The Histone Deacetylase Inhibitor Trichostatin a Induces Gadd45c Expression Via Oct and NF-Y Binding Sites

Oncogene (2008) 27, 1263–1272 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE The histone deacetylase inhibitor trichostatin A induces GADD45c expression via Oct and NF-Y binding sites

MR Campanero1, A Herrero2 and V Calvo2

1Instituto de Investigaciones Biome´dicas, CSIC-UAM, Arturo Duperier, Madrid, Spain and 2Departamento de Bioquı´mica e Instituto de Investigaciones Biome´dicas Alberto Sols, UAM-CSIC, Facultad de Medicina de la Universidad Auto´noma de Madrid-UAM, Madrid, Spain

The GADD45c protein is a potential tumor suppressor leukemia cells in response to genotoxic stress or whose expression is reduced in several tumors. However, interleukin (IL)-6 (Takekawa and Saito, 1998; Zhang very little is known about the regulation of its expression. et al., 1999), in hemangioblasts in response to oncostatin We have determined that the most relevant region of its M (Nakayama et al., 1999), in neurons in response to promoter lies between nucleotides À112 and À54, relative nerve growth factor depletion (Kojima et al., 1999) and to the transcription start site. Putative Oct and NF-Y in T cells in response to IL-2 or IL-12 plus IL-18 elements were found in this region and factors belonging (Beadling et al., 1993; Lu et al., 2004). to these families interacted with these elements in vitro Gadd45 proteins have very similar amino acid and with the promoter in vivo. Mutation of these elements sequences (Takekawa and Saito, 1998; Zhang et al., reduced the basal activity of the promoter, suggesting that 1999) and have been implicated in cell cycle arrest (Fan both sites are essential for basal expression. These factors et al., 1999; Zhang et al., 1999; Jin et al., 2002), DNA interact with chromatin modifying proteins and we found repair (Smith et al., 2000), innate immunity (Lu et al., that histone deacetylase 1 or silencing mediator for 2001; Yang et al., 2001), maintenance of genomic retinoid and thyroid hormone receptor overexpression stability (Hollander et al., 1999) and apoptosis (Take- reduced the basal activity of the promoter. In contrast, kawa and Saito, 1998; Zhang et al., 1999, 2001). Gadd45 forced expression of the histone acetylase protein PCAF proteins display a complex array of physical interactions or cell treatment with the HDAC inhibitor trichostatin A with cellular proteins involved in cell cycle regulation increased GADD45c mRNA levels and induced and the response to extrinsic stress, including p21, Cdc2/ GADD45c promoter activity through its Oct and NF-Y cyclinB1 and components of the JNK and p38 stress- elements. Moreover, ectopic expression of a dominant- induced kinase pathways (Liebermann and Hoffman, negative version of NF-YA strongly inhibited trichostatin 2003). In particular, Gadd45g regulates apoptosis in A-induced activation of the promoter. Our data strongly several cell types (Takekawa and Saito, 1998; Zhang suggest that inhibition of deacetylase activity could et al., 2001; Chung et al., 2003) and cytokine production potentially be used for treatment of tumors where in T cells (Lu et al., 2001). Moreover, Gadd45g GADD45c expression is reduced. negatively regulates progression through the cell cycle. Oncogene (2008) 27, 1263–1272; doi:10.1038/sj.onc.1210735; This effect is likely mediated by its capacity to inhibit published online 27 August 2007 cyclinB1-cdk1 activity (Vairapandi et al., 2002). Gadd45g expression is reduced in several tumors, Keywords: GADD45g; HDAC inhibitor; TSA; Oct; NF-Y such as pituitary adenomas (Zhang et al., 2002), anaplastic thyroid cancer (Chung et al., 2003), hepatocellular carcinoma (Sun et al., 2003) and lymphoma and nasopharyngeal carcinoma cell lines (Ying et al., 2005) Introduction in comparison to normal tissues. A low Gadd45g expression could be functionally relevant to tumor The GADD45 family of genes, including GADD45a, growth because transient transfection of Gadd45g GADD45b and GADD45g, is induced by genotoxic induced growth arrest and/or apoptosis in a number of stress and a subset of cytokines in many different cell tumor cell lines (Takekawa and Saito, 1998; Zhang types (Liebermann and Hoffman, 2003). GADD45g et al., 2001, 2002; Chung et al., 2003; Jiang and Wang, (CR6/OIG37/GRP17) expression is induced in myeloid 2004; Ying et al., 2005). These observations suggest that GADD45g can be considered as a functional tumor suppressor gene. Although several studies have addressed the Correspondence: Dr V Calvo, Departamento de Bioquı´ mica, Facultad transcriptional regulation of the GADD45a and b genes de Medicina, Universidad Auto´ noma de Madrid, Arzobispo Morcillo (Zhan et al., 1994; Constance et al., 1996; Marhin et al., 4, 28029 Madrid, Spain. E-mail: [email protected] 1997; De Smaele et al., 2001; Jin et al., 2001; Takahashi Received 6 November 2006; revised 21 June 2007; accepted 17 July 2007; et al., 2001), very little is known about the regulation of the published online 27 August 2007 GADD45g gene. In fact, to the best of our knowledge, Induction of GADD45c by HDAC inhibitor MR Campanero et al 1264 there is only a single study addressing GADD45g gene regulation where it was suggested that transcription - 818 factors C/EBPb and d might be involved in GADD45g promoter activation during differentiation of M1 cells - 482 (Jung et al., 2000). A general mechanism of gene regulation relies on the - 220 interactions of transcription factors with co-activators - 112 that contain histone acetyltransferase (HAT) activity and co-repressors that contain histone deacetylase - 54 (HDAC) activity (Berger, 2002). HDAC inhibitors, such as trichostatin A (TSA; Sigma, Tres Cantos, Madrid, pGL3-B Spain), induce many genes coding for proteins involved in cell cycle control, including GADD45a and GADD45b 0 20 40 60 80100 120 (Della-Ragione et al., 2001; Chen et al., 2002; Hirose Relative luciferase activity et al., 2003), and stop tumor growth by eliciting either cell cycle arrest or apoptosis in tumor cells (Johnstone, Figure 1 Deletion analysis of the murine GADD45g promoter. Several 50-deleted promoter fragments were cloned into pGL3- 2002). Basic (pGL3-B). The resulting plasmids (125 ng) were transiently Since GADD45g expression is reduced in tumors and transfected into NIH3T3 cells and luciferase activities were its forced expression has tumor-suppressive properties, analysed, normalized to protein concentration and b-galactosidase a detailed knowledge of the molecular mechanisms activity, and referred to the normalized luciferase activity of the À818 promoter reporter (relative luciferase activity). Data are regulating its expression might help to develop new shown as means (bars, s.e.) of six independent experiments. antitumor therapeutic strategies. We therefore analysed the mechanisms of GADD45g transcriptional regulation. Our results indicate that functional Oct and NF-Y elements are essential for the promoter activity, that -818 -482 -220 -112 -54 +15 GADD45g promoter can be regulated through histone acetylation/deacetylation, and that GADD45g expression can be readily induced in tumor cells by HDAC Oct1 NF-Y TATA inhibitors possibly through the NF-Y and Oct elements of its promoter.

