Oncogene (2002) 21, 3253 ± 3257 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Gene expression di€erences between the microsatellite instability (MIN) and chromosomal instability (CIN) phenotypes in colorectal cancer revealed by high-density cDNA array hybridization

Donncha S Dunican1,2, Peter McWilliam1, Orna Tighe1, Anne Parle-McDermott1,3 and David T Croke*,1

1Department of Biochemistry, Royal College of Surgeons in Ireland, 123, St Stephens' Green, Dublin 2, Ireland

Two distinct pathways of tumorigenesis exist in sporadic loss of heterozygosity (LOH). The microsatellite colorectal cancer. The microsatellite instability pathway instability (MIN) phenotype is not as prevalent in (MIN), which is characterized by widespread micro- sporadic colorectal cancer, accounting for 15% of such satellite instability due to aberrant mismatch repair malignancies and has been ®rmly linked to a faulty machinery, accounts for 15% of all sporadic colorectal DNA mismatch repair system. Adenomatous polyposis cancers. The chromosomal instability (CIN) phenotype, coli (APC)orb-catenin mutations are the most which accounts for 85% of sporadic colorectal cancers, common initial molecular lesions in the CIN phenotype is characterized by gross chromosomal lesions but the (Kinzler and Vogelstein, 1996). A link between underlying mechanism remains unclear. We have microsatellite instability and mutant DNA repair genes addressed di€erences in gene expression between the was discovered (Strand et al., 1993). Subsequently ®rm MIN and CIN colorectal cancer phenotypes in vitro by associations between DNA mismatch repair and colon the use of high density cDNA ®lters to compare gene cancer were demonstrated (Fishel et al., 1993; Bronner expression patterns between MIN and CIN colorectal et al., 1994). Unrepaired DNA mismatches in cancer cell-lines yielding a panel of 73 consistently transforming growth factor beta I receptor II di€erentially expressed genes. Nine of these genes were (TGFbIRII) and insulin-like growth factor receptor II subjected to con®rmatory analysis by independent (IGFRII) (Parsons et al., 1995; Souza et al., 1996) are methods, of which six were con®rmed as being thought to be the driving force of tumorigenesis in di€erentially expressed; PLK, RanBP2 and CCNA2 MIN tumours. p53 was initially a candidate gene in the were overexpressed in CIN lines while BTF3, H2AZ and molecular mechanism of CIN. However, a number of PTPD1 were overexpressed in MIN lines. These six cancers with mutant p53 have been shown to be diploid genes are involved in diverse processes, such as with no obvious chromosomal defects and to exhibit maintenance of chromatin architecture, DNA-damage MIN due to impaired mismatch repair (Eshleman et checkpoint and cell cycle regulation, which may al., 1998). Disruption of mitotic checkpoint assembly contribute to the CIN and MIN phenotypes. genes can lead to CIN because checkpoint defective Oncogene (2002) 21, 3253 ± 3257. DOI: 10.1038/sj/ cells can complete mitosis with inappropriately aligned onc/1205431 chromosomes (Paulovich et al., 1997). Recent data strongly support the hypothesis that carcinogen Keywords: colorectal cancer; microsatellite instability; exposure determines the type of instability in MIN chromosomal instability; gene expression and CIN cancers (Bardelli et al., 2001). Despite these ®ndings, the mechanism of CIN progression remains unclear. A number of studies have used cell lines derived The majority of colorectal tumours (85%) are sporadic from sporadic colon tumours as in vitro models of the in origin yet they exhibit close similarities to