GENETIC CONTROL OF SUSCEPTIBILITY TO TESTICULAR CANCER

By

MAN-YEE JOSEPHINE LAM

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Dissertation Advisor: Dr. Joseph H. Nadeau

Department of Genetics

CASE WESTERN RESERVE UNIVERSITY

May, 2005 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

______

candidate for the Ph.D. degree *.

(signed)______(chair of the committee)

______

______

______

______

______

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein. TABLE OF CONTENTS

TABLE OF CONTENTS ...... 1 LIST OF TABLES ...... 3 LIST OF FIGURES ...... 4 ACKOWLEDGEMENTS ...... 5 ABSTRACT ...... 7 CHAPTER 1: INTRODUCTION...... 9 Diagnosis and treatment of TGCTs...... 10 TGCT susceptibility in humans...... 11 Genetics factors...... 13 Mouse models of TGCTs ...... 14 Primordial germ cells and TGCT development ...... 15 Composition of TGCTs ...... 18 Genetic control of TGCTs in mice...... 18 Mendelian mutations that affect TGCT susceptibility ...... 20 Mendelian mutations to sensitize linkage analysis...... 29 Substitution Strains...... 32 Congenic strains ...... 35 Research aims...... 35 CHAPTER 2: PAIRWISE INTERACTION TEST BETWEEN TGCT SUSCEPTIBILITY LOCI IN MICE...... 38 CHAPTER 2 ...... 38 Abstract...... 39 Introduction ...... 40 Materials and methods...... 43 Results...... 46 Discussion ...... 72 CHAPTER 3: P53 AND STEEL-J DELETION INTERACT TO SUPPRESS TESTICULAR GERM CELL TUMOR SUSCEPTIBILTY...... 86 Abstract...... 87

1 Introduction ...... 88 Materials and Methods...... 94 Results...... 97 Discussion ...... 105 CHAPTER 4: SEQUENCE ANALYSIS OF THE HUMAN ORTHOLOG OF A MOUSE TGCT SUSCEPTIBILITY ...... 117 Abstract...... 118 Introduction ...... 119 Materials and Methods...... 127 Results...... 128 Discussion ...... 129 CHAPTER 5: SUMMARY AND FUTURE DIRECTIONS ...... 134 Summary ...... 135 Future Directions...... 140 APPENDIX 1 ...... 147 BIBLOGRAPHY...... 150

2 LIST OF TABLES

CHAPTER 1 1.1 Effects of genes on incidence of teratomas………………………………….22

CHAPTER 2 2.1 Heterogeneity among the tumor frequency of all single-mutants in control and interaction crosses……………………………………………………...…47 2.2 Parental effects of various interaction crosses………………………………54 2.3 Interaction test between M19 and SlJ…………………………………………56 2.4 Interaction test between Ter and SlJ…………………………………………..61 2.5 Interaction test between Ter and p53……………………………………. …..63 2.6 Interaction test between M19 and Ter………………………………………...65 2.7 Interaction test between M19 and Ay………………………………………….68 2.8 Laterality of TGCTs in different double-heterozygous mutants…………….71

CHAPTER 3 3.1 Interaction test between p53 and SlJ in double heterozygous mutants…..99 3.2 Interaction test between p53 and SlJ in homozygous-heterozygous mutants…………………………………………………………...………….....101 3.3 Control cross to test whether double heterozygous parents affect TGCT susceptibility……………………………………………………………104 3.4 Expression ratio of p53 cDNA using Q-PCR……………………………….111

CHAPTER 5 5.1 Effects of Avy on TGCT frequency in the 129-Chr19M……………………..142

3 LIST OF FIGURES

CHAPTER 1

1.1 Histological section of 4-week old mouse testis……………………………..19

1.2 Construction of a chromosome substitution strain…………………………..34

CHAPTER 2 2.1 The location of the congenic segment A2 and C2…………………………..58 2.2 Relationship between the observed and predicted bilateral tumors as a function of the overall TGCT frequency…………………………………………....71 2.3 Genomic structure of the AW and Ay mutant allele………..…………………79 2.4 Network of interaction between M19, Ter, p53, Ay and SlJ…...... 85 2.5 Roles of M19 in the tumorigenesis pathway…………………………………85

CHAPTER 3 3.1 Western analysis of p53 in 4-week old testis……………………………….113

CHAPTER 4 4.1 Gene structure of human ortholog dnd1…………………………………….123

4 ACKOWLEDGEMENTS

First, I would like to thank my mentor, Joe Nadeau, for all the valuable guidance and opportunities in the lab. Although he has many responsibilities and commitments, yet he always finds time for his students.

I would like to thank my parents and my brother for their continue support during the last 5 years. There have been many times that I doubted my abilities to finish my degree but they always believed in me. I am thankful that my mother tolerated my endless complaints about graduate school on the phone and never once did she tell me to stop whining. Thank you for all their love, support and encouragements.

As a graduate student, we spend countless hours in the lab therefore; it is a bonus when people in the lab are also your good friends. I’m very lucky that I’ve made four very dear friends in the lab, Kirsten, Lesil, Sheila and Toshi.

They became my family away from home. We cheered each other on during the hard times and complained together of the hardships. I would specially like to thank Toshi, who has been my bench mate for the last 4 years. He is one of the smartest people I know. He was like my personal encyclopedia that I can always turn to for references. He has also been great at tolerating many of my stressful moments in the lab. All four of them have taught me a lot in science and in life. And I thank them for their friendship from the bottom of my heart.

5 Finally, I would like to thank my fiancé, Ricky, for all his love, support and encouragements. I am so glad that I came to CWRU for my degree. If not, I would have never met the love of my life. People say you’ll find the “one” when you least expect it. Well, that is definitely the case with Ricky. We had mutual friends and he sat directly above me for 3.5 years before we finally got to know each other. I never thought the quiet guy from the Hassold lab who never says hello in the hallway or the elevator would turn out to be my true love. He is the pillar of my life. He makes me stronger and brings out the best in me.

6 Genetic Control of Susceptibility to Testicular Cancer

Abstract

By

Man-Yee Josephine Lam

Testicular germ cell tumors (TGCTs) are the most common solid tumors affecting young to middle aged men. It is the fifth most rapidly increasing type of cancer. Therefore, a better understanding is urgently needed of risk factors, early diagnosis, and treatment before metastasis. To date, there has not been a susceptibility gene identified in humans. We study a mouse model, the 129/Sv inbred mouse strain, in which TGCTs arise spontaneously by 3-weeks of age.

Linkage studies with the 129/Sv strain, which has a low rate (1-5%) of TGCTs, reveal exceptional complexity in the genetic control of susceptibility. Various mutations that are inherited as Mendelian traits in laboratory mice affect susceptibility to spontaneous TGCTs on the 129/Sv genetic background. These genes, alone or in combination, provide important clues to the control of susceptibility and resistance to tumorigenesis. We undertook a systematic survey to test pairwise interactions of all known TGCT susceptibility loci, by comparing the frequency of TGCTs in double-mutant mice to identify combinations that show evidence of enhancer or suppressor effects. We

7 identified three pairs of genes that enhanced the susceptibility of TGCTs and two pairs of genes that suppress the development of TGCTs. These mouse models can help simplify the complexity of TGCTs.

We can extrapolate from the findings in the mouse model to help identify candidate susceptibility genes in human. Recently, we identified the Ter mutation to be a premature stop codon in the Dnd1 gene that causes significantly increased susceptibility to TGCTs. The human ortholog of Dnd1 maps to Chr5q31. We performed an association study to test whether the human ortholog of Dnd1 contributes to susceptibility of TGCTs in human. We did not identify a mutation within the coding region in the cases, which indicates that Dnd1 may be a low frequency variant in human population.

8

CHAPTER 1

INTRODUCTION

9 Testicular germ cell tumors (TGCTs) are the most common cancer in

males aged 20-35 year (BISHOP 1998). TGCTs affect 1 in 500 men and the

incidence varies among geographical areas. The highest testicular cancer

rates are reported in the white populations of northern and western Europe,

especially Denmark and Switzerland (BUETOW 1995). In Eastern Europe, the

incidence has more than doubled in the last 50 years, suggesting that

environmental as well as genetic factors are involved in the development of

TGCTs (BUETOW 1995). The increase in incidence have been predominantly among young men, with little change observed in men aged 65 years and over or in children (SCHOTTENFELD et al. 1980).

Diagnosis and treatment of TGCTs

TGCTs often present clinically as a painless swelling or lump. The tumor may cause mild discomfort because of its weight, but rarely pain. Most testicular cancer cases are unilateral and predominately affect the right testis

(BOSL and MOTZER 1997; BUETOW 1995). Bilateral tumors are rare at diagnosis,

but in two percent of patients with unilateral TGCT, a metachronous new tumor

will develop in the remaining testis.

TGCTs are often malignant and metastasize readily if not treated early.

The major route of metastasis is lymphatic, the lumbar and mediastinal nodes

being commonly involved. Right testicular tumors usually metastasize to nodes

between the aorta and the inferior vena cava and left testicular tumors

metastasize to nodes lateral to the aorta (RAY et al. 1974).

10 This type of tumor is very sensitive to platinum-based chemotherapy,

making TGCTs-the most readily treated form of cancer (BUETOW 1995).

Although TGCTs can be easily treated, the side effects of the treatment can lead to infertility and premature heart attack (VAUGHN et al. 2002). It is therefore of great importance to understand the genetic and molecular mechanisms of

TGCT susceptibility to develop better markers and treatments for TGCTs.

TGCT susceptibility genes in humans

Risk factors for TGCTs include testicular dysgenesis, undescended testis, infertility, and previously diagnosed TGCT. Family history is a significant feature of testicular cancer; brothers and sons of TGCT patients show an increased risk of 8 to 10-fold and 4-fold, respectively, which implies a strong genetic control (FORMAN et al. 1992; LINDELOF and EKLUND 2001). Despite the prevalence of TGCTs, evidence for genetic factors, and increased risk among relatives of TGCT cases, little is known about the genetic control of susceptibility.

To date, a TGCT susceptibility gene has not been identified in humans.

Linkage studies have been difficult because of the rarity of multigenerational pedigrees with several affected individuals, sterility resulting from chemotherapies, and the complexity of genetic control. Mapping studies of families collected by the International Testicular Cancer Linkage Consortium revealed weak linkages on 3, 4, 5, 11, 12 and 18 (BISHOP 1998;

LEAHY et al. 1995). The increased risk of TGCT between fathers and sons is

11 less than the risk between brothers which suggests that there may be X-linked

inheritance. In fact, linkage studies extended to the X chromosome

demonstrated the first significant linkage on Xq27 (RAPLEY et al. 2000).

Interestingly, the LOD score increased when linkage analysis were performed only on families with at least one bilateral tumor case (RAPLEY et al. 2000). This data suggest that the locus on Xq27 to be linked to predisposition to bilateral tumors.

The age distribution of testicular cancer is unusual in comparison to most cancers. In most instances, cancer incidence increases with age, however, this is not the case for TGCT. The incidence of TGCTs peaks at about 20-40 years of age and declines with advancing age thereafter (Nicholson and Harland

1995). TGCTs usually occur in three age groups and each group has characteristic tumor types. A small number of males develop pediatric germ cell tumors before puberty and these are histologically classified as teratomas, teratocarcinomas, and yolk sac tumors. The highest incidence occurs in males after puberty between the ages of 10-34 and these tumors are often malignant and are usually a mix of seminomas and non-seminomas. Males over 50 years old develop tumors derived from spermatocytes and these tumors are histologically classified as seminomas.

TGCTs originate from different stages of primordial germ cell (PGC) development based on evidence from studies of immunohistochemistry, loss of imprinting, methylation profiles, and cytogenetics (LOOIJENGA et al. 1998;

RODRIGUEZ et al. 1992; VOS et al. 1990). Tumorigenesis appears to be very

12 complicated and requires two stages. The first stage is the formation of

carcinoma in situ (CIS). CIS cells have been recognized as the precursor cells

of all types of TGCTs, except spermatocytoma (RAJPERT-DE MEYTS et al. 1996).

Studies by Rajpert-De Meyets demonstrate variation in the immunocytological phenotype of CIS cells, suggesting that these cells have several functionalities.

CIS cells can act as pluripotent stem cells, and give rise to non-seminomas, or they can lose stem cell potential with age and give rise to more limited cell and tissue types (RAJPERT-DE MEYTS et al. 1996). The second stage of tumorigenesis is a somatic event that triggers the CIS to develop into TGCTs.

It is not evident how or when CIS arises, but the CIS resembles the embryonal carcinoma (EC) cells found in the early stage of mouse TGCTs.

An understanding of the normal and abnormal development of PGCs will provide insights into the development of TGCTs. A model system is needed to study the biology of germ cell development, pathways that direct germ cell fate during embryonic, fetal and postnatal development, and the tumorigenesis pathway of TGCTs.

Genetic factors

TGCT are commonly hyperdiploid and more frequently triploid or tetraploid. These events usually involve chromosome 12, and excess 12p genetic material is found in all germ cell tumors. In addition, genetic loss is frequent at two common sites on the long arm of chromosome 12, suggesting that a tumor suppressor gene may exist on 12q (BOSL and MOTZER 1997).

13 However, these chromosomal abnormalities arise from somatic mutations in

TGCTs and they should not be considered as inherited predisposing genetic factors for testicular cancer.

Mouse models of TGCTs

Dissecting the genetic control of susceptibility to TGCT in human has been difficult. As a result, experimental model systems to study the genetic control of TGCT susceptibility are essential. Testicular teratomas are rare in mice and occur spontaneously at a modest but measurable frequency of 1-5% in the 129/Sv inbred strain. The 129/Sv inbred mice are the best characterized mammalian system for studying TGCTs (Stevens and Hummel 1957).

Spontaneous TGCTs in 129/Sv inbred mice resemble human pediatric germ cell tumors that originate during fetal development as they are composed of various cell and tissue types at various stages of differentiation.

There are many advantages of studying TGCTs in the 129/Sv inbred mouse strain. Mice are available in large numbers, they are genetically identical therefore experiments can be repeated, the generation time is only 10 weeks which is shorter than other mammals, and the size of the mouse testis makes it feasible to study serial sections of the entire gonads for histopathology and immunochemistry studies.

14 Primordial germ cells and TGCT development

Leroy Steven’s pioneering work from mid-1950’s established

fundamental information about TGCTs in the 129/Sv inbred strain including the cell of origin to be primordial germ cells (PGCs) (STEVENS 1967b). PGCs first arise during gastrulation around embryonic day 7 (E7) near the base of allantois

(GINSBURG et al. 1990). They become incorporated into the developing hind-gut

endoderm at E8, migrate through the dorsal mesentery of the gut, and arrive at

the genital ridges at E11.5 (MOLYNEAUX et al. 2001). While migrating, PGCs proliferate from a population of ~100 cells at E7 to about 25,000 cells at E13.5

(Tam and Snow 1981). After the arrival of male PGCs at the genital ridges, sexual differentiation of the fetal gonads begins, the PGCs enter mitotic arrest at E13.5 and remain mitotically arrested until a few days after birth (DONOVAN et

al. 1998; LIN 1997).

A series of genital ridge grafting experiments was performed by L.C.

Stevens to experimentally induce TGCTs in testes of adult 129/Sv mice which lead to his discovery of the cell of origin and the onset time of TGCTs (STEVENS

1967b). Mice homozygous for the Steel (Sl) mutation lack PGCs and die due to severe anemia after E12. Teratomas do not develop after genital ridges from

E12 homozygous SlJ/SlJ fetuses are grafted to 129/Sv adult testes. In contrast,

75% of 194 testes of 129/Sv males developed teratomas from genital ridges grafted from SlJ/+ and wild-type fetuses, which have a modest number of PGCs

(STEVENS 1967b). This showed that PGC is the cell of origin for TGCTs. To establish the onset of TGCTs, genital ridges from E11 to E13 of wild-type

15 fetuses were grafted into 129/Sv males to test which genital ridge developed

into TGCTs. The ability of the genital ridges to form TGCTs peaked in grafts

from E11 to E12.5 fetuses and incidence decreases in grafts after E12.5

(STEVENS 1966). These experiments demonstrated that the onset of TGCTs probably occurs between E11 and E12.5.

It has been proposed that TGCTs develop because some PGCs fail to enter mitotic G1 arrest and continue to divide for several days giving rise to

pluripotent stem cells called embryonal carcinoma (EC) cells that become

disorganized and form tumors of various cells and tissues (Stevens and

Mackensen 1961). Therefore, cell cycle regulation is likely to play a major role

in TGCT tumorigenesis. An alternative hypothesis is germ cell tumors may

arise from germ cells that do not migrate properly (STALLOCK et al. 2003).

During embryogenesis, a significant number of PGCs fail to migrate correctly to the genital ridges. These PGCs usually die in ectopic locations. Cell death is also a feature of germ cells that arrive successfully in the genital ridges. Male germ cells undergo apoptosis between E13.5 and E17 (COUCOUVANIS 1993), followed by a second wave of apoptosis around the time of birth which depletes the number of male germ cells by 50% (Roosen-Runge and Leik 1968). Mouse germ cell death is due to apoptosis but it is unclear how germ cells that fail to migrate properly to the genital ridges undergo cell death. Studies have shown that MGF-KIT interaction is necessary for germ cell survival. Mgf stimulates

PGCs to survive or proliferate in culture (DOLCI et al. 1991; MATSUI et al. 1991)

and in vivo, mice lacking Mgf or c-Kit receptor show severe PGC deficiency

16 (Mintz and Russell 1957) and the germ cells die by apoptosis. Bax has also

been shown to be required for germ cell death during and after migration of

germ cells that do not reach the genital ridge (RUCKER et al. 2000). The mechanism by which the PGCs are eliminated when they do not arrive at the genital ridges is still unknown. It is still unclear if germ cells that do not migrate towards the genital ridges are the culprit that causes TGCT. It can be argued against the hypothesis that germ cell tumors arise from germ cells that do not migrate properly because it may be possible that germ cells that migrated away from the genital ridges may require specific growth factors which are not present at the ectopic locations and the lack of proper environment and signals will most likely cause PGCs to undergo cell death. In addition, if PGCs that do not migrate properly do survive and transform into tumors, these tumors would most probably not be TGCTs since the PGCs never arrived at the genital ridges and instead migrated to other ectopic locations.

Studies in mouse models have revealed many different molecular controls of PGC development. TGCTs may result from dysfunctions in many different aspects of PGC development including proliferation, migration, cell cycle regulation, differentiation and pluripotency. Therefore, it is necessary to determine the function of all the different molecular controls of PGC development with respect to TGCT development.

17 Composition of TGCTs

TGCTs are composed of all three primary germ layers in different stages

of differentiation. The composition of teratomas is highly variable (Figure 1.1).

The tumors contain a disorganized collection of cells and tissues in various

stages of differentiation (STEVENS 1967a; STEVENS and HUMMEL 1957). In the mouse, neural tissue is the predominant component in TGCT. TGCTs are a chaotic collection of tissue, with occasional tissue found in their expected association. Skin with hair and sebaceous glands or muscle attached to cartilage and bone can be seen in teratomas (STEVENS 1967a; STEVENS and

HUMMEL 1957). The earliest recognizable TGCT can be observed at E15

(STEVENS 1967a). At E15, clusters of foci derived from EC cells rupture the seminiferous tubules and invade into the interstitial spaces.

Genetic control of TGCTs in mice

Spontaneous TGCTs occur at a measurable frequency only in 129/Sv inbred strains which provides strong evidence for genetic control. Backcrosses between 129/Sv mice and other inbred strains were used to evaluate the genetic complexity of susceptibility. Remarkably, of more than 11,000 backcross male progeny examined, only one had a spontaneous TGCT

(STEVENS 1967a). This low frequency is consistent with a threshold trait requiring the action of at least six independently segregating genes with additive and recessive effects. Despite the availability of a genetically defined strain, the low frequency of spontaneous TGCTs in segregating populations

18

A B

Figure 1.1 Histological section of 4-week old mouse testis.

A. Histological section of a wild-type testis. B. Histological section of a testis affected with TGCT. The mix of different tissue type include: bone, bone marrow, neuroepithelium and secretory epithelium.

19 makes it difficult to dissect the control of susceptibility because tens of

thousands of mice must be screened to obtain a sufficient number of affected

mice to map and clone susceptibility genes. To overcome these difficulties,

alternative genetic approaches are required to identify susceptibility genes. The

novel approaches that have been developed to study TGCTs include,

characterization of Mendelian mutations on the 129/Sv background, the use of

Mendelian mutations to sensitize the linkage analysis of 129-derived TGCT

genes (COLLIN et al. 1996), and the development of genetic resources such as chromosome substitution strains to simplify the genetic analysis of TGCT

(MATIN et al. 1999). Each approach has yielded important insights into TGCT susceptibility and can be applied to the study of many other complex traits.

Mendelian mutations that affect TGCT susceptibility

Several single Mendelian mutations affect TGCT susceptibility but only on the 129/Sv genetic background: Steel (Sl), Steel-J (SlJ), Ter, Phosphate and

tesin homolog deleted from chromosome 10 (PTEN) and Transformation related 53 (p53) increase TGCTs incidence, whereas Lethal yellow (Ay ) decreases tumor incidence .

Mgf-Steel (Sl) and Mgf-SteelJ (SlJ)

The Steel (Sl) and White-spotting (W) mutants were first identified as

coat color mutations (Silvers 1979). The W locus has been shown to encode c-

Kit, a receptor glycoprotein in the platelet-derived growth factor receptor family

20 with tyrosine kinase activity encoded in a split kinase domain (CHABOT et al.

1988). W locus mutations include large deletions, rearrangements and point

mutations which affect the amount of c-Kit protein expressed and the level of

kinase activity (DUBREUIL et al. 1990). The ligand for the c-Kit receptor is mast cell growth factor (Mgf) or Kit ligand (KL) or Steel factor (SLF), which is encoded at the Sl locus (ANDERSON et al. 1990; COPELAND et al. 1990). Mgf is produced as a membrane bound growth factor that undergoes proteolytic cleavage to generate a soluble form. The Sl locus is highly mutable and many induced (X-irradiation) and spontaneous mutations have been described

(BEDELL et al. 1996; COPELAND et al. 1990). Mutant mice are defective in

melanogenesis, gametogenesis and hematopoiesis (COPELAND et al. 1990).

