Reles, Angela : MOLECULAR GENETIC ALTERATIONS IN OVARIAN CANCER The Role of the p53 Tumor Suppressor Gene and the mdm2 Oncogene

1

Kapitel 1. INTRODUCTION

1.1 Clinical and histopathological aspects of ovarian tumors

1.1.1 Diagnosis and clinical aspects

Ovarian cancer is the leading cause of death from gynecologic malignancies, and the fifth most common malignant condition among women in the United States, with an annual incidence of 25.400 and approximately 14.500 deaths each year (Landis et al. 1998). The majority of women, approximately 70%, present with advanced stage disease with either regional or distant metastases at the time of diagnosis (Landis et al. 1998).

The incidence of ovarian cancer increases with age and peaks in the eighth decade. The rate increases from 15.7/100,000 in the 40 to 44 age group to a peak rate of 54/100,000 in the 75 to 79 age group. The median age at diagnosis is 61 years (Averette et al. 1995). The risk of developing ovarian cancer in a lifetime is only 1.4% in women with negative family history, but can increase to 14.6%- 32.2% in women with a family history and germline mutations of the BRCA1 gene at the age of 60 years (Tortolero-Luna et al. 1994, Laplace-Marieze et al. 1999, Whittemore et al. 1997).

Epidemiologic studies have shown weak or mixed degrees of correlation with increased risk for ovarian cancer for tobacco smoke, radiation exposure, talc in genital hygiene, psychotropic medication, the mumps virus, high level physical activity, and dietary factors (as reviewed by Holschneider and Berek, 2000). Infertility has been found to be a significant risk factor for ovarian cancer, while parity, lactation, and oral contraceptive use have a protective effect (as reviewed by Holschneider and Berek, 2000). This supports the „incessant ovulation„ hypothesis for ovarian cancer. According to this hypothesis, ovarian cancer develops from an aberrant repair process of the surface epithelium which is ruptured and repaired during each ovulatory cycle (Fathalla, 1971). Long term use of oral contraceptives reduces the risk of developing ovarian cancer by 50% (Whittemore et al. 1992).

The diagnosis of ovarian cancer is mainly based on bimanual palpation and transvaginal sonography. CT and MRT can be useful to obtain information about the spread of the disease, lymph node involvement and liver metastases. The ultrasound criteria of an adnexal mass suspicious for malignancy are the presence of cystic or polycystic tumors with solid structures, completely solid structures, thick irregular septae, ascites, and evidence of infiltration into adjacent organs (Reles et al. 1997, 1998). The sensitivity of transvaginal ultrasound for the diagnosis of ovarian cancer in this tumor cohort was as high as 95% and specificity was found to be 82% (Reles et al.


2

1998). Transvaginal color Doppler sonography is also routinely used for the diagnosis of ovarian cancer since the blood vessel resistance in malignant ovarian tumors is significantly lower than in benign tumors due to tumor neovascularization. The value of the method though is limited by a relatively low specificity of 69% (Reles et al. 1998).

1.1.2 Histological classification of epithelial ovarian tumors

Malignant epithelial ovarian tumors, including tumors of borderline malignancy, account for almost 90% of all ovarian carcinomas (Scully, 1979). The tumors of "common epithelial origin„ arise from the surface epithelium of the ovary. During embryonic life the celomic cavity forms and is lined by a layer of mesothelium, parts of which become specialized to form the serosal epithelium covering the gonadal ridge. By a process of invagination, this same mesothelial lining gives rise to the müllerian ducts, from which arise the fallopian tube, uterus, and vagina (Kurman, 1987).

As the ovary develops, the surface epithelium extends into the ovarian stroma to form inclusion glands and cysts (Clement, 1987). The epithelium, in becoming malignant, exhibits a variety of müllerian type differentiations: serous (resembling fallopian tube), mucinous (resembling the endocervix), endometrioid (resembling endometrium), and clear cell (cells resembling endometrial glands in pregnancy) tumors. The ovarian surface epithelium, within its repertoire of differentiation, can also exhibit urothelial differentiation in the form of transitional cells. Table 1 presents the histologic classification scheme which has been developed and continuously updated under the auspices of WHO, International Federation of Gynecology and Obstetrics (FIGO), International Society of Gynecologic Pathologists, and Society of Gynecologic Oncologists (Scully, 1979).

1.1.3 Therapeutic aspects of epithelial ovarian tumors

The standard of care for newly diagnosed advanced ovarian cancer currently includes cytoreductive surgery, followed by combination chemotherapy with platinum and paclitaxel. Optimal tumor debulking is critical in the management of ovarian carcinoma and residual disease has been shown to be one of the most important factors influencing survival in patients with advanced ovarian carcinoma (Makar et al. 1995, Eisenkop et al. 1998). Patients in stage IIIc and IV ovarian carcinoma have a


3

Table 1: Histologic classification of common epithelial tumors of the ovary

SEROUS TUMORS

 

Benign:

Cystadenoma and papillary cystadenoma

 

Surface papilloma

 

Adenofibroma and cystadenofibroma

Borderline:*

Cystadenoma and papillary cystadenoma

 

Surface papilloma

 

Adenofibroma and cystadenofibroma

Malignant:

Papillary adenocarcinoma and papillary cystadenocarcinoma

 

Surface papillary carcinoma

 

Malignant adenofibroma and cystadenofibroma

MUCINOUS TUMORS

 

Benign:

Cystadenoma

 

Adenofibroma and cystadenofibroma

Borderline:

Cystadenoma

 

Adenofibroma and cystadenofibroma

Malignant:

Adenocarcinoma and cystadenofibroma

 

Malignant adenofibroma and cystadenofibroma

ENDOMETRIOID TUMORS

Benign:

Adenoma and cystadenoma

 

Adenofibroma and cystadenofibroma

Borderline:

Adenoma and cystadenoma

 

Adenofibroma and cystadenofibroma

Malignant:

Adenocarcinoma

 

Adenoacanthoma

 

Adenosquamous carcinoma

 

Malignant adenofibroma and cystadenofibroma

Epithelial-stromal

 

and stromal:

Adenosarcoma

 

Stromal sarcoma

 

Mesodermal (müllerian) mixed tumors, homologous and heterologous

CLEAR CELL (MESONEPHROID) TUMORS

Benign:

Adenofibroma

Borderline

 

Malignant:

Adenocarcinoma (carcinoma)

TRANSITIONAL CELL TUMORS

Benign

Brenner Tumor

Borderline:

Proliferating Brenner Tumor

Malignant

Malignant Brenner tumor

 

Transitional cell carcinoma

MIXED EPITHELIAL TUMORS

Benign

 

Borderline

 

Malignant

 

UNDIFFERENTIATED CARCINOMA

UNCLASSIFIED EPITHELIAL TUMORS

* Tumors of borderline malignancy / Carcinoma of low malignant potential (LMP)


4

significantly better survival if postoperatively no residual tumor remains as compared to those who have a residual tumor of 10 mm or more (Eisenkop et al. 1998).

As a standard surgical approach a midline incision extending from the symphysis to above the umbilicus or to the xiphoid is performed. After obtaining histological confirmation by frozen section, bilateral salpingo-oophorectomy, hysterectomy, infragastric omentectomy, and pelvic and paraaortic lymphadenectomy are performed. Appendectomy is optional. Resection of the small intestine, colon, peritoneal surfaces, partial gastrectomy, partial liver resection, and urinary tract surgery are performed as necessary, depending on the extent of metastasis. Diffuse peritoneal spread of the carcinoma is treated by infrared light thermocoagulation. The incidence of lymph node metastasis has been reported to be as high as 64%-75% in advanced stage carcinomas, but even in stage I carcinomas, lymph node metastasis have been found in up to 24% of the cases (Burghardt et al. 1991, Petru et al. 1994). Pelvic and paraaortic lymphadenectomy is therefore routinely performed at the Department of Gynecology, Charité Campus Virchow-Klinikum as part of the primary surgery. The aim of surgery is optimal debulking with resection of all visible tumor.

Surgery is followed by adjuvant combination chemotherapy in all cases of FIGO-stage Ic-IV. In stage Ia and Ib disease, the decision depends on the histologic grading of the tumor. Stage Ia/Ib ovarian carcinomas with grade I do not need to receive adjuvant chemotherapy while patients with G3 tumors would be treated with chemotherapy. In stage Ia/Ib, grade II tumors, decision should be made individually.

The standard combination chemotherapy for adjuvant treatment of primary tumors is Carboplatin and Paclitaxel. Application of Cisplatin (75 mg/m²) and Paclitaxel (135 mg/m², 24h infusion) has been shown to be superior to Cisplatin in combination with Cyclophosphamide (750 mg/m²) (McGuire et al. 1996). Because of the toxicity profile of Cisplatin, especially nephro- and neurotoxicity, and similar reponse rates of Cisplatin and Carboplatin, the present regimen for most hospitals as first line treatment of ovarian cancer is Carboplatin AUC 6 and Paclitaxel 175 mg/m² (3-h-infusion). However, resistance to chemotherapy remains a complex problem. Despite high overall clinical response rates of up to 80% including a high proportion of complete reponses achieved with platinum-based therapy, most patients subsequently relapse and require additional treatment (McGuire and Ozols, 1998).

