A52 AGA ABSTRACTS GASTROENTEROLOGY Vol. 118, No.4

472

ELEVATED C-MYC EXPRESSION IS AN EARLY EVENT IN HU­MAN COLORECTAL TUMORIGENESIS AND CORRELATESWITH NUCLEAR LOCALIZATION OF B-CATENIN.Hanlin Wang, Shih-Fan Kuan, Marc Bissonnette, Thomas Brasitus, JohnHart, The Univ of Chicago Hospitals, Chicago, IL; Univ of Chicago,Chicago, IL.

Background: Nuclear accummulation of a cytoplasmic-membrane-boundadhesion molecule, J3-catenin, and subsequent activation of the J3-cateninlTef pathway secondary to mutations of the APC tumor suppressor and/orJ3-catenin itself have been implicated in the initiation of most sporadichuman colorectal epithelial neoplasms. A recent study utilizing culturedcell lines has demonstrated c-myc as a direct target gene of J3-cateninlTefsignaling (Science 281: 1509,1998). Whether the expression of c-Myc isindeed altered by J3-catenin dysregulation in the early stage of humancolorectal tumorigenesis has not been studied. Materials and Methods:Formalin-fixed paraffin-embedded tissues from 17 human colorectal ade­nomas and 5 hyperplastic polyps were used in this study. The size ofadenomas varied from microscopic (involving only a few crypts) to 4.5 em.All tissue sections from adenomas also contained adjacent normal-appear­ing mucosa for comparison. The expression of J3-cateninand c-Myc wereevaluated immunohistochemically using 2 monoclonal antibodies (clone14 from Transduction Laboratories and clone 9ElO from Santa CruzBiotechnology). Results: Weakly positive membranous (basolateral) andcytoplasmic staining for J3-catenin was present in normal-appearing co­lonic mucosa. Fifteen adenomas (88%) showed an increase in J3-cateninlevels with at least focal unequivocal nuclear staining. Among these cases,II (73%) also showed an increase in the expression levels of c-Myc whencompared with adjacent non-neoplastic epithelium. In the remaining 4cases, the c-Myc levels were comparable to that in normal-appearingmucosa. In those 2 adenomas where nuclear staining for J3-cateninwas notobserved, the c-Myc levels were either lower than or comparable to that innon-neoplastic epithelium. These changes were not tumor-size-dependent.When compared with normal-appearing mucosa, hyperplastic polyps didnot show significant changes in the levels and distribution patterns for bothJ3-catenin and c-Myc. Conclusion: Increased c-Myc expression occursearly in human colorectal tumorigenesis, which correlates with nuclearlocalization of J3-catenin. These effects do not appear to be proliferation­dependent since similar changes are not observed in hyperplastic polyps.

474

COX-2 SELECTIVE INHIBITORS INDUCE APOPTOSIS BYBOTH COX-2 DEPENDENT AND COX-2 INDEPENDENT MECH·ANISMS.Banke Agarwal, Prabhakar Swaroop, Petr Protiva, Ram Chuttani, WilliamG. Ramey, Peter R. Holt, Beth Israel Deaconess Med Ctr, Boston, MA;Brigham & Women's Hosp, Boston, MA; St Luke's-Roosevelt Hosp Ctr,New York, NY.

Background: Cox-2 selective inhibitors induce apoptosis in colon cancercells and have a chemopreventive effect against colon cancer, in animalmodels. We studied the role of Cox-2 inhibition in apoptosis induced by theCox-2 selective inhibitor, SC236 a structural analogue of Celecoxib. Meth­ods: Two colon cancer cell lines, HCTll6 (which lack Cox-2) and HT29(which express Cox-2) were used. Apoptosis was quantified by flowcytometry. Lovastatin (Lova) 15-30/LM was used to increase expressionand curcumin (Cure) 25-50/LM to decrease expression of Cox-2. Results:SC236 reduced cell number and induced apoptosis in both HCTll6 andHT29 cells, implying that SC236 can inhibit proliferation and induceapoptosis by a Cox-2 independent mechanism. More apoptosis occurred inHT29 than in HCTll6 cells, suggesting that in cells that express Cox-2,Cox-2 inhibition can induce apoptosis separately from the Cox-2 indepen­dent apoptosis. We then studied the effect of changes in Cox-2 expressionon SC236-induced apoptosis. Lova increased and Cure decreased Cox-2expression in HT29 cells. Cox-2 is undetectable in HCTll6 cells andremains so after Lova or Cure treatment. Lova increased SC236-inducedapoptosis in Cox-2 lacking HCTll6 cells consistent with Cox-2 indepen­dent augmentation of apoptosis (consistent with our previous publisheddata). In HT29 cells, an increase in Cox-2 expression prevented theanticipated increase in apoptosis by Lova, suggesting that increased Cox-2was protective against apoptosis. Conversely, Cure decreased Cox-2 ex­pression and markedly increased SC236-induced apoptosis in HT29 cells,confirming the important role of Cox-2 in SC236 induced apoptosis. InHCTll6 cells, Cure decreased SC236-induced apoptosis, probably byinhibiting activation of AP-l. (table) Conclusion: The selective Cox-2inhibitor, SC236 induces apoptosis by two mechanisms, one Cox-2 inde­pendent and the other Cox-2 dependent. The latter is activated by inhibitionof Cox-2 in Cox-2 expressing cells.