Mice mutated at either locus show defective PGC development, which result in sterility. The MGF-KIT signaling pathway is required for PGC survival and proliferation during PGC development (DOLCI et al. 1991). The mechanism by which Mgf binding to the c-Kit receptor mediates PGC growth and survival is still unknown.

L.C. Stevens transferred various Steel and White-spotting mutations onto the 129/Sv inbred strain to study tumorigenesis and found that only Sl and SlJ alleles result in a significant increase in spontaneous testicular tumors, whereas other alleles of Sl did not affect susceptibility (i.e.: Sld). By contrast, none of the

W mutants (W3J, Wx , Wv) that Stevens tested as congenics on the 129/Sv

inbred background affected susceptibility (Table 1.1).

21

Table 1.1 Effects of genes on incidence of teratomas

Various Steel and White-spotting mutations were transferred on the 129/Sv inbred strain by L.C. Stevens. Sl and SlJ are the only alleles that significantly increase susceptibility of TGCTs. +/+ are wild-type mice and M/+ denotes heterozygous mutants of the various Steel and White-spotting mutations.

% Teratomas Mutation Sample Size +/+ M/+ Sl 280 2.0 14.0 SlJ 2389 2.5 6.9 Sld 1614 5.0 5.0 W 650 4.0 2.0 W3J 122 2.0 0.0 Wx 232 2.0 0.0 Wv 896 2.0 3.0

22 SlJ is a spontaneous mutation that results from a 640kb deletion on chromosome 10 that includes the Mgf (BEDELL et al. 1996). Homozygous

SlJ/SlJ mutants are PGC deficient and embryonic lethal due to severe anemia.

Heterozygous 129/Sv-SlJ /+ males have a tumor incidence more than doubled

(14%) their wild-type littermates (5%) (STEVENS 1967a). However, loss of

PGCs during development is not sufficient to cause TGCTs because other alleles of Sl such as Sld, which is an intragenic point mutation within the Mgf

gene, do not increase tumor incidence even though Sld homozygotes are germ cell deficient and sterile. It is speculated that mutated Mgf is not the cause of

TGCTs, but perhaps a gene near Mgf with haplosufficient effects leads to teratocarcinogenesis.

Ter

Ter is a single gene mutation that reduces germ cell numbers in all inbred strains and significantly increases TGCT incidence in 129/Sv mice

(NOGUCHI and NOGUCHI 1985). The Ter mutation causes PGC deficiency at around E8 and PGC numbers remain reduced throughout the remainder of the

PGC migratory and proliferative stages (SAKURAI et al. 1994). The small number of PGCs that arrive at the genital ridges do not enter G1 mitotic arrest and continue to divide until E15.5 (NOGUCHI and NOGUCHI 1985). It is hypothesized that the small numbers of PGCs that arrive at the genital ridges of

Ter mutants on the 129/Sv inbred background develop into TGCTs at an extraordinary frequency. Homozygous Ter/Ter mutants on the 129/Sv inbred

23 background have a tumor incidence of 94%, 75% of which are bilateral, and

17% of Ter/+ heterozygotes are affected with TGCTs, of which only 10% are

bilateral (Noguchi and Noguchi 1985). Ter/Ter homozygotes are PGC deficient,

therefore males are sterile and females have few eggs in their ovaries and

rarely produce offspring.

Ter was previously mapped to chromosome 18 near Fgf1 (ASADA et al.

1994). The lab have recently cloned the Ter mutation by carrying out further

high resolution genetic mapping using testis weight as a quantitative phenotype

for PGC deficiency and were able to narrow the critical region to 0.14cM

(Youngren et al., in press). To further define the critical segment, complementation tests with bacterial artificial chromosome (BAC) transgenics were used. A series of transgenic mice were generated with four overlapping

BACs that span the critical region. The transgenic mice were crossed to

B6.129-Ter congenic mice to generate Ter/Ter mutants carrying the BAC. Two transgenic mice with overlapping BACs rescued the germ cell deficiency in the

Ter homozygous mutant. There are 4 genes within the overlapping region of the two BACs. Exons and 5’ and 3’ flanking regions of each candidate gene were sequenced. A single base change was identified in the Dnd1 gene that results in a premature stop codon. Dnd1 is an ortholog to the deadend gene in zebrafish, which is required for PGC survival and migration. Dnd1 may be a component of RNA editing. It shows homology to several that have an

RNA recognition motif including apobec-1 complementation factor. In order to confirm that Dnd1 is Ter, transgenic mice carrying only the Dnd1 transgene

24 were crossed to the Ter/+ mutants to obtain Ter/Ter mutants with the

transgene. The Dnd1 transgene showed partial rescue of the germ cell

deficiency. Although many seminiferous tubules remained germ cell deficient,

some contained immature and mature sperm, which is not observed in the

Ter/Ter homozygous mutants. Northern and western analysis verified that

Dnd1 is expressed in testis of wild-type adult mice and not in testis of Ter homozygous.

Dnd1 may be an RNA editing protein and the increased susceptibility of

TGCTs in the Ter mutants may be a consequence of adversely affected RNA editing. Several features of RNA biology play a role in the development of PGC lineages such as noncoding RNAs that repress transcription (MARTINHO et al.

2004), RNA and RNA-binding proteins that regulate translation (KNAUT et al.

2002; MOORE et al. 2003) and maintain totipotency of germ cells (CRITTENDEN et al. 2002; ZHOU and KING 2004). The mechanism by which Dnd1 is implicated in

tumorigenesis is currently under investigation.

PTEN

The tumor suppressor gene, PTEN, is a lipid phosphatase which plays crucial roles in the regulation of cellular proliferation, differentiation, apoptosis, adhesion and migration (CANTLEY and NEEL 1999; DI CRISTOFANO and PANDOLFI

2000; YAMADA and ARAKI 2001). PTEN is one of the most frequently mutated tumor suppressor genes in human cancer, and the overall frequency of PTEN mutations in sporadic human cancers is similar to p53 (STOKOE 2001). PTEN

25 knockout mice have bilateral TGCTs, which result from impaired mitotic arrest

and outgrowth of cells with immature characteristics. Experiments using PTEN-

null PGCs in culture showed that the mutated PGCs have an increased

proliferative capacity and enhanced pluripotent germ cell colony formation

(KIMURA et al. 2003). PTEN appears to be essential for germ cell differentiation

and is an important factor in TGCT development.

p53

p53 is a tumor suppressor gene that arrests cells to allow proper repair

of DNA damage (DONEHOWER et al. 1992). Mutations in p53 are the most commonly observed genetic lesion in spontaneous human cancers. Loss of p53 is associated with various types of cancers such as lymphomas, lung, breast, colon, liver, brain, bladder and ovary. Curiously, it is not associated with testicular cancer. To study the role of p53 in tumorigenesis, a null mutation was made by homologous recombination in mouse embryonic stem cells

(DONEHOWER et al. 1992). Mice homozygous for the null allele appear normal but are prone to a variety of tumors, with lymphomas being the most common

(DONEHOWER et al. 1992). The spectrum of tumors differs among p53-deficient mice on various genetic backgrounds (DONEHOWER et al. 1995; HARVEY et al.

1993a). In a mixed C57BL/6 and 129/Sv genetic background, only 9% of the

p53-deficient mice develop testicular tumors, whereas in a pure 129/Sv genetic

background, 35% of the p53-deficient mice develop testicular tumors (HARVEY et al. 1993a).

26 Spontaneous germ cell death is a common process in the testis,

although the function of this process is still unclear. It is known that p53 mediates spontaneous testicular germ cell apoptosis and failure to remove defective germ cells by this mechanism results in increased number of abnormal sperm and results in reduced fertility, which is a risk factor for testicular cancer. p53 regulates cell cycle control and it is hypothesized that cell cycle regulation plays a major role in TGCT tumorigenesis, therefore it is important to investigate the role of p53 in the development of TGCTs.

Lethal Yellow (Ay)

Ay is one of the oldest identified mutations of the agouti coat color locus.

Ay is a dominant mutation that was first described by Cuénot (Cuenot 1902) at

the agouti locus on chromosome 2. As with other dominant agouti mutations,

Ay leads to the constitutive production of yellow pigment. It is associated with obesity, features of type II diabetes, and an increased predisposition to develop a variety of spontaneous and induced tumors (CARPENTER and MAYER 1976;

DICKERSON and GOWEN 1947; FENTON and CHASE 1951; PLOCHER and POWLEY

1976).

The phenotype of Ay /+ mice results from ~150kb deletion of the coding region of the agouti and several other closely linked genes (MICHAUD et al.

1994). This deletion brings the coding region of agouti under the transcriptional regulation of the Raly promoter, resulting in ectopic and ubiquitous expression of the agouti gene product. Homozygous Ay/Ay mice are embryonic lethal

27 (Eaton and Green 1963). Remarkably, transfer of the Ay mutation to the 129/Sv genetic background resulted in a 8-fold reduction in the frequency of

y spontaneous TGCTs (STEVENS 1967a). We have made a new 129-A congenic strain and verified the reduced TCGT susceptibility (LAM et al. 2004).

The agouti gene has been cloned and it is thought to encode a signaling

molecule that directs follicular melanocytes to switch from the synthesis of

eumelanin, the black pigment, to phaeomelanin, the yellow pigment (BULTMAN et al. 1992). Agouti RNA is expressed only in the skin during synthesis of phaeomelanin. Ay is associated with the ubiquitous expression of agouti

(MILLER et al. 1993) and it has been suggested that some or all of the

pleiotropic effects are due to ectopic expression of the agouti protein.

MOLF-Chromosome 19

At least one susceptibility locus on Chr19 of MOLF/Ei inbred strain

predisposes to TGCTs (COLLIN et al. 1996). A chromosome substitution strain

(CSS) was generated by replacing MOLF-derived Chr19 on the 129/Sv inbred

background (MATIN et al. 1999). Homosomic mice with two copies of MOLF-

Chr19 have a tumor incidence of 80% and heterosomic animals with one copy

of MOLF-Chr19 have a tumor incidence of 20%. Further discussion of this CSS

will be later in the section.

The various mutations discussed above can affect tumor frequency by

increasing penetrance, by reducing the number of 129-derived genes that are

28 required for tumorigenesis, or by activating alternative and genetically simpler

pathways leading to TGCTs. The propensity of these mutations to cause

TGCTs only on the 129/Sv background suggests that they interact with 129-

derived genetic variants. Except for p53 and Ter, which was recently cloned,

the other Mendelian mutants that act as susceptibility genes have not yet been

cloned; therefore much remains to be learned about their functions and their

effect on TGCT susceptibility. In many other systems, double-mutant mice

show suppressed phenotypes, possibly due to one mutant suppressing the

phenotype of the other. In some instances, novel phenotypes are found

(MONTES DE OCA LUNA et al. 1995; PARANT et al. 2001). These suppressed or

novel phenotypes may provide important clues to gene interactions and may

provide clues to the pathways that control tumorigenesis. Double-mutant mice

for several TGCT susceptibility genes may provide insights into their effects on

tumor frequency and laterality as well as their roles in PGC development.

Mendelian mutations to sensitize linkage analysis

The polygenic nature of complex disease and the subtlety of the effects of individual genes on disease progression challenge the ability to discover genes involved in complex traits. In addition, a major obstacle in identifying susceptibility genes in complex traits is the usually modest number of affected individuals that are available for linkage analysis. Single gene mutations can be used to simplify the genetic analysis of complex traits - a method known as

‘sensitized polygenic trait analysis’ (Matin and Nadeau 2001). These mutations

29 increase the frequency of affected mice thereby increasing the statistical power

of the analysis. They appear to act by converting genes with weak background

effects into semi-Mendelian traits that are readily mapped and presumably

readily cloned.

Sensitized polygenic trait analysis and TGCT susceptibility

Two mutations, Ter and p53, have been used to map novel TGCT

susceptibility loci. The Ter mutation was used to sensitize the linkage analysis

for TGCT susceptibility genes in crosses between the 129/Sv-Ter and MOLF/Ei

strain (COLLIN et al. 1996). The sensitized linkage cross resulted in a 1700-fold increase in the number of affected mice in comparison to linkage crosses without Ter. A QTL analysis was performed on progeny affected with TGCTs

and there was no evidence of linkage to susceptibility genes. However, when

linkage analysis was performed only on mice affected with bilateral tumors,

linkage was detected on the central region of Chr19 with a bias for MOLF-

derived alleles. Additionally, mice affected with unilateral tumors showed an

excess of 129-derived alleles on distal Chr19. These results suggest that there

is at least one TGCT susceptibility gene on Chr19 (COLLIN et al. 1996). To further characterize the TGCT susceptibility genes on Chr19, a novel approach of constructing a chromosome substitution strain was used and this will be discussed later in detail.

Another example involves a mutation of p53 that was used to sensitize the linkage cross between the 129/Sv-p53 and C57BL/6 strains. This analysis

30 resulted in a 100-fold increase in TGCT susceptibility and revealed a novel

TGCT susceptibility locus on Chr13 named primordial germ cell tumor locus 1

(pgct1) (MULLER et al. 2000).

Regardless of whether Ter and p53 increase penetrance, interact with a subset of genes, or activate a novel pathway, the increased frequency of affected mice made it possible to map novel TGCT susceptibility genes in sensitized crosses. It is striking that different mutations and different inbred

strains revealed distinct linkages. Systematic use of this paradigm may

therefore prove to be a robust method to identify numerous TGCT susceptibility genes.

Sensitized polygenic analysis for TGCTs in humans

Sensitized polygenic trait analysis can be used to find linkages for TGCT susceptibility genes in humans. For example, TGCTs are frequently observed in

Li-Fraumeni families (HARTLEY et al. 1989). In these families, individuals that inherit mutations in the p53 gene are susceptible to a variety of cancers including lymphomas, breast cancer, osteosarcoma, colon cancer and TGCTs

(MALKIN et al. 1990). In mice, the cancer spectrum in p53 deficient mice

depends on the genetic background, suggesting that other genes influence the

susceptibility of particular cells to tumorigenesis (HARVEY et al. 1993a). Thus, the variable occurrence of TGCTs in Li-Fraumeni families may result in part from stochastic occurrence of second-hit mutations in somatic cells as well as from the influence of the genetic background. Because sensitized polygenic

31 trait analysis appears to convert background genes with weak effects into

genetically simple, semi-Mendelian traits that can be readily mapped, it may be

fruitful to undertake a complete linkage analysis with genome scans on Li-

Fraumeni cases that show TGCTs. The same approach can also be used with

TGCT cases that carry the Xq27 variant that contributes to susceptibility to

bilateral TGCTs (RAPLEY et al. 2000). Identification of TGCT susceptibility genes has been hampered by the modest number of informative cases.

Perhaps sensitized polygenic trait analysis will increase the power of these studies and reveal novel linkages.

Chromosome substitution strains

The traditional approach to validate preliminary linkage results in experimental organisms is to obtain additional affected intercross and backcross progeny. This approach is limited because a larger sample size may not necessarily verify linkage results. A novel alternative approach is to generate a chromosome substitution strain (CSS) or a complete genome panel of CSSs (SINGER et al. 2004). These engineered strains can be used to test a

defined segment of the genome on a population of genetically identical

individuals that share a uniform genetic background. A CSS is made through a

systematic process of repeated backcrosses and selection for mice with non-

recombinant chromosomes. This process substitutes the entire chromosome of

interest in the host strain with the corresponding chromosome from the donor

32 strain. The result is a strain that is genetically identical to the host strain except

for the substituted chromosome, which is intact and homozygous on an inbred

background (Figure 1.2). There are several advantages of CSSs as a

complement to large mapping crosses: 1) genes with weaker effects are

detected in fewer mice, 2) congenic strains can be readily made from CSSs,

and 3) they are genetically defined strains that can be used systematically for

replicated genetic and functional studies. A powerful method to test the entire

genome for trait-controlling genes is to generate a panel of 21 CSSs, one for

each chromosome. A significant difference in trait values between a CSS and

its inbred host strain demonstrates that at least one susceptibility locus is

located on the substituted chromosome.

A CSS to map TGCT genes

To verify the candidate linkages that were discovered in the conventional

linkage testing cross between 129/Sv-Ter/+ and MOLF/Ei (COLLIN et al. 1996), chromosome 19 of the MOLF/Ei strain was substituted on the 129/Sv inbred

M background to generate the 129-Chr19 CSS (MATIN et al. 1999). In the

conventional linkage cross, nearly 250 backcross mice were examined and the

test score for linkage was only 9.6 (COLLIN et al. 1996), whereas in only 28 129-

M Chr19 CSS mice, the chi square score was 287.9 (MATIN et al. 1999). 71% of

the 129-Chr19M were affected, 54% of which were unilateral and 46% were

bilateral. Ten-fold fewer CSS mice were needed for linkage tests but they

detected linkage for susceptibility genes with a 30-fold greater test score. Thus,

33

• X • Strain A Strain B 10 …. …. generations

…. F1 hybrid Select mice heterosomic for chr. of interest

Intercross N15 to homozygose the chr.

….

Figure 1.2 Construction of a chromosome substitution strain (CSSs).

CSSs are strains in which a whole chromosome of a host strain is substituted with that of a donor strain. They are generated from an initial cross between two different inbred strains followed by at least 10 backcross generations from the donor strain onto the recipient strain. 22 CSS lines (19 autosomes, 2 sex and 1 mitochondrial chromosomes) can be made.

34 by controlling the genetic noise from independently segregating TGCT genes

and by testing a genetically uniform population, conclusive evidence was readily

obtained that MOLF/Ei-derived TGCT genes are located on chromosome 19.

Congenic strains

When a significant difference in trait values is mapped to a particular

chromosome, congenic strains can be derived from the CSS to localize the

susceptibility genes. Congenic strains have segments of the substituted

chromosome on the same inbred host strain background and have many of the

attributes and advantages of CSSs.

To establish the number and location of TGCT genes on chromosome 19

in the 129-Chr19M CSS, a panel of 10 single- and double-congenic strains was

M derived from 129-Chr19 (YOUNGREN et al. 2003). Results from these congenic

strains showed that at least five loci, some of which act alone and some in

combination was needed to control susceptibility to unilateral and bilateral

TGCTs (YOUNGREN et al. 2003).

Research aims

Difficulties in mapping TGCT susceptibility genes in human may be a result of the lack of multigenerational pedigree which is informative for linkage analysis. We can use the mouse model to help us gain a better understanding of the genetics of TGCT susceptibility, and to identify better diagnostic markers and targets for the treatment of TGCTs in human. TGCT is a multigenic trait in

35 mice. The low tumor frequency of the 129/Sv inbred strain makes it difficult to

perform linkage analysis to map novel susceptibility genes. I took the approach

of performing pairwise interaction tests between susceptibility loci to learn more

about the tumorigenesis pathway and explore novel interactions that control the

genetics of TGCTs.

Chapter Two describes the pairwise interaction tests between SlJ, M19, p53, Ter and Ay. Two pairs of genes were identified to have additive effects,

while three pairs of genes interact, resulting in both enhancer and suppressor

effects of TGCT susceptibility. The results of the double-mutant mice suggest

that there are loci that are involved on the same tumorigenesis pathway.

Further characterization of the interaction is needed to learn more about the

fundamentals of TGCTs, especially the ability of the Ay mutation to suppress the

development of TGCTs.

Chapter Three describes the interaction between p53 and SlJ. Double

heterozygous mutants have a 4-fold reduction in TGCT development than

expected additive tumor frequency if the pair of susceptibility genes did not

interact. Protein and RNA expression levels of p53 were determined in an

attempt to test how this pair of susceptibility genes interacts to suppress tumor formation. However, in the homozygous cross, mice that are homozygous for p53 and heterozygous for SlJ have the expected additive rate of tumor

frequency. They did not have the ability to suppress the development of

36 TGCTs. The interaction to suppress TGCT susceptibility seems to be dosage

dependent. These results provide important clues about the tumorigenesis

pathway and the role of p53 in the development of TGCTs.

Chapter Four investigates whether the human ortholog of the mouse Ter gene influences the development of bilateral tumors in patients. Several genes that increase TGCT incidence in mice are conserved in the . In the lab, we have cloned the mouse Ter mutation that significantly increases

TGCT susceptibility. I have sequenced all the exons of the human ortholog of

Ter and there were no sequence differences in the coding region of the gene

between affected and unaffected individuals.

37

CHAPTER 2

PAIRWISE INTERACTION TEST BETWEEN TGCT SUSCEPTIBILITY LOCI IN MICE

38 Authors: Man-Yee Josephine Lam, Kirsten K. Youngren and Joseph H. Nadeau

References: 1. Genetics 166:925-933. 2004.

2. Lam, MY, Youngren, KK, Kawasoe, J and Nadeau, JH. Genetics Interaction of Ter. Manuscript in preparation.

ABSTRACT

Susceptibility to spontaneous testicular germ cell tumors (TGCTs) shows unusual genetic complexity including multiple genes, gene interactions and low penetrance. Despite remarkable progress in the genetics analysis of susceptibility to many cancers, TGCT susceptibility genes have not yet been identified in human. An international effort revealed many weak linkages and one significant linkage on chromosome X involving predisposition to bilateral tumors. Various mutations that are inherited as Mendelian traits in laboratory mice affect susceptibility to spontaneous TGCTs on the 129/Sv inbred genetic background. These genes, alone or in combination, provide important clues to the control of susceptibility and resistance to tumorigenesis. We undertook a systematic survey to test pairwise interactions of most TGCT susceptibility loci by comparing the frequency of TGCTs in double-mutant mice to identify combinations that show evidence of enhancer or suppressor effects. Both enhancer and suppressor interaction were found. The lower than expected

TGCT frequency in 129-Chr19M consomic mice that were heterozygous for the

Ay mutation suggest either that these loci complement each other to restore

39 normal functionality in TGCT stem cells or that together these genes activate

mechanisms that suppress incipient TGCTs. By contrast, the higher than

expected TGCT frequency was observed between pairwise interaction crosses

of M19, SlJ and Ter suggest that these mutants exacerbate each others effects.

Together these results provide clues to the genetic and molecular basis for

susceptibility to TGCTs in mice and perhaps in humans.