A consensus for the treatment of recurrent ovarian cancer has not been established so far. Patients with late recurrency (>12 months after completion of primary chemotherapy) benefit from surgical tumor debulking in terms of median survival and response to a second line chemotherapy. In a multidisciplinary approach in a team of gynecologists, surgeons and urologists we try to achieve optimal tumor debulking by surgery for recurrent disease followed by second line chemotherapy.


5

Patients who were initially sensitive to platinum-based chemotherapy have demon-strated relatively high response rates (40%-50%) to second-line platinum-based therapy (Markman et al. 1991).

Promising results have been found for Topotecan, a topoisomerase I inhibitor (Hycamtin, SmithKline Beecham Pharmaceuticals, Philadelphia, PA) with response rates of up to 31% in recurrent ovarian cancer, depending on initial response to platinum-based chemotherapy (ten Bokkel Huinink et al. 1997). Another substance that may play an important role for second line chemotherapy in the future is Gemcitabine, a novel nucleoside analog, which has been shown to have response rates as high as 19% in patients with recurrent ovarian cancer who had been resistant to previous first line Cisplatin chemotherapy (Lund et al. 1994).

Other substances for chemotherapy of recurrent ovarian cancer include Etoposide, Treosulfan, pegylated liposomal Doxorubicin, and Docetaxel. The primary goal of therapy in relapsed ovarian cancer is to extend survival time by maximizing all available therapies while minimizing side effects and preserving quality of life.

1.1.4 Clinical course and prognosis

Although relative 5-year survival is 83%- 88% in stage FIGO I, and 59-64% in FIGO II, the more advanced stages III and IV have a survival of approximately only 30% and 18% respectively (Kosary, 1994). Overall, earlier diagnosis by vaginal ultrasound, radical debulking surgery and chemotherapy have contributed to a significant improvement in the 5-year survival rate from 36% in 1974-1976 to 47% in the years 1986-1993 (Landis et al. 1998).

Established prognostic factors of ovarian cancer are FIGO stage, residual tumor after surgery, histological grading, lymph node status, age, and amount of ascites in primary surgery (Kosary, 1994, Makar et al. 1995, Eisenkop et al. 1998, Reles et al. 2001, as reviewed by Holschneider and Berek, 2000). The role of the histological type as a prognostic factor remains controversial (Kosary, 1994, Makar et al. 1995). Other factors such as DNA-ploidy, decline of CA125 level after surgery, expression of oncogenes and growth factor receptors (Her-2/neu, EGFR), and alterations of tumor suppressor genes (p53, Rb) are currently under investigation.

Within FIGO-stages, grade and lymph node status have a large impact on survival. Survival declines from 94.6% for well differentiated FIGO stage I tumors to 69.7% for poorly differentiated. For FIGO stage II, this decline is from 78.3% to 48.2%, for FIGO stage III from 76.2% to 23.9%, and for FIGO stage IV from 50.2% to 13.8% (Kosary, 1994). For patients diagnosed with stage I ovarian tumors, 5-year survival is 85.1% without lymph node metastasis, but survival drops to 51.9%, if


6

lymph node metastasis are present. Age also has a strong impact on survival. A women at age 40 with stage IV ovarian cancer has a 50% 5-year survival probability, while this declines to 7.7% in stage IV for a patient >70 years of age (Kosary, 1994).

Multivariate analysis usually reveals stage as well as residual tumor after surgery, and in some studies histological grade of differentiation and age as independent prognostic factors in epithelial ovarian cancer (Eisenkop et al. 1998, Reles et al. 2001).

1.2 Molecular genetic alterations in ovarian tumors

1.2.1 Hereditary ovarian tumors

Approximately 10% of ovarian cancers are thought to be due to autosomal dominant hereditary syndromes (Lynch et al. 1978). The risk of ovarian cancer in first and second degree relatives of women with ovarian cancer is increased 3.6- and 2.9-fold respectively, compared to women who have no family history of ovarian cancer (as reviewed by Berchuck et al. 1999). Three syndromes have been recognized so far (Table 2):

Table 2: Genetic alterations in hereditary ovarian cancer syndromes

Hereditary ovarian cancer syndrome

Gene affected by mutation

 

 

Site specific ovarian cancer

BRCA1, BRCA2

Hereditary breast and ovarian cancer (HBOC)

BRCA1, BRCA2

Hereditary Non-Polyposis Colon Carcinoma
(HNPCC)

MSH2, MLH1, PMS1, PMS2
(DNA mismatch repair genes)

(1) site-specific ovarian cancer syndrome (familial predisposition to ovarian cancer only), (2) hereditary breast and ovarian cancer (HBOC) (predisposition to both breast and ovarian cancers), and (3) hereditary non-polyposis colorectal cancer syndrome (HNPCC), also known as Lynch II syndrome (predisposition in men for colorectal, cancer and in women for breast, colorectal, endometrial and ovarian cancer) (as reviewed by Berchuck et al. 1999).


7

It has been estimated that the breast-ovarian cancer syndrome accounts for the majority of hereditary ovarian cancer cases (65-75%), out of which, approximately 10-15% are site specific. Site-specific and breast-ovarian cancer syndromes have been linked to mutations in the BRCA1 and BRCA2 gene.

The BRCA1 gene is located on chromosome 17q21-12 and it is thought to act as a tumor suppressor gene (Hall et al. 1990, Miki et al. 1994). Initial studies in families selected on the basis of strong family history and early age of onset, suggested that germline mutations in BRCA1 were responsible for about 50% of hereditary breast cancers and 90% of hereditary ovarian cancers (Ford et al. 1995). Constitutional mutations in BRCA1 were found in 34-58% of hereditary ovarian cancer families from North American and European origin (as reviewed by Aunouble et al. 2000). In families with BRCA1 mutations, the lifetime risk of developing ovarian cancer was estimated to be as high as 60% (Easton et al. 1995). More recently it has been suggested that initial estimates of BRCA1 penetrance may have been too high because of selection bias. Population-based case-control studies estimated a cumulative risk of 14.6% respectively, 32.2% for BRCA1 mutation carriers to develop ovarian cancer at the age of 60 years (Laplace-Marieze et al. 1999, Whittemore et al. 1997).

The second major breast/ovarian cancer susceptibility gene, BRCA2, was identified on chromosome 13q12 (Wooster et al. 1995), and may account for 10% to 35% of hereditary ovarian cancer. BRCA2 mutations were found in 7-14.5 % high risk families (as reviewed by Aunoble et al. 2000), and the penetrance of the gene was estimated to be 27% by age 70 (Ford et al. 1998). Approximately 80-90% of BRCA1 and BRCA2 mutations result in truncated protein products, while missense mutations are rare (as reviewed by Berchuck et al. 1999).

The HNPCC syndrome (Hereditary Non-Polyposis Colon Cancer) is suspected to account for 10-15% of the hereditary ovarian cancer cases. Affected families typically present with colon and endometrial carcinomas, but some ovarian cancers occur too. HNPCC is associated with germline mutations in a family of genes involved in DNA repair (MSH2, MLH1, PMS1, PMS2) (as reviewed by Berchuck et al. 1999).

At present, testing for BRCA1 and BRCA2 mutations is recommended, if two individuals in a family either had ovarian cancer at any age, or breast cancer before age 50, and are first or second degree relatives. When these conditions are met, there is a 10% to 20% probability of finding a mutation. The most notable founder mutations described thus far are the BRCA1 185delAG, the BRCA1 5382insC and the BRCA2 6174delT mutations (as reviewed by Berchuck et al. 1999).

Besides genetic testing, ultrasound and Doppler screening as well as serum CA125 marker testing are prevention tools for high risk families. Prophylactic oophorectomy followed by hormonal replacement therapy in women in their late 30s


8

or 40s is believed to be a reasonable intervention to consider in BRCA1 and BRCA2 mutation carriers. However, it has been noted that approximately 2% of these women develop primary peritoneal papillary serous carcinoma after oophorectomy (Piver et al. 1993). Oral contraceptive use (six years or more) by women out of families with hereditary ovarian cancer decreases the risk by as much as 60% and therefore seems an attractive alternative for cancer prevention (Narod et al. 1998).

In general, genetic counseling and testing, as well as the decision about the best strategy for prevention or early detection of ovarian cancer in high risk families, requires a multidisciplinary approach by geneticists, gynecologists, and psychologists. Health care, family planning, psychological aspects, and potential health insurance problems are important considerations for these patients.

1.2.2 Sporadic ovarian tumors

Molecular genetic alterations are frequent in sporadic ovarian cancers and these alterations include chromosomal deletion, oncogene amplification and overexpression, mutation of tumor suppressor genes, and alternative and aberrant RNA splicing. Some alterations have been found to correlate with tumor histopathological criteria and clinical outcome.

