ANTICANCER RESEARCH 25: 4287-4292 (2005)

Expression Analysis of a Subset of Coregulators and Three Nuclear Receptors in Human Colorectal Carcinoma

ROMINA GIANNINI and ALDO CAVALLINI

Laboratory of Biochemistry, National Institute for Digestive Diseases, I.R.C.C.S. "Saverio de Bellis", 70013-Castellana Grotte (BA), Italy

Abstract. In human colorectal tissue samples, the Coregulators are that interact with ERs in a expressions of 4 coactivators, p300, pCAF, TIF-2 and TRAP ligand-dependent manner and assemble in a multi- 220, and 7 corepressors, N-CoR, REA, MTA1, MTA1L1, complex, thereby facilitating chromatin remodelling and HDAC1, HDAC2 and HDAC3, linked to estrogen receptors regulating . The coregulators are (ER), were revealed by traditional RT-PCR. Cofactors ER·, functionally divided into two subclasses: the coactivators ER‚ and ERR· mRNA levels were then measured in 40 that enhance the ER transcriptional mechanism by their tumor tissue samples matched with respective normal mucosa histone acetyltransferase (HAT) activity, and the by real-time PCR. The decline of mRNA levels of all corepressors that have histone deacetylase (HDAC) coactivators and the increase of NCoR, HDAC1, HDAC2 activity, which decreases or inhibits the ER transcriptional and MTA1 were observed from normal to tumor tissue, mechanism (6). Therefore, changes in the coregulator whereas REA, HDAC3 and MTA1L1 expressions were pattern or expression levels can affect the transcriptional similar in both tissue compartments. The gene expression of activity of the estrogens and, hence, cause disorders of the ER‚ correlated with those of p300, TIF-2 and REA in ER target tissues (7). normal mucosa, and with that of REA in tumor tissue only. The estrogen -related receptor · (ERR·), No association was found between ER· and coregulators and another related to the ER·, belongs to a between each coregulator and different clinical parameters. class of nuclear orphan receptors since no natural ligands Our findings suggest that the co-induction of ER‚ and some have been discovered for it (8). The ERR·, as well as ERs, cofactors may play an important role during the development recruit several cofactors in the modulation of the ER- of human colorectal carcinoma. activated transcriptional activity (9). Considerable progress has been made in deciphering the The genomic action of the estrogens is mediated by two molecular mechanisms of the coregulators from a variety (ER) subtypes, ER· and ER‚, that of in vitro studies, but there are few studies that have regulate growth, development and differentiation of explored the correlations between ERs and their cofactors different human tissues (1). in human tissues (10-12), including human colorectal The definitive role of ERs in human colorectal mucosa (13). physiology and during malignant transformation has not Therefore, in the first phase of our study, the presence yet been well clarified, but recent data suggest that ER‚, of a series of coregulators was investigated at the mRNA rather than ER·, may play a key role in the regulation of level in several human colorectal tissue samples from 16 growth and differentiation of neoplastic tissue (2-5). patients by the traditional reverse phase and amplification (RT-PCR) method. Since many cofactors bind the hormone-activated ERs, here, 4 coactivators [p300, pCAF (p300/CBP-Associated Factor), TIF2 (Transcriptional Correspondence to: Dr. Aldo Cavallini, Laboratory of Biochemistry, Intermediary Factor 2), TRAP 220 (Thyroid hormone National Institute of Digestive Diseases, I.R.C.C.S. "Saverio De Receptor Associated Protein)], 7 corepressors [N-CoR Bellis", Via Della Resistenza, nc; 70013 Castellana Grotte (BA), (Nuclear Receptor Corepressor), REA (Repressor of Italy. Tel: +39-080-4960342, Fax: +39-080-4960313, e-mail: Estrogen receptor Activity), MTA1 (Metastasis Tumor [email protected] Associated protein-1) MTA1L1 (Metastasis Tumor Key Words: Coregulators, estrogen receptor ·, estrogen receptor ‚, Associated 1-like 1)] and 3 histone deacetylases [(HDAC1, estrogen receptor-related receptor ·, colorectal carcinoma. HDAC2, HDAC3)] related to ERs (7, 14) were chosen.