-818

Results Octm NFYm Deletion and point mutation analysis of the murine GADD45g promoter OctmNFYm To gain insight into the molecular mechanisms regulating GADD45g expression, we first identified the motifs pGL3-B required for basal activity of the murine GADD45g promoter. To this end, a set of 50 serially deleted promoter fragments, spanning from nucleotides À818 to 0 20406080100 120 þ 15 according to the transcription start site (Balliet Relative luciferase activity et al., 2003), were cloned upstream from the luciferase Figure 2 Mutational analysis of the GADD45g promoter. A gene in pGL3-Basic. The resulting reporter constructs schematic representation of the GADD45g promoter showing the were transiently transfected in NIH3T3 cells and the approximate positions of the Oct, NF-Y and TATA box elements luciferase activity was assayed for each construct. As is shown. Reporter plasmids (125 ng) containing either the wild- shown in Figure 1, although the promoter activity type promoter (nucleotides À818 to þ 15) or promoters devoid of either the Oct site (Octm), the NF-Y site (NFYm) or both sites decreased when the region À818 to À112 was deleted, an (OctmNFYm) were transiently transfected into NIH3T3 cells and almost complete ablation of its activity was achieved luciferase activities were analysed. Relative luciferase activities only upon deletion of the À112 to À54 region. These were calculated as in Figure 1. Data are shown as means (bars, s.e.) results suggest that although elements within the À818 of ten independent experiments. to À112 can contribute to this promoter regulation, the region between À112 and À54 is essential for its activity. A search for putative transcription factor binding sites using MatInspector (Cartharius et al., 2005) revealed the presence in the À112 to À54 region of or combined. As shown in Figure 2, mutation of the potential Oct (ATTTGCAT), and NF-Y (CCAAT) individual sites or their combination decreased the binding sites. Site-directed mutagenesis was used to activity to 35–50% of the wild-type promoter activity, generate mutated versions of the À818 construct devoid suggesting that these elements significantly contribute to of the Oct (Octm) and NF-Y (NFYm) sites, individually promoter activation.

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1265 Identiﬁcation of proteins interacting with the Oct and antibody, partially decreased formation of the DNA– NF-Y sites of the murine GADD45g promoter protein complex observed with the Oct probe and Electrophoretic mobility shift assays (EMSAs) were presented a supershifted band (Figure 3a). Employing performed using probes containing either the Oct nuclear extracts from HEK293-T cells transfected with (nucleotides À116 to À93) or the NF-Y site (nucleotides an Oct-1 expression plasmid, anti-Oct-1 addition caused À92 to À69). A major DNA–protein complex was a supershift of the majority of the complex observed detected with each probe whose formation was elimi- with the Oct probe, indicating that this probe can nated by an excess of unlabeled self Oct or NF-Y actually bind Oct-1 and that the antibody used in these competitor probe (Figures 3a and b), indicating that experiments works well in these assays. Because the anti- formation of these complexes was speciﬁc. These factors Oct-1 antibody had only a partial effect employing appear to correspond to Oct-1 and NF-Y factors since NIH3T3 extracts, it is likely that other members of the addition of an excess of oligonucleotides known to bind Oct family might also interact with the GADD45g Oct either Oct (Control-Oct) or NF-Y (Control-NF-Y) element in vitro. proteins also ablated formation of the corresponding To directly test whether Oct-1 and NF-Y were complexes, while an irrelevant oligonucleotide (E2F-1A) associated with the murine GADD45g promoter had no effect (Figure 3a). Moreover, anti-NF-YB, but in vivo, chromatin immunoprecipitation (ChIP) assays not anti-Maz addition, abolished formation of the were performed with anti-Oct-1, anti-NF-YB or DNA–protein complex observed with the NF-Y probe Rabbit anti-mouse IgG antibodies. The products of (Figure 3a). An anti-Oct-1, but not an anti-Maz PCR reactions performed with DNA eluted from the

Figure 3 Binding of nuclear proteins to the Oct and NF-Y sites of the GADD45g promoter. (a) EMSAs were carried out with nuclear extracts prepared from NIH3T3 cells, HEK293-T cells transfected with the pCG empty vector (293-T, Mock) or HEK293-T cells transfected with the pCG-Oct-1 expression plasmid (293-T, Oct-1) as indicated. The end-labeled oligonucleotides containing the sequence of either the NF-Y site (GADD45g-NFY, nucleotides À92 to À69) or the Oct site (GADD45g-Oct, nucleotides À116 to À93) were used as probes and the unlabeled E2F1-A, GADD45g-NFY, Control-NFY, GADD45g-Oct or Control-Oct oligonucleotides were used as competitors. Nuclear extracts were incubated for 15 min at RT with antibodies (anti-Maz, anti-NF-YB or anti-Oct-1) before the addition of the probes as indicated. Arrowheads and arrows point to specific retarded complexes and supershifted complexes, respectively. The position of the free probe is indicated. (b) NIH3T3 cells were analysed by ChIP assay with Rabbit anti- mouse IgG, anti-Oct-1 or anti-NF-YB. Different dilutions of input chromatin DNA (Input) or DNA extracted from the indicated immunoprecipitates (Ab) were subjected to PCR amplification using oligonucleotides flanking the proximal GADD45g promoter and containing the Oct and NF-Y sites (Specific) or oligonucleotides flanking a region about 3000 nucleotides upstream from the GADD45g transcription initiation site (Control). The products separated by agarose gel electrophoresis are shown.