tumours MIN and CIN phenotypes (Cahill et al., 1998). We ONCOGENOMICS resulting from inherited colorectal cancer syndromes. report a global gene expression pro®ling approach used The chromosomal instability (CIN) phenotype, which to identify gene expression di€erences within a panel of accounts for 85% of sporadic cases, exhibits gross MIN and CIN cell lines (selected on the basis of chromosomal abnormalities such as aneuploidy and Heinen et al., 1995). This utilized Gene Discovery Array (GDA) v1.3 ®lters (IncyteGenomics, USA) and SMART cDNA (Clontech, USA) to determine expres- *Correspondence: DT Croke; E-mail: [email protected] sion di€erences between MIN (LS174T, SW48 and Current addresses: 2Department of Biomedical Sciences, Genes and HCT116) and CIN (Caco2, SW480 and HT29) Development Group, The University of Edinburgh, Hugh Robson sporadic colorectal cancer cell lines. The GDA ®lters 3 Building, Edinburgh EH8 9XD, UK; Department of Biochemistry, harbour over 18 000 cDNA species arrayed in Trinity College, Dublin 2, Ireland Received 2 October 2001; revised 19 February 2002; accepted 20 duplicate, of which approximately 4000 are known February 2002 genes and 14 000 are expressed sequence tags (ESTs). Gene expression profiles in colorectal cancer DS Duncan et al 3254 In order to rule out tissue-speci®c di€erences, three previously to identify gene expression di€erences pair-wise comparisons were performed between (in between HCT116 and Caco-2 (Zhang et al., 1997). each case) representative MIN and CIN colorectal The Unigene unique identi®er of each transcript was cancer cell-lines. Brie¯y, radiolabelled SMART total entered into the SAGE database and we were thus able cDNA preparations for each cell line pair were to determine if our results concurred with those hybridized in parallel to identical GDA v1.3 ®lters, obtained from the established technology of SAGE. expression patterns were visualized by phosphorima- The expression di€erences observed by GDA ging (100 mM resolution) and imager ®les were analysed analysis were independently veri®ed by Northern blot using the Gene Discovery Array software (IncyteGe- and RT ± PCR for nine transcripts, selected on the nomics, USA). Individual expression pro®les were basis of current literature reports concerning the evaluated based upon three criteria: (a) consistency in transcripts and previous MIN/CIN ®ndings (see expression di€erence across three pair-wise MIN-CIN above). The choice of veri®cation method was made comparisons; (b) an expression di€erence of ®vefold or based on EST and SAGE tag abundances associated greater as recommended by IncyteGenomics; and (c) with a given transcript. These analyses con®rmed the only known genes were considered. A simple Perl script di€erential expression of six of the nine genes: polo program (GeneMatch v1.0) was coded which allowed (Drosophila) like Kinase (PLK), ran binding protein 2 us to identify the most signi®cant expression di€erences (RANBP2) and cyclin A2 (CCNA2) were overexpressed present in all three experiments. Approximately 70 in CIN; basic transcription factor 3 (BTF3), histone genes were found to be consistently overexpressed 2AZ (H2AZ) and protein tyrosine phosphatase D1 either in MIN or in CIN cell lines (Tables 1 and 2). As (PTPD1) were overexpressed in MIN. Di€erential a control for our expression data we exploited the expression of H2AZ, BTF3, PTPD1 and RANBP2 public Serial Analysis of Gene Expression (SAGE) was evident using either method; expression di€erences database (Lal et al., 1999). SAGE has been used ranged from 2 ± 3-fold (BTF3 and RANBP2) to greater