Loss of heterozygosity (LOH) is defined as a deletion of a portion of a chromosome that contains a putative tumor suppressor gene. LOH implies loss of the normal polymorphism present at a given locus. Tumor DNA has a propensity to loose one of its two alleles of various genes, compared with DNA from normal cells of the same patient. LOH has been frequently identified in ovarian carcinomas. Several investigators reported that chromosomes 17p, 17q, and 6q show high frequencies of LOH, as high as 31%-83%, 39%-74%, and 39%-74% respectively, in ovarian carcinomas (as reviewed by Chuaqui et al. 1997). Other loci potentially important in ovarian carcinogenesis are chromosomes 4p, 7p, 12p, 12q, 18q, and 19p (as reviewed by Chuaqui et al. 1997).

Several oncogenes have been found to be altered in ovarian cancer, including HER-2/neu, k-ras, fms, c-myc, and c-fos. Most important among these are HER-2/neu, k-ras, and c-myc. The HER-2/neu gene maps to 17q21-22 and encodes a 185-kd transmembrane glycoprotein receptor (p185 HER2), which has partial homology with the epidermal growth factor receptor. It has tyrosine kinase activity and promotes cell proliferation and differentiation. The p185 HER2 protein is a member of the family of growth factor receptors, which include also EGF-R, HER-3, and HER-4.

In an early study by Slamon et al. 1989, HER-2/neu was found to be amplified and overexpressed in 26% of ovarian carcinomas. The number of gene copies


9

significantly correlated with clinical outcome, >5 gene copies giving the poorest outcome. The prognostic value of HER-2/neu overexpression still remains contro-versial though. Only one author has reported an independent correlation with survival in 275 cases (Meden et al. 1994), while others found correlations only in univariate analysis or no correlation at all, including 174 cases analyzed at the Department of Gynecology, Charité-Campus Virchow-Klinikum (as reviewed by Aunoble et al. 2000, Schmider, 1999).

Table 3: Molecular genetic alterations in sporadic ovarian cancer

Structure

Type of Alteration

Affected Chromosomes
or Genes

 

 

 

Chromosomes

Loss of heterozygosity (LOH)

17p, 17q, 6q
4p, 7p, 12p, 12q, 18q, 19p
6q, 13q, 19q (serous
ovarian tumors)

Oncogenes

Amplification/Overexpression
Mutation

HER-2/neu, c-myc, c-fos
k-ras

Tumor suppressor
genes

Mutation
Protein overexpression

p53, BRCA1, BRCA2
Rb

Other growth
regulatory genes

Alternative splicing
Overexpression/Polymorphisms

mdm2
p21 WAF1/CIP1

More important than the prognostic value is the fact that this protein is used as a therapeutic target for a recombinant humanized anti-HER2 monoclonal antibody (rhuMAb HER2 [trastuzumab] ). RhuMAb HER2 has achieved an objective response rate of 15% as a single agent in patients with HER-2/neu overexpressing metastatic breast cancer who had received extensive prior therapy (Cobleigh et al. 1999). The therapy was well tolerated and the most common adverse effects were infusion-associated fever and chills. Several protocols of a combination therapy of trastuzumab with chemotherapeutic agents are currently under investigation. The results of the rhuMAb HER2 clinical trials demonstrate that a better understanding of genetic alterations in cancer can lead to new targeted approaches to cancer treatment.


10

The c-myc oncogene is located on chromosome 8q24 and is involved in cell cycle regulation by activation or inhibition of several genes including cyclins (A,D or E) and p53 (Dang, 1999). Amplification and/or overexpression of the wild-type sequence has been demonstrated in 26-37% of ovarian cancer, but a correlation to survival has only been demonstrated in one study in co-expression with HER-2/neu and p21ras (as reviewed by Aunoble et al. 2000).

K-ras was localized to 12p12.1 and is a member of the family of ras genes which code for serine/threonine protein kinases. Mutated K-ras protein looses the ability to become inactivated and thus stimulates growth and differentiation autonomously (as reviewed by Aunoble et al. 2000). In ovarian cancer, mutations of the K-ras oncogene have been detected in 11-75% in mucinous and 5-36% in nonmucinous tumors (as reviewed by Aunoble 2000 et al.). These results suggest that K-ras mutational activation is an early event of mucinous ovarian tumorigenesis. Little is known about the prognostic value of K-ras. While most studies find no correlation, only one study showed a significant relationship between K-ras alterations and poor clinical outcome (Scambia et al. 1997).

Among the tumor suppressor genes which are altered in sporadic ovarian cancer, p53 plays the most important role, while BRCA1 and BRCA2 are only rarely mutated. p53, located at 17p13.1, codes for a 53 kilodalton transcription factor which regulates cell cycle control and apoptosis. It has been found to be mutated in approximately 40-80% and overexpressed in 32%-84% of epithelial ovarian cancers (as reviewed in chapter 1.3.7). The value of p53 mutations and protein overexpression as prognostic markers are controversial (as reviewed in chapter 1.3.7). Since efficacy of platinum-based chemo-therapy is believed to depend on p53-induced apoptosis (re-viewed in chapters 1.3.8 and 5.1.7), restoration of p53 function has become an attractive target for gene therapy (as reviewed in chapter 1.2.3).

Further genes that are frequently altered in ovarian cancer are the mdm2 gene, which tightly regulates p53 function (as reviewed in chapter 1.4.3), and the CIP1/WAF1 gene, which mediates p53 induced G1 cell cycle arrest. (see chapter 1.3.3). The CIP1/WAF1 gene was shown to have no mutations but an AGCrarrAGA (serinerarrarginine) polymor-phism at codon 31 was detected in 15% of ovarian cancer cases (Lukas et al. 1997). We found overexpression of the p21 protein in 61% of ovarian carcinomas, and high overexpression (>50% of the nuclei) was correlated significantly with early stage and favourable prognosis of ovarian cancer (Schmider et al. 2000).

The Rb/cyclin D1/p16 pathway may also be involved in ovarian carcinogenesis, but the mechanisms remain to be further analyzed. Sequence alterations of the retinoblastoma gene are rare, but pRb expression was found in 71% of ovarian carcino-mas as opposed to 41% of benign tissues (Dong et al. 1997). Mutations of the


11

p16ink4A tumor suppressor gene which encodes a cyclin-dependent kinase inhibitor are infrequent in ovarian cancers, but methylation of the gene has been noted in 4% of ovarian cancers (Wong et al. 1999). Recently aberrant RNA splicing of the p16ink4A gene has been described in ovarian cancer (Suh et al. 2000).

As opposed to hereditary ovarian cancer, alterations of the BRCA1 and BRCA2 tumor suppressor gene are rare in sporadic ovarian cancer. Although LOH at the BRCA1 and BRCA2 locus is seen in 80% respectively more than 50% of sporadic ovarian carcinomas, mutations of BRCA1 have been found in only 5%, and mutations of BRCA2 in only 4.6% respectively 8% of these tumors (as reviewed by Aunoble et al. 2000).

1.2.3 Gene therapy

Because p53 is a central regulator in cell cycle control and mutations of the p53 gene are the most frequent abnormalities identified in human tumors, restoration of p53 function has become a major focus of research for gene replacement therapy. Re-introduction of wild-type p53 gene into cells that have lost normal p53 function can induce either growth arrest or apoptosis (El-Deiry et al. 1993, Yonish-Rouach et al. 1991, Wills et al. 1994, Nielsen and Maneval, 1998).

For the delivery of tumor suppressor genes into cancer cells, adenovirus vectors (Ad) have been shown to be most efficient. Adenoviruses can be produced in high titers and can infect both dividing and non-dividing cells at high efficiencies, thus resulting in high levels of transgene expression. Adenovirus vectors do not normally integrate into the host genome, so transgene expression is transient (as reviewed by Pützer 2000).

In vitro studies of cell cultures derived from various human malignancies have shown that delivery of wild-type p53 in cells previously mutant or null for p53 results in a dose-dependent inhibition of cell proliferation, usually associated with cell death via apoptosis (Wills et al. 1994, Yonish-Rouach et al. 1991, Nielsen and Maneval, 1998, Hirai et al. 1999). p53 tumor suppressor gene therapy, applied to an intraperitoneal model of ovarian cancer in mice, demonstrated an increase in survival in the treated compared to the control group (Mujoo et al. 1996, Kim et al. 1999).

Considering the short half-life of wild-type p53 and the fact that only a small percentage of tumor cells can be transduced, some of the growth suppressing effect of adenovirus mediated p53 gene therapy is thought to be due to the so-called bystander effects. These effects possibly include transfer of metabolic products through gap junctions, phagocytosis of apoptotic vesicles containing dead tumor cells that mediate


12

apoptosis, induction of an immune reponse against the tumor, and inhibition of angiogenesis (as reviewed by Pützer 2000).