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1266 immunoprecipitates are shown in Figure 3b. A product Role of co-repressors and co-activators in the regulation containing the Oct and NF-Y sites was specifically of the murine GADD45g promoter amplified with oligonucleotides flanking the proximal NF-Y and Oct-1 can interact with HAT-containing co- GADD45g promoter (Specific in Figure 3b) employing activator complexes and HDAC-containing co-repres- both anti-Oct-1 and anti-NF-YB immunoprecipitates, sor complexes (Nagy et al., 1997; Mantovani, 1999; but not when the anti-mouse IgG immunoprecipitate Kakizawa et al., 2001; Park et al., 2002; Peng and was used. In contrast, no product was amplified with Jahroudi, 2003; Huang et al., 2005). We therefore oligonucleotides flanking a region about 3000 nucleo- assessed the contribution of these factors to GADD45g tides upstream from the GADD45g transcription initia- promoter activity regulation. Silencing mediator for tion site (Control in Figure 3b) using any of the immuno- retinoid and thyroid hormone receptors (SMRT) inter- precipitates, while these oligonucleotides amplified the act with Oct-1 and acts as a transcriptional co-repressor product from input chromatin DNA. These results show (Kakizawa et al., 2001). To analyse the possible role of that Oct-1 and NF-Y can bind to the murine GADD45g Oct-1-associated SMRT in the regulation of the proximal promoter in vivo. GADD45g promoter, a SMRT expression plasmid was Co-transfection assays were performed to confirm the co-transfected with the À818 to þ 15 promoter reporter involvement of Oct-1 and NF-Y in the regulation of the plasmid. As shown in Figure 5a, SMRT co-transfection murine GADD45g promoter. The luciferase reporter resulted in a dose-dependent inhibition of transcrip- plasmids used were either the deleted À220 to þ 15 tional activity. promoter construct, which contains the Oct and NF-Y Because SMRT co-repressor activity can be exerted sites, or a mutated version of this construct devoid of through its presence in a macromolecular complex these sites (À220OctmNFYm). As shown in Figure 4, containing the HDAC1 (Nagy et al., 1997) and NF-Y Oct-1 or NF-YA overexpression slightly increased the inhibits von Willebrand factor promoter activity activity of the À220 to þ 15 reporter, but not that of the through recruitment of HDACs (Peng and Jahroudi, À220OctmNFYm construct or that of the promoterless 2003), we investigated the involvement of HDAC1 in the reporter (pGL3-B). However, co-transfection of both regulation of the GADD45g promoter. To this end, an Oct-1 and NF-YA strongly increased the activity of HDAC1 expression plasmid was co-transfected with the the À220 to þ 15 reporter, but barely affected the À818 to þ 15 promoter reporter plasmid. As shown in activity of the pGL3-B or the À220OctmNFYm Figure 5b, HDAC1 co-transfection reduced GADD45g reporters (Figure 4). promoter activity almost twofold, as compared to co- Altogether, these results show that transcription transfection with the control vector or HDAC3 expres- factors Oct-1 and NF-Y can interact with the murine sion plasmid. GADD45g promoter in vivo and that they might be NF-Y recruits the histone acetylase PCAF to important for the basal transcriptional activity of the promoters (Park et al., 2002; Huang et al., 2005). The promoter. participation of PCAF in the regulation of the GADD45g promoter was addressed by co-transfecting 300 a PCAF expression plasmid with the À818 to þ 15 promoter reporter plasmid. As shown in Figure 5c, PCAF 250 co-transfection resulted in a dose-dependent induction 200 of GADD45g promoter activity.

150

100 Transcriptional induction of the GADD45g gene by TSA The results described above suggest that HATs and 50 deacetylases (HDACs) can participate in the regulation Relative luciferase activity Relative luciferase 0 of the murine GADD45g promoter. To conﬁrm this hypothesis, we used the HDAC inhibitor TSA. Treat- Mock Mock Mock Oct-1 Oct-1 Oct-1 ment of NIH 3T3 cells with TSA for 24 h increased NF-YA NF-YA NF-YA GADD45g mRNA expression levels in a dose-dependent manner, as determined by northern blot analysis Oct-1 + NF-YA Oct-1 + NF-YA Oct-1 + NF-YA (Figure 6a). Because TSA causes G1 phase arrest in the pGL-3B -220 -220OctmNFYm human T-cell leukemia cell line, Jurkat (Yokota et al., 2004) and the increase of GADD45g expression is able Figure 4 Oct-1 and NF-YA expression induces murine to induce growth arrest, we assessed whether Jurkat cells GADD45g promoter activity. NIH3T3 cells were co-transfected treatment with TSA could induce GADD45g mRNA with: pGL3-B (50 ng) or reporter plasmids containing either the wild-type À220 to þ 15 sequence of the murine GADD45g expression in these cells. As shown in Figure 6b, promoter (À220) or the À220 to þ 15 sequence devoid of the Oct GADD45g mRNA expression levels, but not those of and NF-Y sites (À220OctmNFYm) plus either the empty expres- the g-actin mRNA, were notably increased in these cells sion plasmid pCG (100 ng, Mock) or expression plasmids for Oct-1 following treatment with TSA. These results suggest (pCG-Oct-1) or NF-YA (pCG-NF-YA), as indicated. Normalized luciferase activities were referred to the activity of extracts from that G1-phase arrest induction by TSA in the leukemic cells co-transfected with pGL-3B and pCG. Data are representative line Jurkat might be mediated, at least in part, through of three independent experiments with similar results. induction of GADD45g expression. To rule out that

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1267 140

120

100

Relative luciferase activity luciferase Relative 20

0 0 25 50 125 SMRT (ng)

120 Figure 6 Trichostatin A (TSA) increases GADD45g mRNA levels. NIH3T3 (a) or Jurkat (b) cells were treated with different 100 concentrations of TSA for 24 h and GADD45g mRNA levels were analysed by northern blotting. The blots were rehybridized with the 80 probe for g-actin mRNA to normalize the amount of loaded RNA.

