The Cohesin SMC3 Is a Target the for (cid:1) -Catenin/TCF4 Transactivation Pathway*

The structural maintenance of chromosome protein SMC3 is a component of the cohesin complex that medi-ates sister chromatid cohesion and segregation in pro-karyotes and eukaryotes. It is also present extracellularly in the form of a chondroitin sulfate proteoglycan known as bamacan. We have found previously that SMC3 expression is elevated in a large fraction of human colon carcinomas. The additional finding that the protein is significantly increased in the intestinal polyps of Apc Min/ (cid:2) mice has led us to hypothesize that SMC3 expression is linked to activation of the APC/ (cid:1) -catenin/ TCF4 pathway. The immunohistochemical analysis of colon adenocarcinomas from clinical specimens revealed that (cid:1) -catenin and SMC3 antigens co-localize with maximal stain intensity within the transformed areas. Cloning and sequencing of 1578 bp of the human SMC3 promoter unveiled the presence of seven putative consensus sequences for (cid:1) -catenin/TCF4 binding, two of which are conserved in the mouse Smc3 promoter. Transient transfection experiments in HCT116 and SW480 human colon carcinoma cells using deletion and mutated promoter constructs in luciferase reporter vectors confirmed that the putative sites, the first located at (cid:3) 48 bp and the second located at (cid:3) several conserved (cid:1) -catenin/TCF4 binding consensus sequences that are responsible for the transactivation of the gene. We found that SMC3 promoter transcriptional activity is increased by elevated (cid:1) -catenin levels and suppressed by a lowered (cid:1) -catenin level. Immunohistochemical analysis of (cid:1) -catenin and SMC3 in serial sections of colon from patients with invasive adenocarcinoma and of the intestinal adenomas from Apc Min/ (cid:1) mice revealed that the two antigens co-localize with maximal stain intensity at the tumoral sites. The results are consistent with the idea that SMC3 is a target for the (cid:1) -catenin/TCF4 transactivation pathway.

From the Departments of ‡Pathology and Cell Biology and ¶Microbiology and Immunology, Thomas Jefferson University, Philadelphia, Pennsylvania 19107 The structural maintenance of chromosome protein SMC3 is a component of the cohesin complex that mediates sister chromatid cohesion and segregation in prokaryotes and eukaryotes. It is also present extracellularly in the form of a chondroitin sulfate proteoglycan known as bamacan. We have found previously that SMC3 expression is elevated in a large fraction of human colon carcinomas. The additional finding that the protein is significantly increased in the intestinal polyps of Apc Min/؉ mice has led us to hypothesize that SMC3 expression is linked to activation of the APC/␤-catenin/ TCF4 pathway. The immunohistochemical analysis of colon adenocarcinomas from clinical specimens revealed that ␤-catenin and SMC3 antigens co-localize with maximal stain intensity within the transformed areas. Cloning and sequencing of 1578 bp of the human SMC3 promoter unveiled the presence of seven putative consensus sequences for ␤-catenin/TCF4 binding, two of which are conserved in the mouse Smc3 promoter. Transient transfection experiments in HCT116 and SW480 human colon carcinoma cells using deletion and mutated promoter constructs in luciferase reporter vectors confirmed that the putative sites, the first located at ؊48 bp and the second located at ؊701 bp, are susceptible to ␤-catenin/TCF4 transactivation. Co-transfection with a ␤-catenin expression vector enhanced the promoter activity whereas E-cadherin had the opposite effect. Binding of ␤-catenin/TCF4 complexes from SW480 nuclear extracts to these sequences was confirmed by electrophoretic shift and supershift mobility assays. Altogether these results are consistent with the idea that the ␤-catenin/TCF4 transactivation pathway contributes to SMC3 overexpression in intestinal tumorigenesis.
Mutations in the APC (adenomatous polyposis coli) gene are found in more than 70% of human intestinal adenomas (1). APC is regarded as the gatekeeper of the tumorigenesis process in the intestine, because the mutation and the loss of heterozygosity of APC occurs as the earliest event in hereditary and sporadic adenocarcinoma (1). The key role of APC in tumorigenesis has been attributed to its function in targeting ␤-catenin for degradation (2). The finding that ␤-catenin accumulates throughout the cell in the intestinal adenomas and that ectopic expression of APC in colon tumor cells lowers its level and inhibits tumor growth points to a link among APC mutation, ␤-catenin cytoplasmic level, and colon carcinoma (2)(3)(4). Further evidence for a key role of ␤-catenin in intestinal tumorigenesis has been provided by the findings that mutation in ␤-catenin glycogen synthase kinase 3␤-dependent phosphorylation sites makes it resistant to proteolytic degradation, giving rise to a phenotype that is indistinguishable from that of APC null mutations (5,6). Additionally, half of the sporadic colorectal cancer cells with wild-type APC have mutated ␤-catenin (4). ␤-Catenin acts as a transcriptional cofactor by migrating to the nucleus and associating with members of a family of DNAbinding proteins known as T cell factors, of which TCF4 is expressed in the intestinal epithelium (7,8). The targets of ␤-catenin/TCF4 transactivation include development-related genes activated through the Wingless/Wnt signaling pathway (9). More recently, a number of genes relevant for colorectal tumor formation and progression have been identified as being transcriptionally activated by the ␤-catenin/TCF4 complex. Some are relevant for growth control and cell cycling (c-Myc, cyclin D1, c-Jun, fra-1, gastrin, ITF2), some are implicated in cell survival (Id2, and MDR1), and some are implicated in tumor invasion and metastasis (matrilysin, VEGF) (10 -19).
