Fuchtbauer EM, Johnson Jpn J Cancer Res 92: 249–256. KR, Solter D et al. (1997). Dev Dyn 209: 217–226. Natrajan R, Louhelainen J, Williams S, Laye J, Knowles MA. Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor (2003). Cancer Res 63: 7657–7662. A et al. (2005). CA Cancer J Clin 55: 10–30. Nishida N, Fukuda Y, Ishizaki K, Nakao K. (1997). Histol Jensen J, Pedersen EE, Galante P, Hald J, Heller RS, Ishibashi Histopathol 12: 1019–1025. M et al. (2000). Nat Genet 24: 36–44. Nishida N, Fukuda Y, Komeda T, Kita R, Sando T, Kee BL, Quong MW, Murre C. (2000). Immunol Rev 175: Furukawa M et al. (1994). Cancer Res 54: 3107–3110. 138–149. Peppelenbosch MP, Qiu RG, de Vries-Smits AM, Tertoolen Kenny FS, Hui R, Musgrove EA, Gee JM, Blamey RW, LG, de Laat SW, McCormick F et al. (1995). Cell 81: 849–856. Nicholson RI et al. (1999). Clin Cancer Res 5: 2069–2076. Perk J, Iavarone A, Benezra R. (2005). Nat Rev Cancer 5: Knight B, Yeoh GC, Husk KL, Ly T, Abraham LJ, Yu C et al. 603–614. (2000). J Exp Med 192: 1809–1818. Pestell RG, Albanese C, Reutens AT, Segall JE, Lee RJ, Kreider BL, Benezra R, Rovera G, Kadesch T. (1992). Science Arnold A. (1999). Endocr Rev 20: 501–534. 255: 1700–1702. Pin CL, Rukstalis JM, Johnson C, Konieczny SF. (2001). Lassar AB, Davis RL, Wright WE, Kadesch T, Murre C, J Cell Biol 155: 519–530. Voronova A et al. (1991). Cell 66: 305–315. Recknagel RO, Glende Jr EA, Dolak JA, Waller RL. (1989). Ledda-Columbano GM, Pibiri M, Concas D, Cossu C, Pharmacol Ther 43: 139–154. Tripodi M, Columbano A. (2002). Hepatology 36: Rivera R, Murre C. (2001). Oncogene 20: 8308–8316. 1098–1105. Rivera RR, Johns CP, Quan J, Johnson RS, Murre C. (2000). Lee C, Wang Q. (2005). Brief Bioinform 6: 23–33. Immunity 12: 17–26.

Oncogene GCIP suppresses hepatocarcinogenesis in transgenic mouse WMaet al 4216 Roskams TA, Libbrecht L, Desmet VJ. (2003). Semin Liver Takami T, Terai S, Yokoyama Y, Tanimoto H, Tajima K, Dis 23: 385–396. Uchida K et al. (2005). Gastroenterology 128: 1369–1380. Ruzinova MB, Benezra R. (2003). Trends Cell Biol 13: Terai S, Aoki H, Ashida K, Thorgeirsson SS. (2000). 410–418. Hepatology 32: 357–366. Shirakata M, Paterson BM. (1995). EMBO J 14: 1766–1772. Theise ND, Yao JL, Harada K, Hytiroglou P, Portmann B, Sierralta J, Mendoza C. (2004). Brain Res Brain Res Rev 47: Thung SN et al. (2003). Histopathology 43: 263–271. 105–115. Venables JP. (2004). Cancer Res 64: 7647–7654. Sikder H, Huso DL, Zhang H, Wang B, Ryu B, Hwang ST Vesselinovitch SD. (1987). Toxicol Pathol 15: 221–228. et al. (2003). Cancer Cell 4: 291–299. Voronova AF, Lee F. (1994). Proc Natl Acad Sci USA 91: Spicer DB, Rhee J, Cheung WL, Lassar AB. (1996). Science 5952–5956. 272: 1476–1480. Wroblewski F. (1959). Am J Med 27: 911–923. Stamm S, Ben-Ari S, Rafalska I, Tang Y, Zhang Z, Toiber D Xia C, Bao Z, Tabassam F, Ma W, Qiu M, Hua S et al. (2000). et al. (2005). Gene 344: 1–20. J Biol Chem 275: 20942–20948. Sugihara S, Nakashima O, Kojiro M, Majima Y, Tanaka M, Yao Y, Doki Y, Jiang W, Imoto M, Venkatraj VS, Warburton Tanikawa K. (1992). Cancer 70: 1488–1492. D et al. (2000). Exp Cell Res 257: 22–32. Sun XH, Copeland NG, Jenkins NA, Baltimore D. (1991). Yokota Y, Mori S. (2002). J Cell Physiol 190: 21–28. Mol Cell Biol 11: 5603–5611. Zebedee Z, Hara E. (2001). Oncogene 20: 8317–8325. Sun Y, Nadal-Vicens M, Misono S, Lin MZ, Zubiaga A, Hua Zhao XF, Aplan PD. (1999). J Biol Chem 274: 1388–1393. X et al. (2001). Cell 104: 365–376. Zhuang Y, Kim CG, Bartelmez S, Cheng P, Groudine M, Suzui M, Masuda M, Lim JT, Albanese C, Pestell RG, Weintraub H. (1992). Proc Natl Acad Sci USA 89: Weinstein IB. (2002). Cancer Res 62: 3997–4006. 12132–12136.

Oncogene