0250-7005/2005 $2.00+.40 4287 ANTICANCER RESEARCH 25: 4287-4292 (2005)

Table I. Primer sequences, PCR products and published gene sequences. forward/reverse (f/r) Primers PCR product GenBank

Coactivators

TIF-2 (f) 5'-TAATGCACAGATGCTGGCC-3' TIF-2 (r) 5'-TCTGTGTATGTGCCATTCGG-3' 314 bp NM 006540 p300 (f) 5'-GGACCGCTTTGTCTACACCT-3' p300 (r) 5'-CCAGAAGATGAAGCGGGTTG-3' 305 bp NM 001429 TRAP 220 (f) 5'-GGGAGGACTTCAGCAAGGTGTCTCA-3' TRAP 220 (r) 5'-AGATTACCACCTGAGAAACCAAAGC-3' 331 bp NM 004774 pCAF (f) 5'-AGAACATTGCTTCGCTCGG-3' pCAF (r) 5'-TGCCTCAAGTCCAGAAGAGG-3' 345 bp U 57317

Corepressors REA (f) 5'-CTTGAAGGACTTGGCGGGACG-3' REA (r) 5'-CCATCTGTAGGTCTTTGGAGCCTGT-3' 269 bp AF 150962 NCoR (f) 5'-GCTGATGAGGATGTGGATGG-3' NCoR (r) 5'-TTGGACTCTTGGATGTTGCC-3' 349 bp AF 044209 MTA1 (f) 5'-TGGCCGACAAGCACGCAACC-3' MTA1 (r) 5'-GGAGCGGGAGGATTTCTTCT-3' 270 bp NM 004689 MTA1L1 (f) 5'-TTTGAATCATTACCAGCCACCC-3' MTA1L1 (r) 5'-AGTTTCTTGTGGTGGCCCGAGC-3' 317 bp NM 004739 HDAC1 (f) 5'-TATCGCCCTCACAAAGCCAATG-3' HDAC1 (r) 5'-GCATCTGGCTTCTGTTACGTCA-3' 301 bp NM 004964 HDAC2 (f) 5'-GAATCCGCATGACCCATAACTT-3' HDAC2 (r) 5'-AGATTGTCCAGCGTTTGATGGA-3' 215 bp NM 001527 HDAC3 (f) 5'-GCACCATGCCAAGAAGTTTGA-3' HDAC3 (r) 5'-CGCTACTACTGTCTGAACGTGCC-3' 266 bp NM 003883

The primers used to detect ER·, ER‚, ERR· and ‚-actin primers mRNAs have been published previously (3, 15). The abbreviation for each gene is explained in the text.

Subsequently, the mRNA expression levels of the same localized in the right colon, 7 in the left colon and 20 in the series of cofactors and nuclear receptors (ER·, ER‚, sigma-rectum. Histologically, 5 tumors were well-differentiated, ERR·) were measured in colorectal tumor tissue samples 19 moderately-differentiated and 15 poorly-differentiated; 18, 14 and 8 tumors were considered to be stage II (TNM II), stage III matched with adjacent normal mucosa obtained from 40 (TNM III) and stage IV (TMN IV), respectively. patients by real-time PCR (QPCR). The ER and ERR· The quality of the tumor and surrounding normal tissue mRNA levels were correlated with those of the cofactors specimens and the relative cellular composition were determined and, then, the coregulator mRNA levels were correlated by histopathological assessment. within them. Finally, the possible associations between All tissue specimens were stored at –80ÆC until evaluation of each cofactor expression and clinical or pathological the mRNA. parameters (age, sex, histological grade, TNM stage and Qualitative and quantitative PCR analysis. Before the PCR tumor side) were also investigated. quantitative analysis, conventional RT-PCR analysis was performed to investigate the presence of 11 cofactor mRNAs in Patients and Methods the colorectal tissue samples of 16 patients. The RNA extraction and RT-PCR method have been Patients and tissue samples. A total of 40 patients from a large described previously (3). In brief, 5 Ìl of the RT reaction, primary list of patients operated consecutively in our Institute was corresponding to 0.5 Ìg of total RNA, were subjected to the PCR enrolled. The inclusion criteria were: (a) a negative anamnesis for in 25 Ìl of final volume for 45 cycles. The thermocycler program familial colorectal carcinoma, (b) females patients >59 years old was: 30 sec at 95ÆC and then 10 sec at 94ÆC, 10 sec at 55ÆC and (women in pre- and perimenopausal period were excluded), (c) 30 sec at 72ÆC. The PCR products were electrophoresed in 2% no chemo- or radiotherapy before tissue collection, and (d) no agarose gel, stained with ethidium bromide, and the PCR bands hormonal therapy in female patients. were then visualized by UV. The clinical parameters of the patients and the tumor The identity of each PCR product was confirmed by complete characteristics were as follows: 27 males (median age 71 years, sequencing of the purified products. The primers used are range 60-82), 13 females (66 years, range 56-76); 13 tumors were illustrated in Table I.