SMC3 (formerly called Bamacan, Cspg6, HCAP, SmcD, or Mmip1) is a member of the multimeric cohesin complex that plays a key role in establishing chromatids cohesion and is also involved in chromosomal DNA repair (20 -22). Cohesins associates with the AT-rich sequences of chromosomal DNA near the centromeres and also along the arms with a periodicity of 9 to 15 kb. At the onset of anaphase, the complex is cleaved by the cysteine protease separin allowing chromosomal segregation (23). Interference with this process generates chromosomal instability (24). A post-translationally modified form of SMC3 carries chondroitin sulfate chains and is secreted as a proteoglycan known as bamacan (25). We have reported previously (26) that SMC3 mRNA level is elevated in about 70% of tumors from patients with colon carcinoma and that Smc3 protein is specifically increased in the intestinal polyps of Apc Min/ϩ mice in which loss of heterozygosity of the Apc gene causes a large increase of intracellular ␤-catenin (27). In addition, NIH-3T3 and BALB/c 3T3 murine fibroblasts stably transfected with Smc3 display a transformed phenotype, suggesting that this protein is involved in tumorigenesis (26). These findings have led us to hypothesize that SMC3 overexpression in intestinal carcinoma is linked to ␤-catenin/TCF4 transcriptional activation and that its subsequent overexpression is a relevant pathogenetic event.
In this paper we provide evidence in support of this hypothesis. Both the human and the mouse SMC3 promoters contain several conserved ␤-catenin/TCF4 binding consensus sequences that are responsible for the transactivation of the gene. We found that SMC3 promoter transcriptional activity is increased by elevated ␤-catenin levels and suppressed by a lowered ␤-catenin level. Immunohistochemical analysis of ␤-catenin and SMC3 in serial sections of colon from patients with invasive adenocarcinoma and of the intestinal adenomas from Apc Min/ϩ mice revealed that the two antigens co-localize with maximal stain intensity at the tumoral sites. The results are consistent with the idea that SMC3 is a target for the ␤-catenin/TCF4 transactivation pathway.

EXPERIMENTAL PROCEDURES
Materials-Goat anti-human SMC3 and goat anti-human SMC1 antibodies were purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). Anti-human ␤-catenin and anti-␤-integrin monoclonal antibodies were obtained from BD Biosciences. Immunohistochemistry kits were from DAKO (Carpenteria, CA) and Vector Laboratories (Burlingame, CA). Tri-Reagent, horseradish peroxidase-conjugated anti-rabbit and anti-mouse IgG antibodies, and anti-␣-tubulin monoclonal antibody were purchased from Sigma. The West-Dura Supersignal ECL 1 detection system, the NE-PER nuclear-cytoplasmic extraction kit, and the bicinchoninic acid-based protein assay kit were from Pierce (Rockford, IL). Heat inactivated fetal calf serum and cell medium were purchased from Mediatech (Herndon, VA). BioMax x-ray films were from Eastman Kodak Co. (Rochester, NY). [␥-32 P]dATP and [␣-32 P]dCTP were from ICN Biochemical (Irvine, CA). The gt11 HeLa genomic library was from Clontech (Palo Alto, CA). Turbo-Pfu polymerase was from Stratagene (La Jolla, CA). The DNA restriction and modifying enzymes, pGL3-Firefly luciferase empty vector, phRL-SV40 Renilla luciferase reporter vector, the Dual luciferase assay kit, and the Tfx cell transfection agent were from Promega (Madison, WI). Quick-spin oligo columns and proteinase inhibitors were from Roche Molecular Biochemicals. The reverse-transcriptase reagents were from Qiagen (Valencia, CA). The PCR reagents were from Takara (Madison, WI). All the chemicals were of ACS or biological purity and were purchased from Sigma or Fisher.
Immunohistochemistry and in Situ Hybridization Analysis-Polyps and normal tissue from the intestines of seven 6-month-old Apc Min/ϩ mice were dissected under a stereo microscope, fixed in 3.7% buffered formalin, and embedded in paraffin. Archival paraffin blocks of tumoral and matched normal colon tissue from 10 patients that had undergone surgery for colon carcinoma were obtained from the Surgical Pathology service of the Jefferson Hospital. For the immunohistochemistry, serial tumor and normal tissue sections of 5 m were subjected to antigen retrieval by microwaving in 0.1 M citrate solution, pH 6.0, for 10 min, and the immunostaining was performed on a Biogenex Optimax Autostainer. Sections were then incubated for 1 h at room temperature with either monoclonal ␤-catenin antibody (1:1000 to 1:5000) or goat anti-human SMC3 antibody (1:100 to 1:800). SMC3 and ␤-catenin immunocomplexes were visualized in the clinical specimens using Vecstatin ABC kits and a multilink horseradish peroxidase-conjugated secondary antibody. Mouse Smc3 was detected using a biotinylated rabbit anti-goat antibody, whereas ␤-catenin was detected using a DAKO ARK kit designed for the immunohistochemical staining with mouse primary antibodies of mouse tissue sections. Mouse and human stained specimens were examined under a light microscope. For the statistical analysis of the results we tested the hypothesis of coincidence of staining. Adenocarcinoma and normal specimens from the same subject were analyzed separately. We computed a 95% lower confidence bound (onesided confidence interval) based on the exact binomial distribution for the proportion of samples for which the SMC3 result was coincident with the ␤-catenin result (28). The lower confidence bound gives us a value that we can be 95% confident is exceeded by the actual coincidence rate.
Primer 5Ј-Extension-Total RNA was extracted from confluent human skin fibroblasts, HCT116, and HeLa cells using Tri-Reagent, and ethanol was precipitated and dissolved in 10 mM Tris-HCl, pH 7.6, 1 mM Na 2 EDTA. The oligonucleotide 5Ј-AAGGAGACCTGCCCCGGAG-CAGCA-3Ј, complementary to a region flanking the 3Ј-side of the translational start site of the human SMC3 gene, was end-labeled with [␥-32 P]ATP using T4-polynucleotide kinase. After purification on Quickspin columns, 1 pmol of 32 P probe was hybridized with 2 g of total RNA in 40 mM PIPES, pH 6.4, 400 mM NaCl, 1 mM EDTA, at 58°C for 20 min. Reverse transcription was initiated by addition of 20 units of Superscript reverse-transcriptase. Samples were incubated at 37°C for 30 min. The extension product was analyzed on a sequencing 6% polyacrylamide gel containing 8 M urea, and the bands were evidenced by autoradiography.
