138 Abstracts

35 MODEL OF COMPLEX DYNAMICS OF BREAST CANCER KAYROTYPE: SIMULATION AND IN VITRO EXPERIMENTS. Yuriy Gusev, Alfred0

Tendler. Arlene Ashcraft, Wilham Dooley. The Johns Hopkins Univ., School of Med., Baltimore, MD 21287

We have proposed that karyotype heterogeneity in tumors is a complex dynamic process. The result IS relatively “chaotic” karyotypes with preservation of certain karyotypic clonal patterns. This chaos may depend on only a few non-random events within the cell cycle of each individual cell. We have modeled the effects of each of the known MITOTIC abnormalities. Current results of computer simulation show the impact of each particular mechanism alone, and In combination with other mechantsms. Non-dIsJunctIon alone may produce the classic unstable, chaotic dynamics of chromosome number distribution seen In tumors. It may be a major source of variabihty between individual cells and, at the same time, stability of the population DNA distribution. The model also shows an important role of chromosomally- dependent cell death regulation with regard to shift of modal chromosome number and increase of variation of the chromosome number dlstrlbutlon. The simulations implementing alternative mechanisms of aberrant mitoses have produced outcomes that are distinguishable from each other and have been tested by comparison with data of rn vitro observation of karyotype distribution for two human breast cancer cell lines lT47-D, SK-BR-3) and human normal breast fibroblasts line HS5788ST. Results of in vitro experiments have confirmed the important model prediction of the short-term and long-term effects of non-disjunction as a malor contributor to the chaotic karyotype seen within clonal tumor cell populations.

37 ANALYSIS OF DNA SEQUENCE COPY NUMBER VARIATIONS BY COMPARATIVE GENOMIC HYBRIDIZATION (CGH I

D. Pinkelt. Y.P. Shit. J. Gray’. G. .Mohapatra). B. Feuerstem’. D. Sudar*. J. Mullikin? D. Moore’. L. Godley), H. Varmus-‘. L. Donehower’. tUmversny of Califomta. San Franctsco. *Lawrence Berkeley Laboratory. JNattonai lnstnutes of Health, 4Baylor College of Medicine.

The human genome contams approxtmate!y 3x 1 O9 base paus of DNA sequence. Techmques which allow effictent scanning for indrcations of abnormalny are requtred for studtes of mahgnancy smce tt is clear that many new cancer genes rematn to be discovered. Comparative genomic hybrtdtzation ICGH) was developed for detection of regions of aberrant DNA sequence copy number anywhere in the genome. which may indicate the presence of oncogenes or suppressor genes. CGH involves the simultaneous hybridtzatton of differenttally labeled genomic DNA from tumor and normal cells to normal metaphase chromosomes. The ratto of the binding of the two genomic DNAs to a location on the target chromosomes IS proporttonal to the ratto of the copy numbers of sequences tn the two genomes that bind there. Since CGH requtret only genomtc DNA from the tumor analysis can be performed directly on fresh or fixed tissue. CGH provtdes data on the average genomic properties of the cell population bemg studted. includmg whatever normal cells may be mcluded in the spectmen. Thus tt does not distinguish between dtfferent clonal populations tn the tumor.

CGH can be used for a variety of studies such as detectton of previously unknown regions of vanant copy number, determination of their prevalence in different tumor types. correlation of specific abnormahttes with patient outcome, and measurement of some aspects of genornic instabihty at the chromosomal level. An example of the latter ts measurement of copy number changes tn mammary tumors in Wnr-I transgenic mtce with 0.1 and 2 functional coptes of p53 in the germ line. Tumors in mice with 2 germ line copies had few regions of copy number abnormality. while those with 0 had more. The highest frequency of copy number changes occurred in tumors m mice with I germ line copy of p53 that was lost in the tumor. Work supported by National Cancer Institute Grants CA59967, CA54897, CA25215. CA45919. CA39832.

36 EXTENT OF 20413 AMPLICONS M PRIMARY BREAST CANCER AND ASSOCIATION WITH INDICATORS OF POOR PROGNOSIS

Mittna Tanner’. Mika Tiikkonen’. Anne Kallioniemi’. KaiJa Hall’. Cohn Collins’. Joe W Gray’. Jorma Isola’. O-P Kallioniemi’ I ) Lab. Cancer Genetics. Tampere Univ. Hospital and Institute Medical TechnoloE. Tarnpere. Finland. 2) Div Molecular Cytometry. Uti. California. San Francisco. CA.