INTRODUCTION

Testicular germ cell tumors (TGCTs) are the most common cancer

affecting young men. The incidence of TGCTs has doubled within the last 50

years, predominately in Eastern Europe, indicating that environmental factors

affect susceptibility (Buetow 1995). Genetics also contribute to TGCT

susceptibility, with an increased risk of 8- to 10-fold among brothers and 4-fold

among sons of affected individuals (FORMAN et al. 1992; LINDELOF and EKLUND

2001). Despite its prevalence, little is known about the genetic control of TGCT

susceptibility. To date, TGCT susceptibility genes have not yet been identified

in humans. Linkage studies have been difficult in part because of the limited

number of multigenerational pedigrees with sufficient numbers of affected

individuals because treatment of testicular cancer can lead to sterility and the disease is sporadic.

40 A mouse model of pediatric germ cell tumor in humans

The 129/Sv inbred strain of laboratory mice is an important model for

unraveling the complexity of the genetic control of susceptibility and resistance

to TGCTs (Stevens and Hummel 1957). Critical information about TGCTs in

the 129/Sv inbred strain has been established. TGCTs in mice originate from

primordial germ cells (PGCs) between embryonic day 11 (E11) and E12.5

(Stevens 1966). PGCs are transformed by unknown mechanisms to give rise to

embryonal carcinoma cells.

TGCT susceptibility genes

Several mutations that are inherited as Mendelian traits affect TGCT

susceptibility in 129/Sv mice: SlJ, Ter, M19, p53 and Ay. SlJ is a 650kb deletion

J within the Sl locus that includes the Mgf gene (BEDELL et al. 1996). Sl /+

mutants have a tumor frequency double that of wild-type animals (STEVENS

1967a). Ter is a spontaneous mutation that causes germ cell deficiency and increase susceptibility to TGCTs (Stevens 1973). A genome scan suggested that at least one gene on Chr19 of the MOLF/Ei strain dramatically increases

M susceptibility to TGCTs (COLLIN et al. 1996). A 129-Chr19 chromosome substitution strain (CSS) was generated to place a MOLF-derived Chr19 on the

129/Sv inbred background (MATIN et al. 1999) and it significantly increased

TGCT susceptibility. Inactivation of p53 on different inbred genetic background results in a spectrum of tumors and an increased incidence of TGCTs is

41 observed on the 129/Sv strain (DONEHOWER et al. 1995; HARVEY et al. 1993a).

Ay is ~150kb deletion of the coding region of the Raly gene that causes the

Agouti gene to be transcriptionally regulated by the Raly promoter and results in the ubiquitous expression of agouti (MICHAUD et al. 1994). All mutations must be congenic on the 129/Sv inbred genetic background to exert their influence on TGCT frequency, demonstrating that these mutant genes act together with

129/Sv-derived genes to control susceptibility.

Rationale

Testicular cancer is a complex trait that may requires as many as six independently segregating genes with additive and recessive effects (STEVENS

1967a). The low penetrance and multigenic basis of spontaneous TGCTs in the 129/Sv inbred strain make it difficult to dissect the genetic control of susceptibility to TGCTs. We used single gene mutations that are inherited as

Mendelian traits to gain clues to the nature of the genes and pathways that affect TGCT susceptibility. The various mutations on the 129 genetic background increase or decrease penetrance perhaps by activating novel pathways involved in tumorigenesis. The interaction of these Mendelian traits provides a unique opportunity to study TGCT development. In other model systems, interactions between different pairs of genes have provided unique insights into developmental pathways and cancer susceptibility (MONTES DE OCA

LUNA et al. 1995; MOORE et al. 1990; PARANT et al. 2001). We therefore tested

interactions between pairs of TGCT susceptibility genes to evaluate their effect

42 on tumor frequencies, laterality of the tumors, and parental effects to reveal

more about the molecular and developmental pathways leading to

tumorigenesis.

MATERIALS AND METHODS

Mice

129S1/SvImJ (JR002448, previously known as 129/SvJ and 129S3/SvImJ),

129S1/Sv-p+Tyr+KitlSl-J/+ (JR000090), B6.Cg-AY (JR000021), and 129-

Trp53tm1Tyj (JR002080) were obtained from the Jackson Laboratory (Bar

Harbor, ME, USA). The nomenclature for 129 substrains has been revised by the Jackson Laboratory

(www.informatics.jax.org/mgihome/nomen/strain_129.shtml) and the

recommended designations were used in this paper. The 129-Chr19M chromosome substitution strain (CSS) (N15F2+) was described previously

(MATIN et al. 1999) and was obtained from our research colony. Mice were

maintained in the CWRU Animal Resource Center on a 12:12hr light: dark cycle

and fed Lab Diet 5010.

Construction of a 129-Ay congenic strain

To transfer the Ay mutation onto the 129/Sv background, C57BL/6J-Ay mice were crossed to 129S1/SvImJ for 11 generations. Genes in the 129 background are essential for tumorigenesis and therefore to perform the

43 interaction tests, Ay must be transferred to and tested on the 129/Sv

background.

Construction of a 129-Ay / +, MOLF-Chr 19 congenic consomic mice

129-Ay mice were crossed to the 129-Chr 19M CSS. Heterosomic 129-Chr 19M mice carrying Ay were then backcrossed to 129-Chr 19M. Ay progeny from these backcrosses were typed for simple sequence length polymorphism

(SSLP) along the length of Chr 19. Ay progeny that were homosomic for MOLF-

Chr19 were selected to establish the test colony and wild-type littermates were used as controls.

Genotyping

DNA for PCR genotyping was obtained from samples of tail tissue. Each tail sample was digested in 89 ul of water, 10 ul 10X PCR buffer and 1 ul

Proteinase K (10mg/ml). The reaction was incubated at 100°C for an hour.

p53

In this study, we purchased from the Jackson Laboratory the p53/+ mutant strain that was developed by Tyler Jacks et al. (JACKS et al. 1994). This strain has a different p53 mutant allele than that reported by Donehower et al.

(DONEHOWER et al. 1992).

A three primer PCR assay was used to distinguish wild-type from heterozygous p53 animals. The three primers are X7 5’TATACTCAGAGCCGCCT 3’, Neo19

44 5’CTATCAGGACATAGCGTTGG 3’, and X6.5 5’

ACAGCGTGGTGGTACCTTAT 3’. Primers X7 and Neo19 amplify a 600bp

fragment identifying the Neo insert. Primers X7 and X6.5 amplify a 400bp

fragment from the untargeted p53 allele. PCR amplification was carried out in a

96-well block MJ Research PTC-200 thermal cycler. The reagents were 0.15ul

(0.75U) Taq polymerase (Invitrogen), 2.5ul 10X PCR buffer (magnesium free),

0.3ul 10mM dNTPs, 1ul 25mM magnesium chloride, 0.2ul of each primer

(0.1uM), 1ul DNA (25ng), and 19.35ul dH2O in a final volume of 25ul. PCR

conditions were as follows: initial denaturation step for 94°C for 2 min followed by 35 cycles of 94°C for 1 min, 60°C for 1 min, 72°C for 1 min, final extension of

72°C for 5 min and then 4°C for 15 min. PCR products were resolved on a 2% agarose gel and visualized with ethidium bromide.

SlJ

The breakpoints of the SlJ deletion are not known and as a result a PCR

genotyping assay is not available for the SlJ mutation. SlJ /+ mutant heterozygotes have a light coat color in the belly and the tips of the tail and digits are pink.

Ter

Genomic DNA was amplified with standard PCR condition using the primers

4.13a-F (5'-gtagttcaggaactccacttgtg-3) and 4.13a-R (5'-gcctaatgatgaccttcagtgg-

3') to flank the mutation in exon 2 to give a 431bp PCR product. The PCR

45 product was digested with Dde1 enzyme and resolved on a 7% polyacrylamide gel. Dde1 digestion produces fragments of the following sizes:180bp, 154bp and 97 bp products for +/+ genotype and 180 bp, 123 bp, 97 bp and 31 bp products for Ter/Ter genotype.

Ay

Ay progeny were identified based on their yellow coat color.

RESULTS

TGCT frequencies in single-mutant mice

If mutant genes have functionally independent and additive effects on tumorigenesis, the expected tumor frequency in double-mutant mice should be the sum of the frequencies in each of the single-mutant mice. Rather than simply using TGCT frequencies reported in the literature as controls, we measured the frequencies of SlJ/+ (129/Sv x SlJ/+), M19/+ (129/Sv x M19/M19) and Ter/+ (129/Sv x Ter/+) single-mutants in our colony. The control crosses were generated in both parental directions to test whether gender of the parent affected TGCT susceptibility among male progeny. Within the various pairwise interaction crosses, single-mutant littermates were also generated. To increase the confidence levels of the tumor frequency of the single-mutants used to calculate the expected additive tumor frequencies of the double-mutant mice, we tested heterogeneity among the tumor frequencies of the single-mutant

46 Table 2.1: Heterogeneity among the tumor frequency of all single-mutants in control and interaction crosses

Raw data of each genotype in all control and interaction crosses. A contingency test was used to test if the number of affected mice of each genotype of each cross can be pooled to generate the baseline for expected additive rate. A. SlJ/+ heterozygous mutants of all crosses Cross N Unilateral Bilateral Affected Unaffected 129/Sv x SlJ/+ 170 11% (19) 1% (2) 12% (21) 88% (149) Ter/+ x SlJ/+ 100 13% (13) 2% (2) 15% (15) 85% (85) p53/+ x SlJ/+ 87 11% (10) 0 11% (10) 89% (77) p53/+ SlJ/+ x 129/Sv 88 8% (7) 1% (1) 9% (8) 91% (80) p53/+ SlJ/+ x p53/+ 70 9% (6) 0 9% (6) 91% (64) Summary 515 11% (55) 1% (5) 12% (60) 88% (455) X2=1.52, df=4, P=0.82, pooled tumor frequency of SlJ/+ = 12%

B. M19/+ heterozygous mutants of all crosses Cross N Unilateral Bilateral Affected Unaffected 129/Sv x M19/M19 300 29% (86) 4% (13) 33% (99) 67% (201) M19/M19 x SlJ/+ 225 26% (58) 6% (14) 32% (72) 68% (153) Ter/+ x M19/M19 95 23% (22) 6% (6) 29% (28) 71% (67) Summary 620 27% (166) 5% (33) 32% (199) 68% (421) X2=0.41, df=2, P=0.81, pooled tumor frequency of M19/+ = 32%

C. p53/+ heterozygous mutants of all crosses Cross N Unilateral Bilateral Affected Unaffected Ter/+ x p53/+ 91 15% (14) 0 15% (14) 85% (77) p53/+ x SlJ/+ 75 7% (5) 0 7% (5) 93% (70) p53/+ SlJ/+ x p53/+ 125 9% (11) 1% (1) 10% (12) 90% (113) p53/+ SlJ/+ x 129/Sv 82 18% (15) 1% (1) 18% (15) 82% (67) Summary 373 12% (45) 1% (2) 12% (46) 88% (327) X2=6.57, df=3, P=0.087, pooled tumor frequency of p53/+ = 12%

D. Ter/+ heterozygous mutants of all crosses Cross N Uni Bil Affected Unaffected Ter/+ x Ter/+ 40 15%(6) 0 15%(6) 85% (34) 129/Sv x Ter/+ 77 19% (15) 0 19% (15) 81% (62) Summary of 117 18% (21) 0 18% (21) 82%(96) control crosses Ter/+ x SlJ/+ 107 30% (32) 7% (7) 39% (39) 64% (68) Ter/+ x p53/+ 105 29% (30) 7% (7) 35% (37) 65% (68) Summary of 212 29% (62) 7% (14) 36% (76) 64% (136) interaction crosses X2=11.91, df=3, P=0.0077, Tumor frequencies can not be pooled

47 Table 2.1 (continued): Heterogeneity among the tumor frequency of all single- mutants in control and interaction crosses

E. +/+ (wild-type) of all crosses Cross N Unilateral Bilateral Affected Unaffected 129/Sv x SlJ/+ 191 8% (15) 0 8% (15) 92% (176) Ter/+ x 129/Sv 107 8% (9) 0 8% (9) 92% (98) Ay/+ x 129/Sv 50 12% (6) 0 12% (6) 88% (44) Ter/+ x SlJ/+ 94 5% (5) 0 5% (5) 95% (89) Ter/+ x p53/+ 113 8% (9) 0 8% (9) 92% (104) p53/+ x SlJ/+ 93 2% (2) 0 2% (2) 98% (91) p53/+ SlJ/+ x p53/+ 97 7% (7) 1% (1) 8% (8) 92% (89) p53/+ SlJ/+ x 129/Sv 85 7% (6) 0 7% (6) 93% (79) Summary 830 7% (59) 0 7% (60) 93% (770) X2=6.36, df=7, P=0.50, pooled tumor frequency of +/+ = 7%

48 littermates of the various interaction crosses and control crosses by a

contingency test (Table 2.1). The tumor frequencies of the single-mutant

of the various pairwise interaction crosses and control crosses were not

significantly different except for the Ter/+ heterozygotes, therefore the

frequencies of all the single-mutants were pooled except for the Ter/+ mutants.

SlJ

SlJ/+ heterozygotes typically have a TGCT frequency of 12% – 14%, based on surveys of exceptionally large numbers of mice (STEVENS 1967a).

The tumor frequencies of the SlJ/+ heterozygotes of the five crosses were not statistically different (X2=1.52, P>0.5), the pooled tumor frequency is 12%

(tumor frequency range: 9-15%) (Table 2.1A).

129-Chr19M (M19)

The tumor frequency of 129-Chr19M homosomic mice with two copies of

MOLF-Chr19 was reported to be 82% and heterosomic males with only one copy of MOLF-Chr19 was reported to be 24% (MATIN et al. 1999). The tumor

frequencies of the M19/+ heterozygotes of the three crosses were not

statistically different (X2=0.41, P>0.5), the pooled tumor frequency of

heterosomic mice is 32% (tumor frequency range: 29-33%) (Table 2.1B).

49 p53

p53/+ heterozygotes were reported to rarely develop TGCT (DONEHOWER et al. 1995). A control cross to measure the tumor frequency of p53/+ mutants was not generated, but the tumor frequencies of the p53/+ littermates in the four pairwise interaction crosses were not statistically different (X2=0.66, P>0.05)

and the pooled tumor frequency of p53/+ heterozygotes is 12% (tumor

frequency range: 7-18%) (Table 2.1C).

Ter

The tumor frequencies of Ter/+ heterozygotes of the control crosses and

pairwise interaction crosses were statistically different (X2=11.91, P<0.05)

(Table 2.1D), therefore the expected additive tumor frequency of the double-

heterozygous mutants was calculated by using the tumor frequency of the Ter/+

littermates of the specific interaction cross.

Ter/+ heterozygotes were reported to have a tumor frequency of 17%

and homozygous Ter/Ter mutant have a tumor frequency of 94% (Noguchi and

Noguchi 1985). A small survey of our colony was performed to measure the

affected rate of the Ter/+ mutants in our colony. In the control cross between

Ter/+ mutants, 90% of the Ter/Ter homozygous mutants developed TGCTs

(N=20) and 14% of the Ter/+ heterozygous mutants developed TGCTs (N=40).

We confirmed that the affected rate of the Ter mutants in our colony is similar to

the published rate (X2=0.17 and X2=0.53, respectively; P>0.5). The tumor

50 frequencies of Ter/+ heterozygotes in the two pairwise interaction crosses were

higher than the published rate of 17% (Table 2.1D).

For the interaction crosses, all the susceptibility genes are on the

129/SvIMJ except for the Ter mutation which is on the 129T2/SvEmsJ

substrain. We wanted to test whether interactions between the two substrains

affected TGCT susceptibility. Therefore, we generated a control cross between

the Ter subline and wild-type 129/SvIMJ. The control cross was generated in

both directions to test for parental effects. There was no evidence for parental

effects (X2=1.02, P<0.1). The heterozygous mutants had a TGCT frequency of

19% (N=77) and the wild-type mice had a tumor frequency of 8% (N=107)

(Table 2.1D). We also generated a control cross between the wild-type

129T2/SvEmsJ and 129/SvIMJ to test for background effects. The wild-type progeny of this control cross had a tumor frequency of 4% (N=114). Therefore, the interaction of the two different sublines did not affect TGCT frequency.

The tumor frequency of the Ter/+ heterozygotes between the 4 crosses can be partioned into two groups. The pooled tumor frequency of the Ter/+ heterozygous mutants in the two control crosses is 18% (tumor frequency range: 15-19%) and the pooled tumor frequency of the Ter/+ heterozygotes in the two interaction crosses is 36% (tumor frequency range: 35-39%) (Table

2.1D)

51 Ay

The tumor frequency of 129/Sv mice carrying the Ay mutation is ~1%,

whereas their wild-type littermates have a frequency of ~8% (STEVENS 1967a).

To verify Stevens’ discovery, we made a 129/Sv- Ay congenic strain and then

transferred the Ay mutation onto the 129-Chr19M CSS background. This CSS strain has a tumor frequency of ~80% and therefore serves as a statistically powerful way to quickly determine whether Ay suppresses tumorigenesis,

detecting a 10-fold reduction from the 5% TGCT frequency in 129/Sv would

require a survey of significantly more mice than is required to assess TGCT

susceptibility in 129-Chr19M Ay/+ consomic congenic mice. In parallel, a small

survey was conducted with 129-Ay/+ mice. We found a tumor frequency of 4%

in the Ay/+ heterozygotes (N=40) and 12% in the wild-type littermates (N=40).

In comparison to the wild-type littermates, 129-Ay/+ mice have a ~3-fold but not

significant reduction (X2 = 2.3; P<0.5). We continue to survey these mice to obtain a better estimate of TGCT frequencies.

Wild-type littermates

The tumor frequency of the wild-type 129/Sv inbred strain was reported to be 1-10% (STEVENS and HUMMEL 1957). The tumor frequencies of the wild-

type littermates of the eight crosses were not statistically different (X2 = 6.36;

P=0.5). The pooled tumor frequency of the wild-type 129/Sv mice is 7% (tumor frequency range: 2-12%) (Table 2.1E).

52 Double-mutant interaction tests

Double-mutant mice were used to test for interactions between TGCT

susceptibility genes. We evaluated tumor frequency, tumor laterality, and

parental effects in the double mutants. If there was an interaction between the pairs of susceptibility genes, the tumor frequency of the double-mutant would vary significantly from the tumor frequency contributed by each mutation alone.

In contrast, if the tumor frequency of the double mutant did not vary significantly from the tumor frequency contributed by each mutant, then the pair of susceptibility genes has an additive effect and they are not on the same pathway. The expected additive tumor frequency of each double-mutant was calculated by adding the pooled tumor frequencies of the single-mutants minus the frequency of the 129/Sv baseline (wild-type genotype) and comparing with the observed tumor frequency of the double-mutant.

Parental effects

All the crosses were generated in both parental directions to test for parental effects. Among all the crosses, there were no significant differences between the tumor frequencies of the reciprocal parental directions and the data between the two parental crosses were pooled (Table 2.2). Only in the cross between the C2 congenic strain and SlJ/+ heterozygote was a modest

parental effect detected (X2 = 4.0; P<0.025). The chi-square score was

borderline significant, therefore it may be a false positive. In addition, we did

53

Table 2.2: Parental effects of various interaction crosses

No parental effects were found in the reciprocal parental directions of the various interaction crosses

Reciprocal crosses Test Score M19/M19 x SlJ/+ X2=0.83, P>0.1 M19/M19 x M19/M19 SlJ/+ X2=3.65, P>0.05 Ter/+ x SlJ/+ X2=1.23, P>0.1 Ter/+ x p53/+ X2=2.0 P>0.1 M19/M19 x Ter/+ X2=0.06 P>0.5

54 not correct for multiple testing and it further reinforce that the parental effect is

probably not significant.

SlJ and M19 double heterozygotes:

The double heterozygous mutants of the interaction test between SlJ and

M19 revealed a higher than expected TGCT susceptibility (Table 2.3A). If M19

and SlJ acted additively, the expected tumor frequency in the double

heterozygous mutants would be 37% (= 32% [M19/+] + 12% [Sl /+] – 7% [+/+]).

The observed TGCT frequency of 57% in the 215 double heterozygous mutants was significantly higher than expected (X2 = 35.05; P<0.005). The enhanced frequency in TGCT susceptibility resulted largely from an increase in the frequency of bilateral tumors, which increased from 6% in M19/+ +/+ mice to

19% in M19/+ SlJ/+ mice (X2 = 56.0; P<0.005). By contrast, the frequencies of

unilateral TGCTs in these two groups of mice were similar, suggesting that M19

and SlJ interacted to increase the frequency of bilateral tumors only.

SlJ and two M19 congenic strains:

Two congenic strains, A2 and C2, (YOUNGREN et al. 2003) were used in heterozygous state to localize the region on MOLF-Chr19 that interacted with

SlJ to increase susceptibility to bilateral tumors. A2 congenic mice have the proximal and the middle segment of MOLF-Chr19, whereas C2 congenic mice

55 Table 2.3: Interaction tests between M19 and SlJ A. Interaction test between M19 and SlJ in double heterozgyous mutants B. A2 congenic cross to localize the region on MOLF-Chr19 that interacts with SlJ C. C2 congenic cross to localize the region on MOLF-Chr19 that interacts with SlJ D. M19 homosomics and SlJ congenic heterozygotes

A. Tumor frequency of M19/+ SlJ/+ and wild-type controls

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals M19/+ SlJ/+ 215 41% (90) 19% (40) 57% (122) M19/+ +/+ 225 26% (58) 6% (14) 32% (72)

B. Tumor frequency of A2/+ SlJ/+ and wild-type controls

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals A2/+ SlJ/+ 130 23% (30) 7% (9) 30% (39) A2/+ +/+ 163 15% (24) 1% (1) 15% (25)

C-1. Interaction cross: C2/C2 x SlJ/+

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals C2/+ SlJ/+ 97 8% (8) 2% (2) 10% (10) C2/+ +/+ 84 13% (11) 0% 13% (11)

C-2. Interaction cross: SlJ/+ x C2/C2

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals C2/+ SlJ/+ 89 18% (16) 2% (2) 20% (18) C2/+ +/+ 71 8% (6) 0% 8% (6)

D. Tumor frequency of M19/M19 SlJ/+ and wild-type controls

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals M19/M19 SlJ/+ 39 33% (13) 59% (23) 92% (36) M19/M19 +/+ 50 36% (18) 46% (23) 82% (41)

56 have the distal and the overlapping middle segment (with A2) of MOLF-Chr19

(Figure 2.1). If the A2/+ SlJ/+ mutants have a tumor frequency that is similar to

the observed tumor frequency of the M19/+ SlJ/+ mutants, it can be concluded

that SlJ interacts with the proximal end of the MOLF-Chr19. Whereas, if the

C2/+ SlJ/+ mutants have a tumor frequency that is similar to the M19/+ SlJ/+

double heterozygous mutants then SlJ is interacting with the distal end of

MOLF-Chr19. However, if both A2/+ SlJ/+ and C2/+ SlJ/+ mutants have tumor

frequencies that are similar to the double heterozygotes, then SlJ is interacting

with the overlapping region of the two congenic strains on MOLF-Chr19 to

increase susceptibility of TGCTs.