Isolation and Identification of the Human SMC3 Promoter-dEST clones matching the human SMC3 cDNA sequence at its 5Ј-end were identified by blasting the dEST data base with the published cDNA sequence of the gene (GenBank TM accession number NP-005436). Identified clones were purchased from ATCC and used as PCR templates. A 202-bp 32 P-labeled probe spanning the putative 5Ј-untranslated region of human SMC3 was used for screening of a HeLa genomic gt11 phage library. Briefly, after titration, 10 6 plaque-forming units of bacteriophages were plated at a density of 500 plaque-forming units per cm 2 in 15-cm dishes. Positive clones were identified by colony lifting on nitrocellulose membranes followed by hybridization with 32 P-labeled probes at 65°C in Denhardt's-saline/sodium phosphate/EDTA buffer. Positive plaques were isolated and plated at a lower plaque-forming unit density for a second round of screening. Colonies that remain positive after the quaternary screening were expanded, and the packaged human genomic DNA was characterized by digestion with a panel of restriction enzymes. DNA fragments hybridizing with the 32 P probes were subcloned in pBS-KS and sequenced at the SP6 or T7 end to confirm their identity. A 2.1-kb clone was found to span the translational start site of SMC3 at its 3Ј-end. The sequence of the amplified fragment matched that of human clone Hs10 -24204 (Human Genome Sequencing Project) on chromosome 10 and corresponded to the 5Ј-flanking region of the SMC3 gene. The cloning, sequencing, and partial functional characterization of the murine Smc3 promoter have been reported previously (29).
Generation of Promoter Reporter Vectors and of Deletion Constructs-A 1.6-kb insert (SMC3(Ϫ1578/ϩ56)) corresponding to the SMC3 promoter freed of the gt11 phage staffing sequences was generated by PCR from the pBS-KS clone using Taq polymerase and primers of sequence 5Ј-AGCTCTACAAAAACAAAAAAAAGCTC-3Ј and 5Ј-AAGGAGACCTGCCCCGGAGCAGCA-3Ј, respectively. The SMC-3(Ϫ438/ϩ56) insert spanning the 3Ј-end of the SMC3 promoter was likewise generated by PCR using a forward primer of sequence 5Ј-GG-CTAGCACAGTGCGCGCGAGGTC-3Ј. SMC3(Ϫ1578/Ϫ273) was generated using the reverse primer 5Ј-GTCCGTACCACCTCCGAGCG-CGGG-3Ј. These inserts were cloned in pGL3 basic vector by standard molecular biology techniques. pGL3-SMC3(Ϫ207/ϩ56) was generated by taking advantage of a unique EcoRI restriction site at position Ϫ206 of the promoter sequence and by digesting pGL3-SMC3(Ϫ438/ϩ56) with XhoI and EcoRI followed by relegation. The putative ␤-catenin binding site TTTTGTT contained in the SMC3 Ϫ207/ϩ56 was mutated to TCCTGTT by taking advantage of a unique BbvCI restriction site at position Ϫ41 of the promoter. The mutated insert was generated by PCR using the reverse primer 5Ј-CCCCTCAGCCAAACAGGATGGCG-GCGCTCGT-3Ј and the forward primer extending from base Ϫ438. The insert was then digested with EcoRI and BbvCI and ligated at the same restriction sites in pGL3-SMC3(Ϫ438/ϩ56). The new construct was cleaved at XhoI and EcoRI and relegated to yield MT-pGL3-SMC3(Ϫ207/ϩ56). Wild-type human ␤-catenin cloned in pcDNA3 was the kind gift of Dr. Byers (George Washington University, Washington, D. C.). Mouse E-cadherin cloned in pBSII-SK was a gift from Dr. Takeuchi (Kochi Medical School). The E-cadherin coding sequence was retrieved by digestion with XbaI and KpnI and inserted at the same sites in pcDNA3.1. The expression vector (pHR-hTCF4) harboring the wild-type human TCF4 gene driven by the cytomegalovirus promoter was a generous gift from Dr. Kinzler (The Johns Hopkins University, Baltimore, MD).
Transient Transfection and Luciferase Activity Assay-Cells cultures at 70% confluence in 12-well plates were used in all the experiments. The transfection mix contained 100 ng/ml of phRL-SV40 plasmid to correct for transfection efficiency and 1 g/ml of the designed plasmids. Plasmids were mixed in medium 199, followed by the addition of 36 g/g DNA of Tfx-50 transfection agent according to the manufacturer's directions. Mixtures were incubated for 15 min at room tempera-ture prior to addition to the cultures. Three hundred l of the transfection mix was added to each plate well. After 1 h of incubation at 37°C in a humidified incubator, the cells were supplemented with 1 ml of growth medium. Twenty-four h later the cell medium was discarded, and the cultures were washed with ice-cold phosphate-buffered saline. After cell solubilization in 200 l of lysis buffer, a 20-l aliquot was assayed using a dual-luciferase kit. Sample luminescence was read on a Zylux Sirius luminometer (Oak Ridge, TN). Firefly luciferase activity readings were corrected for transfection efficiency using the Renilla luciferase readings. All experiments were carried out with triplicate samples. The statistical differences between groups of data were analyzed by Student's t test.