475

HYPOMETHYLATION OF THE CYCLOOXYGENASE-2 PRO­MOTER IN GASTRIC ADENOCARCINOMAS WITH MICROSAT­ELLITE INSTABILITY.Mahmood Akhtar, Romina M. Magno, Jing Yin, Albert S. Fleisher, GenTamura, Stephen 1. Meltzer, Keith T. Wilson, Univ of Maryland Sch ofMedicine, Baltimore, MD; Yamagata Univ, Yamagata, Japan.

Background: Hypermethylation of CpG-rich regions of gene promoters hasbeen shown to downregulate expression of DNA mismatch repair genes ingastric carcinomas manifesting microsatellite instability (MSI). Elevatedcyclooxygenase(COX)-2 expression is linked to gastrointestinal carcino­genesis and has been described in H. pylori gastritis and gastric cancers.Aim: To determine whether altered COX-2 promoter methylation occurs ingastric cancers with MSI or in gastritis tissues. Methods: DNA wasextracted from gastric carcinomas and matching normal gastric tissuesfrom 40 patients, and from inflamed and normal gastric tissues from 3patients with H. pylori gastritis. Tumor MSI status was assessed byanalysis of 5 NIH consensus loci. COX-2 promoter methylation status wasdetermined by methylation-specific peR performed on bisulfite-modifiedDNA using primers specific for unmethylated or methylated promotersequences. Results: An unmethylated COX-2 PCR product was present inall tissues, consistent with results obtained on other genes studied by ourgroup and others. When compared with paired normal tissues, COX-2promoter hypomethylation was observed in 45% of tumors with MSI (seeTable). In contrast, in tumors without MSI and in gastritis, COX-2 pro­moter hypomethylation occurred in 5% and 0% of cases, respectively.Conclusions: In gastric tumors with MSI, hypomethylation of the COX-2promoter may represent an alternative pathway of COX-2 induction in

Effect ofmodulation ofCox·2 onSC236·induced apoplosis

lovastatlnCox·2 Apoptosis

CurcuminCox·2 Apoplosis

t.j,

.j,absent

nochanget

tabsent

+++

absent

BaselineCox·2

HT29HCT116

473

GASTRIN GENE EXPRESSION DURING MALIGNANT TRANS­FORMATION MAY BE ACTIVATED BY A MUTATION IN THEAPCGENE.Susan A. Watson, Andrew Smith, Christopher Parasceva, Dov Michaeli,Univ of Nottingham, Nottingham, United Kingdom; Univ of Bristol, Bris­tol, United Kingdom; Aphton Corp, Woodland, CA.

Introduction:Gastrin gene expression has been observed in normal as wellas malignant intestinal mucosa in the APCM in and APCI638 mouse modelsof familial adenomatous polyposis. Gastrin gene expression has also beenshown in normal mucosa taken from patients with colonic adenocarcinoma.Aim: To determine whether early activation of the gastrin gene during theadenoma:carcinoma sequence in the colon was mediated via a mutatedAPC gene Methods: Three human adenoma cell lines with known muta­tional status were assessed for gastrin gene expression. AAC115BlOC andAACI had truncating APC mutations whereas RGC2 expressed full lengthAPC protein. The human colonic cell line, HT29, with a zinc-inducibleAPC gene was kindly provided by Professor Bert Vogelstein. To inducewild type APC gene expression, the cell line was cultured in mediumcontaining lOO-200/LM ZnCI2 in serum-containing medium. HT29 with azinc-inducible J3-galactosidase gene was used as the control. Total cellularRNA was extracted from cell pellets and reverse-transcribed from randomhexamer primers. Real Time PCR was performed using the 5700 SequenceDetection System (PE Applied Biosystems). Each PCR was performedwith a reaction buffer containing SYBR Green. The fluorescence of theSYBR green dye bound to the gastrin and GAPDH PCR products wasmeasured after each cycle and the cycle number was recorded when theaccumulated signal crossed an arbitrary threshold (Ct value). The relativegene expression for each sample was determined using the formula 2et