Parental effects between the reciprocal A2/+ congenic crosses were not

significant (X2 = 0.04; P>0.5). The pooled tumor frequency in the double heterozygous mutants A2/+ SlJ/+ was 30% (Table 2.3B), which is significantly lower than the tumor frequency observed in the double heterozygous mutant

(M19/+ SlJ/+ = 57%) in the cross between SlJ/+ and 129-Chr19M (with the entire length of MOLF-Chr19) (X2 = 12.8; P<0.001). In addition, the double heterozygous mutants of the interaction cross between SlJ/+ and 129-Chr19M

has a high frequency of bilateral tumors (19%), while A2/+ SlJ/+ had a

significantly lower frequency of bilateral tumors (7%) (X2 = 7.58; P>0.5).

57

Figure 2.1 The location of the congenic segment A2 and C2.

The two congenic strains are homozygous. M indicates chromosome segments that were derived from 129-Chr19M and I segments were derived from 129/Sv. t 32 28 110 60 111 40 73 86 30 132 46 81 64 13 134 133 5 118 19 20 17 71

MM MM MM MM MM MM MM MM MM MM MM MM MM MM MM II II II II II II II A2

II II II II II II II II II II MM MM MM MM MM MM MM MM MM MM MM MM C2

58

In contrast, a modest parental effect was detected in the interaction

cross between C2/+ and SlJ/+ mice (X2 = 4.0; P<0.025). Among 97 double heterozygous mutants of parental crosses between C2 females x SlJ/+ males were examined and 10% of the double heterozygous mutants developed

TGCTs, of which only 2% were bilateral tumors (Table 2.3C-1). In the reciprocal cross of SlJ/+ females x C2 males, 20% of 89 double heterozygous mutants developed TGCTs, of which only 3% were bilateral tumors (Table

2.3C-2). Again, the C2/+ SlJ/+ mutants had a significantly lower tumor

frequency than the M19/+ SlJ/+ mutants. Importantly, double heterozygous

mutants from both congenic crosses rarely developed bilateral tumors, whereas

a significant increase in bilateral tumor incidence was observed in the M19/+

SlJ/+ mutants. It appears that SlJ did not interact separately with either the

proximal or the distal segments of the MOLF-Chr19 to increase the incidence of

bilateral tumors.

M19 homosomics and SlJ congenic heterozygotes:

Homosomic-congenic mutants were generated by transferring SlJ to the

129-Chr19M strain. If M19 and SlJ acted additively, the expected tumor frequency in the M19/M19 SlJ/+ mice would be 87% (= 82% [M19/M19] + 12%

[SlJ/+] - 7% [+/+]). The TGCT frequency in the homosomic congenic-

heterozygous mutants of the reciprocal crosses was similar (X2 = 3.65; P>0.05).

The homosomic-congenic mutant had the expected high tumor frequency with

59 an affected rate in the pooled data of 87% (Table 2.3D), which was not

significantly different from the expected additive frequency of 94% (X2 = 0.28;

P>0.5). The M19/M19 +/+ littermates had a TGCT frequency of 80%, which was not significantly different than the published tumor frequency (MATIN et al.

1999).

Ter and SlJ double heterozygotes:

There was no evidence for parental effects between the reciprocal crosses. The tumor frequency of the Ter/+ littermates was 36% (Table 2.4), which is higher than the published frequency, therefore the expected additive tumor frequency of the double heterozygotes is the sum of the tumor frequency of the Ter/+ littermates of the interaction cross and the pooled tumor frequency of the SlJ/+ mutant (Table 2.1A) minus the pooled tumor frequency of the wild-

type background. The pooled tumor frequency of the double heterozygotes

Ter/+ SlJ/+ was 64%, which is significantly higher than the expected additive frequency of 41% (=36% [Ter/+] + 12% [SlJ /+ ] – 7% [+/+]) (X2 = 21.4; P>0.5).

Therefore, Ter and SlJ seems to interact and may act on the same tumorigenesis pathway. The incidence of bilateral tumors of the double heterozygous mutant is 25% which is higher than expected because both Ter/+ and SlJ/+ single-mutants rarely develop bilateral tumors, therefore Ter and SlJ may interact to increase the incidence of bilateral tumors.

60

Table 2.4: Interaction test between Ter and SlJ in double heterozygous mutants

Overall tumor frequency of Ter/+ SlJ/+ and littermates Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals Ter/+ SlJ/+ 102 38% (39) 25% (26) 64% (65) Ter/+ +/+ 107 30% (32) 7% (7) 36% (39) +/+ SlJ/+ 100 13% (13) 2% (2) 15% (15) +/+ +/+ 94 5% (5) 0% 5% (5)

61 . The wild-type and SlJ/+ littermates had the appropriate tumor

frequencies, but the Ter/+ littermate in this interaction cross had a higher tumor

frequency than 17% which was previously published (Noguchi and Noguchi

1985) and also within our colony (Ter/+ x Ter/+) (Table 2.1D). The high tumor

frequency of the Ter/+ littermates does not appear to be an interaction between

the two different 129 substrains because Ter/+ heterozygotes of the control

cross between 129/Sv and Ter/+ has a tumor frequency which is similar to the

published frequency. Therefore, we are uncertain as to why the Ter/+

littermates in this interaction cross had a higher tumor frequency than expected.

Ter and p53 double heterozygotes:

There were no parental effects observed between the two reciprocal

crosses. Similar to the interaction cross between Ter/+ and SlJ/+ mutants, the

Ter/+ littermates in this interaction cross had a higher tumor frequency of 35%

(Table 2.5) which is significantly higher than the published tumor frequency of

17%. As a result, the tumor frequency of the Ter/+ littermates of this interaction

cross was used to calculate the expected additive tumor frequency of the

double heterozygotes. The expected additive tumor frequency of the double

heterozygous mutants is 40% (=35% [Ter/+] + 12% [p53/+] – 7% [+/+]) and the pooled tumor frequency of the double heterozygotes is 49%, which is not

62

Table 2.5: Interaction test between Ter and p53

Overall tumor frequency of Ter/+ p53/+ and littermates Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals Ter/+ p53/+ 75 31% (23) 19% (14) 49% (37) Ter/+ +/+ 105 29% (30) 7% (7) 35% (37) +/+ p53/+ 81 5% (4) 1% (1) 6% (5) +/+ +/+ 113 8% (9) 0% 8% (9)

63 significantly different from the expected additive rate (X2 = 2.35; P>0.5) (Table

2.5). The additive effect between this pair of susceptibility genes suggests that

Ter does not interact with p53 on the same pathway.

The p53 mutation is on the 129/SvIMJ background and the Ter mutation is on the 129T2/SvEmsJ background. From our controls crosses, we already determined that there is no interaction between the two different substrains of

129. The Ter/+ mutants with the two mixed sublines had a tumor frequency of

19% which is similar to the published affected rate (Table 2.1D). It is puzzling as to why the Ter/+ littermates in this interaction cross had a significantly higher

tumor frequency than expected (X2=13.5; P<0.001). In addition, it is interesting

that the tumor frequency of the Ter/+ littermates was similar in both interaction

crosses between Ter/+ and SlJ/+ as well as Ter/+ and p53/+.

Ter and M19 double heterozygotes:

Parental effects were not observed between the two reciprocal crosses.

The interaction cross was generated by crossing M19/M19 with Ter/+ and

resulted in double heterozygous mutants and M19/+ littermates. There were no

Ter/+ littermates to help calculate the expected additive tumor frequency of

M19/+ Ter/+ double heterozygotes. The Ter/+ littermates of the interaction

crosses between Ter/+ and SlJ/+ and Ter/+ and p53/+ resulted in higher TGCT

frequency than reported in published literature, therefore we decided to

64

Table 2.6: Interaction test between M19 and Ter A. Interaction test between M19 and Ter in double heterozygous mutants B. A2 congenic cross to localize the region on MOLF-Chr19 that interacts with SlJ C. C2 congenic cross to localize the region on MOLF-Chr19 that interacts with SlJ

A. Tumor frequency of M19/+ Ter/+ and littermate control

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals M19/+ +/+ 95 23% (22) 6% (6) 29% (28) M19/+ Ter/+ 95 32% (30) 44% (42) 76% (72)

B. Tumor frequency of A2/+ Ter/+ and wild-type control

Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals A2/+ +/+ 107 9% (10) 0% 9% (10) A2/+ Ter/+ 94 45% (42) 16% (15) 60% (57)

C. Tumor frequency of C2/+ Ter/+ and wild-type control Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals C2/+ +/+ 93 11% (10) 1% (1) 11% (10) C2/+ Ter/+ 107 32% (34) 8% (1) 40% (42)

65 calculate the expected additive tumor frequency of the double heterozygous

M19/+ Ter/+ for both scenario, using the published tumor frequency of 17% of the Ter/+ heterozygotes and the high tumor frequency of 35% pooled between the two interaction crosses.

The expected additive tumor frequency of the double heterozygous using the published tumor frequency of Ter/+ is 42% (=17% [Ter/+] + 32% [M19/+] -

7% [+/+]). The observed tumor frequency of 95 double heterozygotes was 76%

(Table 2.6A) which is significantly higher than the expected additive rate of 42%

(X2=27.5; P<0.005). Alternatively, the expected additive tumor frequency of the

double heterozygous mutant is 60% (=35% [Ter/+] + 32% [M19/+] - 7% [+/+]) if we assume that the Ter/+ genotype contributed a high tumor frequency as observed in the previous two interactions between Ter and SlJ, and p53 and

Ter. The observed tumor frequency of the double heterozygotes is also significantly higher than the additive rate of 60% (X2=4.36; P<0.05) (Table

2.6A). It appears that the double heterozygous mutants have a higher tumor

frequency than expected regardless if we used the published tumor frequency

of the Ter/+ heterozygotes or the high tumor frequency of the Ter/+ littermates

in the interaction crosses to calculate the expected additive tumor frequency.

Of the total affected Ter/+ M19/+ double heterozygotes, 32% developed unilateral tumors and 44% developed bilateral tumors. Bilateral tumors were rarely observed in the Ter/+ and M19/+ single-mutants, suggesting that Ter may interact with M19 to increase the susceptibility of TGCT and the incidence of bilateral tumors.

66 Ter and two M19 congenics:

To localize the region on MOLF-Chr19 that interacts with Ter, we

crossed Ter/+ mutants with two congenic strains (A2 and C2) generated in the

lab (YOUNGREN et al. 2003) (Figure 2.1). Parental effects were not observed in

the reciprocal crosses of either (A2 or C2) congenic crosses. 60% of 94 A2/+

Ter/+ double heterozygote developed tumors, of the tumors, 45% were

unilateral tumors and only 16% were bilateral tumors (Table 2.6B). The

affected rate of A2/+ Ter/+ is not significantly different than the tumor frequency

of the M19/+ Ter/+ (X2=3.37 ; P>0.05), but the susceptibility of bilateral tumors

was modest compared to M19/+ Ter/+.

In the C2 congenic cross, 40% of 107 C2/+ Ter/+ mutants developed

TGCTs which is significantly different from the affected rate of M19/+ Ter/+

(X2=17.1; P<0.001) (Table 2.6C). It appears that Ter may interact with A2 which is the proximal end of MOLF-Chr19 to increase the susceptibility of

TGCTs, however it does not increase the incidence of bilateral tumors to the high rate observed in Ter/+ M19/+.

Ay and M19:

129-Chr19M mice carrying the Ay mutation have a significant reduction in

TGCT frequency compared 129-Chr19M wild-type littermates (Table 2.7). No evidence for significant parental effects was observed among progeny from reciprocal crosses. 104 male progeny of 129-Chr19M carrying the Ay mutation were examined and 40% developed TGCTs, in comparison to 65% of 105 129-

67

Table 2.7: Interaction test between M19 and Ay

Tumor frequency of M19/M19 Ay/+ and littermates Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals M19/M19 +/+ 105 40% (42) 26% (26) 65% (68) M19/M19 Ay/+ 104 32% (33) 9% (9) 40% (42)

68 Chr19M without the Ay mutation that developed TGCTs (X2 = 9.6; P< 0.005).

These results showed that Ay not only suppresses TGCT frequency in the

129/Sv inbred strain (STEVENS 1967a) but also reduces tumor frequency by

50% on the 129-Chr19M CSS.

Unilateral versus bilateral tumors

Among the various interaction crosses that we observed interactions

between the pairs of susceptibility genes to significantly increase TGCT

frequency, there also seems to be an increased incidence of bilateral tumors in

the double heterozygous mutants in comparison to the sum of the bilateral

tumors of the single-mutants. We tested if the interaction between the pairs of

susceptibility genes affects specifically the susceptibility of bilateral tumors in

the double-heterozygous mutants. We tested the relationship between

unilateral and bilateral tumors of all the double heterozygous mutants to

evaluate if different combinations between pairs of susceptibility genes interact

to cause unilateral or bilateral tumors primarily, or whether the pairs of

susceptibility genes interact to increase susceptibility to TGCT and the bilateral

tumor is result of a unilateral tumor on the left and right testes.

Using the methods of Youngren et al. (YOUNGREN et al. 2003) in

Appendix 1, we first calculated the frequency of TGCTs in the left and right testes separately, regardless of presence of a TGCT in the contralateral testis.

Each testis was counted independently. We then calculated the expected frequencies of unaffected testis, unilateral and bilateral tumors. The underlying

69 model tests the hypothesis that bilateral tumors result from independent occurrence of TGCTs in the left and right testes. Comparison of the expected and observed numbers tests whether bilateral cases occur more (or less) often than expected (Table 2.8). Among the four double heterozygous mutants, the observed number of unaffected mutants was significantly lower than the calculated expected frequency and they all have significantly more bilateral tumors than expected. Interestingly, Ter and p53 did not interact to increase the susceptibility of TGCTs. The observed number of affected Ter/+ p53/+ double-mutants was not significantly different from the expected additive rate.

However, it seems that this pair of susceptibility gene interacted to increase the susceptibility of bilateral tumors. The other double heterozygous mutants

(M19/+ SlJ/+ , Ter/+ SlJ/+ and Ter/+ M19/+), each had significantly higher TGCT frequencies than the expected additive rate and it appears that the three pairs of susceptibility genes interacted to control for the susceptibility of bilateral tumors. The disagreement between the expected and observed numbers of bilateral tumors suggest that the bilateral tumors of the double heterozygous mutants is not the result of the independent occurrence of a unilateral tumor in each testis but is instead, the result of distinct genetic control for bilateral tumors (Table 2.8).

Similar to Youngren et al., we also examined the relationship between the proportion of bilateral tumors as a function of the combined frequency of unilateral and bilateral tumors in the double heterozygous (Figure 2.2). The expected frequency of bilateral cases was calculated assuming that bilateral

70

Table 2.8: Laterality of TGCTs in different double-heterozygous mutants

Mutants No TGCT Unilateral Bilateral N Test Score M19/+ SlJ/+ Observed 93 90 40 215 106.8** Expected 138 69 9 Ter/+ SlJ/+ Observed 37 39 26 102 88.2** Expected 61 35 5 Ter/+ M19/+ Observed 23 30 42 95 121** Expected 47 40 9 Ter/+ p53/+ Observed 38 23 14 75 10.3** Expected 56 18 2 ** P<0.001

Figure 2.2: Relationship between the observed and predicted bilateral tumors as a function of the overall TGCT frequency.

The blue line represents the expected bilateral cases if unilateral and bilateral tumors are under the same genetic control. The pink data points correspond to the number of bilateral cases in the four double-heterozygous mutants.

70% e 60% ar t a

h 50% Expected number of es t 40% bilateral cases l a r cas

e 30% T

t Double-heterozygous C a l mutants G 20% bi T f

o 10% n o i

t 0% ac

r 0% 20% 40% 60% 80% 100% F Combined TGCT frequency

71 and unilateral tumors are under the same genetic control (bilateral tumors are

the result of two independent unilateral tumors), and that the ratio of TGCTs in

the left versus the right testis was 2:1(OLIVER 1990; SKAKKEBAEK et al. 1998;

WEIR et al. 2000). The expected proportion of bilateral cases plotted in Figure

2.2 illustrate that the number of bilateral cases increased non-linearly as the overall tumor frequency increased. The number of bilateral cases of the four double heterozygous mutants is plotted in pink on Figure 2.2 and they do not fit on the expected line. This reinforces that the bilateral cases of the double heterozygous mutants are not the occurrence of two unilateral tumors, instead bilateral tumors are under a distinct genetic control from the unilateral tumors.

DISCUSSION

The low incidence of spontaneous TGCTs in the 129/Sv inbred strain makes it difficult to study the genetic control of TGCT susceptibility in these mice. To find susceptibility genes, we would have to screen thousands of backcrossed progeny for mapping studies which makes it a prohibitive task.

The complex genetics of TGCT susceptibility raises the possibility that gene interactions influence susceptibility. In this paper, single gene Mendelian mutations that are known to individually increase or decrease the frequency of

TGCTs were used to test interactions as a way to learn more about the pathways that control TGCT susceptibility.

72 Parental effects

Among the different interaction crosses, we also tested for parental

effects to test whether imprinting, epistasis, or in utero environment affected

TGCT susceptibility. Among the different susceptibility genes, parental effects

did not affect TGCT susceptibility. Among the various interaction crosses, only

the congenic cross between C2/+ and SlJ/+ mutants showed a modest parental effect. The chi-square score between the two parental directions was 4.0, which may be a false positive because it is at the borderline between significant and non-significant. We can continue to collect data on more animals to increase our sample size, which may give us a more accurate estimate of the parental effect observed between the C2 congenic segment and SlJ.

Laterality

It is believed that unilateral and bilateral tumors are under different

genetic control. In humans, linkage for a gene controlling susceptibility to

bilateral but not unilateral TGCTs was reported (RAPLEY et al. 2000). In the mouse model, there is circumstantial evidence from linkage crosses that suggests distinct genes might control susceptibility to unilateral and bilateral

M tumors (COLLIN et al. 1996). Ter homozygotes and 129-Chr19 mice both have a high frequency of bilateral tumors. As a result, it may be possible that unilateral and bilateral TGCTs are under distinct genetic controls. An alternative hypothesis proposed by Youngren et al. suggests that bilateral cases result from coincidental co-occurrence of unilateral cases and that co-

73 occurrence is more likely with increasing strength of the TGCT genes

(YOUNGREN et al. 2003). According to their hypothesis, the frequencies of unilateral and bilateral tumors can be predicted from the overall frequencies of

TGCTs in each testis. From their analysis of the 129/Sv, 129-Chr19M and

mutant mice, they found that unilateral and bilateral tumors are under the same

genetic control (YOUNGREN et al. 2003). Using the formula of Youngren et al.

(Appendix 1), the expected frequency of bilateral tumors (if a bilateral tumor results from two unilateral tumors) was calculated for each of the double heterozygous mutants. The expected and observed numbers of bilateral tumors were in disagreement between all the double heterozygous mutants

(Table 2.8). The various single-mutants rarely developed bilateral tumors but the different pairs of susceptibility genes interact to increase the susceptibility of bilateral tumors and the bilateral tumors of the double heterozygous mutants

are not the result of two unilateral tumors under the genetic control of a strong

TGCT susceptibility gene.

Ter/+ littermates of the pairwise interaction crosses has a higher tumor

frequency than expected

Interestingly, the Ter/+ littermates in both interaction crosses (Ter/+ and

SlJ/+, Ter/+ and p53/+) had a significantly higher tumor frequency than the published rate of 17% (NOGUCHI and NOGUCHI 1985). Initially, we hypothesized that it was the interaction between the two different 129 substrains that caused the increase susceptibility of TGCTs in the Ter/+ littermates. However, data

74 from the control crosses between the two 129 substrains suggest that the combination of the two 129 substrains does not account for the increased susceptibility to TGCT observed in the Ter/+ littermates. Significant increases of tumor frequency to the same levels in the two interaction crosses argue against statistical artifacts.

Additive effects

The pairwise interaction tests showed that one pair of susceptibility genes acted in an additive manner, whereas several pairs of susceptibility genes interact to enhance or suppress the development of TGCTs. The tumor frequency of the Ter/+ p53/+ double heterozygotes were similar to the expected additive rate, therefore, this pair of genes have an additive effect and do not appear to act on the same pathway.

Gene interactions

Three pairs of genes, M19 and SlJ, Ter and SlJ and also Ter and M19, interacted to increase susceptibility to TGCTs. The M19/+ SlJ/+ double heterozygotes had a 20% higher TGCT frequency than the expected additive rate (Table 2.3A). The TGCT frequency of Ter/+ SlJ/+ double heterozygous mutants was significantly higher than the expected additive tumor frequency regardless if the expected frequency was calculated using the published or the pooled frequency of other interaction crosses (Table 2.4). Similarly, Ter/+

M19/+ double heterozygotes also interacted to increase TGCT frequency by

75 16% or 34% depending on whether the published TGCT frequency of the Ter/+

heterozygotes or the pooled tumor frequency of 35% of the interaction crosses

were used to calculate the expected additive tumor frequency (Table 2.6A).

We also identified a pair of susceptibility genes, Ay and M19, which

interact to suppress the development of TGCTs. 129-Chr19M CSS with the Ay

mutation have 50% lower tumor frequency than the 129-Chr19M CSS (Table

2.7). The interactions observed among SlJ, M19 and Ay susceptibility genes

indicate that they may be acting on the same tumorigenesis pathway.