Cell culture experiments have shown that the sensitivity of tumor cells to various chemotherapeutic agents such as adriamycin, etoposide, doxorubicin, cisplatin and others depends on the efficient induction of apoptosis mediated by a functional p53 protein. Therefore, loss of p53 can enhance resistance to chemotherapy (Lowe et al. 1993, Vasey et al. 1996, Vikhanskaya et al. 1997). In addition to the anti-tumor effect of p53 tumor suppressor gene transduction, sensitization to chemotherapy by re-introduction of wt p53 may play an important role for gene therapy. Other potential targets for gene therapy include the WAF1/CIP1, a universal inhibitor of the cyclin dependent kinases, the mdm2 gene and other genes which are involved in cell cycle regulation.

A synergistic efficacy of adenovirus-mediated p53 gene therapy and chemotherapy, both for Cisplatin and Paclitaxel, have been demonstrated in xenograft models of colon, ovarian, prostate and breast cancer (Ogawa et al. 1997, Nielsen et al. 1998, Gurnani et al. 1999).

Promising results have been found in the first clinical p53 gene therapy study (Schering Plough Research Institute/ Essex Pharma) for ovarian carcinomas with p53 mutations. In a phase I clinical trial, re-introduction of wild-type p53 via a recombinant adenovirus was investigated in recurrent ovarian cancer. Intraperitoneal administration of Ad p53 resulted in successful p53 gene transfer and expression as well as objective tumor response in part of the cases (I.B. Runnebaum, personal comunication).

Currently, the Department of Gynecology of the Charité-Campus Virchow-Klinikum participates in a multinational, multicenter, randomized phase II/III trial of chemotherapy alone (Taxol/Carboplatin) versus chemotherapy plus p53 gene therapy in newly diagnosed stage III ovarian cancer and primary peritoneal cancer with le 2cm residual disease following surgery, which is conducted by the Schering-Plough Research Institute (Project Director: Jo Ann Horowitz, M.D., USA).

The present data suggests that gene replacement therapy provides a potentially effective anti-tumor strategy. The increasing knowledge about regulation of cell division and apoptosis opens new perspectives for treatment strategies. By means of molecular genetic techniques such as the chip technology for analyzing mutations and gene expression patterns, a precise genetic characterization of tumor cells and an individual choice of the most efficient therapy may be possible in the future.


13

1.3 The p53 tumor suppressor gene

1.3.1 p53 domains: structure and function

The p53 gene is a tumor suppressor gene and has been named "the guardian of the genome„ (Lane, 1992) or the "cellular gatekeeper of growth and division„ (Levine, 1997). The p53 protein is a transcription factor which plays a central role in the regulation of cell cycle arrest and apoptosis following DNA damage.

The p53 gene contains eleven exons which encode for a 2.8 kb mRNA, translated into a 53kD protein (Matlashewski et al. 1984, Harlow et al. 1985). Exon 1 is always non-coding. In human p53, a very large intron of 10 kb with unknown biological function is located between exon 1 and exon 2 (Soussi and May, 1996).

Fig. 1: Structural and functional regions of the p53 protein (redrawn from May and May, 1999) Functional regions and corresponding amino acid residues are shown on top. In the middle, evolutionary highly conserved domains I-V are shown. The bottom represents the tertiary structure of the site-specific DNA-binding. L1, L2, L3 indicate loops, and LSH indicates a loop-sheet-helix structure. The tertiary structure is shown in more detail in Fig. 2.

The human p53 protein contains 393 amino acids and can be divided into three regions: 1) the amino-terminal region, which contains the transcriptional trans-activation domain, 2) the carboxy-terminal region, which contains the functional regions for nuclear localization, tetramerization, and both non-specific DNA-binding


14

and recognition of primary DNA damage, and 3) the central region of the protein (amino acid residues 102 to 292), which contains most of the sequence specific DNA binding domains (Soussi and May, 1996) (Fig. 1).

Fig. 2: Topological diagram of the secondary structure elements of the core domain of the p53 protein (redrawn from Cho et al. 1994). The residues at the start and the end of each secondary structure element are indicated. The DNA-binding regions of the protein (L1-S2-S‘, L2, L3, S10-H2) correspond to the conserved regions of the p53 gene and are colored yellow for region II (residue 117-142), red for region III (residue 171-181), blue for region IV (residue 234-258) and green for region V (residue 270-286). The boundaries of the two ß-sheets that make up the ß-sandwich are shaded. The scaffolding regions are white.

The functional form of the p53 protein is a tetramer, which is assembled by the amino acid residues 323-356 of the carboxy-terminal region of the protein (Wang et al. 1994). It has been shown that some p53 mutants exhibit a dominant negative phenotype and are able to associate with the wild-type p53 protein, which is expressed by the remaining wild-type allele, to induce the formation of an inactive heterooligomer (Soussi and May, 1996).


15

Within the p53 sequence five regions have been identified, which have been highly conserved throughout evolution (Soussi et al. 1990). The conserved domain I (amino acid residues 12-23) is located within the aminoterminal region and corresponds to the binding site of the MDM2 protein. The conserved domains II (residues 117-142), III (residues 171-181), IV (residues 234- 258) and V (residues 270-286) are located in the central region of the protein and are almost identical to the DNA interactive regions of the protein as described in the crystallographic structure of the p53 protein by Cho et al. (1994) (Fig 2).

The crystallographic analysis of the p53 protein has revealed the following motifs in the central region of the gene: 1) two antiparallel beta-sheets composed of four and five beta-strands that hold the other elements, 2) a loop-sheet-helix-motif (LSH) containing three beta-strands, an alpha-helix, and the L1 loop (residues 112-141), 3) an L2 loop containing a small helix (residues 163-195), and 4) an L3 loop composed mainly of turns (residues 236-251) (Cho et al. 1994) (Fig 2). The LSH motif and the L3 helix are involved in direct DNA interaction (Cho et al. 1994).

There is remarkable correspondence between these structural elements and the conserved domains II-V of the gene. More than 90% of the missense mutations in the p53 gene have been found clustered in so-called "hotspots„ in the highly conserved regions of the II-IV of the gene, which have also been identified as the DNA binding regions (Greenblatt and Harris, 1994). It has been suggested to distinguish between class I mutations, which affect amino acid residues directly involved in the protein-DNA interaction (residues in LSH and L3), and class II mutations, which affect amino acids involved in the stabilization of the tertiary structure of the protein (residues in L2) (Soussi and May, 1996). While class I mutations result in defective contacts with the DNA and loss of the ability of p53 to act as a transcription factor, class II mutations disrupt the structural basis of the p53 core domain and may, therefore, indirectly affect its function.

1.3.2 p53 response to cellular stress: upstream events

Under normal conditions, the p53 protein is present in extremely small quantities in most cells and displays a rapid turnover rate which is on the order of minutes. Rapid degradation of p53 in normal cells is critical to efficiently dampen p53 activity and is regulated in large part by MDM2 via ubiquitin-mediated proteolysis (Maki and Howley, 1997, Haupt et al. 1997, Kubbutat and Vousden, 1998).

In its role as a tumor suppressor gene, p53 serves as a "guardian of the genome„ (Lane, 1992). The amount of p53 protein increases in response to a variety of signals mediated by cellular stresses, such as DNA damage produced by gamma-irradiation


16

or ultraviolet radiation (Guidos et al. 1996), oncogene activation, hypoxia (Graeber et al. 1996), nucleotide depletion (Linke et al. 1996), and metabolic and pH changes (Fig. 3). The p53 protein level increases proportionally to the extent of DNA damage through a prolonged half-life of the protein (Levine, 1997). Furthermore, the DNA-binding activity of p53 is increased. p53 stabilization and enhanced activity are regulated by several mechanisms.

1) p53 protein stabilization can occur through N-terminal phosphorylation and decreased MDM2-binding, for example after exposure to DNA-damaging ionizing radiation or the topoisomerase inhibitor etoposide (Siliciano et al. 1997). Two candidate kinases for p53 phosphorylation have been identified. The ATM gene which recognizes DNA damage phosphorylates p53 on serine 15, and DNA-PKs (DNA-activated protein kinase) which phosphorylates p53 at both serine 15 and serine 37 (Woo et al. 1998, Siliciano et al. 1997, Canman et al. 1998, Shieh et al. 1997). Phosphorylation of p53 at serines 15 and 37 impairs the ability of MDM2 to inhibit p53-dependent transactivation, most likely due to a conformational change of p53 (Shieh et al. 1997).

Shieh et al. (1999) have also shown that in addition to serine 15, serine 20 is phosphorylated in response to DNA damage. Recently it has been shown that mutation of serine-20 renders p53 less stable and more prone to MDM2 mediated degradation. p53 peptides phosphorylated at serine-20 are less efficient competitors of the p53/MDM2 interaction than non-phosphorylated peptides (Unger et al. 1999). These observations are particularly interesting, given that serine-20 resides in the region of p53 which binds to MDM2 (Kussie et al. 1996, Picksley et al. 1994). p53 protein that has been phosphorylated at the N-terminus binds poorly to MDM2 and is, therefore, stabilized through a decrease in its degradation rate.

2) Acetylation of the C-terminal domain of p53 which occurs on lysine 382 by p300 or lysine 320 by PCAF causes a conformational change at the C-terminus and activates p53-dependent DNA-binding (Sakaguchi et al. 1998).