_

(GAPDH)-ct(gastrin) Results: The human colonic adenoma cell lines, AACIand AAC1l5BlOC, expressed the gastrin gene with ~aCts of lxlO- 2 and1.5xlO-', respectively. RGC2 had no detectable gastrin gene expression.HT29-APC cells were shown to express the gastrin gene (aaCt 0.57) asdid HT29-J3gal cells (~~Ct 0.52). Nine hours after zinc induction, at apoint when approx. 20% of cells were apoptosing, the ~~Ct of HT29-APCcells was reduced to 30% of that of HT29-J3gal (p=O.OOI). This wasmaintained at 24 hours when >95% of the cells were undergoing apopto­sis. Conclusions:A mutation in the APe gene may activate gastrin geneexpression directly or indirectly. This may occur at an early phase of theadenoma: carcinoma sequence and may provide a viable therapeutic targetfor the inhibition of colorectal carcinogenesis.

April 2000

gastric cancer. COX-2 promoter hypomethylation does not occur fre­quently in gastritis or in gastric tumors without MSI.

Frequency ofhypomethylation ofCOX-2 promoter ingastric cancers vs. adjacent normal tissue

Tissue Type Hypometh % P*

MSI.high Iumors 7/14 50 0004MSI·low tumors 2/6 33 NSMSltotal 9/20 45 0008MSI·neg 1/20 5Gastritis 0/3 0 NS

'Fisher's exact test, vs. MSI-neg

476

EXPRESSION OF CYCLO-OXYGENASE-2 PROTEIN IS ASSOCI­ATED WITH DYSPLASIA, K-RAS MUTATION AND EXPRES­SION OF P53 IN COLORECTAL ADENOMAS.Robert Benamouzig, Elisabeth Longchampt, Helen Yoon, Eric Jullian,Antoine Martin, Thierry Coste, Daniel Couturier, Jacques Rautureau,Stanislas Chaussade, Hosp Avicenne, Bobigny, France; Hosp Cochin,Paris, France.

Cyclo-oxygenase-2 (COX-2) is an inducible enzyme that catalyzes theconversion of arachidonic acid to prostaglandins and thromboxanes.COX-2 protein expression is observed in colorectal neoplasms and may bea target for anticolorectal cancer activity of NSAIDs and aspirin in humans.The purpose of this study was to evaluate COX-2 protein hyperexpressionin 182 adenomas obtained from the 116 first consecutive patients (73males, 43 females, age 58 +/- 9 years) included in a large prospectivemulti-center randomised study aimed to evaluate the effect of long termdaily use of low dose aspirin in reducing the occurence of new adenoma­tous polyps. All patients have had colonoscopy to the cecum with adequatepreparation resulting in clearance of either a single adenoma> 10 mm insize or 3 adenomas of any size. Methods : All biopsy specimens wereblindly assessed for architectural pattern and dysplasia by 2 independentpathologists. Immunohistochemistry was carried out on formalin-fixedparaffin-embedded sections with specific anti-COX-2 antibody and anti­P53 antibody. Staining intensity was scored 0-3 by two blinded indepen­dent observers. After DNA extraction from these specimens, PCR ampli­fication and sequencing were performed to detect K-ras mutation. Results: COX-2 epithelial cells staining was heterogenous both between cryptsand inside a crypt. Superficial COX-2 staining was also observed ininterstitial cells. An high to moderate hyperexpression of COX-2 wasobserved in 36% of adenomas (score 2 and 3), COX-2 hyperexpression wasmore frequent in adenomas exhibiting high grade dysplasia (p<0,05) andintense p53 immunostaining (p<O,05). This hyperexpression seems alsomore frequent in K-ras mutated adenomas (p=0,07) and in left sidedadenomas (p=0,08). No relation with age, sex, BMI, endoscopic polypstype (pedunculated or sessile) and size, histological types (tubular, tubulo­villous or villous) nor personal or familial history of colorectal polyps orcancer were observed. Conclusion: COX-2 protein hyperexpression isfrequently observed in colorectal adenomas. The correlation betweenCOX-2 protein hyperexpression and dysplasia suggest a role of COX-2 incolorectal carcinogenesis from the early steps. The absence or presence ofthis expression may have implications for the putative chemopreventiveactivity of aspirin or COX-2 inhibitors.

477

CYCLOOXYGENASE-2 PROTEIN VARIES WITH TUMOR PRO­GRESSION IN EXPERIMENTAL COLON CANCER.Charlene Compher, John L. Rombeau, Noel N. Williams, Univ of Penn­sylvania, Philadelphia, PA.

Introduction: Cyclooxygenase-2(COX-2)is upregulated in colon tumorsand cell lines, and may be an early event in carcinogenesis progression.