Enhancers of TGCT susceptibility

Both Ter and SlJ appear to interact with MOLF-Chr19 to increase

susceptibility to TGCTs. Congenic strains of 129-Chr19M were used to further

dissect the interaction between Ter and M19. It was determined that Ter may

interact with the proximal end of MOLF-Chr19 to increase susceptibility to

TGCTs because Ter/+ A2/+ mice have a tumor frequency of 60% which is not

significantly different from the tumor frequency of Ter/+ M19/+. However, Ter/+

A2/+ mutants did not have an increase incidence of bilateral tumors as

observed in the Ter/+ M19/+ mutants (that have the entire length of MOLF-

Chr19). Positional cloning of Ter identified a point mutation that introduced a

termination codon in Dnd1, which has a RNA recognition motif and has the

highest similarity to mouse apobec-1 complementation factor (Acf).

Interestingly, a candidate gene on the proximal end of MOLF-Chr19 is Acf.

76 From the interaction between Ter and the A2 congenic strain, Ter may be

interacting with Acf on MOLF-Chr19 to increase the susceptibility of TGCTs.

The results from the congenic cross and SlJ complement the analysis of

the panel of congenic strains used to characterize susceptibility genes in the

M 129-Chr19 CSS (YOUNGREN et al. 2003). A survey of TGCT frequencies in this

panel of congenic strains suggests that several genes on MOLF-Chr19 affect

TGCT susceptibility in both additive and epistatic manners and act as

enhancers or suppressors. The various congenic strains with only one segment

of MOLF-Chr19 had lower TGCT frequency than the 129-Chr19M CSS, whereas double-congenic strains with both the proximal and distal end of MOLF-Chr19, but missing the small middle segment resulted in a TGCT frequency similar to

M that of 129-Chr19 CSS (YOUNGREN et al. 2003). Double-congenic strains showed epistasis between TGCT susceptibility genes on MOLF-Chr19. The increased TGCT frequency observed in the double-heterozygous mutant, SlJ/+

M19/+, may result from an interaction between the susceptibility gene and an

interacting set of genes on MOLF-Chr19 rather than a single gene on Chr19.

Ter and SlJ interact to increase TGCT susceptibility. Both of the susceptibility genes cause a defect in germ cell numbers. Ter homozygous mutants are PGC deficient in all inbred background, and similarly, SlJ

homozygous mutants are PGC deficient and embryonic lethal due to severe

anemia. The SlJ deletion deletes the MGF gene which causes defects in melanogenesis, gametogenesis and hematopoiesis. The two susceptibility genes may be on the same pathway that controls for PGC survival and when

77 they interact with modifiers on the 129 inbred genetic background, it causes an

increased susceptibility of TGCTs. The possibility that Ter and SlJ are on the

same tumorigenesis pathway may be reinforced by the interaction that they

both interact with M19 to increase the susceptibility of TGCTs and specifically

for bilateral tumors.

Suppressors of TGCT susceptibility

Tumor suppression in mice such as the Ay /+ M19/+ double-mutant, with their significantly reduced TGCT frequencies, may reveal clues to the genetic and molecular basis for modulating TGCT susceptibility. Previous studies showed that Ay mice have a greatly increased susceptibility to many different

types of tumors (WOLFF 1987; WOLFF et al. 1986). However, 129/Sv male mice

that are heterozygous for the Ay mutation show an 8-fold reduction in TGCTs

(STEVENS 1967a). The mechanism for TGCT suppression was not pursued and

Stevens’ congenic strain was lost many years ago.

To verify findings of L.C. Stevens, we created a new 129-Ay congenic strain. 129/Sv mice normally develop TGCTs at a rate of 1%-5%, therefore

Stevens needed to examine a large number of males in order to find the 8-fold reduction. To detect a stronger effect, with a smaller number of mice, we crossed Ay congenic mice to a strain that has a high TGCT incidence, 129-

M Chr19 (MATIN et al. 1999). M19/M19 mice have a TGCT incidence of 80%

(MATIN et al. 1999) and therefore serve as an excellent strain to test for TGCT suppression with Ay. We found that the Ay mutation does indeed suppress

78

AW alle le

Raly Eif2s2 agouti Ahcy

ps1 1A1A’ 1 1 23 4 Ay allele A B C ()150 kb deletion

’ 1A1A1A 1 1 23 4 y B C

Figure 2.3: Genomic structure of the Aw and Ay mutant allele.

The Ay allele has a 150kb deletion that places the agouti gene under the transcriptional regulation of the Raly promoter. Transcription from the Raly promoter generates a ubiquitously expressed Ay-specific transcript containing the first 5’ untranslated exon of Raly. The Ay mutation does not affect the coding region of agouti.

79 TGCT formation and could serve a key role in understanding TGCTs and

tumorigenesis in general. The suppressive effect of Ay on TGCTs is exciting,

but little is known about the ways in which this mutation affects tumorigenesis.

Ay is a ~150kb deletion that include the deletion of Raly, Eif2s2 and the promoter of the agouti gene (Figure 2.3). The Ay deletion brings the coding

region of agouti (a) under the transcriptional regulation of the Raly promoter, resulting in the ubiquitous expression of agouti and the chronic inhibition of signal transduction through melanocortin receptors (DUHL et al. 1994; MICHAUD et al. 1994). The reduced tumor frequency observed in the Ay/+ M19/+ can be explained by two models. The first model suggests that TGCT suppression may be linked to the ubiquitous expression of agouti which may result in TGCT suppression as a consequence of the unique expression pattern involving the melanocortin receptor signaling pathway. Agouti is normally expressed during

development, in neonatal skin, and the testis (YEN et al. 1994) and is responsible for the wild-type coat color seen in mice (Silvers 1979). Agouti expression results in the switching of eumelanin to phaeomelanin production by inhibition of α-melanocyte-stimulating hormone due to antagonism of melanocortin receptor 1 (MC1-R) (BULTMAN et al. 1992; CONE et al. 1996; LU et

al. 1994). However in Ay mutants, ectopic agouti expression interferes with this switching pattern by preventing eumelanin production leaving only phaeomelanin production and this results in the uniform yellow coat associated with the mutation (CONE et al. 1996).

80 Ectopic expression of the agouti gene product leads to chronic inhibition

of signal transduction through the MC1-R and MC4-R (BULTMAN et al. 1992;

CONE et al. 1996). Characterization of the agouti gene revealed multiple transcripts of varying sizes that were expressed in the testis but not in any other adult tissues sampled. None of the testis-specific transcripts were expressed in neonatal skin or during development and they therefore play an undetermined role independent of pigmentation and development (BULTMAN et al. 1992). In the Ay mutant, a unique transcript is ectopically expressed in all tissues

examined (MILLER et al. 1993). It is possible that the overexpression of this larger agouti transcript interacts with the testis-specific transcripts leading to a suppression of TGCTs.

An alternative explanation for suppression is the deletion of Raly or

Eif2s2. The second model suggests that TGCT suppression results from loss- of-function of genes deleted in the 150kb region upstream of agouti. It is possible that deletion of Raly or Eif2s2 suppresses TGCTs. Raly is normally ubiquitously expressed and belongs to a family of RNA-binding proteins involved in pre-mRNA processing and embryonic developmental regulation

(KHREBTUKOVA et al. 1999; MICHAUD et al. 1994). Raly functions in preimplantation development and its deletion may account for the embryonic

y lethality associated with the A mutation (DUHL et al. 1994). Another gene located in the deletion between Raly’s promoter and Agouti that could account for TGCT suppression is Eif2s2. Eif2s2 is the beta subunit of translation

81 initiation factor eIF2, which, through its phosphorylation and dephosphorylation,

is a key regulatory component of protein synthesis (Gebauer and Hentze 2004).

Further studies are necessary to test the two models that may cause the

suppression of TGCTs. We are currently crossing the Avy mutant to the 129-

Chr19M to help clarify how Ay suppresses TGCT susceptibility. The Avy mutant

has ubiquitous expression of the Agouti gene product due to an intra-cisternal A

particle (IAP) retrotransposon inserted upstream of the agouti protein. The Avy

mutant has yellow coat color, obesity, diabetes and increased susceptibility to

vy various types of tumors (MORGAN et al. 1999). If the A mutation suppresses the development of TGCTs in the 129-Chr19M mice similar to our observations

using the Ay mutation, it can then be concluded that the suppression of TGCTs

is due to the ubiquitous expression of the agouti gene product rather than the

deletion of Raly or Eif2s2.

RNA biology

Various aspects of RNA biology have a function in the development of the PGC lineage. In PGCs, noncoding RNAs repress transcription (MARTINHO et al. 2004) and RNA and RNA-binding proteins that regulate translation (KNAUT et al. 2002; MOORE et al. 2003) and PGC totipotency (CRITTENDEN et al. 2002;

ZHOU and KING 2004). We recently cloned the Ter mutation and identified a

premature stop codon in the Dnd1. Dnd1 has significant sequence similarity to

Acf, which encodes an RNA-binding subunit of the RNA editing enzyme complex that converts specific cytidines to uridines in target mRNAs. Raly and

82 Eif2s2, which are deleted within the Ay mutation also have distinct functions in

RNA biology. Raly belongs to a family of RNA-binding proteins involved in the processing and transport of pre-mRNAs. Eif2s2 is the beta subunit of translation initiation factor eIF2. It is still unknown if it is the deletion of a tumor suppressor gene within the SlJ locus that causes the increased susceptibility to

TGCTs. We have further narrowed down the SlJ deletion from ~640kb to

gb gb ~120kb using the Sl allele (BEDELL et al. 1996). We backcrossed the Sl

deletion on the 129/Sv inbred background and found that Slgb/+ mutant also

doubled the tumor frequency of the wild-type animal. In silco analysis showed that within the ~120kb deletion, only Mgf and a small pseudo gene is deleted.

We are currently generating a Mgf knock-out mouse to determine if the deletion of Mgf increases TGCT susceptibility. Interestingly, we have recently identified

a microRNA (miRNA) that is deleted with the SlJ locus (Kawasoe and Nadeau, unpublished data) and it has been shown that miRNAs have a role in maintenance of the pluripotent cell state and in the regulation of early mammalian development (BARTEL 2004; HOUBAVIY et al. 2003). Maybe, the

deletion of the miRNA plays a critical role in the susceptibility of TGCT instead

of the deletion of the Mgf gene because point mutation within the Mgf gene has

been shown to not affect TGCT susceptibility on the 129/Sv mice.

We made a network diagram to summarize the data collected from our

pairwise interaction tests (Figure 2.4). Ter and SlJ single-mutants both have

similar phenotypes, and they interact to increase TGCT susceptibility in Ter/+

SlJ/+ double heterozygous mutants therefore, they may act on the same

83 pathway. MOLF-Chr19 interacts to increase or decrease TGCT susceptibility with each of the mutations that may play a role in RNA biology (Ay, SlJ, Ter). It may be possible that M19 is the initiator that activates the three RNA related genes and these target genes then act in different pathways to modulate TGCT susceptibility (Figure 2.5A). Alternatively, the three RNA related genes act on different pathways and converge to act upon M19 which is their target (Figure

2.5B). We are still currently characterizing the TGCT susceptibility genes on

MOLF-Chr19 and we are currently investigating if there are any RNA related genes on Chr19 that may contribute to TGCT susceptibility and how the may interact with the other mutations to influence TGCT frequency. From this network, we can draw general conclusions to the pathway of tumorigenesis but further analysis is needed to characterize each susceptibility gene and how it functions within the pathway.

84 Figure 2.4: Network of interaction between M19, Ter, p53, Ay and SlJ. Indicated by the red line, M19 interacts with all the mutations that is known to play a role in RNA biology to increase or decrease TGCT susceptibility .

RNA M19

Non-RNA

Interaction p53 Ter Additive

SlJ AY

Figure 2.5: Roles of M19 in the tumorigenesis pathway. A. M19 is upstream of SlJ, AY and Ter and acts as the initiator. B. M19 is downstream in the pathway of SlJ, AY and Ter act as the target

A. B.

INITIATOR INITIATOR

M19 SlJ AY Ter

SlJ AY Ter M19

TARGET TARGET 85

CHAPTER 3

P53 AND STEEL-J DELETION INTERACT TO SUPPRESS TESTICULAR GERM CELL TUMOR SUSCEPTIBILTY

86 Authors: Man-Yee Josephine Lam and Joseph H. Nadeau

References: 1. Genetics 166:925-933. 2004

ABSTRACT

Testicular germ cell tumors (TGCTs) are the most common cancer in

young men. The genetic control of susceptibility is complex. To date, there has

not been a susceptibility gene identified in human. We use mouse models to

study the genetic control of susceptibility to TGCTs. The 129/Sv mouse is the

only inbred strain that has an appreciable frequency of spontaneous TGCTs.

TGCT susceptibility in mice is a multigenic trait. Different approaches have

been used to map novel susceptibility genes but, we still need a better

understanding of the genetic and developmental control of TGCT susceptibility.

The approach that I used was to maintain a uniform 129/Sv inbred background

and test the pairwise interaction between p53 and SlJ deletion to test whether

they are involved in the tumorigenesis pathway. From our heterozygous

interaction cross, the tumor frequency of the double-heterozygous mutants

indicated that p53 and SlJ interact to suppress 4-fold the development of

TGCTs from the expected additive affected rate. In the homozygous cross,

p53/p53 SlJ /+ mutants did not suppress development of TGCTs indicating that the suppressor effect maybe a dosage affect. Preliminary studies have begun to study the expression profile of p53 and Mgf in the double heterozygous mutants in an attempt to characterize the interaction between apoptosis and the

87 MGF-KIT signal transduction pathways and determine how they modulate

susceptibility to TGCTs.

INTRODUCTION

Alterations in three types of genes, oncogenes, tumor-suppressor genes,

and stability genes are responsible for tumorigenesis. Oncogenes induce

uncontrolled cell proliferation. Oncogene activations can result from

chromosomal translocations from gene amplifications, or from subtle intragenic

mutations affecting crucial amino acids that regulate the activity of the gene

product. Tumor-suppressor genes control cell proliferation or activate the

apoptotic pathways. Stability genes keep genetic alterations to a minimum and

when they are inactivated, mutations in other genes are tolerated at a higher

rate. Mutations in these three classes of genes can occur in the germline,

resulting in hereditary predispositions to cancer, or in somatic cells, resulting in

sporadic tumors.

Perhaps the most important finding in cancer biology is that almost all

tumor inducing DNA viruses in experimental animals and humans encode

proteins that inactivate both Rb and p53 (zur Hausen 2001). In addition to the

Rb and p53 pathways, there are other genes that play a role in many tumor types such as adenomatous polyposis coli (APC), glioma-associated oncogene

(GLI), hypoxia-inducible transcription factor (HIF), phosphoinositide 3-kinase

(PI3K), SMADs and receptor tyrosine kinases (RTKs). Although many cancer

88 genes and susceptibility pathways are known, mutations in p53 occur in about

half of all human cancers.

Mutations of the p53 tumor suppressor gene are the most frequently observed genetic lesion in human cancers (LEVINE et al. 1991). Loss of function of p53 either through point mutation or deletion plays an important role in the progression of many types of human cancers (NIGRO et al. 1989). In addition to

point mutations and deletions, germ-line mutations in p53 occur in families with

Li-Fraumeni syndrome, which is an autosomal dominant disease characterized

by significant predisposition to various types of cancer (SRIVASTAVA et al. 1990).

By 30 years of age, 50% of Li-Fraumeni patients develop neoplasms, the most

common types being brain tumors, soft tissue sarcomas, osteosarcomas,

breast tumors, and leukemias (MALKIN et al. 1990).

Functions of p53

The p53 gene encodes a transcriptional regulatory protein involved in regulation of the cell cycle and apoptosis in response to DNA-damage or imbalance in intracellular growth signals. It is speculated that p53 serves as a

‘guardian of the genome’ to prevent proliferation of a cell that has sustained genetic damage (LEE and BERNSTEIN 1993). Its most important involvement is cell-cycle regulation and apoptosis. P53 is also involved with differentiation, gene amplification, DNA recombination, chromosomal segregation and cellular senescence (OREN and ROTTER 1999). Under stress conditions such as DNA damage and hypoxia, the P53 protein is stabilized leading to an increase in

89 cellular P53 levels. There are various cellular responses to p53 activation and the response depends on factors such as cell type, cell environment and other oncogenic alterations that are sustained by the cell. The effect of p53 activation is to inhibit cell growth through cell cycle arrest or apoptosis to prevent tumor formation. The apoptotic and cell cycle arrest activities of p53 can be independent. P53 is involved in both the extrinsic and intrinsic pathways of apoptosis (ZHIVOTOVSKY and KROEMER 2004). The intrinsic pathway is

characterized by mitochondrial-membrane permeabilization (MMP). p53 also acts as transcription factor, transactivating a series of pro-apoptotic genes from the BCL2 family which induce MMP and release apoptogenic factors from the mitochondrial intermembrane space that causes bioenergetic failure. The extrinsic pathway activates caspase. p53 can stimulate the expression of death receptor and transactivates genes that encode MMP-inducing proteins.

Mechanisms for p53 loss in cancers

p53 is mutated in about half of all cancers, resulting in loss of apoptosis and cell cycle arrest (LEVINE et al. 1991). Tumor-associated mutations in p53

are mainly point mutations that result in single amino-acid substitutions which

result in frameshift mutations that result in the complete loss of P53 protein

expression. It is also possible that point mutated P53 protein is more stable

than wild-type P53 and present at high levels in the tumor cell. The mutated

p53 can act as dominant-negative inhibitor of wild-type p53. An alternative is

90 that many tumors that have a point mutated p53 also show loss of

heterozygosity and eliminate the wild-type allele of p53.

p53 mouse model

Mice deficient for p53 are developmentally normal but have enhanced

susceptibility to spontaneous tumors. The p53 model that we chose to use in

our experiments was generated by Jacks et al. and it carries a germline

disruption of the gene (JACKS et al. 1994). The mutation removes approximately 40% of the coding region of p53 and eliminates the synthesis of the P53 protein. p53/p53 homozygous mice are viable but are highly predisposed to a variety of tumors at a very early age. These mice develop many different types of tumors, but the most frequently observed tumor is lymphoma.

Genetic background alters the spectrum of tumors that develop in p53 deficient mice (DONEHOWER et al. 1995; HARVEY et al. 1993a). The p53 mutation was placed on the 129/Sv background to monitor tumor development and was compared to the tumor spectrum of p53-deficient mice on a mixed genetic background of C57BL/6 and 129/Sv. The spectrum of tumors of the two strains is similar except 129/Sv-p53/p53 mice have a significantly increased incidence of teratocarcinomas (DONEHOWER et al. 1995; HARVEY et al. 1993a).

Nearly 50% of the 129/Sv-p53/p53 males developed teratomas, whereas the mixed inbred p53/p53 males rarely develop this tumor type. These teratomas

develop very early in age and testicular tumors were not observed in the 129/Sv

91 heterozygous and wild-type males. This background effect observed is

interesting because p53 is not mutated in human testicular cancer patients

(FLEISCHHACKER et al. 1994).

TGCT development in mice

In mice, primordial germ cells (PGCs) are the earliest recognizable precursors of gametes and arise outside the gonads (MOLYNEAUX et al. 2001).

During embryogenesis, PGCs are first identified around embryonic day 7 (E7)

at the base of the primitive streak. The PGCs leave the yolk sac, migrate

through the dorsal mesentery of the gut, and arrive at the genital ridges by

E11.5. As the PGCs migrate towards the genital ridges, they proliferate in number from ~40 at E7 to ~20,000 at E11.5 when they arrive at the genital ridges where sexual determination occurs. At E13.5, male PGCs enter G1 mitotic arrest and remain arrested until a few days after birth (DONOVAN et al.

1998). This period from E11.5-E13.5 is when critical events occur in TGCT development. It has been hypothesized that some PGCs fail to enter G1 mitotic arrest and continue to divide into pluripotent stem cells called embryonal carcinoma (EC) cells that in turn become tumors composed of various cells and tissue types (JIANG and NADEAU 2001; STEVENS 1967a).

Rationale

Testicular germ cell tumors (TGCTs) are rare in most inbred mouse

strains and occur spontaneously in the 129/Sv inbred strain at a rate of 1-5%

92 (STEVENS and HUMMEL 1957). Several Mendelian mutations are known to affect the susceptibility of TGCTs. Interestingly, these mutations must be congenic on the 129/Sv inbred genetic background in order to observe an effect on TGCT susceptibility. This demonstrates that these mutant genes act together with

129-derived genes to control susceptibility. In 129/Sv-p53/p53 homozygous mutants, the increased incidence of TGCTs results from interaction between p53 and 129-modifier genes. We wanted to identify which pathways p53 may be involved in for tumorigenesis, and how p53 interacts with the SlJ deletion to

affect the susceptibility of TGCTs because it is known that there is a

relationship between Mgf and p53 during germ cell development (ABRAHAMSON et al. 1995).

The SlJ deletion is ~640kb at the Sl locus and includes the deletion of the mast cell growth factor gene (Mgf) (BEDELL et al. 1996). L.C. Stevens transferred different Sl alleles on to the 129/Sv inbred strain and found that only the Sl and SlJ deletion on the 129/Sv inbred background cause a significant

J J increase in the frequency of TGCTs (STEVENS 1967a). Sl /Sl homozygous mutants are PGC deficient and embryonic lethal due to severe anemia.

Heterozygous SlJ/+ males have a tumor incidence more than doubled the wild-

type littermates (STEVENS 1967a). We wanted to test the interaction between this pair of susceptibility genes to test their effect on tumor frequency, laterality

and parental effects to dissect the pathways to tumorigenesis. If p53 and SlJ interact on the same pathway, the expected tumor frequency of the double- mutant would vary significantly from the additive tumor frequency of each

93 mutation alone. However, if the tumor frequency of the double-mutant is similar

to the additive frequency of each mutation alone, then the genes act

independently and probably do not act on the same pathway.

MATERIALS AND METHODS

Mice

129S1/SvImJ (JR002448, previously known as 129/SvJ and 129S3/SvImJ),

129S1/Sv-p+Tyr+KitlSl-J/+ (JR000090), B6.Cg-AY (JR000021), and 129-P53tm1Tyj

(JR002080) were obtained from the Jackson Laboratory (Bar Harbor, ME,

USA). The nomenclature for 129 substrains has been revised by the Jackson

Laboratory (www.informatics.jax.org/mgihome/nomen/strain_129.shtml) and the

recommended designations were used in this paper. The 129.MOLF-Chr19

CSS (N15F2+) was described previously (MATIN et al. 1999) and was obtained

from our research colony. Mice were maintained in the CWRU Animal Resource

Center on a 12:12hr light:dark cycle and fed Lab Diet 5010.