Table 1 Genes exhibiting elevated expression in MIN cell lines Accession SAGE tags per SAGE tags per Gene transcript number Tentative gene function million in HCT116 million in Caco2

Hemoglobin, alpha 1 W04832 Oxygen transport ± ± Heat shock 27 kD protein 1 W69499 Activates apoptosis/casases ± ± Cell division cycle 2-like 1 N98592 Cell division 645 48 H2AZ HISTONE W45695 Nucleosome assembly 530 407 Spectrin, beta, non-erythrocytic 1 N70585 Cytoskeletal component of red blood cells ± ± Metallothionein-Ie H25252 Heavy metal transport 82 0 Transcription factor AP-2 N63770 Transcription factor/gene expression ± ± Isovaleryl Coenzyme A H37754 Metabolism 363 264 ETS-related transcription factor ELF-1 N74959 Transcription factor/gene expression ± ± Cathespin S precursor N53585 Cysteine protease; protein degradation/processing 64 72 Collagen, type V, alpha 1 W52829 Structural protein; bone architecture 0 32 BASIC TRANSCRIPTION AA203284 Transcription factor/gene expression 629 505 FACTOR 3 CTP synthetase H11731 Pyrimidine metabolism 49 32 Retinol-binding protein 3 H64488 Vit A regulation (in eye cells) ± ± Carbonic anhydrase I R46266 Hydration of CO2 ±± 40S ribosomal protein S19 AA046433 Monocyte chemotactic factor ± ± DNA primase R43686 DNA polymerase/DNA replication 0 48 Kinase insert domain receptor R35713 VEGF-A tyrosine kinase receptor ± ± Dual specificity phosphatase 5 R43931 Cell signalling 49 16 Integrin, alpha L T83460 T-cell adhesion to target cells ± ± HLA class I-histocompatability R74335 Cell signalling/cell communication 447 44 ADP-ribosylation factor 4 N52015 Purine metabolism 66 97 Microfibrallar protein R01211 Cytoskeletal organisation 0 16 Alu RNA transcript W31678 Unclassified ± ± Trans-1,2,dihydrobenzene-1,2-diol W63727 Aldehyde and ketone oxidation 0 32 dehydrogenase Ribonuclease, Rnase A family, R62231 RNA metabolism 545 0 1 pancreatic Elastin T79116 Cytoskeletal ± ± Prepro form of corticotropin R33777 Cell signalling/communication ± ± PROTEIN TYROSINE R23593 Gene/protein expression 32 0 PHOSPHATASE D1

A summary of the genes showing consistently elevated expression in the microsatellite instability (MIN) sporadic colorectal cancer cell lines LS174T, SW48 and HCT116 with respect to the chromosomal instability (CIN) cell lines Caco-2, SW480 and HT29. Tentative gene functions were assigned to each gene where possible. Genes highlighted in bold text were selected for further expression analysis by RT-PCR and Northern blot. SAGE tag numbers per million tags are detailed where available in HCT116 and Caco2

Oncogene Gene expression profiles in colorectal cancer DS Duncan et al 3255 Table 2 Genes showing elevated expression in CIN lines Accession SAGE tags per SAGE tags per Gene transcript number Tentative gene function million in HCT116 million in Caco2