20 Control TSA 0.01 µM Relative luciferase activity luciferase Relative 0 Control HDAC1 HDAC3

5 TSA 0.03 µM TSA 0.3 µM 4

2 Events PCAF induction fold 1

0 0 500 1000 TSA 1 µM TSA 3 µM PCAF (ng)

Figure 5 Role of co-repressors and co-activators in the regulation of the murine GADD45g promoter. (a) SMRT expression inhibits GADD45g promoter activity. NIH3T3 cells were co-transfected with the À818 promoter reporter plasmid (125 ng) and increasing amounts of pCMX-SMRT. Normalized luciferase activities were referred to the activity of extracts from cells transfected in the absence of SMRT. Data are shown as means (bars, s.e.) of three FL2-A independent experiments. (b) HDAC1 expression inhibits the Figure 7 TSA is not toxic to NIH3T3 cells. Asynchronous GADD45g promoter. NIH3T3 cells were co-transfected with the NIH3T3 cell cultures were treated with the indicated TSA À818 promoter reporter plasmid (50 ng) and 200 ng of control concentrations for 24 h. The DNA content was analysed by flow vector, pCMX-HDAC1 or pCMX-HDAC3. Normalized luciferase cytometry. Propidium iodide fluorescence intensity is represented activities were referred to the activity of extracts from cells on the x axis on a linear scale (FL2-A). transfected with the control vector. Data are shown as means (bars, s.e.) of three independent experiments. (c) PCAF expression induces the GADD45g promoter. NIH3T3 cells were co-transfected with the À818 promoter reporter plasmid (125 ng) and toxicity caused by TSA treatment had a major role in increasing amounts of pCX-PCAF. Normalized luciferase activities were referred to the activity of extracts from cells transfected in the GADD45g expression regulation, propidium iodide absence of PCAF. Data are shown as means (bars, s.e.) of seven staining was used to assess DNA content by flow independent experiments. cytometry analysis. As shown in Figure 7, none of the different TSA doses employed in our experiments caused the appearance of apoptotic cells with a sub-G1 DNA content.

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1268 Oct and NF-Y sites participate in the induction of the 120 GADD45g promoter by TSA To analyse the possible contribution of Oct and NF-Y 100

sites to GADD45g promoter induction by TSA, the n transcriptional activity of different site-directed mutant 80 reporters was analysed in NIH3T3 cells untreated or treated with 0.1 mM TSA. Wild-type Oct and NF-Y 60

elements appear to be essential for GADD45g promoter inductio TSA activation by TSA because mutation of both the Oct and 40 NF-Y sites ablated TSA induction of the À818 to þ 15 promoter (Figure 8a). In fact, similar results were Relative 20 obtained with the promoterless luciferase vector pGL3-B (Figure 8a). Mutation of either site alone did 0 not inhibit promoter induction by TSA (Figure 8a). The 0 25 50 125 transcriptional activity of a minimal version of the NFY-Am29 (ng) promoter, spanning from nucleotides À220 to þ 15 and Figure 9 Dominant-negative NF-YA expression inhibits TSA containing the Oct and NF-Y sites, was also strongly induction of the GADD45g promoter. NIH3T3 cells were co- induced by TSA (Figure 8b). This induction was transfected with the À818 promoter reporter plasmid and increas- dependent on the Oct and NF-Y sites, since mutation ing amounts of either empty pSG5 or pSG5-NF-YAm29. Cells of both sites reduced TSA induction to almost 10% to were treated with the indicated TSA concentrations for 24 h. Normalized luciferase activities were referred to the activity of that of the wild-type minimal promoter (Figure 8b). extracts from cells transfected with the control empty pSG5 vector. These results indicate that the Oct and NF-Y sites are Data are shown as means (bars, s.e.) of three independent experiments.

-818 essential for the induction of the murine GADD45g promoter by TSA. Octm The role of NF-Y transcription factor in TSA induction of the murine GADD45g promoter was further NFYm analysed by using the dominant-negative NF-YA expression plasmid, NF-YAm29, which is devoid OctmNFYm of DNA-binding activity (Mantovani et al., 1994). Co- pGL3-B transfection of NF-YAm29, but not that of the control plasmid, with the À818 to þ 15 promoter reporter resulted in a dose-dependent inhibition of TSA induc- 0 5 10 15 20 25 tion (Figure 9), further implicating NF-Y in the TSA- induced activation of the GADD45g promoter. -220

-220OctmNFYm Discussion pGL3-B Despite the importance of GADD45g as a tumor 0 20 40 60 80 100 120 140 160 suppressor, almost nothing was known about the TSA induction fold mechanisms governing its expression. It was only known Figure 8 The Oct and NF-Y sites are essential for GADD45g that GADD45g promoter is induced by IL-6 or during promoter induction by TSA. (a) NIH3T3 cells were transfected M1 cells differentiation, likely through a noncanonical with reporter plasmids (125 ng) containing either the wild-type C/EBP binding element (Jung et al., 2000). Notably, promoter (nucleotides À818 to þ 15) or promoters devoid of either very little is known about the mechanisms eliciting to its the NF-Y site (NFYm), the Oct site (Octm) or both sites (OctmNFYm), and treated or not with TSA (0.1 mM) for 24 h. loss of expression in tumors. The murine GADD45g Extracts were prepared from these cells and normalized luciferase promoter shares some structural features with the activities were determined. Each fold induction by TSA was murine GADD45a promoter, namely Oct and NF-Y calculated by dividing the normalized luciferase activity of TSA- putative binding sites. Mutation of these sites in the treated cells by that of untreated cells. Data are shown as means human GADD45a promoter decreases its basal activity (bars, s.e.) of seven independent experiments. (b) Reporter plasmids (125 ng) containing either the wild-type À220 to þ 15 sequence (Hirose et al., 2003). Similarly, we have shown that (À220) or the À220 to þ 15 sequence devoid of the Oct and NF-Y mutation of either the Oct or NF-Y sites decreases the sites (À220OctmNFYm) were transfected into NIH3T3 cells. Cells murine GADD45g promoter basal activity and that were treated with TSA (0.1 mM) for 24 h and normalized luciferase expression of Oct-1 and NF-YA increases its activity. activities were determined. Each fold induction by TSA was calculated by dividing the normalized luciferase activity of TSA- We have also shown that these factors interacted with treated cells by that of untreated cells. Data are shown as means these sites in vitro and with the GADD45g promoter (bars, s.e.) of three independent experiments. in vivo. Therefore, the Oct and NF-Y sites shared by the