Genotyping

DNA for PCR genotyping was obtained from samples of tail tissue. Each tail

sample was digested overnight in 89 ul of water, 10 ul 10X PCR buffer and 1 ul

Proteinase K (10mg/ml). The digestion was incubated for 1hr at 100°C to inactivate the reaction.

94 p53

In this study, we purchased from the Jackson Laboratory the p53/+ mutant

strain that was developed by Tyler Jacks et al. (JACKS et al. 1994). This strain has a different p53 mutant allele than that reported by Donehower et al.

(DONEHOWER et al. 1992).

A three primer PCR assay was used to distinguish wild-type from heterozygous p53 animals. The three primers are X7 5’TATACTCAGAGCCGCCT 3’, Neo19

5’CTATCAGGACATAGCGTTGG 3’, and X6.5 5’

ACAGCGTGGTGGTACCTTAT 3’. Primers X7 and Neo19 amplify a 600bp fragment identifying the Neo insert (mutant). Primers X7 and X6.5 amplify a

400bp fragment from the untargeted p53 allele (wild-type). PCR amplification was carried out in a 96-well block MJ Research PTC-200 thermal cycler. The reagents were 0.15ul (0.75U) Taq polymerase (Invitrogen), 2.5ul 10X PCR buffer (magnesium free), 0.3ul 10mM dNTPs, 1ul 25mM magnesium chloride,

0.2ul of each primer (0.1uM), 1ul DNA (25ng), and 19.35ul dh2O in a final volume of 25ul. PCR conditions were as follows: initial denaturation step for

94°C for 2 min followed by 35 cycles of 94°C for 1 min, 60°C for 1 min, 72°C for

1 min, final extension of 72°C for 5 min and then 4°C for 15 min. PCR products were resolved on a 2% agarose gel and visualized with ethidium bromide.

SlJ

The breakpoints of the SlJ deletion are not known and as a result a PCR

genotyping assay is not available for the SlJ mutation. SlJ /+ mutant

95 heterozygotes have a light coat color in the belly and the tips of the tail and digits are pink.

Northern Analysis

Total RNA was isolated from testis and spleen of the mice using Trizol. 60ug of mRNA was loaded per lane; the RNA was fractionated on a 1% formaldehyde/argarose gel, transferred to nylon membrane and hybridized with a 2.3kb cDNA probe. The membrane was stripped and a GAPDH probe was used as a control for loading. We followed the standard protocol of

NorthernMax kit (Ambion).

Western blotting

Testes were homogenized in buffer containing 50 mM Tris, 150 mM NaCl, 0.1%

SDS, 1% NP-40, 1 mM PMSF and 1 complete tablet (Roche). Protein concentration was determined using BCA assay kit (Pierce). Protein at a concentration of 20 micrograms was loaded on 12% tris-glycine gels

(Invitrogen) before transfer onto PVDF membrane. Membranes were blocked with 5% non-fat dry milk in TBS-Tween (20 mM Tris, 137 mM NaCl, 0.5%

Tween–20, pH 7.6) for 1hr followed by overnight incubation at 4°C with 1/2000 diluted anti-p53 monoclonal (Ab-1, Oncogene Research Products).

Membranes were washed 3 times for 5 minutes each in TBS-Tween. Diluted

1/10000 conjugated anti-mouse IgG was added to the membrane and incubated for 1 hour at room temperature. Membranes were washed 3 times

96 for 5 minutes in TBS-Tween. The membrane was developed using the

enhanced chemiluminescence (ECL) system (Amersham-Pharmacia Biotech).

Quantitative PCR

RT-PCR was performed on the total RNA to obtain cDNA. Q-PCR was

performed using the following primers: p53QExon3/4-F 5’

GCAACTATGGCTTCCACCTG 3’ and p53QExon3/4-R 5’

ACTGCACAGGGCACGTCT 3’. These primers were designed to flank the end

of exon 3 and the beginning of exon 4 of the p53 gene which resulted in a product of 118bp to ensure that there were no genomic DNA contaminations.

Q-PCR was performed using MJ research DyNAMO kit.

RESULTS

p53/+ and SlJ/+ double-mutant heterozygote:

The interaction test between p53 /+ and SlJ/+ mutants revealed

unexpected results. The tumor frequencies in the reciprocal parental crosses

were not significantly different (X2 =0.5; P>0.1). The expected additive tumor frequency of the double heterozygous mutants was calculated using the pooled tumor frequency of the various control and interaction crosses (Table 2.1). If the two mutant genes acted additively, the expected additive tumor frequency would be 17% (=12% [p53/+] + 12% [SlJ/+] – 7% [+/+]). Surprisingly, in the

pooled data of 75 double heterozygous mutants, only 7% developed a TGCT

97 (Table 3.1). The double heterozygous mutants had more than 2-fold lower

tumor frequency than expected (X2 = 5.95; P>0.01), suggesting that partial

deficiency of the tumor suppressor gene p53 and heterozygosity for the SlJ deletion interact to suppress development of TGCTs.

p53/p53 and SlJ/+ double mutant:

We next tested whether the double homozygous mutants also

suppressed the development of TGCTs. p53 homozygous mutants have a

significantly increased TGCT susceptibility compared to p53/+ mutants

(DONEHOWER et al. 1995; HARVEY et al. 1993a). Therefore, if p53 interacts with

SlJ to suppress the development of teratomas, the double homozygous mutants would be informative because we might observe a stronger effect if suppression depends on the extent of p53 deficiency. SlJ/SlJ homozygous mutants are

embryonic lethal (COPELAND et al. 1990), therefore we could not obtain p53/p53

SlJ/SlJ double homozygous mutants. In addition, homozygous p53/p53 mutants

do not breed well therefore, we had to set up our interaction cross by crossing

p53/+ SlJ/+ and p53/+ mutants to obtain p53/p53 SlJ/+ mice.

Parental effects were not found except in the double heterozygous

littermates (X2=4.35; P<0.025). This interaction cross resulted in littermates

98

Table 3.1: Interaction test between p53 and SlJ in double heterozygous mutants

Overall tumor frequency of p53/+ SlJ/+ and littermates controls Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals p53/+, SlJ/+ 75 7% (5) 0% 7% (5) p53/+, +/+ 91 15% (14) 0% 15% (14) +/+, SlJ/+ 87 11% (10) 0% 11% (10) +/+, +/+ 93 2% (2) 0% 2% (2)

99 with five different genotypic combinations of genotype contributed from the

parents, which each act as internal controls. Homozygous p53/p53 mutants

had a tumor frequency of 29%, which was not significantly different from the

2 published tumor frequency of 35% (X =1.03; P>0.1) (HARVEY et al. 1993a). The

TGCT frequency of the p53/p53 littermates (Table 3.2) and the pooled tumor frequency of the SlJ/+ of the control crosses and interaction crosses (Table

2.1A) were used to calculate the expected additive affected rate for p53/p53

SlJ/+ mutants. If the two genes acted in an additive manner, the expected additive tumor frequency is 34% (=29% [p53/p53] + 12% [SlJ/+] – 7%[+/+]).

Among the pooled 45 p53/p53 SlJ/+ double mutants, 26% developed TGCTs

and the affected rate was not significantly different from the expected additive

rate (X2=0.9; P>0.1). Although the pair of susceptibility genes did not interact to

significantly suppress the development of TGCTs in the p53/p53 SlJ/+ mutants,

yet it was lower but not significant lower than the expected additive tumor

frequency. The p53/p53 SlJ/+ mutants have a TGCT frequency of 26% which is

similar to the 29% of p53/p53 mutants that develop TGCT therefore, it appears that the SlJ/+ mutation had no effect on TGCT susceptibility in the p53/p53 SlJ/+ mutants. However, the sample size is small, therefore we need to continue to survey these mice to obtain a better estimate of TGCT frequency to clearly determine if 1) p53/p53 SlJ/+ mutants have an additive tumor frequency, 2) SlJ/+ mutation had no effect on TGCT susceptibility in the p53/p53 SlJ/+ mutants, or

3) p53/p53 SlJ/+ mutants have a significantly lower tumor frequency than

expected.

100

Table 3.2: Interaction test between p53 and SlJ in homozygous-heterozygous mutants

Overall tumor frequency of p53/p53 Sl J/+ and littermates controls Genotype Sample Unilateral Bilateral Total Affected Size Tumors Tumors Animals p53/p53 Sl J/+ 42 17% (7) 10% (4) 26% (11) p53/p53 +/+ 41 24% (10) 5% (2) 29% (12) p53/+ +/+ 125 9% (11) 1% (1) 10% (12) p53/+ Sl J/+ 125 22% (27) 4% (5) 27% (32) Sl J/+ +/+ 70 9% (6) 0% 9% (6) +/+ +/+ 97 7% (7) 1% (1) 8% (8)

101 In the heterozygous cross, the p53/+ SlJ/+ mutants have at least 2-fold

lower tumor frequency than the expected additive rate. In this interaction

across, a modest parental effect was observed within the p53/+ SlJ/+ littermates

in the two reciprocal crosses (X2=4.35; P<0.05), the Chi-square score is

borderline significant, therefore the data were pooled. The observed affected

rate of the double heterozygous mutants is 27% which is significantly higher

than the expected additive rate of 17% (X2=6.92; P>0.01) (Table 3.2). It was

surprising that the double heterozygous mutants of the cross between p53/+

SlJ/+ and SlJ/+ mutants did not have a significantly lower tumor frequency than

the expected additive rate as observed in the heterozygous cross. The

inconsistency of the double heterozygous mutants between the two interaction

crosses (p53/+ x SlJ/+ and p53/+ SlJ/+ x p53/+) may be explained by the nature of the interaction crosses. The additive tumor frequencies of the double heterozygous mutants observed in the cross between p53/+ SlJ/+ and p53/+ may be contributed by the genotype of the parent which is either a p53/+ SlJ/+ male or p53/+ SlJ/+ female depending on the direction of the parental cross. As

a result, we generated a control cross to test whether the genotype of the

parent affected suppression of TGCTs in p53/+ SlJ/+ mutants.

Control cross to test whether double heterozygous parents contribute an effect on TGCT susceptibility:

Double heterozygous littermates in the cross between p53/+ SlJ/+ and

SlJ/+ mutants did not suppress TGCTs as expected from the double

102 heterozygous cross between p53/+ and SlJ/+. We used control crosses

between p53/+ SlJ/+ and 129/Sv (in both parental directions) to test whether the

genotype of the double heterozygous parent differentially affect TGCT

susceptibility in their progeny. Parental effects were not detected in the

reciprocal crosses and the data were pooled. Among 78 p53/+ SlJ/+ mutants,

14% developed TGCT which is not significantly different than the expected

additive rate of 17% (X2=0.37; P>0.5) (Table 3.3). As a result, the genotype of the parent in the interaction cross may play a role in the susceptibility of TGCTs in the progeny. We observed a decrease in TGCT incidence in the double heterozygous mutants of the heterozygous cross, while p53/p53 SlJ/+ and

p53/+ SlJ/+ mutants of the p53/+ SlJ/+ x SlJ/+ cross have the expected additive

tumor frequency and it could be explained by the genetics of the parent.

Parental effects

All the crosses were tested for parental effects by generating each

cross in reciprocal parental directions. Parental effects were not found and the

data for each cross was pooled. In the cross between p53/+ SlJ/+ and p53/+ mutants, the double heterozygous mutants have a modest parental effect

(X2=4.35; P<0.05). It can be a false positive, therefore the tumor frequencies

between the two parental crosses were pooled.

103

Table 3.3: Control cross to test whether double heterozygous parents affect TGCT susceptibility

Overall tumor frequency of p53/+ SlJ/+ and littermates controls Genotype Sample Unilateral Bilateral Total affected size tumors tumors animals p53/+ SlJ/+ 78 14% (11) 0% 14% (11) p53/+ +/+ 82 18% (15) 0% 18% (15) +/+ SlJ/+ 88 8% (7) 1% (1) 9% (8) +/+ +/+ 85 7% (6) 0% 7% (6)

104 Laterality

The single-mutants develop mainly unilateral tumors therefore, we

expected the double-mutants to rarely develop bilateral tumors. In the double

heterozygous cross, the double-mutants did not develop bilateral tumors. In the

cross between p53/+ SlJ/+ and p53/+ mutants, there was also a low frequency of bilateral tumors observed in the p53/p53 SlJ/+.

DISSCUSSION

p53 is mutated in half of all human cancers but rarely if ever mutated in

human testicular cancer (LEVINE et al. 1991). p53 deficient mice develop

normally, but tumors arise rapidly and all develop tumors by 10 months of age

(DONEHOWER et al. 1992). Heterozygous mice also develop tumors, but the age of onset is delayed. Genetic background alters the spectrum of tumors

(DONEHOWER et al. 1995; HARVEY et al. 1993a), but only on the 129/Sv inbred background does p53 deficiency increase the susceptibility of TGCTs, demonstrating that p53 interacts with 129-derived genes to modulate TGCT susceptibility.

p53 and MGF-KIT signaling pathway

In the interaction cross between p53/+ and SlJ/+ mutants, the double

heterozygous mutants had a lower tumor frequency (~2-fold decrease) than the

rate expected from the additive effects of the single mutants. These results

105 suggest that interactions between apoptosis and MGF-KIT signal transduction

may modulate susceptibility to testicular cancer.

p53 mediates apoptosis and cell cycle arrest at the G1/S stage (LEE and

BERNSTEIN 1993), whereas the MGF-KIT signaling pathway prevents apoptosis

and promotes cell division and growth of melanocytes, mast cells and germ

cells (DOLCI et al. 1991). In cell culture, MGF inhibits p53-mediated apoptosis and differentiation, but not G1/S cell cycle arrest (ABRAHAMSON et al. 1995). In contrast to most studies that examined the relationship between KIT-MGF signaling and p53 involved cell culture systems (ABRAHAMSON et al. 1995;

MATSUI et al. 1991), Jordan et al. used an in vivo model system to study the affect of p53 on melanocytes, mast cells and germ cells by taking advantage of

W-v the Kit mutant which is homozygous viable (JORDAN et al. 1999).

Mouse mutants of the KIT receptor and its ligand, MGF, are defective in melanogenesis, gametogenesis and hematopoiesis. Although the phenotype of most of the Kit and Sl mutants are similar, L.C. Stevens found contrasting results for several Kit and Sl alleles on the 129/Sv inbred background. Two Sl alleles with large deletions (Sl and SlJ) interacted with 129/Sv-derived

susceptibility genes to increase TGCT frequency, whereas none of the Kit and

Sld (point mutation within Mgf) mutants showed an increased susceptibility to

W-v TGCTs (STEVENS 1967a). In a separate study, the Kit mutant was crossed to p53 deficient mutant to study the role of p53 in regulating cell death in the absence of MGF-KIT signaling (JORDAN et al. 1999). The double-mutants that were deficient in P53 function did not increase the survival of melanocytes and

106 mast cells, but showed instead an increased number of germ cells and restored

fertility (JORDAN et al. 1999). Normally, MGF-KIT signals are induced between

E7.5 and E13 (Manova and Bachvarova 1991) to inhibit P53 function during

W-v PGC proliferation (ABRAHAMSON et al. 1995). The Kit mutants lack MGF-KIT signals and therefore can not inhibit p53, leading to apoptosis of PGC and sterility. Jordan et al. demonstrated that double mutants deficient in p53 and Kit rescued sterility defects because apoptosis of male germ cells is p53- dependent but apoptosis of melanocyte and mast cells is regulated by a p53- independent pathway. This experiment explains how fertility can be restored in mutants that lack function of KIT and MGF but it does not readily explain the suppression of TGCT development in our p53/+ SlJ/+ double heterozygous mutants.

Male PGCs enter G1 mitotic arrest at E13.5 and remain mitotically inactive until birth (DONOVAN et al. 1998). It has been hypothesized that PGCs

which fail to enter G1 mitotic arrest may become pluripotent stem cells resulting in tumors of various cell and tissue types. The double heterozygous mutants have partial deficiency of P53, which may allow germ cells to escape G1 cell cycle check points and could result in uncontrolled cell growth leading to the development of TGCTs. In addition, the double heterozygous mutants have the

SlJ deletion which can involve the loss of a tumor suppressor gene or Mgf that results in an increased of TGCT frequency. Therefore it is surprising that the partial deficiency of p53 interacts with the SlJ mutation to correct the tumor

107 phenotype that each mutation alone possess and together they suppress the

development of TGCTs to the baseline rate in 129/Sv.

p53 dosage effects

Tumor suppressor genes are viewed as being recessive at the cellular

level, therefore mutation or loss of both tumor suppressor alleles is required for

tumor formation. p53/+ heterozygous mice have a ~50% chance of developing

tumors by 18 months of age and almost all are dead by 2 years of age. About

half of the tumors that develop in p53/+ mice do not show loss of the wild-type

allele (also known as loss of heterozygosity) and in these mice, the wild-type

p53 allele appears structurally and functionally intact (VENKATACHALAM et al.

1998). Susceptibility to tumors in the p53/+ mice illustrates that a reduction in

p53 level is sufficient to alter the response to genotoxic stress in specific cell

types in a dosage sensitive manner. Dosage effects in p53/+ cells in vitro are evident with respect to various biological properties, including cell proliferation

and induction of apoptosis (HARVEY et al. 1993b). It is possible that p53 functions as a tetramer. If the level of total p53 is halved in a cell because of an

allele loss, then the effective concentration of tetramers may be more than

halved, depending on the affinity of the p53 monomers for each other

(FRIEDMAN et al. 1993). Therefore, the level of functional p53 may be reduced by more than 50% in the heterozygous state (VENKATACHALAM et al. 1998).

Yoon et al. also demonstrated that the dosage of p53 affects the transcriptional regulation of target genes (YOON et al. 2002). The authors identified 35 genes

108 whose expression is significantly correlated to the dosage of p53. The various genes are involved in a variety of cellular processes including signal transduction, cell adhesion and transcription regulation. As a result, examination of the expression of p53 and Mgf in the p53/+ SlJ/+ mice may uncover clues to the interaction that suppresses development of TGCTs in the double heterozygous mutants. It is possible that the wild-type allele of the p53/+ mutant is activated by the SlJ deletion to cause the intact p53 allele to express P53 protein that induces apoptosis and cell cycle arrest, which corrects the TGCT phenotype.

Preliminary studies of the expression of p53 in the testis of the double heterozygous mutants

Northern analysis suggests expression of the p53 transcript in the 4- week old testis of the p53/+ SlJ/+ may be higher than in the p53/+, SlJ/+ and wild-type littermates. However, because technical problems occurred during the rehybridization of the northern blot with GAPDH as the loading control, we wanted to confirm our data using quantitative PCR (Q-PCR). RT-PCR was used to generate cDNA from total RNA of 4-week old testis of the double heterozygous mutants, p53/+, SlJ/+ and wild-type littermates. A pair of Q-PCR primers were generated to flank the end of exon 3 and the beginning of exon 4

(118 bp product) to ensure that there was no genomic DNA contamination when amplifying the cDNA product. Q-PCR was performed on two sets of mice in duplicate. GAPDH was used as the endogenous control to normalize the initial

109 amount of cDNA. For each set of mice, the average was taken between the

two replicate experiments to determine the average normalized expression of

p53 for each genotype. The averaged expression of p53 for each genotype was then divided by the value of the wild-type mouse to generate a normalized ratio of p53 expression for each genotype relative to the wild-type mouse.

Lastly, the expression ratios of two set of mice were averaged to give an overall expression ratio of p53. The p53/+ and SlJ/+ littermates each had an averaged

expression ratio of 0.3 and the double heterozygous mutant had an average

expression ratio of 0.5 (Table 3.4).

It was expected that the expression of p53 in p53/+ mutant would be

≤50% of the wild-type littermates, but it was surprising that the expression of

p53 in the SlJ/+ littermates was also ~50%. The lower than expected

expression of p53 in the SlJ/+ littermates may due to the relationship between

p53 and MGF-KIT signaling pathway that controls the proliferation of germ cells during development but further analysis is needed to test whether the relationship between p53 and the MGF-KIT signaling pathway affects transcription of p53. The double heterozygous mutants have a p53 transcript expression ratio of 0.5 which is not overexpressed in comparison to the wild- type littermates but it is slightly higher than the single-mutant littermates. From the ratio of the p53/+ and SlJ/+ littermates, we expected the double

heterozygous mutants to have an expression ratio of 0.15 (half of 0.3). Instead,

it seems that p53 and SlJ interact to increase the expression of p53 in the

double heterozygous mutants. As a result, we hypothesized that the increase

110

Table 3.4: Expression ratio of p53 cDNA using Q-PCR

The expression of p53 was measured in two sets of animals in duplicate. The expression levels of p53 were first averaged between the duplicate experiments to calculate the average expression ratio for each set of mice. The ratio for the expression of p53 was generated by dividing the number of copies of p53 in each genotype by the copy number of the wild-type animal to normalize the data. The ratio of the two set of animals were then averaged to determine the overall average p53 expression ratio.

Genotype Ratio1 Ratio2 Average Ratio +/+ 1 1 1 p53/+ 0.37 0.26 0.3 SlJ/+ 0.45 0.23 0.3 p53/+ SlJ/+ 0.66 0.42 0.5

111 of p53 transcript in the double heterozygous mutant may be from the wild-type

allele of p53, which results in an elevated level of P53 protein to activate the

programmed cell death pathway in a quantitative manner to kill germ cells

destined to become TGCTs.

Western analysis was performed to test our hypothesis: if P53

expression is higher in the double heterozygous mutants in comparison to the p53/+ and SlJ/+ littermates. Preliminary results show the expression of P53

protein is similar between the p53/+ and SlJ/+ littermates and lower than the wild-type littermates, as expected for p53/+ but not for SlJ/+ littermates.

Interestingly, the double heterozygotes have a slightly higher P53 expression

than p53/+ and SlJ/+ littermates but still lower than the wild-type littermates

(Figure 3.1). If the data from the western blot are correct, they correspond

closely to the Q-PCR analysis that the amount of p53 transcript is lower in the

heterozygous littermates but an interaction between p53 and SlJ result in higher

P53 expression in the double heterozygous mutants. However, to accurately

measure the expression of P53 levels, we need to perform quantitative western

analysis using a series of dilutions to accurately detect the expression levels in

each animal.