4288 Giannini and Cavallini: ER Coregulators in Human Colorectal Cancer

Figure 1. Schematic representation of 11 cofactor mRNAs in human colorectal tissues. Four coactivators (p300, pCAF, TIF-2, TRAP 220) and 7 corepressors (NCoR, REA, HDAC1, HDAC2, HDAC3, MTA1, MTA1L1) mRNAs were found in the colorectal tumor tissue (T) and adjacent normal mucosa (N) samples of 16 patients. After total RNA extraction and reverse transcriptase reaction, aliquots of cDNA (corresponding to 0.5 Ìg of total RNA) were subjected to the traditional PCR method. The PCR bands were then visualized by 2% gel-electrophoresis and photographed. The identity of each PCR product was confirmed by complete sequencing of the purified products. The primer sequences and expected size of each PCR product are shown in Table I. M=Molecular weight marker (Marker IX, Roche Diagnostics, Monza, Italy). The abbreviations for the are explained in the text.

The QPCR methodology was essentially the same as the conventional RT-PCR described above. However, the master mix containing SYBR Green I was used (iQ SYBR Green Supermix, BioRad, Milan, Italy) and a further melt-curve step (melt-curve started at 55ÆC and the temperature was increased 0.5ÆC per cycle for 80 cycles) was added in the thermocycler program at the end of the last extension phase of 45 cycles to control the presence of the secondary non-specific PCR products or primer- dimer products that can severely reduce the amplification efficiency and sensitivity. Both qualitative and quantitative PCR reactions were performed by iCycler Thermal Cycler System apparatus (BioRad). For quantitative analysis, the external calibration curves, one for each target gene examined and one for the ‚-actin gene, as the control gene, were used. Each calibration curve was performed using the amplicons of the correspondent gene transcript obtained from the conventional RT-PCR. With this approach, after purification ( microClean kit; West Sussex, UK) and UV quantitation, serial dilutions of known copy number of amplicons (102-107 molecules for each PCR reaction) were used to generate a single standard curve. For each tissue sample, the absolute concentration of an unknown target gene expression was divided by that of the endogenous ‚-actin expression. This ratio expressed the relative mRNA level of each target gene. Figure 2. The distribution of ER, ER‚, ERR· and 11 coregulator mRNA levels in human colorectal tissue samples of 40 patients. The mRNA Statistical analysis. The relative gene expression levels between expression of ER·, ER‚, ERR·, 4 coactivators (p300, pCAF, TIF-2, TRAP tumor tissue and surrounding normal mucosa were tested using 220) and 7 corepressors (NCoR, REA, HDAC1, HDAC2, HDAC3, MTA1, the two-tailed Wilcoxon signed-rank. MTA1L1) were quantified by real-time PCR (QPCR) and expressed as target The Spearman’s rank correlation was used to study the gene/‚-actin ratio. Data are evaluated by two-tailed Wilcoxon signed-rank relationships between two different gene mRNA levels, and the test and presented as means±SEM. The white and black columns represent non-parametric Kruskal-Wallis test was used for the associations the distribution of each genes’ mRNA levels in normal and pathological between each gene expression level and clinical or tumor tissues, respectively. A. The ER· , ER‚, ERR· mRNA expression levels (1, parameters. p=0.019; 2, p=0.029; 3, p=0.009). B. The 11 coregulator mRNA expression The relative mRNA levels were expressed as means±SEM, levels (1, p=0.048; 2, p=0.047; 3, p=0.013; 4, p=0.040; 5, p=0.029; 6, and two-sided p≤0.05 was considered as statistically significant. p=0.024). The abbreviations for the genes are explained in the text.