AGAA53

COX-I is constitutively expressed in colonocytes. We hypothesized thatCOX-2 expression would be upregulated during tumor initiation and pro­motion, while constitutive COX-I would remain unchanged. Methods: Fortumor initiation F344 rats, induced with 20 mgfkg ip azoxyrnetha­ne(AOM), were harvested after 5 days. For tumor promotion, harvest was8 weeks after the intitial injection of 2 weekly AOM doses. Tissue wasstained for COX-2 and COX-l by IHC. Results: COX-2 was not seen invehicle-treated rats(data not shown). Data are mean +/- SEM(n);likesuperscripts, p<0.05, unpaired t-test.(table) Conclusion: Increased COX-2protein during tumor initiation may relfect a generalized inflammatory orcarcinogen activation response, which also occurs to a lesser degree withCOX-I. The high rates of COX-2 staining in proximal colon at tumorinitiation had resolved by tumor promotion. At tumor promotion, increasedCOX-2 in distal versus proximal colon is associated with increased aber­rant crypt foci (ACF) and typical of the location of colon tumors in thismodel and human sporadic colon cancer. The increased ACF may reflectCOX-2 tumorigenic effects. We conclude that COX-2 expression mayserve different functions in the same tissue site during colon cancerprogression.

(% total epithelial cells I proximal middle distal totalcolonsection)

Initiation %-t(;ells/cryptCOX-2 (14) 12.2+1-0.67' 9.82+/-0.6 8.17+/-0.6 9.66+/-0.44b

COX·1 (8) 7.7+/-0.76<.1 5.15+/-0.36 6.07+/-0.3'·, 6.1 +/-0.32'·hPromotionCOX·2 (16) Qa,i 4.66+/-0.31 6.35+/-0.37' 3.67+/-019b

COX·1 (9) 1.08+/-0.17' 4.34+/-0.27 3.88+/-0.13' 3.24+/-0 15'Vehicle Control COX·1 (5) 3.09+/·0.6' 4.94+/-0.23 3.25+/-0.33'·, 3.89+/-025hACF(nlcm'l 2.18+/-0.04(30)' 6.22+1-0.1 (38)'

478

INVOLVEMENT OF MAPK PATHWAYS IN INDUCED APOPTO­SIS AND THE UPREGULATION OF CYCLOOXYGEANASE-2 BYTHE COX-2 SELECTIVE INHmITOR NS-398 IN HUMAN COLO­RECTAL TUMOR CELL LINES.Douglas J. Elder, Dawn E. Halton, Christos Paraskeva, Univ of Bristol,Bristol, United Kingdom.

Background: Non-steroidal antiinflammatory drugs (NSAIDs) are chemo­preventive for colorectal cancer. An important component of their anti­neoplastic action is the ability to inhibit proliferation and to induce apo­ptosis of colorectal tumor cells. The signalling pathways by which theseoccur require further definition. We have previously shown the COX-2selective NSAID NS-398 to induce COX-2 protein expression and apo­ptosis in the HT29 colorectal carcinoma cell line. The MAPK pathwayshave been shown to positively regulate COX-2 protein expresssion. Aim:To investigate the involvement of MAPK pathways in the regulation ofCOX-2 expression and induction of apoptosis by NS-398. Methods: Sub­confluent cultures of human colorectal tumor cells were treated with theCOX-2 selective inhibitor NS-398 (20·IOOJ.LM) and/or inhibitors of thekinases p38 (SB203580, 5J.LM) and MEK (UOI26, 1-1OJ.LM) for up to 96h.Following treatment, the cell yield and extent of apoptosis was determinedin control (vehicle) and treated cultures. Cell lysates were analysed byWestern blotting. MEK activity was assessed by determining the phos­phorylation of ERK using polyclonal phospho-specific antibody. COX-2expression was determined using a polyclonal antibody to COX-2. Results:NS-398 upregulated COX-2 protein expression in a dose-dependent man­ner in the colorectal adenoma cell lines AAICI and RRlCI as well as in thecarcinoma line HT29. This was further examined in HT29 cultures wherethe presence of SB203580 or UOl26 completely inhibited the upregulationof COX-2 by NS-398. At a concentration of I J.LM, U1026 had no effect onthe proliferation (determined by the cell yield) or apoptosis of HT29 cellsand, although it did not fully inhibit MEK activity, inhibited the anti­proliferative effect of NS-398 (40J.LM) by 53± II %. This was associatedwith the inhibition of NS-398-induced apoptosis. Conclusions: The up­regulation of COX-2 protein expression by the COX-2 selective inhibitorNS-398 in human colorectal tumor cells involves the ERK and p38 sig­nalling pathways. The inhibition of NS-398-induced apoptosis by blockadeof the ERK pathway suggests an important role for this pathway insignalling apoptosis induced by this class of NSAID.