3) C-terminal dephosphorylation of p53 at serine 376 creates a consensus binding site for interaction with 14-3-3sigma, which enhances the ability of p53 to bind to DNA (Waterman et al. 1998). A phosphorylation-acetylation cascade model has been discussed. It suggests that following DNA damage, N-terminal phosphorylation directs C-terminal acetylation to activate p53 (Sakaguchi et al. 1998). This would result in both higher stability of the p53 protein as well as increased DNA binding ability.


17

Fig. 3: p53 pathways. Upstream activators are shown above in red. Multiple downstream effectors of p53 which play a role for cell cycle arrest and apoptosis are shown. Arrows indicate a positive effect, while a flat line indicates inhibition. Quadrangles indicate specific functions in p53/MDM2 interaction.


18

1.3.3 Downstream mediators of p53-dependent cell cycle arrest

The downstream events mediated by p53 play a major role for cell cycle arrest, apoptosis, and genomic stability. The growth controlling functions of p53 include cell cycle arrest, apoptosis, senescence and antiangiogenesis (Fig. 4). The transcriptional activating function of p53 is a major component of its biological effects and a substantial number of target genes have been identified. These include WAF1/CIP1, GADD45, 14-3-3sigma, bax, FAS/APO1, KILLER/DR5, PIG3, Tsp1, IGF-BP3 and others (as reviewed by El-Deiry 1998). It has been suggested, that there may be as many as 200-400 p53 target genes or perhaps even more (El-Deiry et al. 1992).

Fig. 4: Response to cellular stresses in cells with wildtype p53 versus cells with mutated p53

The G1 checkpoint

p53 directly regulates the expression of p21WAF1/Cip1, a universal inhibitor of the cyclin dependent kinases (Cdk) (Fig.3) (El-Deiry et al. 1993). This 21 kilodalton protein forms a quaternary complex found in normal cells with cyclin/Cdks and the DNA polymerase processivity factor PCNA (Xiong et al. 1993). It binds to the cyclin-Cdk complexes cyclin D1-Cdk4, cyclin E-Cdk2, cyclin A-Cdk2 and cyclin A-Cdc2 and inhibits their kinase activity (Fig. 3). One molecule of p21 per complex appears to permit Cdk activity, while two molecules inhibit its activity and block cell cycle progression (Zhang et al. 1994). The available evidence suggests that p21 acts on cyclin-Cdk complexes and PCNA such that it blocks DNA replication. p21 binds with its greatest affinity to G1 cyclin-CDK complexes, and binds poorly to cyclin B/cdc2. p21 appears to be required for G1 arrest following DNA damage (Waldmann et al. 1995).


19

The G2 checkpoint

While it is clear that p53-deficiency leads to defective cell cycle arrest in G1, the role of p53 in G2 arrest is less clear. Deregulation of the G2 checkpoint is one of the hallmarks of malignant transformation. The G2 checkpoint is the last barrier before the perpetuation of mutations and therefore, very important for the maintanance of genomic stability. Overexpression of wildtype p53 protein has been found to arrest cells in G2 and inhibit entry into mitosis (Stewart et al. 1994). This property of p53 is important in a novel cell cycle checkpoint that controls entry into mitosis when DNA synthesis is blocked. p53-/- cells do not arrest in response to spindle inhibitors, but undergo multiple rounds of DNA synthesis without DNA segregation. This results in the formation of tetraploid and octaploid cells (Cross et al. 1995).

The GADD45 (growth arrest and DNA-damage inducible gene #45) was initially discussed to play a role in G1 arrest. The gene is induced when cells are subjected to DNA damage, leading to arrest in the G1 phase of the cell cycle (Kastan et al. 1992). The GADD45 protein was found to interact with the replication and repair factor PCNA and to inhibit the entry of cells into the S-phase (Smith et al. 1994). More recent evidence has suggested that cyclin B/cdc2 may be bound and inhibited by GADD45, thereby leading to G2 arrest (Zhan et al. 1999). The cdc2 kinase-cyclin B1 complex, termed Mitosis Promoting Factor (MPF), serves as the primary mediator of the signal for G2-mitosis transition (Zhan et al. 1999).

The 14-3-3sigma proteins have also been suggested to be involved as mediators of p53-dependent G2-arrest. These 14-3-3sigma proteins have been shown to bind to the non-functional phosphorylated form of the cdc25c phosphatase, which in turn, prevents the dephosphorylation of the cdc2 kinase, thus resulting in non-functional MPF and inducing G2 arrest (Conklin et al. 1995). The 14-3-3sigma gene has been shown to be upregulated by p53 (Hermeking et al. 1997).

Another gene which may play a role as a mediator of cell cycle arrest is B99. It is a p53 target gene, which is induced by DNA damage and is specifically upregulated in the G2-phase, during which its protein is localized in the microtubule network. Ectopic overexpression of B99 in p53-null fibroblasts leads to G2 arrest (Utrera et al. 1998).

The present knowledge about p53 suggests, that even though p53 can cause an arrest or prolongation of G2 to enable sufficient repair of DNA defects before the onset of mitosis, p53 also contributes to the efficiency of DNA repair and therefore reduces the G2 delay (as reviewed by Schwartz and Rotter 1998).


20

The spindle checkpoint

p53 is associated with centrosomes and thus may affect centrosome duplication directly (Brown et al. 1994). Embryo fibroblasts from p53-null mice acquire more than two centrosomes, leading to mitosis with more than two spindle poles and frequent mitotic failure (Fukasawa et al. 1996). p53(-/-) cells have also been found to show aberrant DNA replication, termed endoreduplication. Polyploid giant cells were the result. Therefore, this checkpoint prevents the formation of aneuploid cells that are destined to harbor an unstable genome (Cross et al. 1995).

1.3.4 Mechanisms of p53-dependent apoptosis

Overexpression of wild-type p53 can result in a rapid loss of cell viability by a manner characteristic of apoptosis (Yonish-Rouach et al. 1991). Apoptosis is defined as a programmed form of cell death in which the cell "commits suicide„ by disintegrating into membrane vesicles. The determinants of whether a cell undergoes a viable cell cycle arrest or apoptosis in response to p53 activation are not fully understood. The extent of the DNA damage and the levels of p53 have been shown to affect the choice between cell cycle arrest and apoptosis (Chen X. et al. 1996).

Other factors of influence include expression of the retinoblastoma gene (Rb), the transcription factor E2F1, viral protein expression and growth factor availability. The Rb/E2F1 pathway may play a central role in determining the balance between cell cycle arrest and apoptosis. Loss of Rb function by expression of viral proteins or by homozygous gene disruption has been correlated with loss of G1 arrest after DNA damage and apoptosis (Slebos et al. 1994, Howes et al. 1994, Morgenbesser et al. 1994). Inactivation of pRb by either E2F-1 overexpression (Qin et al. 1994) or by caspase-mediated cleavage of p21WAF1 has also been shown to induce apoptosis (Levkau et al. 1998). Furthermore, overexpression of Rb blocks p53-mediated apoptosis (Haupt et al. 1995).

p53-mediated apoptosis is thought to occur through a combination of sequence-specific transactivation-dependent and independent mechanisms acting in concert. Several target genes, such as bax, Fas/APO1, KILLER/DR5, PIG genes, PAG608, and Pidd are induced by p53 through sequence specific transactivation (SST) and have been identified to participate in the induction of apoptosis.

The bax gene encodes a 21-kD protein of the Bcl-2 family and has been found to possess potent pro-apoptotic properties (Oltvai et al. 1993, Miyashita and Reed 1995). The ratio of p21 Bax: Bcl-2 has been suggested as a criteria to determine whether cells live or die (Oltvai et al. 1993). Furthermore, the downregulation of Bcl-


21

2, which normally enhances cell survival, further promotes apoptosis (Miyashita et al. 1994). Thus p21 Bax may be an important target mediator of p53-dependent apoptosis.

The cell surface death receptor gene CD95/Fas/Apo-1 is a potent inducer of apoptosis (as reviewed by Nagata 1997). Once cell surface death receptors bind death ligands, they transmit rapid apoptotic signals. However, the role of CD95/FAS/APO1 seems to be cell-type and signal-dependent (as reviewed by Gottlieb and Oren 1998). p53 stimulates the expression of another death-receptor, KILLER/DR5, which is activated by the death ligand TRAIL (as reviewed by Gottlieb and Oren 1998).

PIGs (p53-induced genes) cause disruption of mitochondrial integrity and subsequent apoptosis through stimulating reactive oxygen species formation (Polyak et al. 1997). Other putative pro-apoptotic target genes of p53 are PAG608 (Israeli et al. 1997), the genes which encode IGF-BP3 (Buckbinder et al. 1995) and p85 (Yin et al. 1998). Recently, Pidd, a new gene regulated by p53, was identified and found to induce apoptosis (Lin et al. 2000).