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1269 human GADD45a and murine GADD45g promoters inhibitor might also induce the expression of these genes appear to be relevant for their transcriptional activities. in NIH3T3 or Jurkat cells. In the human GADD45a Because deletion of the À818 to À112 partially promoter, two sets of Oct and NF-Y sites contribute to decreased the basal activity of the murine GADD45g its activation by TSA (Hirose et al., 2003). However, the promoter, it is likely that this region also contains human GADD45b proximal promoter contains NF-Y positive regulatory elements. In fact, MatInspector sites but no Oct site, as predicted by MatInspector identified potential transcription factor binding sites in analysis (Cartharius et al., 2005). In consequence, this region, such as ATF6, NFAT, Nur77 and myc. although the three GADD45 genes are induced by Besides Oct and NF-Y binding sites, the À112 to À54 TSA, the molecular mechanism underlying this induc- region also contains a TATA box and potential binding tion may be different in the GADD45b gene, as sites for additional factors such as or C/EBP, ZF5 and compared to the GADD45a and GADD45g genes. NF-I. These sites might account for the basal activity of Oct sites are present in the promoters of cell cycle- the À818 to þ 15 reporter devoid of the Oct and NF-Y regulated genes such as histone 2B (Hinkley and Perry, sites (Figure 2). However, the Oct and NF-Y sites in the 1992) and cyclin D1 (Brockman and Schuler, 2005). In À112 to À54 region are the most critical sites for addition, the binding of NF-Y to several cell cycle induction of the promoter by the HDAC inhibitor TSA control genes is strictly regulated during the different (Figure 8). phases of the cell cycle (Caretti et al., 2003). Thus, the Since Oct-1 and NF-Y can interact with HAT- involvement of Oct and NF-Y sites in the regulation of containing co-activator complexes and HDAC-con- the GADD45a and GADD45g genes by TSA, a growth- taining co-repressor complexes (Nagy et al., 1997; arrest inducer (Mariadason et al., 2000), constitutes Mantovani, 1999; Kakizawa et al., 2001; Park et al., another example of Oct and NF-Y-mediated, cell cycle- 2002; Peng and Jahroudi, 2003; Huang et al., 2005), we dependent regulation of gene expression. studied the possible role of SMRT, HDAC1, and PCAF The mechanisms that underlie the participation of the in the regulation of the murine GADD45g promoter. Oct and NF-Y sites in the TSA-induced activation of the SMRT or HDAC1, but not HDAC3, ectopic expression murine GADD45g promoter remain unknown. A reduced GADD45g promoter activity, while forced possible mechanism is that TSA alters the DNA-binding expression of PCAF increased it. These results suggest ability of Oct-1 or NF-Y. In fact, it has been reported that regulation of histone acetylation/deacetylation that the binding activity of Oct-1 to the HLA-DRA levels is essential for murine GADD45g promoter promoter is greatly reduced by TSA treatment (Osborne control. It has been proposed that SMRT-induced et al., 2001). However, EMSAs showed that TSA repression could be exerted through its presence in a treatment did not change the affinity of nuclear factors complex containing HDAC1. However, we have found for the NF-Y or Oct sites in the murine GADD45g that SMRT-induced repression of GADD45g promoter promoter (unpublished observations). This agrees with activity is stronger than that induced by HDAC1. These the lack of effect of TSA on the affinity of NF-Y or Oct- results might suggest that other members of the HDAC 1 for their sites in the human GADD45a promoter family might be involved in SMRT-induced repression. (Hirose et al., 2003). Alternatively, TSA treatment may Alternatively, SMRT might elicit GADD45g promoter affect the interaction of Oct-1 with SMRT-HDAC1, repression through some other mechanism independent and/or the interactions of NF-Y with HDAC1 or of HDAC activity. PCAF. In this regard, TSA releases HDAC2 from the Although HAT and HDAC activities play a general histone–DNA complex at the INK4d promoter (Yokota role in the regulation of gene activity, only a small et al., 2004). Further experiments will be required to proportion of genes is susceptible to regulation by definitely establish the molecular mechanism involved in HDAC inhibitors (Mariadason et al., 2000). A large GADD45g promoter activation by TSA. proportion of HDAC inhibitor-regulated genes encode GADD45g expression is reduced in pituitary adeno- for proteins involved in cell cycle control (Johnstone, mas, anaplastic thyroid cancer and hepatocellular 2002). In this study, we show that the murine GADD45g carcinoma, and lymphoma and nasopharyngeal carci- gene is induced by the HDAC inhibitor TSA and that noma cell lines (Zhang et al., 2002; Chung et al., 2003; the Oct and NF-Y sites present in the murine Sun et al., 2003; Ying et al., 2005) and forcing of its GADD45g proximal promoter are essential for its expression blocks proliferation of several cancer cell regulation by TSA. Mutation of individual Oct or NF- lines (Takekawa and Saito, 1998; Zhang et al., 2001, Y sites in the murine GADD45g promoter did not 2002; Chung et al., 2003; Jiang and Wang, 2004; Ying hamper its activation by TSA, suggesting that either site et al., 2005). In particular, TSA causes G1 phase arrest is sufficient to mediate activation by TSA. This inhibitor in the human T-cell leukemia cell line, Jurkat (Yokota has been previously shown to upregulate GADD45a et al., 2004). The mechanisms eliciting loss of GADD45g mRNA levels in human colon carcinoma and osteosar- expression in tumors are poorly characterized. Our coma cells (Della-Ragione et al., 2001; Chen et al., 2002; results showing that TSA treatment induces GADD45g Hirose et al., 2003) and GADD45b mRNA levels in expression in Jurkat cells, but not in ARO thyroid or human colon carcinoma cells (Chen et al., 2002). MCF-7 mammary cancer cell lines (Figure 6b Because we have not directly investigated whether and unpublished observations) suggest that histone GADD45a or GADD45b expression is induced by deacetylation might account for the mechanism involved TSA in NIH3T3 cells, we cannot rule out that this in GADD45g loss of expression in some tumors.