Western Immunoblotting and Semi-quantitative RT-PCR-For immunoblotting, cells were collected in 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 50 mM Tris-HCl, pH 7.4, lysis buffer supplemented with 50 g/ml antipain, 40 g/ml bestatin, 5 g/ml E-64, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin, 1 g/ml aprotinin, 1 mM Na 2 EDTA. The cell lysate protein (50 g) was electrophoresed on a 10% SDS-PAGE slab gel. Separated proteins were transferred to a nitrocellulose by electroblotting, and the filters were blocked in 5% dry milk followed by incubation for 1 h with goat anti-human SMC3 (1:1000). After washing, the filters were incubated with an anti-goat IgG horseradish peroxidase-conjugated (1:10,000) secondary antibody, and the immunocomplexes were identified using a West-Dura ECL kit followed by autoradiography. To confirm that equal amount of proteins had been loaded and to examine ␤-catenin levels, the immunocomplexes were stripped by incubating the membranes in 2% SDS, 100 mM 2-mercaptoethanol, 62.5 mM Tris-HCl, pH 6.7, at 56°C for 30 min. After blocking in 5% dry milk, the filters were cut at the 90-kDa molecular mass standard mark, and the bottom section was incubated with anti-human ␣-tubulin monoclonal antibody (1:1000). The top portion of the filter was instead incubated with anti-human ␤-catenin monoclonal antibody (1:2000). After reaction with anti-mouse IgG horseradish peroxidase (1:10,000), the immunocomplexes were detected by ECL followed by autoradiography and densitometric scanning. SMC3 and SMC1 transcript levels were assessed by semi-quantitative RT-PCR. For this purpose, the transfected cells were extracted in Tri-Reagent, and 1 g of RNA was reverse-transcribed using the Qiagen Sensiscript kit and oligo(dT) primers. PCR cycling was stopped after completion of 20 and 30 cycles to ensure that products were quantitated during the nonsaturating phase of the reaction. SMC3 and SMC1 cDNA levels were assessed by agarose electrophoresis. Gels were photographed, and the propidinium iodine stained bands were quantitated by densitometric scanning.
Electrophoretic Mobility Shift Assay-The cell nuclear extracts were prepared using the NE-PER extraction kit. Before use, the nuclear extracts were dialyzed at 4°C overnight against 20 mM HEPES, pH 7.9, 75 mM KCl, 0.1% Na 2 EDTA, and protein content was assayed. The following oligonucleotides and their antisense were synthesized: 5Ј-CGCCGCCATTTTGTTTGGC-3Ј, corresponding to the Ϫ57/Ϫ39and Ϫ56/Ϫ37-bp regions of the human and the mouse promoters, respectively, and 5Ј-TAAATCTTTGTTGCAATTGT-3Ј, corresponding to the Ϫ706/Ϫ687-bp region of the human promoter. One g of the sense and the complementary oligonucleotide were combined in 50 l of 100 mM MgCl 2 , 250 mM NaCl, 100 mM Tris-HCl, pH 7.5, and annealed by incubation at 90°C for 3 min and by cooling to 25°C at a 1°C/min rate. Double stranded oligos were end-labeled with T4-polynucleotide kinase using 50 Ci of [␥-32 P]dATP and purified by chromatography through a Quick-spin oligo column. An aliquot (15,000 cpm) was incubated with 5 g of cell nuclear extract in 20 mM HEPES, pH 7.9, 75 mM KCl, 0.1% Na 2 EDTA containing 0.5 g poly(dI⅐dC) in a 50 l final volume for 10 min at room temperature. For the supershift assay, 0.5 g of antibody were added to the incubation mixture. DNA-protein hybrids were analyzed on a 6% non-denaturating polyacrylamide gel (acrylamide/bisacrylamide, 39:1) in 25 mM Tris, 200 mM glycine buffer, and the bands were visualized by autoradiography.

Tissue Localization of SMC3 and ␤-Catenin in Adenomatous
Polyps-The expression of SMC3 in human adenomatous polyps was examined by immunohistochemistry in tumoral and in matched normal tissues from 10 patients that had undergone surgery for colon carcinoma (Fig. 1, A-I). Serial sections were stained with monoclonal anti ␤-catenin and polyclonal anti-SMC3 antibody to assess the degree of co-localization of the two antigens. In normal tissue, ␤-catenin was detected at the cell-cell interface. On the contrary, adenoma and adenocarcinoma cells had diffuse ␤-catenin staining suggestive of an increase of the cytoplasmic antigen level. Adenomatous areas that displayed intense staining for ␤-catenin also had enhanced anti-SMC3 antibody reactivity. Neither ␤-catenin nor SMC3 were found at high levels in the surrounding normal tissue. This finding was invariably observed in all the specimens examined. As for ␤-catenin, the strongest staining for SMC3 was detected intracellularly. Incubation of the tissue sections with a blocking peptide of sequence corresponding to the epitope recognized by the SMC3 antibody effectively neutralized the tissue immunoreactivity. Furthermore, no immunoreactivity was observed by incubation with the secondary antibody alone. With all the specimens displaying an identical SMC3 and ␤-catenin distri- ␤-Catenin and SMC3 distribution in human (A-I) and mouse (J and K) colonic adenomas were detected by immunohistochemistry. A-C, serial sections (ϫ25) of formalin-fixed and paraffin-embedded tissue were immunostained with antibodies against ␤-catenin (A), SMC3 (B), or SMC3 antibodies plus competing SMC3 polypeptide (C). Immunoreactivity was visualized with a peroxidase-conjugated second antibody and using 3,3Ј-diaminobenzidine as substrate. D, F, and H, ␤-catenin immunostained sections from the adenomatous specimens from another patient. D and F illustrate two different areas of the same specimen (ϫ80). The inset in F is shown enlarged in H (ϫ500). E, G, and I, tissue sections immunostained with SMC3 and corresponding to those shown in D-H. Note the intense immunostaining for both ␤-catenin and SMC3 in the transformed areas associated to the intracellular accumulation of the antigen. J and K, consecutive sections (ϫ40) of a polyp from Apc Min/ϩ mice immunostained with ␤-catenin (J) or SMC3 (K) as described for the human specimens.
␤-Catenin/TCF4 Transactivates the Cohesin SMC3 Promoter bution pattern, the positivity rate, i.e. the proportion of samples displaying identical ␤-catenin and SMC3 staining pattern, was 100% (10/10) with a 95% confidence bound of 74.1%. These results are in line with our previous results (26) showing a statistically significant higher expression of SMC3 in the tumoral colon specimens compared with matched normal tissues. Mouse intestinal sections displayed the same pattern of staining as the human tissues. In particular, the greater ␤-catenin staining intensity in the transformed areas was matched by that of Smc3 (Fig. 1, J and K). Polyps from the intestine of seven mice were examined. Given the coincidence of SMC3 and ␤-catenin staining in the mouse samples examined, the positivity rate had a 95% confidence bound of 65.3%.