Inactivation of p53 function may provide a selective advantage for clonal expansion of preneoplastic and neoplastic cells. Of particular interest is the fact that hypoxia can activate p53 and promote p53-dependent apoptosis. A hypoxic environ-ment occurs frequently within the central portion of tumors, particularly prior to sufficient neoangiogenesis. This would explain the hypoxia-driven selection for cells with diminished apoptotic potential within solid tumors and would provide an attractive explanation for the loss of p53 function in tumors (Graeber et al. 1996).

The p53-dependent apoptotic pathway has not only been demonstrated to be critical to the development of tumors but also to their treatment, since the effectiveness of various chemotherapeutic agents depends on the ability of p53 to induce apoptosis (as reviewed in chapter 1.3.8).

1.3.5 p53 and genomic stability

Inactivation of p53 can lead to an increase of mutation frequency resulting from inefficient nucleotide excision repair (NER). Poor NER causes genomic instability. The instability is manifested as gene amplification, aneuploidy, and chromosomal aberrations, associated with malignant progression. p53 can bind to several DNA helicases, which are part of the basal transcription factor TFIIH (Wang et al. 1995). One functional outcome of such interaction may be modulation of nucleotide excision repair (NER) by p53 (Wang et al. 1995).

A model has been proposed in which p53 binds the damaged DNA and stimulates DNA repair (Wang et al. 1996). Upon successful completion of DNA repair, it has been proposed that p53 would be phosphorylated and then released from


22

the DNA (Lu et al. 1997). If repair is not carried out successfully, it is assumed that p53 fails to be released and its extended interaction with DNA helicases triggers apoptosis (Wang et al. 1996).

Furthermore, p53 can recognize several forms of damaged DNA including the mismatched DNA (Lee et al. 1995) and the single-stranded DNA ends (Oberosler et al. 1993, Bakalkin et al. 1994). p53 may act as a sensor that binds to damaged DNA regions and recruits the NER machinery by trapping TFIIH, a major component of the repair complex, at sites where it is needed. This p53-TFIIH complex in turn may facilitate the formation of a functional „repairosome„ (Wang and Harris, 1997).

p53‘s function is frequently lost during the process of tumorigenesis and in the spontaneous immortalization of primary cells. This indicates that loss of p53‘s function promotes genomic instability.

1.3.6 The role of p53 in tumorigenesis

The p53 gene plays a crucial role for cell cycle control, induction of apoptosis, DNA repair, and genomic stability, all of which are central to the prevention of human carcinogenesis. Inactivation of the wild-type p53 function is thought to be a common event during the development of cancer. Several mechanisms of functional inactivation of p53 such as mutation, inhibition, nuclear export, degradation through the MDM2 protein, or aberrant transcription are known.

Mutations which inactivate some or all of p53‘s functions as a tumor suppressor gene have been found in over 50% of all human tumors (Hollstein et al. 1994). Over 85% of these mutations are missense mutations that encode altered forms of the protein with a prolonged half-life. Missense mutations are predominantly located in the DNA-binding regions of the protein and therefore affect the DNA-binding ability of the p53 protein (Cho et al. 1994). As a consequence, p53 can not activate the target genes that regulate cell cycle arrest. p53 protein alterations due to mutations therefore provide a selective advantage for clonal expansion of neoplastic cells (Vogelstein 1994).

In addition to failure to elicit a tumor suppressor response, these mutant p53s are also unable to activate expression of MDM2. As a result, these mutant p53 proteins are unusually stable and accumulate to high levels in the tumor cells (Kubbutat and Vousden, 1998). However, wildtype p53 may also be inactivated through amplification of the mdm2 gene in some tumors, especially sarcomas, since the MDM2 protein rapidly degrades and inhibits p53 by binding to the protein (Oliner et al. 1992, Kubbutat et al. 1997, Haupt et al. 1997).


23

The function of p53 may also be abolished through mechanisms that prevent its entry into the nucleus, or promote nuclear exclusion, and keep it localized in the cytoplasm, as has been described for some breast tumors and neuroblastomas (Moll et al. 1995, Roth et al. 1998, Stommel et al. 1999).

Furthermore in a number of human tumors transcription of p53 has been shown to be reduced or absent. This could result from cis-acting mutations in the regulatory regions of the gene or hypermethylation of the promoter as has been shown for other tumor suppressor genes (Counts and Goodman, 1995, Baylin, 1997).

More common than loss of expression, is elevated expression of p53 due to mutation. The myc-oncogene may play an important role for high p53 expression. C-myc has been found to be elevated as an early event in oncogenesis and is thought to subsequently induce p53 expression, which then leads to cell cycle arrest. A cell population with mutant p53 would not undergo growth suppression as expected but would have a growth advantage under these conditions. Clonal selection of these cells would therefore be favored (as reviewed by Reisman and Loging 1998).

Humans with germline mutations, who are heterozygotes for the wild-type allele of p53, have a very high frequency of developing cancer (90-95%) at an early age (Li-Fraumeni-Syndrom). The tissue distribution of these cancers is, however, not random (sarcomas, breast, adrenal carinomas) and it is not clear what this means for the p53 function (Li and Fraumeni, 1969).

1.3.7 Clinical implications

Ovarian cancer, like most other adult malignancies, is thought to result from accumulation of mutations in multiple genes, which are important for normal function. The p53 tumor suppressor gene is the most frequently mutated gene in human cancers (Greenblatt and Harris, 1994) and plays a critical role in the regulation of cell cycle and apoptosis. It has been found to be mutated in approximately 40-80% of epithelial ovarian cancers (Mazars et al. 1991, Kihana et al. 1992, Milner et al. 1993, Kohler et al. 1993b, Kupryjanczyk et al. 1993, Wertheim et al. 1994, Zheng et al. 1995, Kappes et al. 1995, Casey et al. 1996, Righetti et al. 1996, Skilling et al. 1996, Schuyer et al. 1998). In a previous study of 105 ovarian cancer patients we found mutations in 57% of the cases (Wen et al. 1999). It is thought that p53 protein alterations due to missense mutations, nonsense or frameshift mutations, provide a selective advantage for clonal expansion of neoplastic cells (Vogelstein, 1994).

Only few studies have analyzed the entire open reading frame of the gene (Kihana et al. 1992, Kupryjanzcik et al. 1993, 1995, Casey et al. 1996, Skilling et al. 1996, Sood et al. 1997, Angelopoulou et al. 1998, Wen et al. 1999) and have found


24

mutations in 50% - 79% of the cases. 5%-20% of these mutations were outside exon 5-8 (Kupryjanzcik et al. 1993, Skilling et al. 1996, Casey et al. 1996, Wen et al. 1999). Therefore a study which is limited to exons 5-8 will miss a substantial number of mutations.

Though p53 protein expression has been studied extensively by immuno-histochemistry in ovarian cancer. In those studies which used frozen ovarian carcinoma specimens for immunohistochemical analysis of p53, the percentage of cases with p53 overexpression was 32%-84% with an overall average of 51% (Marks et al. 1991, Kihana et al. 1992, Eccles et al. 1992, Kohler et al. 1993a, Kiyokawa et al. 1994, Henriksen and Oberg 1994, Sheridan et al. 1994, Lee et al. 1995, Casey et al. 1996, Skilling et al. 1996, Buttitta et al. 1997, Geisler et al. 1997, Schuyer et al. 1998, Wen et al. 1999).

The role of p53 alterations as a prognostic factor remains controversial. Several studies have identified p53 overexpression as a prognostic factor (Bosari et al. 1993, Henriksen et al. 1994, Hartmann et al. 1994, van der Zee et al. 1995, Klemi et al. 1995, Levesque et al. 1995, Herod et al. 1996, Viale et al. 1997, Eltabakkah et al. 1997, Geisler et al. 1997, Röhlke et al. 1997, Werness et al. 1999, Anttila et al. 1999), and few studies as an independent prognostic factor of overall survival in multivariate analysis (Klemi et al. 1995, Herod et al. 1996, Geisler et al. 1997, Röhlke et al. 1997, Anttila et al. 1999). p53 mutations have been found to be associated with a significantly shorter overall survival compared to the wildtype p53 sequence in only one study (Wen et al. 1999).

1.3.8 p53 and efficacy of chemotherapeutic agents

Cell culture experiments have shown that the sensitivity of tumor cells to various chemotherapeutic agents depends on the efficient induction of apoptosis mediated by a functional p53 protein. Therefore loss of p53 can enhance resistance to chemotherapy (Lowe et al. 1993, Vasey et al. 1996, Vikhanskaya et al. 1997). It is well documented that drugs such as Adriamycin, Etoposide, Doxorubicin, Cisplatin, and others induce DNA damage and p53-dependent apoptosis.