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1270 Furthermore, our results suggest that it is very likely GADD45gÀNFY-upper, 50-GGGCGCTGGCCAATGG that the TSA anti-proliferative effect in some cancer cell GAGTGCAG-30; lines is mediated, at least in part, by the induction of GADD45gÀNFY-lower, 50-CTGCACTCCCATTGGCCA GCGCCC-30; GADD45g expression. Our results might therefore 0 provide the basis for the development of better Control-NFY-upper, 5 -ACTTTTAACCAATCAGAAA AAT-30; therapeutic approaches based on the induction of this Control-NFY-lower, 50-ATTTTTCTGATTGGTTAAAA tumor suppressor gene. GT-30. E2F1-A-upper, 50-CTAGAGCTCTTTCGCGGCAAAAA GGAG-30 Materials and methods E2F1-A-lower, 50-GATCCTCCTTTTTGCCGCGAAAGA GCT-30 Plasmids The Control-NFY-double-stranded oligonucleotide has been A DNA fragment containing nucleotides between À818 and reported to bind specifically to NF-Y (Inoue et al., 1999). The þ 15 of the murine GADD45g gene (Balliet et al., 2003) was E2F1-A double-stranded oligonucleotide corresponds to the amplified by PCR from a genomic clone provided by Dr J Ihle distal E2F element in the E2F1 promoter (Campanero et al., (St Jude Children’s Research Hospital, Memphis, TN, USA) 2000). The Control-Oct oligonucleotide, double stranded, that and cloned into the pGL3-Basic vector (Promega, Madison, contains the binding site for Oct2A was obtained from Roche WI, USA). The resulting reporter plasmid (À818) was used to (Barcelona, Spain). The sequence of its upper strand is: 50- generate deleted versions of the promoter reporter in pGL3- GATCCAGATGCAAATACGGATCCA. The oligonucleo- Basic. The À482 and À220 reporters (nucleotides between tides were mixed at an equimolar ratio in 10 mM Tris (pH À482 or À220 and þ 15) were generated by using two internal 7.5), 50 mM NaCl, heated to 65 1C and annealed by slow PstI sites. The À112 and À54 reporters (nucleotides between cooling to room temperature (RT). Double-stranded oligonu- À112 or À54 and þ 15) were generated by PCR. Point cleotides (100 ng) were labeled with T4 polynucleotide kinase. mutations that eliminate the Oct or NF-Y sites (named Octm The antibodies used were anti-MAZ (sc-13484 Â , Santa Cruz or NFYm) were introduced in the À818 or À220 reporters by Biotechnology, Santa Cruz, CA, USA), anti-Oct-1 (05-240, using PCR or the Qickchange mutagenesis kit (Stratagene, Upstate, Charlottesville, VA, USA) and anti-NF-YB (sc- Amsterdam, The Netherlands). The upper-strand oligonucleo- 13045 , Santa Cruz Biotechnology). Binding reactions and tides used were Octm, 50-CAGGCCTGCCGCCTCACTTCATG Â gel electrophoresis were performed as described previously CAGAAAGCGGGCGCTGG-30 and NFYm, 50-AAGCGG (Campanero et al., 1999). GCGCTGGCCCATGGGAGTGCAGCC-30.Themutations correspond to the underlined sequences and their presence in the reporters was verified by sequencing. The expression vectors ChIP assay and their providers were pCG-Oct-1, Dr W Herr, Universite´ de Cells were treated with 10 mM dimethyl adipimidate and Lausanne, Switzerland; pCMX-SMRT, pCMX-HDAC1 and 0.25% dimethylsulfoxide in phosphate-buffered saline (PBS) pCMX-HDAC3, Dr RM Evans, The Salk Institute for Biological for 45 min at RT and subsequently treated with formaldehyde Studies, San Diego, CA, USA; pCX-PCAF, Dr T Kouzarides, at a final concentration of 1% for 15 min at RT. Sonication The Gurdon Institute, Cambridge, UK and pSG5-NF-YAm29, was performed to achieve an average DNA length of 200– Dr R Mantovani, Universita` di Milano, Italy. 800 bp. Crosslinked chromatin was precleared with Protein A-Sepharose beads (preincubated with sonicated salmon Cell culture, transient transfection and reporter assay sperm DNA, bovine serum albumin (BSA) and normal rabbit NIH3T3 fibroblasts were maintained in Dulbecco’s modified serum) and immunoprecipitated with Rabbit anti-mouse IgG Eagle’s medium supplemented with 10% calf serum (Sigma). (M7023, Sigma), anti-Oct-1 (05-240, Upstate) or anti-NF-YB (sc-13045 , Santa Cruz Biotechnology). Immunoprecipitated Cells (6 Â 104) were seeded in 24-well plates and transfected Â with luciferase reporter plasmids (25–125 ng), expression complexes were collected with Protein A-Sepharose beads vectors (25–200 ng), pcDNA3-b-gal, LipofectAMINE and preincubated with sonicated salmon sperm DNA and BSA. Plus reagent (Invitrogen, Prat de Llobregat, Barcelona, Spain). The beads were washed and eluted, and DNA was extracted In some experiments, NIH3T3 fibroblasts were incubated with from the eluates according to the ChIP Assay Kit protocol the HDAC inhibitor TSA (Sigma). After 24 h, cells were (Upstate). DNA was subjected to PCR using either specific primers that amplified a product corresponding to positions – harvested into 100 ml of passive lysis buffer (Promega), and 0 luciferase activity was measured according to the manufac- 310 to þ 46 of the murine GADD45g gene (5 -TTCCAGT ATCGTATCAGCAAACAAC-30 and 50-GGTAAAGATTCC turer’s instructions. The luciferase activity was normalized to 0 the protein concentration of the lysate, as determined with the CAGAGTCCGATGC-3 ) or control primers that amplified a product corresponding to positions –2943 to À2680 of the Bradford protein assay (Bio-Rad, Hercules, CA, USA), and to 0 0 the b-galactosidase activity. same gene (5 -CAAGTAGCTGGGATTCTAGGTTTTA-3 and 50-AACCCCATATTATCCGGTGGCTACA-30). The amplified DNA products were separated on 1% agarose gel Nuclear extracts and EMSA and visualized with ethidium bromide. Nuclear extracts from NIH3T3 cells or HEK293-T cells transiently transfected with either the pCG-Oct-1 or the empty expression plasmid were prepared as described previously RNA preparation and northern blot analysis (Campanero et al., 1999). EMSAs were performed with the NIH3T3 fibroblasts were incubated with the HDAC inhibitor complementary oligonucleotides (Invitrogen): TSA for 24 h and the cells were harvested and lysed in Tri- Reagent (Sigma). Total RNA was extracted according to the GADD45g-Oct-upper, 50-GCCGCCTCATTTGCATCAG manufacturer’s protocol and 20 mg were subjected to electro- AAAGC-30; phoresis in an agarose–formaldehyde gel and transferred to GADD45g-Oct-lower, 50-GCTTTCTGATGCAAATGAG Zeta-Probe membranes (Bio-Rad). Membranes were hybri- GCGGC-30; dized with the [a-32P]dCTP probe for murine GADD45g