Amplification of the 2Oqlj retion ia breast cancer was originally found by comparatne genomic hybridization (CGH) Subsequently. the mimmal common region of amplification was narrowed down by interphase FISH to _ 1.5 Mb and most candidate genes were excluded A positional clonin_e effort to idea@ the target gene IS in progress Here. we have determined the prevalence of 2Oq13 amplibcation in 133 primary and 11 recurrent breast carcinomas by FISH using a probe (KMC2OCOO I ) mapping to the critical region of amphhcation. The size of the amplicons in primary tumors and association of amplification with phenotypic properties of the tumor were also evaluated.

The prevalence of KMC2OCOOI amplification by FISH was 23% (30/133) in primary and 36% (5/14) in recment tumors. Nine primary tumors (6%) showed high level (‘3-fold) amplification. 2Oq I3 amplification was more common in high grade (p=O.O07). DNA aneuploid (p=O.O095) and highly proliferative (p=O.O25) tumors. Phosphotyrosine phosphatase gene (PTPNI). located 3-5 Mb cemromeric horn KMCZOCOOI. was co- amplified in six cases with high-level 2Oq 13 amplification. Markers between these two loci were not amplified. However. PTPN I gene was never amplified alone and in the co-amplified cases amplilication levels were usuaJJy lower than those of RMC2OCOO I

In conclusion. the results suggest a complex structure for 2Oq I3 amplilication. PTPN I gene is invotved in the amplicon. and known to be often overexpressed. However. it is located outside the minimal region and is therefore not likely to be the primarv target of amplification. The higher prevalence of 2Oql3 amplification in recurrent tumors. as weU as the correlations with markers of poor prognosis. suggest that the gene(s) activated b! the 20q 13 amplification confer an agaessive phenotype.

38 BCL-2 GENE FAMILY AND THE REGULATION OF PROGRAMMED CELL DEATH, Staniey J. Korsmeyer, Xiao-Ming Yin, Elizabeth Yang, Jiping Zha,

Thomas Sedlak and Zoltdn Oltvai, Howard Hughes Medical Institute, Washington University School of Medicine, St. Louis, MO. The Bcl-2 gene was identified at the chromosomal break point of t(14;18) bearing B cell lymphomas. Bcl-2 has proved to be unique among proto-oncogenes in blocking programmed cell death rather than promoting prolifer- ation. In adults, 6~1-2 is topographically restricted to progenitor cells and longlived cells but is much more widespread in the developing embryo. Transgenic mice that overexpress 2~1-2 in the B cell lineage demonstrate extended cell survival, and progress to high grade lymphomas. 6~1-2 deficient mice complete embryonic development and display relatively normal hematopoietic differentiation but undergo fulminant lymphoid apoptosis of thymus and spleen. Moreover, they demonstrate profound renal cell death and develop polycystic kidney disease and also hair hypopigmenta- tion. A family of 8~1-2 related proteins regulate cell death and share highly conserved BHl and BH2 domains. BHl and BH2 domains of Bcl-2 were required for it to heterodimerize with Bax and repress apoptosis. A yeast two-hybrid assay defined a selectivity and hierarchy of further dimerizations. Eax also hetero-

Mel-1, and Al. Substitutions in dimerizes with,Bcl-x~, BHl of Bcl-xLdisrupt d its heterodimerization with Bax and also abrogated its inhibition of apoptosis. Both yeast two-hybrid system and lambda expression cloning identified a new protein, Bad, which has homology clustered in BHl and BH2 domains. Bad hetero- dimerizes with Bcl-x s_tronger than &l-Z. Bad reverses the protective effec of Bcl-x c restoring apoptosis by preferentially dimerizing w\;h Bcl-x and decreas- ing the amount of Bcl-x /Bax complexes.

b LThe competi-

tion of Bad with Bax fo Bcl-xLresults in more unbound Bax, leadint to unopposed cell death. Thus, the susceptibility to a death stimulus in a given cell is dictated by a complex set point determined by the relative levels and interactions amongst these family members.