Results

RT-PCR analysis. Conventional RT-PCR analysis Results of the QPCR optimization. The real-time PCR confirmed the presence of all cofactors examined. The efficiencies of the standard curves were obtained from the results obtained are shown in Figure 1. given slopes using the formula: E=10(-1/slope) (iQ Thermal

4289 ANTICANCER RESEARCH 25: 4287-4292 (2005)

Cycler software, BioRad). The PCR efficiencies, expressed b) Colorectal tumor tissue: In this tissue compartment, in percent, ranged from 92% to 98% for nuclear receptors, the ER‚ expression correlated only with that of the cofactors and ‚-actin (data not shown). corepressor REA (r=+0.61, p=0.004), whereas the ER· To confirm the accuracy and reproducibility of the or ERR· mRNA levels and each cofactor expression were QPCR, the intra-assay precision, determined in three PCR not significant. Finally, an association between HDAC1 runs of the same RNA sample, was <2.5%. Inter-assay and HDAC2 expressions (r=+0.59, p=0.032) was variation, investigated in 3 different experimental runs observed. performed on 3 days using the aliquots of a pool of the total The gene expression levels among patient subgroups RNA, was <3.7%. defined by clinical and pathological parameters (age, sex, tumor side, histological differentiation and TMN grade) Expression analysis of 11 cofactors and 3 nuclear receptors did not show any association (data not shown). in colorectal tissue samples by QPCR. The distributions of the nuclear receptor and cofactor mRNA levels, Discussion respectively, in tumor tissues and in adjacent normal mucosa of a well-defined cohort of 40 patients, are shown The transcriptional activity of the ERs and other nuclear in Figures 2A and 2B. receptor superfamily members, such as ERR·, is The ER· and ER‚ expressions were lower in malignant modulated by different coregulator proteins. tissue samples than in non-malignant epithelium (p=0.019 Although we can begin to understand the critical role of and p=0.029, respectively), whereas the ERR· expression ER-mediated signaling in the development and was found to be higher in the colorectal tumor tissue than in progression of human colorectal carcinoma, the role of normal mucosa (p=0.009). These data are in accordance to altered expression of the ER coregulators is still elusive for those previously published (15). this tissue. For this reason, our first set of experiments was Among the coactivators, the pCAF, TIF-2, p300 and designed to elucidate, at the mRNA level, whether a TRAP-220 mRNA levels were lower in cancerous tissue than subset of the cofactors was present in human colorectal in adjacent normal mucosa (p=0.048, p300; p=0.047, pCAF; tissue, as already found in other human tissues. p=0.013, TIF-2; p=0.040, TRAP-220). Among the The qualitative PCR analysis revealed the presence of corepressors, the mRNA expression levels of NCoR, all cofactors examined. The presence of these cofactors in HDAC1, HDAC2 and MTA1 were higher in the carcinomas the colorectal tissue as well suggests that the cofactors can than in the normal colorectal mucosa, but statistical be considered ubiquitous in the human organism. significance was reached only for HDAC2 and MTA1 mRNA However, the changes in the expression levels of these levels (p=0.029, p=0.024, respectively). cofactors along with the changes in the ER expressions The expression profile of the other corepressors (REA might be important in determining the tissue-specific HDAC3 and MTA1L1) did not change between malignant response to a particular steroid hormone (10). Therefore, and normal colorectal tissues. in the second series of experiments, the expression levels of the ER cofactors and 3 nuclear receptors were Relationships within gene expression levels. The possible screened in human colorectal tumor tissue matched with relationships between coregulator mRNA levels and those of adjacent normal mucosa samples obtained from 40 ER·, ER‚ and ERR·, either in tumor tissue or in adjacent patients. Even if the small sample number (n=40) of normal mucosa, were analyzed. patients may limit the statistical power of the analyses, a) Colorectal normal mucosa: The ER‚ expression was we have demonstrated the significant overexpression of significantly related with TIF-2 (r=+0.64; p=0.006), p300 ERR· and the up-regulation of 2 ERs from normal to (r=+0.56; p=0.0042) and REA (r=+0.44, p=0.047) tumor tissue, as previously found with a higher number expressions. An inverse correlation was observed between of patients (15). ERR· and TRAP 220 (r=–0.51, p=0.044) mRNA levels, and The same QPCR analysis also showed the decline of the no significant associations were observed between ER· and coactivator and the increase of the corepressor mRNA each coregulator gene expression. levels from normal to tumor tissue and a similar amount of In addition, the following correlations were observed some corepressors in both tissue compartments. These data between coactivators and corepressors: TIF-2 (coactivator) suggest a different role of the coactivators and corepressors and REA (corepressor; r=+0.53, p=0.030), TIF-2 and in normal and in tumor colorectal tissues, respectively, and HDAC1 (r=–0.73, p=0.020) and a trend toward significance it seems that the coactivator action predominates in the was observed between TIF-2 and HDAC2 (r=–0.58, normal tissue compartment, whereas that of the p=0.090). A positive correlation was also found between corepressors in pathological tissue. However, the validity of HDAC2 and HDAC1 mRNA levels (r=+0.60, p=0.040). this hypothesis should be tested in more extensive