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1271 (provided by Dr J Ihle), stripped and reprobed for g-actin Acknowledgements mRNA. We are grateful to Dr RM Evans, Dr T Kouzarides, Dr R Cell cycle analysis Mantovani and Dr W Herr for providing plasmids. We thank Asynchronous NIH3T3 cell cultures were exposed to TSA for Dr AC Carrera for helpful suggestions and M de la Fuente for 24 h and harvested from culture dishes. After washing with excellent technical assistance. This work was supported by the PBS, cells were fixed in 70% ethanol, treated with RNAse A Spanish Instituto de Salud Carlos III RTICCC (FIS03 C03/10) and the nuclei were stained with propidium iodide. Their to VC and Spanish Ministerio de Educacion y Ciencia fluorescence was analysed by flow cytometry using a FACScan (SAF04-05888) and Spanish Instituto de Salud Carlos III (Becton Dickinson). (C03/10 and G03/179) to MRC.

References

Balliet AG, Hollander MC, Fornace Jr AJ, Hoffman B, Liebermann Huang W, Zhao S, Ammanamanchi S, Brattain M, Venkatasubbarao DA. (2003). Comparative analysis of the genetic structure and K, Freeman JW. (2005). Trichostatin A induces transforming chromosomal mapping of the murine GADD45g/CR6 gene. DNA growth factor beta type II receptor promoter activity and acetyla- Cell Biol 22: 457–468. tion of Sp1 by recruitment of PCAF/p300 to a Sp1. NF-Y complex. Beadling C, Johnson KW, Smith KA. (1993). Isolation of interleukin J Biol Chem 280: 10047–10054. 2-induced immediate-early genes. Proc Natl Acad Sci USA 90: Inoue T, Kamiyama J, Sakai T. (1999). Sp1 and NF-Y synergistically 2719–2723. mediate the effect of vitamin D(3) in the p27(Kip1)gene Berger SL. (2002). Histone modifications in transcriptional regulation. promoter that lacks vitamin D response elements. J Biol Chem Curr Opin Genet Dev 12: 142–148. 274: 32309–32317. Brockman JL, Schuler LA. (2005). Prolactin signals via Stat5 and Jiang F, Wang Z. (2004). GADD45g is androgen-responsive and Oct-1 to the proximal cyclin D1 promoter. Mol Cell Endocrinol growth-inhibitory in prostate cancer cells. Mol Cell Endocrinol 239: 45–53. 213: 121–129. Campanero MR, Armstrong M, Flemington E. (1999). Distinct Jin S, Fan F, Fan W, Zhao H, Tong T, Blanck P et al. (2001). cellular factors regulate the c-myb promoter through its E2F Transcription factors Oct-1 and NF-YA regulate the p53-indepen- element. Mol Cell Biol 19: 8442–8450. dent induction of the GADD45 following DNA damage. Oncogene Campanero MR, Armstrong M, Flemington E. (2000). CpG methyla- 20: 2683–2690. tion as a mechanism for the regulation of E2F activity. Proc Natl Jin S, Tong T, Fan W, Fan F, Antinore MJ, Zhu X et al. (2002). Acad Sci USA 97: 6481–6486. GADD45-induced cell cycle G2-M arrest associates with altered Caretti G, Salsi V, Vecchi C, Imbriano C, Mantovani R. (2003). subcellular distribution of cyclin B1 and is independent of p38 Dynamic recruitment of NF-Y and histone acetyltransferases on kinase activity. Oncogene 21: 8696–8704. cell-cycle promoters. J Biol Chem 278: 30435–30440. Johnstone RW. (2002). Histone-deacetylase inhibitors: novel drugs for Cartharius K, Frech K, Grote K, Klocke B, Haltmeier M, Klingenhoff A the treatment of cancer. Nat Rev Drug Discov 1: 287–299. et al. (2005). MatInspector and beyond: promoter analysis based on Jung N, Yi YW, Kim D, Shong M, Hong SS, Lee HS et al. (2000). transcription factor binding sites. Bioinformatics 21: 2933–2942. Regulation of GADD45g expression by C/EBP. Eur J Biochem 267: Chen Z, Clark S, Birkeland M, Sung CM, Lago A, Liu R et al. (2002). 6180–6187. Induction and superinduction of growth arrest and DNA damage Kakizawa T, Miyamoto T, Ichikawa K, Takeda T, Suzuki S, gene 45 (GADD45) alpha and beta messenger RNAs by histone Mori J et al. (2001). Silencing mediator for retinoid and deacetylase inhibitors trichostatin A (TSA) and butyrate in SW620 thyroid hormone receptors interacts with octamer transcription human colon carcinoma cells. Cancer Lett 188: 127–140. factor-1 and acts as a transcriptional repressor. J Biol Chem 276: Chung HK, Yi YW, Jung NC, Kim D, Suh JM, Kim H et al. (2003). 9720–9725. GADD45gamma expression is reduced in anaplastic thyroid cancer Kojima S, Mayumi-Matsuda K, Suzuki H, Sakata T. (1999). and its reexpression results in apoptosis. J Clin Endocrinol Metab 88: Molecular cloning of rat GADD45gamma, gene induction and its 3913–3920. role during neuronal cell death. FEBS Lett 446: 313–317. Constance CM, Morgan IV JI, Umek RM. (1996). C/EBPalpha Liebermann DA, Hoffman B. (2003). Myeloid differentiation (MyD) regulation of the growth-arrest-associated gene GADD45. Mol Cell primary response genes in hematopoiesis. Blood Cells Mol Dis 31: Biol 16: 3878–3883. 213–228. Della Ragione F, Criniti V, Della Pietra V, Borriello A, Oliva A, Lu B, Ferrandino AF, Flavell RA. (2004). GADD45beta is important Indaco S et al. (2001). Genes modulated by histone acetylation as for perpetuating cognate and inflammatory signals in T cells. Nat new effectors of butyrate activity. FEBS Lett 499: 199–204. Immunol 5: 38–44. De Smaele E, Zazzeroni F, Papa S, Nguyen DU, Jin R, Jones J et al. Lu B, Yu H, Chow C, Li B, Zheng W, Davis RJ et al. (2001). (2001). Induction of GADD45beta by NF-kappaB downregulates GADD45gamma mediates the activation of the p38 and JNK MAP pro-apoptotic JNK signalling. Nature 414: 308–313. kinase pathways and cytokine production in effector TH1 cells. Fan W, Richter G, Cereseto A, Beadling C, Smith KA. (1999). Immunity 14: 583–590. Cytokine response gene 6 induces p21 and regulates both cell growth Mantovani R. (1999). The molecular biology of the CCAAT-binding and arrest. Oncogene 18: 6573–6582. factor NF-Y. Gene 239: 15–27. Hinkley C, Perry M. (1992). Histone H2B gene-transcription during Mantovani R, Li XY, Pessara U, Hooft van Huisjduijnen R, Benoist Xenopus early development requires functional co-operation be- C, Mathis D. (1994). Dominant negative analogs of NF-YA. J Biol tween proteins bound to the CCAAT and octamer motifs. Mol Cell Chem 269: 20340–20346. Biol 12: 4400–4411. Marhin WW, Chen S, Facchini LM, Fornace Jr AJ, Penn LZ. Hirose T, Sowa Y, Takahashi S, Saito S, Yasuda C, Shindo N et al. (1997). Myc represses the growth arrest gene GADD45. Oncogene (2003). p53-independent induction of GADD45 by histone deace- 14: 2825–2834. tylase inhibitor: coordinate regulation by transcription factors Oct-1 Mariadason JM, Corner GA, Augenlicht LH. (2000). Genetic and NF-Y. Oncogene 22: 7762–7773. reprogramming in pathways of colonic cell maturation induced by Hollander MC, Sheikh MS, Bulavin DV, Lundgren K, Augeri- short chain fatty acids: comparison with trichostatin A, sulindac, Henmueller L, Shehee R et al. (1999). Genomic instability in and curcumin and implications for chemoprevention of colon GADD45a-deficient mice. Nat Genet 23: 176–184. cancer. Cancer Res 60: 4561–4572.