Structure of the Human SMC3 Promoter-The transcription start site was determined by primer extension (Fig. 2A) and by comparison of the published sequences of dEST cDNAs. Based on our calculations, the site would be located 76 bp upstream of the translational start site. The size of the human gene 5Јuntranslated region is therefore similar to that identified previously for the mouse Smc3 gene and calculated to be 95 bp in size. The mouse and human promoter sequences were examined for the presence of conserved putative transcriptional binding sites utilizing the TFSEARCH data base of published recognition sequences. As in the murine gene, a TATA box was also absent in the human sequence (see Fig. 2B). Both promoters, however, had a conserved region of sequence 5Ј-GCCGC-CATTTTGTTTGGC-3Ј beginning at Ϫ56 in the human gene and Ϫ55 in the murine gene. In addition to this highly conserved sequence, a series of putative transcriptional factor binding sites were identified whose location was conserved in the human and mouse SMC3 promoters (Fig. 2C). A binding sequence for the tumor suppressor MZF1 is located within the initial 40 bp flanking the transcriptional start site. Recognition sites for the ubiquitously expressed Sp1 transcriptional factor are situated in the Ϫ20to Ϫ250-bp region. Binding sites for c-Myc and E2F are found within the Ϫ270to Ϫ380-bp range. Finally, a conserved binding site for the tumor suppressor Cdx is located in the Ϫ670-bp region and is flanked by a CREB-P binding sequence in the Ϫ700-bp neighborhood. This region also harbors one of the two conserved ␤-catenin/TCF4 binding sites located at position Ϫ701 bp in the human promoter and at position Ϫ763 bp in the mouse sequence. The human binding site 5Ј-CTTTGTT-3Ј is a perfect consensus sequence (30), whereas the mouse binding site (5Ј-ATCAAG-3Ј) had sequence complementary to that identified in the human Id2 promoter (10). Both the human and the mouse promoters contained an imperfect TCF4 binding sequence within the highly conserved promoter initiation site (5Ј-tTTTGTT-3Ј) diverging one nucleotide from the canonical T cell factor consensus as indicated by the nucleotide in lowercase. In addition to the consensus sequence starting at Ϫ701, the human promoter harbored a second canonical ␤-catenin/TCF4 binding site at Ϫ504 bp (5Ј-AACAAAG-3Ј), which had no match within the mouse promoter. Finally, both the human and mouse promoters harbored four additional putative ␤-catenin/TCF4 binding sequences located upstream to the conserved distal ␤-catenin/ TCF4 binding site.
Functional Characterization of the SMC3 Promoter-To assess the potential role of ␤-catenin in the regulation of the human SMC3 promoter, a series of terminal and nested deletions of the promoter were cloned into a Firefly luciferase reporter vector (pGL3), and their activity was assayed in cells following transient transfection. We first determined whether the full-length promoter was affected by intracellular ␤-catenin and/or TCF4 levels. For these experiments we selected the human colon carcinoma HCT116 cell line. HCT116 cells have a low ␤-catenin/TCF4 pool (11), and alteration of its level should result in changes of gene transcription. We transfected these cells with a luciferase reporter vector (Fig. 3A) harboring the full-length SMC3 promoter (SMC3 (Ϫ1578/ϩ56 bp)), together with an increasing concentration of ␤-catenin expression vector. A dose-dependent increase in gene transactivation could be evidenced with a 10-fold maximal activation (Fig. 3B). Higher concentration of ␤-catenin negatively affected cell survival at 24 h post-transfection (data not shown) consistent with the report that overexpression of ␤-catenin may initiate apoptosis (31). Co-transfection of the promoter construct with the TCF4 expression vector enhanced the promoter activity and potentiated the effect of ␤-catenin (Fig. 3C). On the contrary, transfection with murine E-cadherin negatively affected the basal promoter activity (Fig. 3D) in line with the idea that this maneuver lowers the available endogenous ␤-catenin pool by sequestering the protein at the cell membrane (32). We next examined whether the conserved 5Ј-tTTTGTT-3Ј-site proximal to the transcriptional start site was functionally active. The ␤-catenin/TCF4 responsive elements of the matrix metalloproteinase Matrilysin (14), Id2 (10), cyclin D1 (11,19), and gastrin (17) genes have been mapped in close proximity to the transcriptional start site. The putative binding sequence identified diverges by one base from the canonical ␤-catenin/TCF4 binding site, but it has been found previously (11) to be relevant for the transactivation of the cyclin D1 promoter. The sequence is located within a perfectly conserved short region in the human and murine promoters. A 263-bp promoter segment (Ϫ207/ϩ56 bp) harboring this sequence (Fig. 4A) was tested in the luciferase reporter assay in HCT116 and SW480 cell lines. The human carcinoma SW480 cells were selected, because this cell line has a high constitutive level of the ␤-catenin/TCF4 transactivation complex (11). As observed for the full promoter, ␤-catenin greatly enhanced luciferase activity in HCT116 cells, and the effect rapidly reached plateau (Fig. 4B). The activity of the truncated promoter was 5-fold greater in SW480 cells as it would be expected if a ␤-catenin/TCF4 responsive element were present within its sequence. In these cells, however, co-transfection with ␤-catenin reduced the luciferase activity by about 55%. As we also observed in HCT116 cells, exceedingly high levels of ␤-catenin appear to have an inhibitory effect on the transcription of SMC3. To confirm the functional significance of this conserved ␤-catenin/TCF4 transactivation site, a two-base substitution mutation was introduced within the putative binding sequence. The new construct, MT-pGL3-SMC3(Ϫ207/ϩ56), displayed the same basal luciferase activity in HCT116 and in SW480 cells (Fig. 4C). In addition, contrary to that observed with the construct harboring the intact promoter sequence, co-transfection with ␤-catenin did not result in enhanced luciferase activity in HCT116 cells. The luciferase reporter construct SMC3(Ϫ438/ϩ56) was tested, because the proximal 450 bp of the human promoter contains binding sites for transcriptional factors that are conserved in the murine promoter and that are likely to be relevant for the constitutive SMC3 promoter activity. Basal activity of the reporter was 4-fold higher (Fig. 4D) compared with that of SMC3(Ϫ207/ϩ56) in HCT116 cells consistent with idea that the additional 231-bp promoter sequence harboring the recognition sequences for the ubiquitous Sp1 transcriptional factor and for c-Myc and E2F are likely responsible for the constitutive transcription of the SMC3 gene. ␤-Catenin further enhanced in a dose-dependent fashion the activity of the truncated promoter, corroborating the idea of a major role of the ␤-catenin/TCF4 transactivation pathway in the regulation of SMC3 expression. Accordingly when tested in SW480 cells, the construct displayed higher activity than in HCT116 cells, and co-transfection with ␤-cate-␤-Catenin/TCF4 Transactivates the Cohesin SMC3 Promoter ␤-Catenin/TCF4 Transactivates the Cohesin SMC3 Promoter 20263 nin did not enhance further the luciferase activity. Finally, we tested the responsiveness of the most distal ␤-catenin/TCF4 binding sites, one of which had a match in the murine promoter. For this purpose, the SMC3(Ϫ1578/Ϫ273) deletion construct was tested (Fig. 4E). The activity of the construct in HCT116 cells was significantly lower (12-fold) than that observed for the SMC3(Ϫ438/ϩ56) construct, suggesting that the distal part of the promoter contains transcription-silencing elements. Nonetheless, co-transfection with ␤-catenin enhanced gene transcription by 3-fold, and luciferase activity was significantly higher (11 times) in SW480 cells. These results confirmed that the distal SMC3 promoter contains functionally active ␤-catenin/TCF4 binding sites.