Q-PCR and western analysis need to be repeated to investigate the

expression of Mgf to test whether the dosage of p53 alters the expression of

Mgf in double heterozygous mutants. In addition, the expression analysis needs to be repeated to examine expression profile of p53 during embryonic development of the double heterozygous mutants. We detected the expression

112

/+ ol ol r J r l t

n o ont + S + c c

/+ J +

-ve Sl +/ p53/ p53/ +ve

p53

Figure 3.1: Western analysis of P53 in 4-week old testis.

20ug of protein extracted from 4-week old testis was loaded on a 12% Tris-cycline gel to determine the expression of P53 between the different littermates.

113 of p53 in 4-week old testis which may be too late in development because the

tumor phenotype has already been rescued. Instead, I propose to look at the

expression of p53 during the critical time point 1) after the PGCs arrive at the genital ridges, 2) after PGCs have entered G1 mitotic arrest, and 3) in

newborns. This may give us a better idea as to how the double heterozygous

mutants suppress the development of TGCTs.

p53/p53 SlJ/+ mutants

It is reported that p53 homozygous mutants have an increased TGCT

incidence of 35% (HARVEY et al. 1993a). We therefore wanted to examine whether the interaction between p53 and SlJ suppresses the TGCT frequency of the p53/p53 SlJ/+ mutants. The tumor frequency of the p53/p53 SlJ/+

mutants was not significantly different than the expected additive tumor

frequency but the tumor frequency had a similar trend as the double

heterozygotes in the heterozygous cross to be lower than the expected additive

rate. We need to increase the sample size to test if p53/p53 SlJ/+ mutants have

a significantly lower TGCT frequency similar to the double heterozygous mutant

in the double heterozygous cross or if the p53/p53 SlJ/+ mutants have the

expect additive rate which indicates that dosage effects must play an important

role in the interaction between p53 and SlJ to suppresses TGCTs susceptibility.

In the cross between p53/+ SlJ/+ and p53/+, the double heterozygous littermates have a tumor frequency that is additive and did not display the reduced development of TGCT as observed in the heterozygous cross. An

114 explanation for the discrepancy between the interaction crosses was that

double heterozygous parents were used. Control crosses between p53/+ SlJ/+

and 129/Sv mice were generated in both parental directions to test whether the

genetics and gender of the parents contribute to TGCT frequency. The tumor

frequency of the double heterozygous mutants of the control cross is not

significantly different than the expected additive rate, demonstrating that the

genetics of the parent exert an effect on the suppression of TGCTs in the

double heterozygous mutants. It is interesting that double heterozygous

mutants of the heterozygous cross have the ability to suppress the development of TGCTs whereas, when the double heterozygous mutant is crossed to a single-mutant, the double heterozygous progeny which is genetically identical to the double heterozygous mutant of the heterozygous cross does not suppress the development of TGCTs instead it has the expected additive tumor frequency. From control crosses, it appears that the genetics of the double heterozygous parent in the cross between p53/+ SlJ/+ and p53/+ may contribute

to the expected additive tumor frequency observed in the double heterozygous

mutants. Further characterization of the double heterozygous mutant is needed

to unravel how it suppresses the development of TGCTs and also test what and

how parental effects of double heterozygous parents affect the TGCT

susceptibility of its offspring.

115 The lower than expected TGCT frequencies in mice with partial deficiencies of P53 and MGF-SLJ suggest either that these genes complement each other to restore normal functionality in TGCT stem cells or that together these genes activate mechanisms that suppress incipient TGCTs. Together these results provide clues to the genetic and molecular basis for susceptibility to TGCTs in mice and perhaps in human.

116

CHAPTER 4

SEQUENCE ANALYSIS OF THE HUMAN ORTHOLOG OF A MOUSE TGCT SUSCEPTIBILITY GENE

117 Authors: Man-Yee Josephine Lam and Joseph H. Nadeau

ABSTRACT

A mutation in the Ter gene causes primordial germ cell (PGC) deficiency

in most inbred mouse strains and significantly increases susceptibility to

testicular germ cell tumors (TGCTs) on the 129/Sv inbred genetic background.

Positional cloning of Ter identified a point mutation that introduced a termination

codon in Dnd1, which has a RNA recognition motif. Our studies showed that

inactivation of Dnd1 causes progression to TGCT development. Ter/+

heterozygous mutants have a tumor frequency of 17% and Ter/Ter

homozygous mutants have a tumor frequency of 94% with a high frequency of

bilateral versus unilateral tumors. The human ortholog of Dnd1 maps to

Chromosome 5q31. Using a case-control study, we wanted to test whether

there is a mutation in the human ortholog of Dnd1 that may contribute to TGCT susceptibility in human. We sequenced the coding region of the human ortholog in 75 patients and 15 controls to determine if there are any DNA variants within the human ortholog of Dnd1 that may contribute to the susceptibility of TGCTs in human. We did not identify a mutation within the coding sequence of the human ortholog Dnd1 between the cases and controls.

118 INTRODUCTION

Ter – a spontaneous mutation that increase TGCT susceptibility in mice

Ter is a spontaneous mouse mutation that causes severe primordial germ cell (PGC) deficiency in both males and females regardless of inbred genetic background. When Ter is congenic on the 129 inbred genetic background, Ter significantly increases susceptibility to TGCTs. Ter/Ter homozygous mutants have a tumor incidence of 94% and Ter/+ heterozygous

mutants have a tumor incidence of 17% in comparison to wild-type mice, which

have a tumor incidence of ~5% (NOGUCHI and NOGUCHI 1985).

In mice, the onset of TGCTs occurs during development around

embryonic day 11.5 (E11.5) to E12 from PGC. PGCs arise during gastrulation during E7 at the primitive streak, become incorporated into the developing hind- gut endoderm, migrate through the dorsal mesentery of the gut and arrive at the genital ridges at E11.5 (MOLYNEAUX et al. 2001). During migration, PGCs

proliferate and increase from a population of 100 cells at E7 to ~25,000 at

E13.5 (Tam and Snow 1981). When the PGCs arrive at the genital ridges at

E11.5, they enter G1 mitotic arrest and remain mitotically silent until a few days

after birth (DONOVAN et al. 1998). In Ter/Ter homozygous mutants, reduction of

PGC numbers begins at E8, prior to sexual determination. At E12.5, only ~12

PGCs are found in the genital ridges of Ter/Ter homozygotes in comparison to

~20,000 PGCs in the genital ridges of wild-type embryos (NOGUCHI and

NOGUCHI 1985). As a result, male and female Ter/Ter mutants are sterile. It is hypothesized that 129-derived genes transform the small number of PGCs that

119 arrive at the genital ridges into embryonal carcinoma (EC) cells, which

proliferate into pluripotent stem cells that give rise to the tumors of various cell

types and tissues.

Ter was previously mapped to Chromosome 18 near Fgf1. In our lab, a

high resolution genetic mapping cross was performed using testis weight as a

quantitative phenotype for PGC deficiency to narrow down the critical region to

0.14cM. To identify candidate genes within the critical region, complementation

tests were performed using BAC transgenic mice. We identified two

overlapping BACs that rescued the PGC deficiency phenotype. Within the

overlapping region of the two BACs, four candidate genes were identified. The

5’ and 3’ flanking regions and exons of the four candidate genes were sequenced to identify mutations. Of all four genes, only a single base change was identified in the Dnd1 gene, which results in a premature stop codon

(Youngren et al., in press). Dnd1 cDNA transgenic mice confirmed that Dnd1 is the candidate gene for Ter. The Dnd1 cDNA partially rescued PGC deficiency in the Ter/Ter homozygous mutants, in contrast to Ter/Ter homozygous mutants without the transgene that never have any seminiferous tubules containing mature or immature sperm. The DND1 protein can not be detected in testes and the Dnd1 transcript is significantly reduced in the testes of the Ter/Ter homozygous mutants. Whole-mount in situ hybridization using a

Dnd1 specific RNA probe detected Dnd1 expression in embryonic tissues, including the hindgut and gonads during the critical period when TGCTs

120 originate. From these experiments, we identified the Ter mutation as Dnd1

gene.

The mouse Dnd1 shows sequence similarity to several proteins that

have RNA recognition motifs (RRMs) with the highest similarity to mouse

apobec-1 complementation factor (Acf). Acf encodes an RNA-binding subunit

of the RNA editing enzyme complex that converts specific cytidines to uridines

in the apoliprotein B transcript and other RNAs (MEHTA et al. 2000). Base

modifications result in the synthesis of alternative forms of the protein which

have different biological functions. The RRM in DND1 is closely related to one

of the three RRMs in the Acf protein. Several RNA binding proteins are

anomalously expressed in certain cancers but it is unclear how they are

involved in tumorigenesis.

Dnd1 is an ortholog of the dead-end (dnd) gene in zebrafish that is

required for PGC survival and migration (WEIDINGER et al. 2003). Dnd

knockdown mutants were generated by injection of dnd-specific morpholino

antisense oligonucleotides to inhibit dnd translation. The mutant shows that dnd is not required for the specification of PGCs but it causes the PGCs to be confined in the outermost cell layer of the knockdown embryos. In addition,

PGCs of dnd knockdown mutants do not migrate as individual cells as observed in the control PGCs. Instead the dnd knockdown PGCs remain in groups of cells that maintain close cell-cell contact. PGCs of dnd knockdown do not show morphological features of motile cells.

121 Dnd is conserved in other vertebrate species. ESTs and genomic

sequences that are closely related to dnd have been identified in Xenopus

laevis, chick and human. Dnd orthologs are expressed in germ plasm of

Xenopus (Hudson and Woodland 1998) and in germ cells of chicks (NOCE et al.

2001). From sequence similarity and expression pattern, it is likely that dnd plays a role in germline development in other vertebrates and it is still necessary to determine its role in TGCT susceptibility.

Human ortholog of Dnd1

The human ortholog maps to Chromosome 5q31.3. It is a small gene that consists of only 4 exons (Figure 4.1).

Association test

Genetic association studies assess correlations between genetic variants and trait differences in a population. The case-control study has been the most widely applied strategy for association studies. There are various published examples of case-control studies that revealed common variants that are associated with diseases such as breast cancer (MACPHERSON et al. 2004), prostate cancer (ZHENG et al. 2002) and rectal cancer (KRISTENSEN et al. 2004).

The advantage of the case-control study is the ease of obtaining cases and comparing to controls, as compared to traditional linkage studies that

122

5 3

Coding untranslated

Figure 4.1 Gene structure of human ortholog dnd1. Schematic of intron and exon structure of human dnd1.

123 require families with several affected individuals. Linkage analysis and genetic association rely on similar principles and assumptions. Both methods rely on the co-inheritance of adjacent DNA variants, but the basis of linkage studies is to identify haplotypes that are inherited intact over generations in families, whereas association studies rely on the retention of adjacent DNA variants over generations in historic ancestries. Disease gene regions that are identified by linkage studies will often be large which may not be informative if they encompass a large number of candidate genes. In contrast, association studies draw from historic recombination so disease regions are small and may encompass only one gene or the DNA variant may be within the candidate gene. Through generations over time, as the disease mutation is transmitted, recombination will cause the mutation to be separated from the specific alleles of its original haplotypes. Particular DNA variants can remain together on ancestral haplotypes for many generations and this type of non-random association of the disease mutation with alleles is known as linkage disequilibrium.

A particular difficulty with the case-control study is to find an optimal control population. In the case-control study design, controls are subjects who are unaffected by the disease of interest and are assumed to be genetically unrelated to cases. The selection of controls is crucial because of population admixture (a population of subgroups having different allele frequencies) and any allele frequency differences between cases and controls can appear as variants that are associated with the disease. Controls should be selected from

124 the same source population as the cases. Since genotype frequencies may

vary across ethnic groups, controls should be matched to cases according to

ethnicity to avoid population stratification bias. Ethnic diversity leading to

population stratification can have great consequences on the power of a case-

control study because the evolutionary history of haplotypes and linkage

disequilibrium patterns will be significantly different in distinct ethnic

populations. Age is a matching factor for controls because some disease risk y

may vary with age. For case-control studies, there is a need for a large patient

population because small populations provides less robust results and create a

substantial risk of associations being described by chance. The optimal control

population for association studies would be family-based controls which would

reduce the effects of population stratification. However, rarely can family-based

controls be obtained and this method would be similar to performing linkage

studies. Linkage studies to map human TGCT susceptibility genes have been

difficult due to the lack of multigenerational pedigrees with several affected

individuals, which are most informative for linkage analysis. There are a limited

number of multigenerational pedigrees with sufficient numbers of affected

individuals because treatment of testicular cancer can lead to sterility and the disease is sporadic. As a result, the case-control study is a better method to map TGCT susceptibility genes.

125 Mutation analysis

Single-nucleotide polymorphisms (SNPs) are preferred over

microstatellite markers for association studies, because SNPs are abundant in

the human genome and the accessibility of high-throughput genotyping. In our

study, Dnd1 is a small gene that consist of only 4 exons and the coding sequence of the gene is only ~1.5kb, therefore I chose to sequence all the exons instead of genotyping my cases and controls with SNPs. I sequenced the coding regions of the Dnd1 gene to test if pathogenic mutations were present in higher frequency in cases compared to controls.

For our cases, we obtained 60 adult patient samples affected with bilateral testicular germ cell tumors from Budapest, Hungary. In addition, we included 15 pediatric teratoma samples from the Cooperative Human Tissue

Network (CHTN) in our association study because the 129/Sv inbred mice are the best characterized mammalian system for studying human pediatric germ cell tumors that originate during fetal development. Therefore, we wanted to include pediatric teratoma samples to test whether the ortholog of Ter increases susceptibility to TGCTs or specifically pediatric germ cell tumors. For comparison against the cases, 15 controls were obtained from Budapest,

Hungary.

126 MATERIALS AND METHODS

Type and source of tissue

Anonymous DNA samples were obtained from patients of Dr. Jozsef Horti from the Centre of Clinical Oncology in Budapest, Hungary. A total of 60 patients diagnosed with bilateral tumors and 15 male controls were sampled. These

DNA samples were extracted from both buffy coat and tumor tissue. We also obtained 15 pediatric teratoma samples from the Cooperative Human Tissue

Network that is funded by the National Cancer Institute.

Polymerase chain reaction (PCR) and gene sequencing

The coding sequences of the human homologue of the Ter gene were sequenced in both the forward and reverse directions, covering 4 exons of approximately 161,000 base pairs. The region to be sequenced was first amplified by PCR. PCR amplification was carried out in a 96-well block MJ

Research PTC-200 thermal cycler. The reagents were 0.15ul (0.75U) Taq polymerase (Invitrogen), 2.5ul 10X PCR buffer (magnesium free), 0.3ul 10mM dNTPs, 1ul 25mM magnesium chloride, 0.2ul of each primer (0.1uM), 1ul DNA

(25ng), and 19.35ul dH2O in a final volume of 25ul. PCR conditions were as follows: initial denaturation step for 94°C for 2 min followed by 35 cycles of 94°C for 1 min, 58°C for 1 min, 72°C for 1 min, final extension of 72°C for 5 min and then 4°C for 15 min. PCR products were resolved on a 2% agarose gel and visualized with ethidium bromide. Exon 3 is a GC-rich region and the PCR for the exon was optimized by adding 5ul of Betaine in each PCR (HENKE et al.

127 1997). 8ul of PCR product was digested with 1.5ul of dH2O, 0.2ul of exo1 and

0.3ul of SAP. The SAP/exo1 reaction was digested for 5 minutes at 37°C and

inactivated at 72°C for 15 minutes. The digested PCR product was sequenced

with dRhodamine Terminator Cycle Sequencing Ready Reaction (PE

Biosystems) by ABI 3100, using the standard protocol.

The following primers were used to sequence the 4 exons:

Exon 1 and 2: TER1F 5’CAGAGCCAATGAGAGCTTGC 3’ TER1R 5’CGGGTAGTTCGCAGTTTCTG 3’

Exon 3: TER2F 5’ CTCAGAATT CGAAAGCGTCG 3’ TER2R 5’ CAGTATAAAGTGCTGGGCAG 3’

Exon 4: TER3AF 5’ CAGGAAATGGG AAGGCTACA 3’ TER3AR 5’ GTCAGCACAACCCAGATGAG 3’ TER3BF 5’ TGGTACCAG GTGGTGATTCC 3’ TER3BR 5’AGATCACTGGGTCAGGTTGG 3’ TER3CF 5’GGCAGCTT AGGTTTGGGTTA 3’ TER3CR 5’AAGTACTTGCCCCGTAGCAG 3’

RESULTS

Sequencing primers were generated to flank the 4 exons of the human

Dnd1. For each individual, 5 sequencing reactions that each flanked ~500bp were needed to span the entire coding region of Dnd1 for each individual. 60 patients diagnosed with bilateral tumors, 15 pediatric TGCT cases and 15 controls were all sequenced in both the forward and reverse direction using dRhodamine dye to identify any polymorphisms between the coding region of

128 the human ortholog of Dnd1. DNA sequences of the patients and controls were aligned and compared using the computer application, Consed (GORDON et al.

1998). In comparison to controls, no mutation or polymorphisms were identified within the 4 exons of the Ter gene among the 60 bilateral tumor patients and 15 pediatric patients. It does not appear that there is any pathogenic mutation within the coding sequence of the Ter gene between the case and control samples.

DISCUSSION

To date, there has not been a TGCT susceptibility gene identified in humans in part due to the lack of multigenerational pedigrees with several affected individuals which is most informative for linkage studies. As a result, mouse models can be useful tools to help understand the complex genetics of

TGCTs. The similarities in biology and pathology between mice and humans are reflected in the genomes. Over 90% of the mouse and human genome can

be partitioned into corresponding regions of conserved synteny, reflecting

segments in which the gene order in the most recent common ancestor has

been conserved in both species (WATERSTON et al. 2002). Comparison between the two genomes by sequence analysis, approximately 40% of the

human genome including coding and non-coding conserved segments can be

aligned to the mouse genomes (WATERSTON et al. 2002). For nearly every gene in the human genome, a counterpart can readily be identified in the mouse.

Genetic manipulation within the mouse to engineer genetic lesions that model

129 those observed in human genetic diseases is common. In some cases, these

manipulations yield a model that closely resembles the pathology of the

corresponding human condition. Understanding the genetic complexity of the

disease in mice will provide important clues for treating the human disease.

Extrapolating from the findings in Ter mutant mice, there may be

important implications for our understanding of the development of testicular cancer in infants and young men. We wanted to test whether the human ortholog of Dnd1 is associated for the increase susceptibility of TGCTs in humans. We performed a case-control study to test whether any DNA variants within the Dnd1 gene that is strongly associated with cases versus controls.

The advantage of the case-control study is the ease of obtaining cases and controls, as compared to traditional linkage studies that require families with several affected individuals.

Sequence analysis of Dnd1 in humans between cases and controls did not reveal a mutation in the coding sequence of Dnd1. Several explanations are possible:

1) Our sample size was too small. Sample size is critical to the success of association studies. These studies require significant numbers of cases and controls to adequately power the study of disease genetics because of the modest effects likely to be contributed by any particular disease genes. In our case-control study, we sequenced 60 adult cases and 15 controls. We need to recruit a larger sample size to test for variants with weaker effects or that are more weakly associated with disease. We also sequenced 15 pediatric cases

130 because in the mouse, Ter causes germ cell deficiency during development and it is a better mouse model for human pediatric germ cell tumors. We did not observe any DNA variants among the 15 pediatric cases, but the sample size of pediatric tumors was small. Therefore, we need to obtain a larger number of pediatric patients for a more rigorous study. In addition, the sampling strategy was not optimal. We compared pediatric cases with controls obtained from

Budapest, Hungary which are not age matched and do not account for population admixture. We need to obtain a better control population that is age matched and from the United States where the case samples were obtained for an optimal analysis. The International HapMap project is a large effort to compare the genetic sequences of different individuals to identify chromosomal regions where genetic variants are shared. Using the database of the HapMap project, we can also compare our cases against the HapMap project.

In addition, in our study we obtained our cases and controls from our collaborator in Budapest and we do not have any knowledge of the samples.

Although the samples originate from the same location, there can still be population admixture within our samples because the cases and controls could be from immigrants to Europe. As a result, our sampling strategy may have reduced the power of our study.

2) Within our cases, Dnd1 may have a low allele frequency and it may not be a mutation within our cases. Cancer susceptibility involves many genes with small effects acting in concert with environmental or other non-genetic factors. For example, 20-25% of breast cancers can be ascribed to mutations

131 in BRCA1 and BRCA2, but the majority of the excess familial risk remains

unaccounted for (Easton 1999). The remaining familial risk could be due to

high-penetrance genes that have not yet been identified, but linkage analysis

has failed to reveal significant linkage at novel loci in recent studies and a

polygenic mechanism appears more plausible (PHAROAH et al. 2002). With the recent discovery of low-penetrance genes that predispose individuals to common cancers (Houlston and Peto 2004), it is evident that risks associated with each allele may be individually modest and they likely act in combination to contribute to a substantial proportion of overall cancer incidence. Linkage analysis of TGCTs in humans revealed many weak linkages, clearly identifying

TGCTs as a polygenic trait (Bishop 1998). In parallel, mapping studies in the

129/Sv mouse model and other inbred strains illustrates similar genetic complexity of TGCT susceptibility (STEVENS 1967a). As a result, TGCT is a polygenic trait that may require multiple genes with subtle effects and Dnd1 might carry a low penetrance susceptibility allele that was not detected in our screen.

3) In our study, we only sequenced the coding exons of the Dnd1 gene.

The mutation within the Dnd1 gene could lie within the non-coding region of the gene that might be involved in regulation (enhancer effects, expression, etc.).

We can also sequence the 5’ and 3’UTR and the introns of the gene for a thorough analysis between the cases and controls. In the NCBI database of single nucleotide polymorphisms (dbSNP), there are 12 known SNPs, 11 of which are within the introns or untranslated regions. There are examples of

132 intronic SNPs that are associated with disease and they have also been shown

to be associated to affect the expression of the gene. (MCWHINNEY et al. 2003;

ZAPATA-VELANDIA et al. 2004). Therefore, we may want to perform an

association study to test whether if any of the intronic SNPs are associated with

TGCT susceptibility.

Although we did not identify a mutation within the human ortholog of

Dnd1 that is associated with TGCT, many other mutations in the mouse model

are known to increase or decrease susceptibility of TGCTs. When other TGCT

susceptibility genes are identified in mice, we can use the same approach that

has been discussed in this paper and perform candidate gene case-control

studies to determine if mutation in the orthologous human gene can predispose

to TGCTs.