4290 Giannini and Cavallini: ER Coregulators in Human Colorectal Cancer experiments at the protein level or in in vitro experiments. may influence the transcriptional machinery of the ER Despite the small number of cofactors examined, target genes or may play an important role in the approximately 50 cofactors have been discovered in human development and progression of human colorectal tissues, and the following associations within gene carcinoma. expression have been found. The ER‚ correlated with TIF2 and p300 (coactivators) and with REA (corepressors) in Acknowledgements normal tissue, whereas the tumor TIF2 expression was associated only to tumor REA expression. Since the ER‚ This research was supported by a grant of the Ministero della transcriptional activity can be modified by selective ER Salute (Prog. Fin. NÆ 132/2002), Italy. modulators (coactivators and/or corepressors), our data suggest that the ER target gene expression levels may be References modulated by a common regulatory pathway that expresses a balance between enhancement and suppressive effects. 1 Nilsson S, Makela S, Treuter E, Tujague M, Thomsen J, This is in accordance with other authors (14) who, for the Andersson G, Enmark E, Pettersson K, Warner M and Gustafsson J A: Mechanism of estrogen action. Physiol Rev 81: first time, suggested a synergic role between ERs and 1535-1565, 2001. cofactors in the development and progression of human 2 Di Leo A, Messa C, Cavallini A and Linsalata M: Estrogen and breast cancer. colorectal cancer. Curr Drug Targets-Immune Endocrine In our study, the HDACs seem not to be involved in Metab Disor 1: 1-12, 2001. the regulation of the ER target genes in the normal 3 Cavallini A, Messa C, Pricci M, Caruso ML, Barone M and colorectal tissue compartment, since we found an inverse Di Leo A: Distribution of estrogen receptor subtypes, correlation between TIF2 and HDAC1 or HDAC2 expression of their variant forms, and clinicopathological characteristics of human colorectal cancer. Dig Dis Sci 47: expressions. The corepressors, such as HDACs, contain 2720-2728, 2002. the histone deacetyltransferase activity that maintains the 4 Konstantinopoulos PA, Kominea A, Vandoros G, Sykiotis GP, ER target genes in an inactive state, and this condition is Andricopoulos P, Varakis I, Sotiropoulou-Bonikou G and generally more expressed in pathological than in normal Papavassilou AG: Oestrogen receptor beta (ER‚) is abundantly tissues (16). Another association was found between expressed in normal colonic mucosa, but declines in colon HDAC1 and HDAC2 mRNA levels, either in the tumor adenocarcinoma paralleling the tumour’s dedifferentiation. Eur tissue or in the surrounding normal mucosa. It has been J Cancer 39: 1251-1258, 2003. 5 Yi P, Driscoll MD, Huang J, Bhagat S, Hilf R, Bambara RA shown recently that these 2 HDACs have a high degree and Muyan M: The effects of estrogen-responsive element- and of homology and coexist within the same protein ligand-induced structural changes on the recruitment of complexes on the DNA structure during the translation cofactors and transcriptional responses by ER alpha and ER process, even if the individual HDAC might function in beta. Mol Endocrinol 16: 674-693, 2002. a distinct manner (17). 6 Klinge CM: Estrogen receptor interaction with co-activators The lack of association between ER· expression and and co-repressors. Steroids 65: 227-251, 2000. cofactors suggests that the ER‚, rather than ER·, is the 7 Baek SH and Rosenfeld MG: Nuclear receptor coregulators: their modification codes and regulatory mechanism by principal ER subtype in the estrogenic transcriptional translocation. Biochem Biophys Res Commun 319: 707-714, activity in the human colorectal tract, but it is also possible 2004. that other cofactors are involved in the activated-ER· 8 Horard B and Vanacker J-M: Estrogen receptor-related activity. The ER· and ER‚ are not equally potent receptors: orphan receptors desperately seeking a ligand. J Mol transcriptional activators on the chromatin structure, and Endocrinol 31: 349-357, 2003. although both ER· and ER‚ bound equally well to an 9 Giguere V: To ERR in estrogen pathway. Trends Endocrinol ERE (Estrogen Responsive Element) sequence present in Metab 13: 220-225, 2002. 10 Girault I, Lerebours F, Amarir S, Tozlu S, Tubiana-Hulin M, the promoter of the ER target genes, the differences Lidereau R and Bieche I: Expression analysis of estrogen between the 2 ERs in their transcriptional activities may receptor · coregulators in breast carcinoma: evidence that be due to a different receptor-dependent chromatin NcoR1 expression is predictive of the response to tamoxifen. remodeling (18). Clin Cancer Res 9: 1259-1266, 2003. Finally, since there is a lack of information on the 11 Vienonen A, Miettinen S, Blauer M, Martikainen PM, Tomas functional role of ERR· in human colorectal mucosa, the E, Heinonen PK and Ylikomi T: Expression of nuclear importance of the inverse association between ERR· and receptors and cofactors in human endometrium and myometrium. J Soc Gynecol Investig 11: 104-112, 2004. TRAP 220 mRNA levels in normal tissue is still unknown. 12 Kurebayashi J, Otsuki T, Kunisue H, Tanaka K, Yamamoto S In conclusion, our study, the first exploratory analysis on and Sonoo H: Expression levels of estrogen receptor-·, the coregulator expression in the human colorectal tract, estrogen receptor-‚, coactivators, and corepressors in breast revealed a co-induction of ER‚ and several cofactors that cancer. Clin Cancer Res 6: 512-518, 2000.