Responsiveness of the SMC3 Gene to ␤-Catenin Level-To provide a quantitative measure of the effect of ␤-catenin on SMC3 gene transactivation, the SMC3 transcript and protein levels were assessed in HCT116 and SW480. The constitutive transcript level of SMC3 was 4-fold higher in SW480 cells than in HCT1116 cells. Transfection with a ␤-catenin expression vector resulted in a dose-dependent increase in SMC3 transcript in HCT116 cells but had virtually no effect in SW480 cells (Fig. 5A). As a control in these studies, we measured the expression of SMC1, a cohesin protein that forms stoichiometric multimeric complexes with SMC3. The analysis of 1.5 kb of the SMC1 promoter we have cloned has revealed that the sequence does not harbor putative ␤-catenin/TCF4 transactivation binding sites (data not shown). Therefore the transcript level of this gene should be insensitive to changes in ␤-catenin levels. This was in fact confirmed by the RT-PCR results. In particular the SMC1 transcript level was virtually identical in HCT116 and SW480 cells and was not influenced by increasing the ␤-catenin level. We next examine whether the relationship between endogenous ␤-catenin and SMC3 transcript levels was mirrored by similar changes in the protein level. Endogenous ␤-catenin protein level was about 8-fold higher in SW480 cells than in HCT116 cells (Fig. 5B). SMC3 protein level was also higher (about 3-fold) in SW480 than in HCT116 cells. Transfection of these cell lines with a ␤-catenin expression vector resulted in a dose-dependent increase of the intracellular ␤-catenin protein level that was much more evident in HCT116. The ectopic expression of ␤-catenin raised SMC3 protein level in HCT116 cells whereas in SW480 cells the endogenous SMC3 protein remained virtually unchanged.
Binding Activity of the Putative ␤-Catenin/TCF4 Sequences-The identity of the conserved ␤-catenin/TCF4 binding sites was confirmed by performing electrophoresis mobility shift and supershift analysis. Double-stranded 19 -20-bp oligomers containing the putative core sequences and their flanking sequences identified within the human SMC3 promoter were tested (Fig. 6). When the labeled Ϫ57/Ϫ39 oligonucleotide corresponding to the sequence of the proximal conserved binding site was incubated with nuclear extract from either HCT116 and SW480 cells, a major retarded band could be easily visualized corresponding to the binding of the transactivating proteins to the labeled probe. A larger amount of the DNA⅐protein complex was formed when the SW480 nuclear extract was used. The complex formation could be prevented by adding an excess (50-fold) of unlabeled probe. The addition of ␤-catenin monoclonal antibody but not of an irrelevant monoclonal antibody (anti-␤-integrin) resulted in a specific supershift of the band confirming that ␤-catenin was part of the complex. Identical results were obtained when the labeled Ϫ706/Ϫ687 oligo-  (30) ␤-catenin/TCF4 binding sites whereas the sites identified with empty arrowheads differ of one base from the perfect consensus sequence. HCT116 human colon carcinoma cells were co-transfected with 1 g/ml of full-length SMC3 promoter cloned in Firefly luciferase vector (pGL3-SMC3(Ϫ1578/ϩ56)) and 0.1 g/ml of phRL-SV40 plasmid carrying the Renilla luciferase gene. Increasing amounts of wild-type human ␤-catenin expression vector without (B) or together with (C) human TCF4 expression vector (1 g/ml) were also co-transfected. The effect of E-cadherin (1 g/ml) in cells transfected with 2.5 g/ml ␤-catenin is illustrated in D. Luciferase activity was assayed as described under "Experimental Procedures" and corrected for the transfection efficiency based on the Renilla luciferase activity. Values were normalized to the luciferase activity of cells transfected with the reporter vector alone. The bars represent the mean Ϯ S.D. of samples run in triplicate. Transfected cells expressed luciferase activity at a significant different level (p Ͻ 0.05) than did untransfected cells.
␤-Catenin/TCF4 Transactivates the Cohesin SMC3 Promoter nucleotide corresponding to the distal conserved binding site was tested, confirming its ability to bind ␤-catenin/TCF4.

