REGULATION OF SIGNAL TRANSDUCTION AND CHANGES IN PATTERNS OF GENE EXPRESSION IN BREAST AND OVARIAN CANCER

Breast cancer is the most common malignancy among women in North America and accounts for approximately 27% of all malignant neoplasms in this group. Approximately 180,000 new cases of breast cancer are being diagnosed every year in the United States alone and almost 45,000 women die every year from this disease. Although fewer women are afflicted with ovarian cancer, which affects approximately 26,000 individuals annually in the United States, more than 14,000 women die every year from this malignancy. A major reason for the poor prognosis of the disease is the fact that, due to its asymptomatic development, ovarian cancer is usually diagnosed at an advanced stage when aggressive surgery and treatment are still unable to circumvent its lethal progression.

The majority of mammary carcinomas originate in the luminal epithelial layer of the mammary gland, while more than 90% of ovarian carcinomas originate in the ovarian surface epithelial monolayer. Despite all efforts, the etiology of both types of cancers still remains elusive. The identification and elucidation of the molecular components and signals that control different biological processes underlying the regulation of cell growth, differentiation and apoptosis of the mammary and ovarian epithelium is of enormous significance in efforts to understand the molecular events leading to the oncogenic development of these two tissues.

The work in our laboratory focuses on two lines of research: 1) determining the regulatory role of key signaling molecules in cell proliferation and apoptosis, and their deregulated function during mammary and ovarian oncogenesis, and 2) identifying the specific patterns of gene expression associated with breast and ovarian cancer development.

We have identified Tpl2, a serine/threonine protein kinase that is associated with the induction and progression of rodent T cell lymphomas. More recently, this gene has also been implicated in the development of murine mammary carcinomas. Our results suggest that Tpl2 mediates tumor necrosis factor (TNF) and other signals involved both in cell proliferation and in the induction of apoptosis. Treatment of cells with TNFa induces rapid and transient activation of Tpl2, concomitant with its subcellular translocation from the insoluble to the soluble cytoplasmic fraction. Although a Cterminally truncated mutant of Tpl2 is highly oncogenic, and promotes cell cycle progression and morphological transformation in a variety of transformed cells, expression of the wild type protein in a number of cell lines, including REF52, 293, MCF7, and T47D, induces apoptosis. Our data indicate that Tpl2 regulates programmed cell death via its interaction with and phosphorylation of Tvl1, a novel ankyrin repeat protein that was shown to interact also with Raf1 and members of the Bcl2 family of apoptosis-related proteins.

More recently, we showed that Tvl1 participates in formation of the pro-apoptotic Apaf1/caspase complexes and promotes TNFa-induced apoptosis. Akt1 is a serine/threonine protein kinase that was shown to be a direct downstream effector of phosphoinositol (PI) 3kinase and to mediate growth factor receptor generated survival signals. Our preliminary data indicate that Akt1 abrogates the association of Tpl2 with the apoptosome complexes by inhibiting its interaction with Tvl1.

Our goal is to determine the stoichiometry and biochemistry of interaction of Tpl2, Tvl1 and Akt1, and perhaps Akt2, with components of the Apaf-1/caspase apoptosome in MCF7 and T47D mammary carcinoma cell lines, as well as the flow of information between these molecules during the induction of apoptosis. The mechanism underlying the deregulated Tpl-2 function following its Cterminal truncation will be also investigated. We will further examine the potential role of Tpl2 and Akt in the regulation of "anoikis"-programmed cell death, which is triggered by the disruption of cell surface-extracellular matrix interactions and mediated through integrin-dependent signaling pathways. Previous work demonstrated that Akt2 is amplified and/or overexpressed in 10-20% of ovarian carcinomas. It is likely that aberrations in the expression/activity of Akt2 contribute to advantageous growth and survival potentials of affected tumor cells that lead to their further malignant progression and clonal selection. We will examine the role of Akt2 in the regulation of the cell cycle and the induction of anoikis in human ovarian surface epithelial (HOSE) cell lines. These cell lines, which represent an in vitro model of HOSE cell malignant transformation, were developed previously by Drs. Godwin and Hamilton at Fox Chase and comprise a series of HOSE cells, transduced with the SV40 large Tantigen. The cell lines range from mortal/non-transformed to immortal/non-transformed and tumorigenic cell lines.