4291 ANTICANCER RESEARCH 25: 4287-4292 (2005)

13 Toh Y, Oki E, Oda S, Tokunaga E, Ohno S, Maehara Y, 17 Huang BH, Laban M, Leung CH, Lee L, Lee CK, Salto-Tellez M, Nicolson GL and Sugimachi K: Overexpression of the MTA1 Raju GC and Hooi SC: Inhibition of gene in gastrointestinal carcinomas: correlation with invasion increases apoptosis and p21(Cip1/WAF1) expression, independent and metastasis. Int J Cancer 74: 459-463, 1997. of histone deacetylase 1. Cell Death Differ 12: 395-404, 2005. 14 Shang Y, Hu X, DiRenzo J, Lazar MA and Brown M: Cofactor 18 Cheung E, Schwabish MA and Kraus WL: Chromatin exposes dynamics and sufficiency in estrogen receptor-regulated intrinsic differences in the transcriptional activities of estrogen transcription. Cell 103: 843-852, 2000. receptors · and ‚. EMBO J 22: 600-611, 2003. 15 Cavallini A, Notarnicola M, Giannini R, Montemurro S, Lorusso D, Visconti A, Minervini F and Caruso MG: Oestrogen receptor- related receptor alpha (ERR·) and oestrogen receptors (ER· and ER‚) exhibit different gene expression in human colorectal tumour progression. Eur J Cancer 41: 1487-1494, 2005. 16 Kawai H, Li H, Avraham S, Jiang S and Avraham HK: Overexpression of histone deacetylase HDAC1 modulates breast cancer progression by negative regulation of estrogen Received May 9, 2005 receptor alpha. Int J Cancer 107: 353-358, 2003. Accepted July 11, 2005

4292