Guanosine 5'-monophosphate synthase R94457 Purine metabolism 49 16 Creatine kinase B AA069373 Cell signalling/cell communication 182 505 Phosphatidic acid R59009 Cell signalling (lipid mediators and 0 16 Phosphatase Type 2A androgens) Pregnancy-specific beta 1- R80286 Foetal development; embryonic growth 0 16 glycoprotein 7 and development TBP-associated factor TAFII80 R44112 General transcription factor 33 16 Receptor protein-tyrosine kinase (EDDR1) R56392 Cell signalling/cell communication 265 114 Ribosomal protein S6 kinase AA039599 Growth factor regulated serine/threonine kinase 0 32 Coagulation factor XI (plasma N52835 Cell/organism defence 0 16 thromboplastin antecedent) Guanine nucleotide binding protein H61281 Cell signalling/G protein intracellular messenger 82 32 Complement component 4-binding protein H53489 Cell/organism defence 0 14 Cholinesterase-related cell division controller R83723 Cell division/regulates passage through mitosis 16 0 Granulocyte-macrophage colony- N99581 Cytokine receptor 7900 7300 stimulating factor receptor Calcium channel L-type alpha 1 subunit AA136049 Cell signalling/cell communication ± ± Iduronate 2-sulphatase N26477 Hunter Syndrome 0 16 Fibronectin 1 N90769 Cell adhesion/morphology 0 717 Chronic granulomatous disease AA149281 Cellular respiration 16 8 (b-245 beta polypeptide) HMGCo-A reductase N93874 Cholesterol biosynthesis 0 65 RYK-receptor like tyrosine kinase W32536 Signaller in MAPK pathway 0 32 RAN BINDING PROTEIN 2 R82691 Nuclear pore complex activator; involved 16 32 in mitosis Collagen, type XVIII, alpha 1 AA156812 Structural peptide; contains motif (endostatin) 49 163 capable of inhibiting proliferation, angiogenesis and tumour growth Electron-transfer-flavoprotein, alpha W85863 Mitochondrial respiration 49 65 polypeptide Monokine induced by a-interferon AA1131406 Chemotactic cytokine ± ± Secretogranin II precursor R41603 Neuropeptide precursor of secretogranin 16 0 Amylo-1,6-glucosidase, 4-alpha- AA017181 Glycogen metabolism 49 16 glucantransferase Small inducible cytokine A2 AA024754 Inflammatory response ± ± Glucose-6-phosphate dehydrogenase AA017135 Glucose metabolism 49 114 Phospholipase C, gamma 2 H48246 Phospholid metabolism 0 32 HE4 AA025750 Possible role in sperm maturation 0 32 Glucosidase II W47502 Glycoprotein metabolism 645 603 Angiodema inhibitor AA056092 Possible role in reproduction ± ± Alpha crystallin A chain H82655 Chaperone protein in lens ± ± POLO (DROSOPHILA) LIKE KINASE R99810 Cell cycle regulator 33 163 Glycogenin N35332 Glycogen biosynthesis 33 16 Gem GTPase AA129719 Ras-related GTPase 33 16 ATPase, Cu++ transporting, beta R41429 Copper metabolism 0 16 Oestrogen sulphotransferase H79311 Sulphate conjugation 0 16 Rearranged immunoglobulin lambda T71328 Cell/organism defence ± ± Prostaglandin E receptor 2 N28920 Prostaglandin biosynthesis ± ± CYCLIN A2 AA001916 Cell cycle regulator; DNA replication 0 32 Phosphoribosyl pyrophosphate synthase H14586 Purine and pyrimidine biosynthesis 16 0 subunit I DNAX activation protein W92376 Protein tyrosine kinase binding protein 0 16 Insulin-like growth factor 2 R32132 Autocrine growth factors 110 587 Solute carrier family 2 N51166 Glut5 transport fructose 0 16

A summary of the genes which showed consistently elevated expression in the chromosomal instability (CIN) cell lines Caco-2, SW480 and HT29 compared to the MIN cell lines LS174T, SW48 and HCT116. Tentative gene functions were assigned where possible and genes in bold type indicate those selected for further expression analysis. SAGE tag numbers per million tags are detailed where available in HCT116 and Caco2

(H2AZ). CCNA2 and PLK1 were assayed by RT ± these ®ndings, it would therefore seem reasonable to PCR demonstrating their low expression level (40 suggest that over 66% of our gene panel may represent cycles of PCR). Finally, although not clearly di€eren- bona ®de expression di€erences. tially expressed between the MIN and CIN cell lines, Potential functions of each transcript were explored Northern blot analysis of HDAC2 revealed decreased using sequence databases (See Websites and Databases expression in MIN (ranging from approximately 2 ± 6- Section). A large number are derived from receptor fold) with the exception of SW48 (Figure 1). From genes, transcription factor genes, cell cycle related