The wildtype p53-expressing A2780 human ovarian cancer cell line acquired cross resistance to Cisplatin and Doxorubicin by transfection with a dominant negative mutant p53 gene, while it retained sensitivity to Taxol (Vasey et al. 1996). In another study Cisplatin caused strong induction of p53, WAF1, and Bax in the Cisplatin sensitive A2780 cell line, while there was no such effect in a Cisplatin resistant cell line A2780-DX3, which furthermore showed a significant proportion of potentially inactive p53 protein located in the cytoplasm instead of the nucleus (Vikhanskaya et


25

al. 1997). Other studies suggest that Cisplatin-resistance in the 2780CP ovarian cancer cell line is caused by a defect in the signal transduction pathway for p53 induction following cisplatin-induced DNA damage (Siddik et al. 1998).

Further strong evidence for the importance of a functional p53 protein for the efficacy of Cisplatin and Carboplatin is given by a database of the National Cancer Institute on drug activity in cell lines which shows a strong correlation between p53 wildtype sequence respectively p53 function and efficacy of Cisplatin and Carboplatin (Weinstein et al. 1997). In contrast to this, antimitotic tubulin-active agents such as Taxol show a strong negative correlation between p53 wildtype as well as p53 function and activity of the drug (Weinstein et al. 1997). Sensitivity to Taxol has been found enhanced through the absence of functional p53 protein because of increased G2M arrest and p53 independent apoptosis (Wahl et al. 1996, Vikhanskaya et al. 1998).

Since dysfunctional p53 can not mediate the apoptotic process, tumors with p53 mutations or altered p53 protein may become resistant to platinum-based chemotherapy (Lowe et al. 1994). This has been initially demonstrated in two clinical studies which showed that patients with FIGO III/IV ovarian cancer had a poor response to platinum-based chemotherapy if their tumors had p53 overexpression (Buttitta et al. 1997) or p53 missense mutations (Righetti et al. 1996). In a further study which analyzed 168 primary stage III-IV ovarian carcinomas, p53 overexpression was significantly correlated with resistance to a platinum based chemotherapy (Ferrandina et al. 1999).

The hypothesis that ovarian cancer cells with functional p53 are more sensitive to Cisplatin is further supported by the findings of gene therapy studies. Introduction of wildtype p53 protein via adenovirus gene transfer into A2780/CP Cisplatinum resistent cells significantly sensitized these cells to platinum cytotoxicity, indicating that p53 was involved in resistance to cisplatin (Song et al. 1997). Similar results were reported for p53-deleted SK-OV-3 ovarian cancer cells (Kanamori et al. 1998).

Since the efficacy of a platinum-based chemotherapy may depend in part on a functional p53 gene, the introduction of a p53 wildtype sequence into ovarian cancer cells via an adenovirus has become a major focus of research (see chapter 1.2.3). Furthermore, molecular genetic analysis of ovarian tumors might in the future identify those patients who are most likely to respond to a chemotherapy.


26

1.4 The mdm2 (murine double minute 2) gene

1.4.1 Molecular structure of the mdm2 gene

The mdm2 (murine double minute) gene was initially described as a proto-oncogene and is amplified and overexpressed in human soft tissue sarcomas and glio-mas. It was originally identified in a spontaneously transformed Balb/cT3 fibroblast murine cell line (Cahilly-Snyder et al. 1987). In these cells mdm2 is localized on double minute chromosomes and is amplified approximately 50-fold (Fakharzadeh et al. 1991).

In humans, the gene is located on chromosome 12q13-14 (Oliner et al. 1992). The full length mRNA encodes for a 90 kD protein of 491 amino acids. The mouse mdm2 gene contains 12 exons, of which exon 3-12 comprise the coding region (Montes de Oca Luna et al. 1996). Comparison of the gene in different species reveals four major conserved regions (as reviewed by Piette et al. 1997 and Freedman et al. 1999) (Fig. 5).

Fig. 5: Structure and functional regions of the MDM2 protein. Above functional domains with corresponding amino acid residues are shown (see Freedman et al. 1999). DNA-PK sites comprise aa 388/389, 395/396 and 407/408. The p53 binding domain and conserved region I are almost identical. Since the human mdm2 exon boundaries are not fully known yet, mouse exons 1-12 are shown with the first codon of each exon indicated. Exon 1 and 2 are noncoding (Montes de Oca Luna et al. 1996).


27

Region I contains 90 aa of the N-terminus and corresponds closely to the p53 binding site (aa 19-102) (Chen et al. 1993). In a crystal structure analysis of the 109-residue amino-terminal domain of MDM2 bound to a 15-residue transactivation domain peptide of p53, it was shown that MDM2 has a deep hydrophobic cleft, on which the p53 peptide binds as an amphiphatic alpha helix (Kussie et al. 1996).

The region II of MDM2 protein contains a central acidic domain, which was shown to interact with the ribosomal L5 protein and its associated 5S rRNA (Marechal et al. 1994, Elenbaas et al. 1996). This suggests a possible function in ribosome biosynthesis or in translational regulation.

Between region I and II, a putative nuclear localization signal has been suggested to be located at amino acid 181 (Fakharzadeh et al. 1991). Recently, a conserved nuclear export signal (NES) sequence has been localized to amino-acid residues 191-199 and mediates the ability of MDM2 to shuttle between the nucleus and the cytoplasm and back (Roth et al. 1998). Region III contains a central zinc-fingerlike sequence of unknown function.

Region IV, which is highly conserved between species, contains a c-terminal RING finger domain, which has been shown to specifically bind to RNA. This suggests a role for MDM2 in translational regulation in a cell (Elenbaas et al. 1996). Sequence analysis defined the RING finger domain as the most conserved region of MDM2. Recently it was shown that the RING finger domain is essential for MDM2 targeted degradation of p53 (Kubbutat et al. 1999). The C-terminal region may also play an important role for the suppression of MDM2 function since it has recently been shown that ARF binds to MDM2 at the C-terminus and degrades the MDM2 protein. Thereby, MDM2‘s inhibition of p53 is indirectly neutralized (Zhang et al. 1998, Pomerantz et al. 1998).

RING finger domains have been hypothesized to participate in protein-protein interactions or DNA-binding, but for MDM2 it has been shown that the RING finger interacts specifically with RNA in vitro (Elenbaas et al. 1996). This means that the RNA-binding activity of MDM2 is likely to be an important function of MDM2. Recently the RING finger of MDM2 was shown to play a role in cell cycle regulation which was independent of p53 degradation (Argentini et al. 2000).

1.4.2 mdm2 expression in tumors

The mdm2 gene has been found amplified and overexpressed mainly in tumor cells of non-epithelial origin, especially in those derived from the mesenchyme. mdm2 amplification has been found in approximately 20% of soft tissue tumors including osteosarcomas, liposarcomas, and malignant fibrous histiocytomas, and furthermore, in


28

about 9% of gliomas, 12% of testicular germ cell tumors (as reviewed by Momand and Zambetti 1997). In carcinomas, except for esophageal tumors, amplification is rare (as reviewed by Momand and Zambetti 1997) and has not been found in any case of ovarian cancer (Foulkes et al. 1995). In primary sarcomas and cell lines that were characterized in more detail, amplification of the mdm2 gene correlated with overexpression of the MDM2 protein (Oliner et al. 1992).

Overexpression has been found in some tumors independently of amplification. In human leukemia samples, mdm2 mRNA was dramatically upregulated while a normal DNA copy number was maintained (Momand and Zambetti, 1997). MDM2 protein overexpression has been found despite normal levels of mdm2 mRNA, most likely due to enhanced translation efficiency (Landers et al. 1994).

Evidence of different sizes of mRNA transcripts in tumors was noted when mdm2 was initially cloned (Oliner et al. 1992) and, later, altered mRNAs resulting from alternative splicing were described in bladder and ovarian carcinomas (Sigalas et al. 1996). These alternatively spliced mdm2 mRNAs were correlated with advanced stage and poor differentiation in ovarian carcinomas and transfection of NIH 3T3 cells with the splice variants showed transforming ability of the RNA transcripts (Sigalas et al. 1996). Smaller protein variants of sizes between 78 kD and 12kD have been noted in non-small cell lung cancer, and are also thought to be derived by differential splicing (Maxwell, 1994).

The finding of these alterations in human tumors indicates that mdm2 acts as a proto-oncogene that promotes the tumorigenicity of a cell through inactivation of the p53 tumor suppression function and possibly by other, as yet unidentified, mechanisms.

1.4.3 The MDM2/p53 autoregulatory feedback-loop

mdm2 has several characteristics of a cellular proto-oncogene and the mechanism by which it promotes the tumorigenicity of a cell became clearer when MDM2 was discovered to bind to the p53 protein and inhibit p53 mediated transactivation (Momand et al. 1992). MDM2 is a potent inhibitor of p53. It binds to the transcriptional activation domain of p53 and blocks its ability to regulate target genes (Picksley et al. 1994, Chen J. et al. 1996) and to exert antiproliferative effects (Chen, C.-Y. et al. 1994, Haupt and Oren, 1996, Chen, J. et al. 1996). Mdm2 expression is upregulated by p53 in an autoregulatory feedback loop (Gottlieb and Oren, 1998).

Several mechanisms of MDM2/p53 interaction are presently known: 1) counteracting the induction of cell cycle inhibitory genes 2) a protective role of


29

MDM2 against apoptosis 3) nuclear export and 4) rapid degradation of the p53 protein by MDM2 (Fig. 6).