We are in the process of constructing animal models for the physiological and oncogenic development of the mammary gland and the ovarian surface epithelium. Such models will be constructed on the basis of the novel approach of "tissue reconstitution," and by in situ gene transfer using high-titer retroviral expression constructs. In the case of the mammary gland, tissue reconstitution techniques have allowed for the successful development of reconstituted ductal epithelium. The reconstituted epithelium exhibits normal hormone-responsiveness and is indistinguishable from the normal ductal tree, both in terms of diameter and branching, as well as histology. This animal model development technique offers a number of important advantages over the germline transgenic approach: 1) it avoids the aberrant expression of the studied transgene(s) during early stages of development or the generalized expression in non-target tissues, 2) it facilitates the interplay between normal and manipulated cells, thus approaching a better approximation of the in vivo situation during the early stages of tumor development, and 3) it allows the choice of adequate, constitutively active promoters according to the desired levels of transgenic expression, as well as promoters that respond to specific inducible factors. This approach will be adapted to develop models of mammary and ovarian oncogenesis by introducing oncogenic mutants of various genes, including Tpl2, Tvl1, Akt1/Akt2 and PI 3K(p110). Primary epithelial cells will be infected in vitro with retroviral constructs produced in vitro at high titers (10⁵--10⁶ cfu/ml), and transferred into the cleared mammary fat pad or the ovarian bursa of syngeneic animals treated with human chorionic gonadotropin (hCG) or hCG in combination with pregnant mare's serum gonadotropin (PMSG), respectively, at regimens promoting high mitogenic activity in these two organs.

Alternatively, retroviral expression constructs of Tpl2, Tvl1, Akt1/Akt2

and PI 3K(p110), packaged in vitro andconcentrated to high titers (10⁶--10⁷ cfu/ml), will be used as vehicles for in situ gene transfer to the mammary glands or the ovarian surface epithelium in rats. Retrovirus suspensions will be directly injected into the main mammary gland duct or ovarian bursa of animals treated with hCG, or with a combination of PMSG and hCG, respectively. Administered at previously established regimens, the PMSG/hCG combination induces superovulation resulting in increased mitogenic activity and repair processes in the ovarian surface epithelium, thereby ensuring highly efficient gene transfer.

Upon successful development of animal models for mammary and ovarian oncogenesis, we will create a resource of premalignant and malignant tissue samples and short-term cell lines that we will use to determine the role of Tpl-2, Tvl-1, Akt-1/Atk-2 and PI3-K(p110) in the regulation of cell growth, differentiation and tumorigenesis. In addition, using cDNA expression array techniques and pattern recognition analysis, we will seek to identify early biomarkers of mammary and ovarian oncogenesis, and to test new generation chemopreventive agents and their in vivo mechanism of action.

A major area of research in our laboratory is the implementation of the recently developed cDNA array techniques for the identification and characterization of genetic changes associated with the development of breast and ovarian cancer. Preliminary data have clearly demonstrated the utility of the Atlas^TM cDNA (Clontech, Palo Alto, CA) expression arrays that contain ~600, previously identified and characterized genes. These genes play key roles in diverse biological processes and are arrayed into functional classes that include: cell cycle and growth regulators, activators and inhibitors of apoptosis, oncogenes and tumor suppressor genes, DNA damage response and repair regulators, genes associated with cell adhesion and motility, regulators of angiogenesis, cell-cell interactors and invasion regulators, and growth factors and cytokines.