Oncogene Gene expression profiles in colorectal cancer DS Duncan et al 3256 to be sites of LOH in renal cell carcinoma (Druck et al., 1995). CIN tumours are recognized by gross chromosomal lesions including loss of heterozygosity (LOH). Expression of PTPD1 was undetectable in CIN cells by RT ± PCR. This suggests either low expression of the transcript or potential LOH at the PTPD1 locus in CIN cells. The group of di€erentially expressed genes found in this study did not contain genes previously associated Figure 1 Con®rmation of GDA v1.3 expression di€erences by with MIN. However, the reciprocal expression of Northern blot and RT ± PCR. (a) Con®rmation of elevated H2AZ and HDAC2 in MIN cells (with the exception expression of three transcripts (H2AZ, BTF and PTPD1)in of SW48) may represent a gene `silencing' pathway in MIN colorectal cancer cells. Northern blot analysis was carried out as described previously (Sambrook et al., 1989). Total RNA MIN tumours. PLK and CCNA2 are attractive (10 ± 20 mg) was resolved on 1.0% formaldehyde agarose gels and potential CIN targets due to their involvement in cell transferred to Hybond-N+TM (Amersham PharmaciaBiotech, cycle and DNA damage checkpoints. PTPD1 is not as UK). Northern blot probes were made by radiolabelling PCR clearly associated with the cell cycle, yet has been 32 products with [a- P]dCTP. Where appropriate, clone inserts were implicated previously in tumour progression and LOH isolated from bacterial clones (Incyte Genomics, USA) and radiolabelled with [a-32P]dCTP. Northern blots were probed with (Partanen, 1996). b-actin to control for RNA loading. (b) Con®rmation of elevated This study has identi®ed alterations in gene expres- expression of three transcripts (RANBP2, CCNA2 and PLK1)in sion patterns speci®c to the MIN and CIN phenotypes CIN colorectal cancer cells. cDNA was synthesised from total in colorectal cancer. A number of genes were found to RNA using Superscript II Reverse Transcriptase (Gibco BRL, UK). PCR was performed using Taq DNA Polymerase (Promega be expressed at elevated levels in MIN cells with Corporation, USA). Contaminating genomic DNA was controlled respect to CIN cells. This phenomenon may be by RT ± PCR analysis using intron-spanning primers. RT ± PCR explained by the occurrence of widespread LOH in products were resolved on 12% polyacrylamide gels and 1.5% the CIN phenotype, which accounts for the loss of agarose gels. b-actin was used as a loading control for RT ± PCR expression of speci®c alleles. It may prove fruitful to and Northern blot. #=RT ± PCR; *=Northern blot determine whether any of the genes we identi®ed exhibit LOH in CIN tumours. Other genes were identi®ed which exhibit elevated expression in CIN genes and genes implicated in chromatin structure. with respect to MIN. Mutations in genes due to the PLK1 can a€ect chromosome number and chromo- MIN phenotype may compromise mRNA stability some segregation (Wolf et al., 1997), while over- leading to increased degradation of mutant transcripts. production of RanBP2 in yeast induces mitotic Careful scrutiny of our panel of gene sequences for chromosome disjunction (Ouspenski et al., 1995). simple sequence repeat polymorphisms may reveal Elevated expression of both PLK1 and RanBP2 may MIN markers. a€ect chromosome number and mitosis in CIN, thereby potentially contributing to chromosomal in- Websites and databases stability. It is an attractive possibility that elevated levels of CCNA2, a cell cycle checkpoint regulator Unigene: http://www.ncbi.nlm.nih.gov; The Institute (Badie et al., 2000), in CIN inhibits a cell-cycle for Genomic Research: http://www.tigr.org; National checkpoint contributing to the abnormalities seen in Centre for Biotechnology Information: http:// these cells. BTF3, a transcriptional activator, was www.ncbi.nlm.nih.gov; Online Mendelian Inheritance observed to be overexpressed in MIN cells but a in Man: http://www.ncbi.nlm.nih.gov/Omim; PubMed: speci®c role for this transcript is unclear. It is http://www.ncbi.nlm.nih.gov/pubmed; Serial Analysis conceivable that overexpression of H2AZ in MIN cells of Gene Expression database: http://www.ncbi.nlm.- may cause chromatin remodelling thereby a€ecting nih.gov/SAGE gene transcription. Deacetylation of p53 by HDAC2 is known to silence p53 activity (Juan et al., 2000) and abrogate p53 cell cycle control. Underexpression of HDAC2 was observed by GDA analysis and Northern blot in a subset of the MIN cell lines examined (data Acknowledgments not shown) which may lead to histone hyperacetyla- We thank Dr Pat Vaughan (Department of Biochemistry, tion. This result and the H2AZ data taken together University College, Cork) for assistance with phosphor- imaging. We thank P Hassell for helping with the coding of may suggest reciprocal roles for HDAC2 and H2AZ in GeneMatch v1.0. We gratefully acknowledge ®nancial gene silencing in MIN cells. It has been suggested that support from the Cancer Research Advancement Board H2AZ transport to the nucleus is regulated by masking of the Irish Cancer Society, The Health Research Board of the H2AZ nuclear localization signal by acetylation. (Ireland) & The Research Committee of the Royal College PTPs (protein tyrosine phosphatases) have been shown of Surgeons in Ireland.