1) MDM2 was shown to inhibit p53-transcriptional activation by „concealing„ the acidic activation domain of p53 from the transcriptional machinery (Oliner et al. 1993). Other experiments suggested that p53 protein, if it is bound in a complex with MDM2, no longer binds to p53 target DNA (Zauberman et al. 1993). The binding site of MDM2 to p53 has been located to amino-acid 18-23 at the N-terminus of the p53 protein by mapping studies (Chen et al. 1993). Only the amino-terminal end of MDM2 is necessary to bind and inhibit p53, suggesting that the full-length protein may carry out additional functions. The binding site for p53 has been localized to aa 19-102 of

Fig. 6: Autoregulatory feedback loop of the p53 tumor suppressor gene and the mdm2 gene. p53 at basal and at increased levels activates the mdm2 gene. The MDM2 protein binds to p53, inhibits its DNA binding ability, and promotes nuclear export and protein degradation.

the mdm2 gene (Chen et al. 1993). MDM2 residues G58, D68, V75 and C77 have been shown to be critical for MDM2 interaction with p53. Mutation of these residues prevents MDM2 interaction with p53 in vitro and MDM2‘s regulation of p53‘s transcriptional activity in vivo (Freedman et al. 1997).

Recent reports also indicate that MDM2 may have a second mechanism to inhibit p53-dependent transcriptional activation at a promoter. In these experiments, deletion mutants of MDM2 defective in p53 binding are able to repress both p53-


30

activated and basal transcription when brought to a promoter by fusion to a DNA-binding domain (Thut et al. 1997).

MDM2 is upregulated in response to p53 activation and has been shown to inhibit p53-dependent G1 arrest in response to irradiation (Chen et al. 1994, Chen J. et al. 1996). Control of p53 activity by MDM2 also plays an important role during embryonic development. The expression of the xenopus laevis homolog xdm2 of mdm2 increases during early embryogenesis from oocyte stage I/II to reach its maximum in oocyte stage V/VI in unfertilized eggs and then becomes undetectable (Marechal et al. 1997). Earlier studies had already demonstrated that the mdm2-null genotype leads to embryonic lethality while mice deficient for both mdm2 and p53 develop normally and are viable (Montes de Oca Luna et al. 1996, Jones et al. 1995). These results strongly suggest a primary developmental role for MDM2 in negative regulation of p53 function.

2) The role of MDM2 in protecting against apoptosis is less clear, but several studies have shown evidence for this function of MDM2. p53 dependent apoptosis induced by c-myc overexpression was inhibited by MDM2 (Chen et al. 1996) and protection from apoptosis required formation of a p53- MDM2 complex in human H1299 cells (Haupt et al. 1997). Rb can overcome the antiapoptotic effect of MDM2 on p53-induced apoptosis by preventing MDM2 from targeting p53 for degradation. Rb-MDM2 interaction, though, does not prevent inhibition of p53-mediated transcription (Hsieh et al. 1999). Recently, experiments with fibroblasts from p53/mdm2 null mice, which were transfected with a retroviral vector carrying temperature-sensitive p53, have shown that loss of mdm2 can induce the p53-dependent aopoptotic pathway in vivo (de Rozieres et al. 2000).

3) MDM2 has been shown to contain a nuclear export signal (NES) and to be able to shuttle across the nuclear membrane in both directions (Fig. 6) (Roth et al. 1998). The findings suggest that MDM2 binds to p53 in the nucleus and transports it to the cytoplasm for degradation (Roth et al. 1998, Tao and Levine, 1999). Cytoplasmic distribution of p53 is thought to result from efficient export of nuclear p53 in combination with MDM2-mediated degradation (Lu et al. 2000).

4) MDM2 was shown to promote rapid degradation of the p53 protein (Haupt et al. 1997, Kubbutat et al. 1997, Kubbutat and Vousden, 1998). MDM2 downregulates the amount of p53 protein but does not reduce p53 mRNA levels (Haupt et al. 1997, Kubbutat et al. 1997). These results indicate that p53 levels are regulated by MDM2 through a post-transcriptional mechanism. Once shuttled out of the nucleus, p53 is thought to be transported to a cytoplasmic proteasome for ubiquitin mediated degradation (Roth et al. 1998, Tao and Levine, 1999). Amino-acids 92-112 of p53 have been identified to function as a degradation signal (Gu et al. 2000).


31

Degradation of p53 by MDM2 depends directly on the ability of MDM2 to shuttle from the nucleus to the cytoplasm (Tao and Levine, 1999). Expression of p19ARF was shown to block the nucleo-cytoplasmic shuttling of MDM2 and could therefore stabilize p53 by protecting it from MDM2 mediated degradation (Tao and Levine, 1999).

Fig. 7: Hypothetic model about the effects of impaired MDM2 function on p53 protein accumulation in the cell. Cellular stresses cause p53 activation, which is downregulated by functional MDM2 in an autoregulatory feedback loop. In case of nonfunctional MDM2, wildtype p53 accumulates in the nucleus.

The MDM2/p53 autoregulatory feedback loop, as described above, can be abolished by p53 as well as MDM2 alterations (Fig. 7). Firstly, p53 mutations have been shown to cause accumulation of p53 protein which is frequently seen in tumors due to a lack of p53-mediated MDM2 induction (Haupt et al. 1997, Kubbutat et al. 1997). Secondly, a dysfunctional MDM2, as it is seen in some forms of alternative splicing, can cause p53 accumulation due to the loss of p53 binding ability (Kubbutat et al. 1997, Kraus et al. 1999).

Physiologically, the MDM2 protein itself is degraded by ARF, which means that p53 cell cycle regulatory functions can be restored by ARF/ MDM2 binding (Zhang et al. 1998, Pomerantz et al. 1998). Despite other, as yet not fully understood, functions of MDM2 the MDM2/p53 interaction seems to play a crucial role in cell cycle regulation. MDM2 ensures effective reduction and termination of the p53 signal


32

after DNA damage induced p53 activation to reverse cell cycle arrest and allow the re-entry of the cell into the cell cycle.

1.4.4 Clinical implications of mdm2 alterations

mdm2 amplification has been frequently noted in tumors of mesenchymal origin, and overexpression, which is present in approximately 20% of soft tissue sarcomas, has been found associated with poor prognosis (Cordon-Cardo et al. 1994). mdm2 alternative mRNA splicing has been found to be associated with poor differentiation and advanced stage in ovarian carcinomas (Sigalas et al. 1996). Since MDM2 controls p53 function in an autoregulatory feedback loop, it plays a major role in tumorigenesis.

It was hypothesized that tumors expressing abnormally high levels of mdm2 could be treated by means of peptides derived from the p53 interaction domain. By binding to MDM2 and preventing its interaction with p53, such peptides could restore p53 function to the tumor cells, presumably preventing further growth (Oliner et al. 1993). Microinjection of a monoclonal antibody against the p53-binding domain of MDM2, 3G5, (Chen et al. 1993) blocks the p53/MDM2 interaction and leads to p53-dependent activation of reporter genes in cell lines that have wild-type p53 (as reviewed by Freedman et al. 1999).

Since MDM2 is assumed to counteract the induction of apoptosis by p53, it may also play a role for the response of tumors to DNA-damaging chemotherapeutic agents. Overexpression of mdm2 was found to decrease the susceptibility of human glioblastoma cells to apoptosis induced by Cisplatin, while expression of antisense mdm2 increased their susceptibility (Kondo et al. 1995). Lowering the levels of mdm2 in cell lines that overexpress p53 also appears to activate the wild-type p53 protein present in such cells. By using antisense oligonucleotides to the mdm2 message in a choriocarcinoma cell line which has wild-type p53 and mdm2 gene amplification, p53 reporter genes can be activated and the cells induced to undergo apoptosis (Chen et al. 1998). Following intraperitoneal administration of anti-mdm2 antisense oligonucleo-tides, in vivo antitumor activity was observed in nude mice bearing osteosarcoma and choriocarcinoma xenografts (Wang et al. 1999). An mdm2 antisense phosphorothioate oligodeoxynucleotide was identified that effectively inhibits mdm2 expression in tumor cells containing mdm2 gene amplification. Antisense inhibition of mdm2 was asso-ciated with a decrease in MDM2/p53 complex formation, increase in p53-inducible gene expression, increase in p53 transcriptional activity and apoptosis (Chen et al. 1998). Recently a novel mechanism of MDM2 degradation by the protein ARF has been identified, which can neutralize the inhibition of p53 by MDM2 and restore


33

the p53 function for cell cycle control and apoptosis (Zhang et al. 1998, Pomerantz et al. 1998).

Further understanding of the functions and regulatory mechanisms in which mdm2 is involved will give important new insights in tumorigenesis and resistance to chemotherapy. Disruption of the MDM2/p53 wildtype interaction and perhaps Rb, E2F1, and DP1 might be important goals for cancer therapy.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Thu Aug 15 15:17:40 2002