Highly purified poly(A+)RNAs, obtained from an SV40 Tantigen transduced mortal HOSE cell line (HIO118) and from its tumorigenic counterpart (HIO118NuTu),

were used to synthesize low-complexity cDNA probes. Each of the probes was hybridized to a pair of identical Atlas^TM cDNA arrays. The gene expression patterns in the two cell lines are shown in Figures 1A and 1B. In order to quantitate the obtained array data, the scanned autoradiographic images were overlayed with a custom-made reference grid, using an Interactive Data Language (IDL) (Research Systems, Inc., Boulder, CO), as shown in the enlarged area (Figure 1C). To determine the level of expression of each double-spotted gene on the array, the signal-intensity contained in each grid-square was automatically determined and the average intensity for each gene calculated. The obtained data were plotted against the location of each gene in the array in the form of a line-graph by connecting each row of the array tail-to-head beginning at the left bottom (Figure 2a and 2b). To determine the changes in gene expression in the two mRNA populations, we subtracted the intensity values in graph b (HIO118NuTu) from the values in graph a (HIO118) (Figure 2c). Using the apparent variance in the housekeeping genes (fluctuations about the trend line), we determined a threshold for significance of the differences observed. Following this procedure, three groups of genes were identified: 1) genes, whose changes are within the threshold noise level, 2) genes, whose levels decrease or fully disappear in the transformed (HIO-118NuTu) cell line, and 3) genes, whose intensities increase in the HIO-118NuTu cell line. These data are presented graphically as a correlation plot in Figure 3. As it would be expected, the data localized within the threshold (open diamonds) is strongly correlated (r²=0.94). The genes whose expression decrease or disappear in the NuTu cell line appear in the lower portion of the graph (filled diamonds), and the genes with higher level of expression in the NuTu cell line appear in the upper portion (filled triangles). The list of genes undergoing major changes during the process of transformation can be automatically constructed by referring to the grid, and hence, the identity and function of the genes whose intensity values are off the middle line can be readily determined. We are currently evaluating the utility of a Clontech software for processing Atlas^TM array derived data, soon to be released by the company.

Other commercially available low-density arrays to be used in our studies are the Atlas^TM 1.2 arrays with 1200 genes, or the Genome Systems (St. Louis, MO) series of human and murine Gene Discovery Arrays (GDAs). The human GDAs contain approximately 27,000 non-redundant cDNAs of known genes and sequences from the EST and TIGR data banks, while the murine GDAs contain up to 18,000 such cDNAs. The use of GDAs may allow the identification of novel genes associated with the physiological and neoplastic development of the mammary and ovarian surface epithelium.

Our ultimate goal is to implement multivariate statistical analysis techniques, developed by Drs. Brown, Ochs and Stoyanova, to the obtained cDNA array data. We have obtained preliminary data with computer simulated cDNA array data sets that demonstrate the power of the applied pattern recognition algorithms. The combined cDNA array/pattern recognition analysis approach should create the necessary methodology for the identification of clusters of genes whose coordinated changes in expression may be associated with specific phenotypic changes. Hence, such gene expression changes could be utilized as genetic markers to determine different stages of malignant transformation of the mammary and ovarian surface epithelium. Thus, this study may generate new important insights with regard to the molecular events underlying mammary and ovarian oncogenesis, as well as the response of pre-malignant or malignant lesions to treatment.

PUBLICATIONS

The following publications represent work carried out previously in P. Tsichlis' laboratory:

MAKRIS, A., LIN, J-H., BEAR, S.E., PATRIOTIS, C., McMAHON, C., PRASAD, V.R., BRENT, R., GOLEMIS, E. TSICHLIS, P.N. Touvlo, a novel protein substrate of Raf-1. J. Mol.Biol. (in press).

TSATSANIS, C., PATRIOTIS, C., BEAR, S.E., TSICHLIS, P.N. The Tpl-2 proto-oncoprotein activates the nuclear factor of activated T cells and induces interleukin-2 expression in T cell lines. Proc. Natl. Acad. Sci. USA 95(7):3827-3832, 1998.

TSATSANIS, C., PATRIOTIS, C., TSICHLIS, P.N. Tpl-2 induces IL-2 expression in T cell lines by triggering multiple signaling pathways that activate NFAT and NF-kB. Oncogene 17:2609-2618, 1998.

^§   Fox Chase researcher

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998