Oncogene Gene expression profiles in colorectal cancer DS Duncan et al 3257 References

Badie C, Itzhaki JE, Sullivan MJ, Carpenter AJ and Porter LalA,LashAE,AltschulSF,VelculescuV,ZhangL, AC. (2000). Mol. Cell. Biology, 20, 2358 ± 2366. McLendon RE, Marra MA, Prange C, Morin PJ, Polyak Bardelli A, Cahill DP, Lederer G, Speicher MR, Kinzler K, Papadopoulos N, Vogelstein B, Kinzler KW, Straus- KW, Vogelstein B and Lengauer C. (2001). Proc. Natl. berg RL and Riggins GJ. (1999). Cancer Res., 59, 5403 ± Acad. Sci. USA, 98, 5770 ± 5775. 5407 . Bronner CE, Baker SM, Morrison PT, Warren G, Smith LG, Ouspenski II, Mueller UW, Matynia A, Sazer S, Elledge SJ Lescoe MK, Kane M, Earabino C, Lipford J, Lindblom A, and Brinkley BR. (1995). J. Biol Chem., 270, 1975 ± 1978. TannergaÊ rd P, Bollag RJ, Godwin AR, Ward DC, Parsons R, Myerof LL, Liu B, Willson JK, Markowitz SD, NordenskjoÈ ld M, Fishel R, Kolodner R and Liskay RM. Kinzler KW and Vogelstein B. (1995). Cancer Res., 55, (1994). Nature, 368, 258 ± 261. 5548 ± 5550. Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Partanen S. (1996). Anticancer Res., 16, 943 ± 946. Markowitz SD, Kinzler KW and Vogelstein B. (1998). Paulovich AG, Toczyski DP and Hartwell LH. (1997). Cell, Nature, 392, 300 ± 303. 88, 315 ± 321. Druck T, Kastury K, Hadaczek P, Podolski J, Toloczko A, Sambrook J, Fritch EL and Maniatis T. (1989). Molecular SikorskiA,OhtaM,LaForgiaS,LasotaJ,McCueP, Cloning: A Laboratory Manual. 2nd edn. Plainview, NY: Lubinski J and Huebner K. (1995). Cancer Res., 55, 5348 ± Cold Spring Harbor Laboratory Press. 5353. Souza RF, Appel R, Yin J, Wang S, Smolinski KN, Eshleman JR, Casey G, Kochera ME, Sedwick WD, Swinler AbrahamJM,ZouTT,ShiYQ,LeiJ,CottrellJ,Cymes SE, Veigl ML, Willson JK, Schwartz S and Markowitz K,BidenK,SimmsL,LeggettB,LynchPM,FrazierM, SD. (1998). Oncogene, 17, 719 ± 725. Powell SM, Harpaz N, Sugimura H, Young J and Meltzer Fishel R, Lescoe MK, Rao MR, Copeland NG, Jenkins NA, SJ. (1996). Nat Genet., 14, 255 ± 257. Garber J, Kane M and Kolodner R. (1993). Cell, 75, Strand M, Prolla TA, Liskay RM and Petes TD. (1993). 1027 ± 1038. Nature, 365, 274 ± 276. Heinen CD, Richardson D, White R and Groden J. (1995). Wolf G, Elez R, Doermer A, Holtrich U, Ackermann H, Cancer Res., 55, 4797 ± 4799. Stutte HJ, Altmannsberger HM, Rubsamen-Waigmann H Juan LJ, Shia WJ, Chen MH, Yang WM, Seto E, Lin YS, Wu and Strebhardt K. (1997). Oncogene, 14, 543 ± 549. CW and Kakinoki YJ. (2000). J. Biol. Chem., 275, 20436 ± Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, 20443. Hamilton SR, Vogelstein B and Kinzler KW. (1997). Kinzler KW and Vogelstein B. (1996). Cell, 87, 159 ± 170. Science, 276, 1268 ± 1272.

Oncogene