DISCUSSION
SMC3 is a component of the cohesin multisubunit complex that holds together sister chromatids during mitosis and enables their segregation to the opposite poles of the cell prior to cytokinesis (22,33,34). In the multimeric complex, SMC3 combines with another SMC cohesin, SMC1, and two non-SMC proteins, named Scc1 and Scc3, after their discovery in budding yeast Saccharomyces cerevisiae (20). SMC proteins share a unique structural motif with globular N-and C-terminal ATPase-like domains separated by a long coiled-coil segment in the center of which is a globular hinge domain (35). Haering et al. (36) have recently proposed a molecular model whereby Scc1 links two heads of a single SMC1/SMC3 heterodimer. This proteinaceous ring holds together sister chromatids until Scc1 is cleaved by the cysteine protease separin. The function of the latter is modulated by securin (37). There is evidence in mammalian cells that interference with the securin/separin/Scc1 system causes chromosomal instability. For example, the homozygous deletion in HCT116 colon carcinoma cells of the hSecurin gene leads to retardation of sister chromatid separation and a high rate of chromosomal loss because of defective Scc1 cleavage (24). On the other hand, overexpression of the murine orthologue of Scc1 (PW29 protein) in mouse fibroblast leads to inhibition of proliferation, implicating this protein and its complex with SMC proteins in the control of mitotic cell cycle progression (21). We have shown previously (26) that overexpression of Smc3 in 3T3 fibroblasts causes cell transformation and enhances cell proliferation. Furthermore, an increased expression of SMC3 in human colon carcinomas and in the tumoral tissue from the intestine of Apc Min/ϩ mice was observed. Because in Apc Min/ϩ mice loss of heterozygosity amplifies ␤-catenin activity, we considered the possibility that Smc3 and ␤-catenin overexpression be linked. By functionally analyzing the organization of the human gene promoter, we have now identified the ␤-catenin/TCF4 pathway as a main transcriptional regulator of SMC3. In support of this conclusion, we have also found that in colon carcinoma cells the SMC3 transcript and protein levels correlate positively with the intracellular ␤-catenin concentration. This finding is mirrored in colon adenocarcinomas by the increased expression of SMC3 at the sites where ␤-catenin is overexpressed. Two conserved transcriptional binding sites for ␤-catenin/TCF4 are present in the human and mouse SMC3 promoters. The first site is located proximal to the transcriptional start site. The promoters of the human cyclin D1, Id2, matrilysin, MDR1, and gastrin genes also harbor an active TCF4 transcriptional binding site FIG. 4. Luciferase expression assay of SMC3 gene promoter deletion constructs. A, schematic representation of the 5Ј-stepwise and of the nested deletion constructs used to test the functional activity of the SMC3 promoter in transient transfection assays. Numbering and symbols are as described for Fig. 3. The mutated promoter sequence is identified by MT. B-E, HCT116 or SW480 human colon carcinoma cells were co-transfected with SMC3 promoter-Firefly luciferase constructs (1 g/ml) and phRL-CMV plasmid (0.1 g/ml), along with increasing concentrations of ␤-catenin expression vector as indicated. Luciferase activity was assayed as described under "Experimental Procedures" and corrected for the transfection efficiency based on the Renilla luciferase activity. The values shown are the normalized mean Ϯ S.D. from three independent samples from a representative experiment. Except for the experiment shown in C, cells transfected with ␤-catenin expressed luciferase activity at significantly different levels (p Ͻ 0.05) than did untransfected cells.
in this same region (10,11,13,14,17). The second conserved binding site for TCF4 is located at Ϫ701 bp in the human promoter in the same region as the ␤-catenin/TCF4 responsive elements identified in the c-MYC and VEGF promoters (12,15).
The comparative analysis of the mouse and the human promoters has enabled us to identify several conserved features that are therefore likely to be relevant for the regulation of the gene and indicative of the functional role of SMC3. In addition to the conserved ␤-catenin/TCF4 binding sites, a number of putative binding sites for tumor suppressors and oncogenes were also identified, emphasizing the concept that the SMC3 gene is targeted in tumorigenesis. A binding site for MZF1 maps to the Ϫ38/Ϫ31 bp region in both the human and the mouse promoter. MZF1 is a transcription factor belonging to the Kruppel family of zinc finger proteins. Mzf1(Ϫ/Ϫ) knockout mice develop lethal neoplasia characterized by the infiltration and complete disruption of the liver architecture by a monomorphic population of myeloid cells (38). A putative recognition sequence for a second tumor suppressor, CDX, maps to the same distal region as one of the conserved ␤-catenin/TCF4 binding sites. Cdx is a homeobox-containing gene. The mouse expresses three related genes named Cdx-1, Cdx-2/3, and Cdx-4. Cdx2-null embryos die before gastrulation, but interestingly, the heterozygous animals develop intestinal tumors (39). Furthermore CDX2 has been found mutated in a colorectal cancer cell line with normal APC/␤-catenin signaling, suggesting that the intact protein is necessary for normal cell behavior (40). A recognition sequence for the oncogene c-MYC is located in the Ϫ400 bp region of the promoters. The site is ϳ50 bp apart from the binding sequence for the E2F transcriptional factor. c-MYC and E2F transcription factors share a number of functional properties including the ability to induce quiescent cells to enter the cell cycle and progress into S phase and to control cell fate by activating the p53-dependent apoptotic pathway (41). c-MYC is also a downstream target for the ␤-catenin/TCF4 transactivation pathway, raising the possibility of an amplifying effect of c-MYC and ␤-catenin on SMC3 expression in colon carcinoma. The presence in the promoter of a conserved CREB protein binding site may be of particular significance. As in the cyclin D1 promoter (11), in the human SMC3 promoter this site is located in close proximity (less than 20 bp) to a conserved ␤-catenin/TCF4 locus. The CREB-binding protein has intrinsic acetyltransferase activity and acetylates TCF4 and ␤-catenin (42)(43)(44). Acetylation of ␤-catenin decreases its transcriptional activity (44) whereas mutation at the acetylation site increases ␤-catenin ability to specifically activate the c-myc promoter. It has been hypothesized (42) that the CREB-binding protein might participate in the recruitment of the basal transcriptional machinery, and the region of the c-myc and SMC3 promoters harboring the ␤-catenin/TCF4 and CREB binding sites may act as the docking site. Interestingly, the Wnt-1-and ␤-catenin-mediated transactivation of the WISP-1 growth factor (Wnt-1 induced secreted protein 1) gene specifically involves a CREB binding site whereas a T cell factor binding site also present in the promoter plays a minor role (45). FIG. 5. Effect of ectopic ␤-catenin overexpression on the SMC3 transcript and protein levels. A, cells were co-transfected with the indicated concentrations of ␤-catenin expression vector, together with human TCF4 expression vector (1 g/ml). After 24 h total RNA was extracted in Tri-Reagent. One g of RNA was reverse-transcribed using oligo(dT) primers, and the generated cDNA was amplified by PCR using SMC3-and SMC1-specific primers. The product generated after 20 reaction cycles was analyzed on a 0.7% agarose gel by electrophoresis. The results shown are representative of the three experiments performed. B, after transfection with the ␤-catenin expression vector, the cells were lysed in buffer containing 0.1% Tween 20 and a mixture of anti-protease inhibitors. Fifty g of proteins were electrophoresed on a 10% SDS-PAGE slab gel and then transferred to nitrocellulose membrane and immunoreacted with a goat anti-human SMC3 antibody. After ECL development and autoradiography, the immunocomplexes were stripped, and the bottom part of the filter containing proteins of 90 kDa or lower size was incubated with a mouse anti-human ␣-tubulin antibody. The top portion of the filter was probed for ␤-catenin using a mouse anti-human ␤-catenin antibody. The results shown are representative of the two experiments performed.