Oncogene Induction of GADD45c by HDAC inhibitor MR Campanero et al 1272 Nagy L, Kao HY, Chakravarti D, Lin RJ, Hassig CA, Ayer DE et al. Takekawa M, Saito H. (1998). A family of stress-inducible GADD45- (1997). Nuclear receptor repression mediated by a complex contain- like proteins mediate activation of the stress-responsive MTK1/ ing SMRT, mSin3A, and histone deacetylase. Cell 89: 373–380. MEKK4 MAPKKK. Cell 95: 521–530. Nakayama K, Hara T, Hibi M, Hirano T, Miyajima A. (1999). A Vairapandi M, Balliet AG, Hoffman B, Liebermann DA. (2002). novel oncostatin M-inducible gene OIG37 forms a gene family with GADD45b and GADD45g are cdc2/cyclinB1 kinase inhibitors with MyD118 and GADD45 and negatively regulates cell growth. J Biol a role in S and G2/M cell cycle checkpoints induced by genotoxic Chem 274: 24766–24772. stress. J Cell Physiol 192: 327–338. Osborne A, Zhang H, Yang WM, Seto E, Blanck G. (2001). Histone Yang J, Zhu H, Murphy TL, Ouyang W, Murphy KM. (2001). IL- deacetylase activity represses gamma interferon-inducible HLA-DR 18-stimulated GADD45 beta required in cytokine-induced, gene expression following the establishment of a DNase I-hypersen- but not TCR-induced, IFN-gamma production. Nat Immunol sitive chromatin conformation. Mol Cell Biol 21: 6495–64506. 2: 157–164. Park SH, Lee SR, Kim BC, Cho EA, Patel SP, Kang HB et al. (2002). Ying J, Srivastava G, Hsieh WS, Gao Z, Murray P, Liao SK et al. Transcriptional regulation of the transforming growth factor beta (2005). The stress-responsive gene GADD45g is a functional tumor type II receptor gene by histone acetyltransferase and deacetylase is suppressor, with its response to environmental stresses frequently mediated by NF-Y in human breast cancer cells. J Biol Chem 277: disrupted epigenetically in multiple tumors. Clin Cancer Res 11: 5168–5174. 6442–6449. Peng Y, Jahroudi N. (2003). The NFY transcription factor inhibits von Yokota T, Matsuzaki Y, Miyazawa K, Zindy F, Roussel MF, Sakai T. Willebrand factor promoter activation in non-endothelial cells (2004). Histone deacetylase inhibitors activate INK4d gene through through recruitment of histone deacetylases. J Biol Chem 278: Sp1 site in its promoter. Oncogene 23: 5340–5349. 8385–8394. Zhan Q, Bae I, Kastan MB, Fornace Jr AJ. (1994). The p53-dependent Smith ML, Ford JM, Hollander MC, Bortnick RA, Amundson SA, gamma-ray response of GADD45. Cancer Res 54: 2755–2760. Seo YR et al. (2000). p53-mediated DNA repair responses to UV Zhang W, Bae I, Krishnaraju K, Azam N, Fan W, Smith K et al. radiation: studies of mouse cells lacking p53, p21, and/or GADD45 (1999). CR6: A third member in the MyD118 and GADD45 gene genes. Mol Cell Biol 20: 3705–3714. family which functions in negative growth control. Oncogene 18: Sun L, Gong R, Wan B, Huang X, Wu C, Zhang X et al. (2003). 4899–48907. GADD45gamma, down-regulated in 65% hepatocellular carcinoma Zhang W, Hoffman B, Liebermann DA. (2001). Ectopic expression of (HCC) from 23 Chinese patients, inhibits cell growth and induces MyD118/GADD45/CR6 (GADD45beta/alpha/gamma) sensitizes cell cycle G2/M arrest for hepatoma Hep-G2 cell lines. Mol Biol Rep neoplastic cells to genotoxic stress-induced apoptosis. Int J Oncol 30: 249–253. 18: 749–757. Takahashi S, Saito S, Ohtani N, Sakai T. (2001). Involvement of the Zhang X, Sun H, Danila DC, Johnson SR, Zhou Y, Swearingen B Oct-1 regulatory element of the GADD45 promoter in the et al. (2002). Loss of expression of GADD45 gamma, a growth p53-independent response to ultraviolet irradiation. Cancer Res 61: inhibitory gene, in human pituitary adenomas: implications for 1187–1195. tumorigenesis. J Clin Endocrinol Metab 87: 1262–12627.

Oncogene