FIG. 6. Identification of ␤-catenin/TCF4 binding site in the SMC3 promoter. A, lanes 1 and 2, 5 g of protein of HCT116 or SW480 cell nuclear extracts were incubated in 20 mM HEPES, pH 7.9, 75 mM KCl, 0.1% EDTA containing 0.5 g poly(dI⅐dC) in a 50-l volume for 10 min at room temperature, with 15,000 cpm (ϳ200 cpm/ng) of [␥-32 P]dATP-labeled oligonucleotides corresponding to the Ϫ57/Ϫ39and Ϫ56/Ϫ37-bp regions of the human and mouse promoter, respectively. After 10 min of incubation at room temperature, DNA-protein hybrids were analyzed on 6% non-denaturating polyacrylamide gel (acrylamide/bisacrylamide, 39:1) in 25 mM Tris, 200 mM glycine buffer, pH 8.6, and the bands were visualized by autoradiography. B, 5 g of protein of SW480 cell nuclear extract were incubated with labeled oligonucleotides corresponding to the human promoter regions Ϫ57/ Ϫ39 (lanes 1-4) and Ϫ706/Ϫ687 (lanes 5-8), as described. For the competition binding, 30-fold excess of unlabeled oligonucleotides were included in the incubation mixture (lanes 2 and 6). For the supershift assay, 0.5 g of an irrelevant monoclonal antibody (anti-␤-integrin in lanes 3 and 7) or anti ␤-catenin monoclonal antibodies (lanes 4 and 8) were added instead.

␤-Catenin/TCF4 Transactivates the Cohesin SMC3 Promoter
It has been reported previously (21) that the bulk of SMC3 undergoes redistribution from the chromosome vicinity to the cytoplasm during prometaphase and back to the chromatin in telophase. Therefore the vast majority of SMC3, along with SMC1, is localized within the cytoplasm during the cell cycle. We have confirmed this observation by investigating SMC3 cytolocalization by confocal microscopy in human fibroblasts and HCT116 colon carcinoma cells (data not shown). In addition, part of the antigen is associated with the cell membrane at all times, a finding consistent with the staining of SMC3 at the cell-cell interface in tissue sections. A glycanated form of SMC3 has been detected extracellularly (25), and Madin-Darby canine kidney cells constitutively secrete chondroitin sulfateconjugated SMC3 in culture (46). No intracellular form of glycanated SMC3 has thus far been identified raising the possibility that the glycanated protein has an extracellular fate. On the other hand there is mounting evidence that intracellular SMC proteins have other functions beside the formation of the cohesin complex. Recently SMC1 has been identified as a downstream effector in the ATM-dependent response to DNA damage (47,48). ATM kinase is responsible for the activation of the G 1 , S, and G 2 /M checkpoints through the phosphorylation of the tumor suppressors p53, CHK2, and BRCA1. SMC1 is an additional substrate for ATM kinase and forms a complex with BRCA1 and SMC3 distinct from the cohesin complex (47). The biological function of this complex is still poorly understood; however mutation of SMC1 that prevents its phosphorylation impairs DNA damage repair. SMC3 and SMC1 also are components of the mammalian recombinase (denoted RC-1), a multimeric complex that is involved in chromosomal recombination and single strand DNA repair (49). Earlier, a truncated form of SMC3 has been found to modulate c-MYC-dependent transcription by complexing and sequestering members of the MAX/MAD c-MYC interacting proteins (50). Most of the recognized functions of SMC3 require the presence of this protein in the nucleus either for directing chromosomal segregation or to repair DNA. The fact that for most of the cell cycle SMC3 resides in the cytoplasm raises the question of the functional significance of this large protein pool. The available information suggests that SMC3 participates in stoichiometric amount to the formation of the cohesin (21), the SMC1/SMC3/BRCA1 (47), and the RC-1 multimeric complexes (49). Elevation of SMC3 alone will then conceivably result in excess free protein that may be relevant for tumorigenesis. Excess SMC3 may sequester other proteins in the cytoplasm thereby preventing their nuclear translocation and the formation of multimeric complexes or may affect their phosphorylation and/or proteolytic processing. Interference with chromosomal segregation and DNA repair would ultimately result in chromosomal instability as demonstrated for the securin-null cells. The fact that overexpression of SMC3 is sufficient to trigger cellular transformation, along with the fact that the gene is activated by ␤-catenin/TCF4, a key transactivation pathway responsible for the initiation and progression of colon carcinoma, argues in favor of a major role of SMC3 in tumorigenesis.