133

CHAPTER 5

SUMMARY AND FUTURE DIRECTIONS

134 SUMMARY

TGCTs are the most common solid tumors affecting young to middle- aged men and they are the fifth most rapidly increasing type of cancer. The genetic control of susceptibility is complex and mutated genes that contribute to inherited risk have not yet been identified. The goal of this project was to use the 129/Sv mouse model to gain a better understanding of the genetics of

TGCT susceptibility and to extrapolate the findings in mice to help us dissect the complex trait of TGCTs in humans. However, the low penetrance and multigenic basis of spontaneous TGCTs in the 129/Sv inbred strain make it difficult to dissect the complex genetic control of susceptibility to TGCTs.

Therefore, we studied single gene mutations that are inherited as Mendelian traits to gain clues to the nature of genes and pathways that affect TGCT susceptibility. We discovered several novel interactions between susceptibility genes.

Chapters Two and Three address the following question: Do the different pairs of susceptibility genes interact with each other? In other model systems, interactions between different pairs of genes provide unique insights into developmental pathways and cancer susceptibility. We performed pairwise interaction tests with various TGCT susceptibility genes to evaluate their effect on tumor frequencies, laterality, and parental effects to reveal more about the molecular identity and developmental pathways leading to tumorigenesis.

135 Chapter Two described pairwise interaction tests among Ter, p53, M19,

SlJ and Ay. The tumor frequency of the double heterozygotes, Ter/+ p53/+, were similar to the expected additive tumor frequency suggesting that each mutation contributed independently and that this pair of susceptibility genes, Ter and p53, do not act in the same pathway. By contrast, three pairs of genes,

M19 and SlJ, Ter and SlJ and separately Ter and M19, interacted to increase susceptibility of TGCTs in double mutant mice. All the double heterozygous mutants interacted to increase the susceptibility of bilateral tumors (Table 2.8).

Contrary to most other mutants, the increased incidence of bilateral tumors in the double heterozygous mutants did not result from coincidental co-occurrence of unilateral tumors, instead bilateral tumors of the double heterozygous mutants are under distinct genetic control from unilateral tumors.

To localize the region of MOLF-Chr19 that the susceptibility genes interact with, the Ter/+ and SlJ/+ single-mutants were each crossed to two congenic strains generated from the 129-Chr19M to further dissect the

interaction with M19. But we were unable to localize the region on MOLF-

Chr19 that interacts with Ter and SlJ. As many as five TGCT genes are widely

distributed on MOLF-Chr19, some with additive effects and others act

epistatically (YOUNGREN et al. 2003). The increased susceptibility to bilateral tumors may require at least three factors, either of the susceptibility genes (Ter or SlJ) together with both the proximal and distal portions of MOLF-Chr19.

M19 also interacts with Ay, significantly suppressing the development of

TGCTs. 129/Sv mice that are heterozygous for the Ay mutation have an 8-fold

136 reduction in TGCTs (STEVENS 1967a). However, the low tumor frequency of 1-

5% in the 129/Sv inbred strain makes it difficult to confirm Stevens’ observation.

An alternative approach, we crossed the Ay/+ mice to the 129-Chr19M (because

of its high tumor frequency) to more readily detect a stronger effect with a

smaller number of mice. 129-Chr19M with the Ay mutation have ~50% lower

tumor incidence in comparison to their 129-Chr19M littermates.

Several models might explain these results. M19 might act as the initiator in the three interactions and these target genes then act in different pathways to modulate TGCT susceptibility (Figure 2.4A). Alternatively, the three susceptibility genes act in different pathways but they all converge at M19 which is the target they act upon (Figure 2.4B).

Interestingly, Ter, SlJ and Ay appear to be genes involved in various aspects of RNA biology and it has been shown that RNA biology plays a role in

PGC lineage. We need to further characterize M19 in order to learn more about the genetic and molecular basis for the interaction among the pairs of susceptibility genes. Is it possible that M19 may also involve RNA related genes? How does RNA biology contribute to TGCT susceptibility?

Chapter Three is devoted exclusively to the interaction tests between p53 and SlJ. In the double heterozygous cross, the tumor frequency of p53/+

SlJ/+ males was 2-fold lower than the expected additive rate. p53/+ heterozygous mutants are reported to rarely develop TGCT, whereas p53/p53 homozygous mutants were reported to have a tumor frequency of 35% (HARVEY

137 et al. 1993a). Therefore we wanted to test whether p53 show stronger

suppression because of p53 homozygosity interacts in p53/p53 SlJ/+ mutants.

In the interaction cross between p53/+ SlJ/+ and p53/+, p53/p53 SlJ/+ mutants

have the expected additive tumor frequency and did not suppress the

development of TGCTs as observed in the p53/+ SlJ/+ heterozygous mutants.

This suggests that the interaction has a dosage effect or that heterozygosity is critical. Although not significant, the tumor frequency of p53/p53 SlJ/+ mutants

was lower than the expected additive tumor frequency, thus showing a similar

trend as the double heterozygous mutants in the heterozygous crosses. A

larger sample size is needed to obtain a better estimate of TGCT frequency.

Interestingly, the p53/+ SlJ/+ littermates in the cross between p53/+ SlJ/+ and

p53/+ did not have the low affected rate that was observed in the double

heterozygous cross and they also displayed a weak parental effect. To test

whether the genotype of the parent of the cross between p53/+ SlJ/+ and p53/+ resulted in the additive tumor frequency in the double heterozygous mutants, a control cross between p53/+ SlJ/+ and wild-type 129/Sv in both parental

directions was made. Parental effects were not found in the control cross and

the pooled tumor frequency of the double heterozygous mutants of the control

cross is similar to the expected additive tumor frequency. The control cross

demonstrated that the genotype of the parent contributed significantly to the

double heterozygous mutants’ inability to suppress the development of TGCTs.

We next tested whether dosage effects of p53 contributed to TGCT

susceptibility in the double heterozygous mutants. Preliminary results indicate

138 that the p53 transcript was higher in the testis of 4-week old double

heterozygous mutants than in the p53/+ and SlJ/+ single-mutant littermates.

We also expected that expression of p53 in the SlJ/+ littermate to be similar to

the wild-type littermates. But the apparently reduced expression of p53 in SlJ/+

heterozygotes may illustrate a functional relationship between p53 and MGF-

KIT signaling pathway. Western analysis also showed that the expression of

P53 protein may be higher in the double heterozygous mutant in comparison to the single-mutant littermate. Further analysis of the expression of p53 and Mgf

at different time points in genital ridge development in single and double

heterozygous mutants is needed to understand the complex interaction

between p53 and SlJ that suppresses the development of TGCTs.

Chapter Four describes the mutation detection test to identify any pathogenic mutation in human ortholog of Dnd1 that is associated with the susceptibility of TGCTs in human. We recently identified a point mutation in the

Dnd1 gene of Ter mice, which results in a premature stop codon. To test whether mutations in Dnd1 in humans are associated with susceptibility of

TGCT, we performed a mutation detection test. We sequenced the exons of the human ortholog of Dnd1 and did not find any DNA variants between our cases and controls. We want to increase our sample size and repeat our screen. Additionally, Dnd1 may have a low allele frequency and it may not be a common mutation within our population. As a result, we should also study other populations.

139

FUTURE DIRECTIONS

Does the Ay mutation suppress the development of TGCTs in other TGCT

susceptible mutants?

We observed a significant decrease in TGCT frequency of 129-Chr19M

CSS males with the Ay mutation in comparison to the CSS littermates without

the mutation. We need to generate additional pairwise interaction crosses to

test whether the Ay mutation can reduce the TGCT susceptibility of other mutation such as SlJ, Ter and p53.

How does the Ay mutation suppress the development of TGCTs?

L.C. Stevens reported that the 129-Ay congenic mice have an 8-fold

reduction in TGCT frequency in comparison to the wild-type littermates. To

detect a stronger effect, we tested if the Ay mutation suppresses the

development of TGCTs on the 129-Chr19M CSS which has a high tumor

frequency of 80%. CSS males with the Ay mutation have a ~50% reduction in tumor frequency compared to the CSS without the mutation. The reduced tumor frequency might be the result of the ubiquitous expression of agouti or the deletion of Raly or Eif2s2. We generated a 129-Chr19M CSS with the Avy

mutation to test whether agouti or Raly/Eif2s2 suppresses the development of

TGCTs. In the Avy mice, there is ubiquitous expression of the agouti gene due to an intracisternal A particle (IAP) insertion in pseudoexon 1A of agouti and

140 both Raly and Eif2s2 are present. Preliminary data suggest that 129-Chr19M

CSS with the Avy mutation have similar tumor frequencies as the 129-Chr19M

CSS without the Avy mutation (Table 5.1). This indicates that the suppression of

TGCT in the 129-Chr19M CSS with the Ay mutation is the result of the deletion

of Raly or Eif2s2 and not the ubiquitous expression of the agouti gene.

To test if the deletion of Raly or Eif2s2 reduces susceptibility to TGCTs,

we can generate bacterial artificial chromosome (BAC) transgenic mice by

selecting BACs that one span for Raly and one for Eif2s2. The different lines of transgenic mice could be crossed to the 129-Chr19M CSS to identify which BAC

restores the high TGCT frequency in the 129-Chr19M CSS. It would also be

necessary to generate Raly and Eif2s2 knock-out mice to test whether it is the

deletion of the genes and not the intragenic sequence that causes the

suppression of TGCTs in the 129-Chr19M.

Do PGCs of TGCT susceptibility strains enter G1 mitotic arrest at E13.5?

It has been established by L.C. Stevens that the onset of TGCTs occurs at E11.5 to E12 and the cell of origin is PGCs. PGCs first arise during gastrulation around E7 near the base allantois, and then they migrate through the mesentery gut toward the genital ridges. As the PGCs migrate, they proliferate and arrive at the genital ridges at E11.5. Male PGCs enter mitotic

141

Table 5.1: Effects of Avy on TGCT frequency in the 129-Chr19M

Overall tumor frequency of Avy/+ M19/+ and littermates controls Genotype Sample size Total affected animals Avy/+ M19/M19 22 86% (19) +/+ M19/M19 19 95% (18)

142 arrest at E13.5 and remain mitotically arrested until after birth. It is believed

that PGCs that do not enter mitotic arrest at E13.5, continue to divide into

pluripotent stem cells and form tumors of various cells and tissues.

To test this hypothesis, we can use different strains such as C57BL/6

(not TGCT susceptible), 129/Sv (TGCT susceptible), and 129-Chr19M (highly susceptible) to test whether PGCs of the different strains enter G1 mitotic arrest

at E13.5. I propose to test embryos at E13, E14 and E15 to determine when

the PGCs enter G1 mitotic arrest in the different strains. Timed-matings are

necessary to determine the exact age of the embryos. Anti-BrdU will be

injected in the pregnant female mice an hour before dissection. Genital ridges

of the embryos will be dissected, fixed, embedded, sectioned and stained to

determine if any brown-black BrdU-positive nuclei are observed in genital ridges

of the TGCT susceptibility strain at E14. In addition, using genital ridges from

embryos after E14, we can determine when the PGCs enter mitotic arrest.

If there are BrdU staining differences between the TGCT susceptible

strains, 129/Sv and 129-Chr19M, and the TGCT resistant strain, C57BL/6, we can determine which gene is responsible for this phenomena. For our anti-

BrdU experiment, we only need to detect staining differences using one of the two genital ridges of the mouse, therefore the other genital ridge can be used to extract RNA for Q-PCR to compare the expression differences of cell-cycle genes between the TGCT susceptible and non-susceptible strains.

Alternatively, we can perform microarray analysis to test many genes and to identify differences between the TGCT susceptible and non-

143 susceptible strain. However, Q-PCR would be an easier and less expensive

experiment and can be used as an initial screen before performing a more

thorough analysis using the microarray chip.

Ter/Ter mutants are PGC deficient. Are the PGCs not migrating towards

the genital ridges or are the PGCs undergoing cell death during

development?

The Ter mutation causes PGC deficiency at around E8 and PGC

numbers remain reduced throughout the remainder of the PGC migratory and

proliferative stages (SAKURAI et al. 1994). Are the PGCs not proliferating and

migrating properly towards the genital ridges, or do the PGCs undergo cell

death? It is difficult to test whether PGCs proliferate during their migratory path towards the genital ridges. We can not utilize the live embryo imaging technique described by Molyneaux at el. (MOLYNEAUX et al. 2001) because it is difficult to track the migration path of single germ cells. The low resolution of the movie makes it difficult to determine whether a germ cell truly divided or another germ cell from another plane happened to come in the plane and the cultured slices of the living embryo can only subsist for a maximum of 12 hours which allows PGCs to proceed through one cycle of division. Instead, we can perform Tunnel assays on embryo slices starting at E9 to E11.5 to examine if there are more Tunnel positive PGCs in the Ter/Ter embryos in comparison to the wild-type embryos to test whether PGCs deficiency in the Ter/Ter embryos

144 is due to cell death. We can also perform BrdU staining on embryo slices to

test whether the PGCs of the Ter/Ter embryos undergo cell divisions.

Identify modifiers on the 129/Sv inbred mice that affect TGCT susceptibility

The 129/Sv strain is the only inbred strain that spontaneously develops

TGCTs at an appreciable frequency of 1-5%. Ter, Ay, SlJ, p53, PTEN and M19

are all mutations known to affect TGCT susceptibility. These mutations must be

congenic on the 129/Sv inbred genetic background to exert their influence on

TGCT frequency, demonstrating that these mutant genes act together with 129-

derived genes to control susceptibility. We can map the modifiers on the

129/Sv strain that affect TGCT susceptibility using a panel of CSS strains. We can generate two genome panels of CSSs (19 autosomal chromosomes, two sex chromosome and a mitochondrial line) using 129/Sv and C57BL/6 strains.

Examining each of the CSSs, we can test which chromosomes of the 129/Sv strain have a modifier for TGCT susceptibility. For example, if B6-Chr2129 has

TGCTs, we can conclude that there is a modifier on chromosome 2 that affects

TGCT susceptibility because C57BL/6 mice do not develop TGCTs. We can generate congenic strains from the CSS to map the modifiers that affect TGCT susceptibility.

145 CONCLUSIONS

Mouse models as experimental systems have made many contributions to disease genes discovery and gene function studies. Studies on mouse

TGCTs pioneered by L.C. Stevens established an important experimental system. The mouse model provides a unique opportunity to dissect the genetic predisposition to TGCTs. Discovering the molecular identity of the different

TGCT susceptibility genes will provide important clues about the genes and pathways that modulate TGCT susceptibility. Genetic interactions between different mutants can be used to evaluate how different genes and pathways work together to cause tumor formation. The questions regarding the control of cell cycle in PGC, as well as the regulation of pluripotency and differentiation of germ cells still needs to be answered. These studies can help guide future studies in human, and may provide better diagnostic markers and targets for the treatment of TGCTs in human.

146

APPENDIX 1

147 Appendix 1: Estimating the expected frequencies of unilateral

and bilateral TGCTs

Let f(L) be the frequency of left testes with a TGCT, regardless of the presence of a TGCT in the right testis, or (number left tumors) / n.

Let

1 - f(L) be the frequency of left testes that did not have a TGCT, regardless of the presence of a TGCT in the right testis.

Let f(R) be the frequency of right testes with a TGCT, regardless of presence of a

TGCT in the left testis.

Let 1 – f(R) be the frequency of right testes that did not have a TGCT, regardless of the presence of a TGCT in the left testis.

The expected frequencies of mice with unilateral-left, unilateral-right, bilateral and unaffected testes is [f(L) + (1 – f(L)][f(R) + (1 – f(R)] = 1.

The frequency of mice that did not have a TGCT (f(NT)) is the product of the frequency of mice that did not have TGCT in the left testis [1 – f(L)] and the frequency of mice that did not have a TGCT in the right testis, i.e. [1 – f(R)], or f(NT) = [1 – f(L)] x [1 - f(R)].

148

The frequency of mice with a TGCT in the left but not the right testis (f(UL)) is f(UL) = f(L) x [1 – f(R)].

The frequency of mice with a TGCT in the right but not the left in (f(UR)) is f(UR) = [1 – f(L)] x f(R).

The frequency of mice with bilateral TGCTs (f(B)) is f(B) = f(L) x f(R).

149

BIBLOGRAPHY

150 ABRAHAMSON, J. L., J. M. LEE and A. BERNSTEIN, 1995 Regulation of p53- mediated apoptosis and cell cycle arrest by Steel factor. Mol Cell Biol 15: 6953-6960.

ANDERSON, D. M., S. D. LYMAN, A. BAIRD, J. M. WIGNALL, J. EISENMAN et al., 1990 Molecular cloning of mast cell growth factor, a hematopoietin that is active in both membrane bound and soluble forms. Cell 63: 235-243.

ASADA, Y., D. S. VARNUM, W. N. FRANKEL and J. H. NADEAU, 1994 A mutation in the Ter gene causing increased susceptibility to testicular teratomas maps to mouse chromosome 18. Nat Genet 6: 363-368.

BARTEL, D. P., 2004 MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116: 281-297.

BEDELL, M. A., L. S. CLEVELAND, T. N. O'SULLIVAN, N. G. COPELAND and N. A. JENKINS, 1996 Deletion and interallelic complementation analysis of Steel mutant mice. Genetics 142: 935-944.

BISHOP, D. T., 1998 Candidate regions for testicular cancer susceptibility genes. The International Testicular Cancer Linkage Consortium. APMIS 106: 64-70; discussion 71-62.

BOSL, G. J., and R. J. MOTZER, 1997 Testicular germ-cell cancer. N Engl J Med 337: 242-253.

BUETOW, S. A., 1995 Epidemiology of testicular cancer. Epidemiol Rev 17: 433- 449.

BULTMAN, S. J., E. J. MICHAUD and R. P. WOYCHIK, 1992 Molecular characterization of the mouse agouti locus. Cell 71: 1195-1204.

CANTLEY, L. C., and B. G. NEEL, 1999 New insights into tumor suppression: PTEN suppresses tumor formation by restraining the phosphoinositide 3- kinase/AKT pathway. Proc Natl Acad Sci U S A 96: 4240-4245.

CARPENTER, K., and J. MAYER, 1976 Am. J. Physiol 193: 499-504.

151 CHABOT, B., D. A. STEPHENSON, V. M. CHAPMAN, P. BESMER and A. BERNSTEIN, 1988 The proto-oncogene c-kit encoding a transmembrane tyrosine kinase receptor maps to the mouse W locus. Nature 335: 88-89.

COLLIN, G. B., Y. ASADA, D. S. VARNUM and J. H. NADEAU, 1996 DNA pooling as a quick method for finding candidate linkages in multigenic trait analysis: an example involving susceptibility to germ cell tumors. Mamm Genome 7: 68-70.

CONE, R. D., D. LU, S. KOPPULA, D. I. VAGE, H. KLUNGLAND et al., 1996 The melanocortin receptors: agonists, antagonists, and the hormonal control of pigmentation. Recent Prog Horm Res 51: 287-317; discussion 318.

COPELAND, N. G., D. J. GILBERT, B. C. CHO, P. J. DONOVAN, N. A. JENKINS et al., 1990 Mast cell growth factor maps near the steel locus on mouse chromosome 10 and is deleted in a number of steel alleles. Cell 63: 175- 183.

COUCOUVANIS, E. C., SHERWOOD, S.W., CARSWELL-CRUMPTON, C., SPACK, E. G., JONES, P. P., 1993 Evidence that the mechanism of prenatal germ cell death in the mouse is apoptosis. Exp. Cell Res. 209: 238-247.

CRITTENDEN, S. L., D. S. BERNSTEIN, J. L. BACHORIK, B. E. THOMPSON, M. GALLEGOS et al., 2002 A conserved RNA-binding protein controls germline stem cells in Caenorhabditis elegans. Nature 417: 660-663.

CUENOT, L., 1902 La loi de Mendel et l'hereditie de la pigmentation chez la souris. Arch Zool Exp Gen 10: 26-30.

DI CRISTOFANO, A., and P. P. PANDOLFI, 2000 The multiple roles of PTEN in tumor suppression. Cell 100: 387-390.

DICKERSON, G., and J. GOWEN, 1947 Science 105: 496-498.

DOLCI, S., D. E. WILLIAMS, M. K. ERNST, J. L. RESNICK, C. I. BRANNAN et al., 1991 Requirement for mast cell growth factor for primordial germ cell survival in culture. Nature 352: 809-811.

152 DONEHOWER, L. A., M. HARVEY, B. L. SLAGLE, M. J. MCARTHUR, C. A. MONTGOMERY, JR. et al., 1992 Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumours. Nature 356: 215-221.

DONEHOWER, L. A., M. HARVEY, H. VOGEL, M. J. MCARTHUR, C. A. MONTGOMERY, JR. et al., 1995 Effects of genetic background on tumorigenesis in p53- deficient mice. Mol Carcinog 14: 16-22.

DONOVAN, P. J., M. DE MIGUEL, L. CHENG and J. L. RESNICK, 1998 Primordial germ cells, stem cells and testicular cancer. APMIS 106: 134-141.

DUBREUIL, P., R. ROTTAPEL, A. D. REITH, L. FORRESTER and A. BERNSTEIN, 1990 The mouse W/c-kit locus. A mammalian gene that controls the development of three distinct cell lineages. Ann N Y Acad Sci 599: 58- 65.

DUHL, D. M., M. E. STEVENS, H. VRIELING, P. J. SAXON, M. W. MILLER et al., 1994 Pleiotropic effects of the mouse lethal yellow (Ay) mutation explained by deletion of a maternally expressed gene and the simultaneous production of agouti fusion RNAs. Development 120: 1695-1708.

EASTON, D. F., 1999 How many more breast cancer predisposition genes are there? Breast Cancer Res 1: 14-17.

EATON, G., and M. GREEN, 1963 Giant cell differentiation and lethality of homozygous yellow mouse embryos. Genetics 34: 155-161.

FENTON, P., and H. CHASE, 1951 Proc Soc Exp Biol 77: 420-422.

FLEISCHHACKER, M., T. STROHMEYER, Y. IMAI, D. J. SLAMON and H. P. KOEFFLER, 1994 Mutations of the p53 gene are not detectable in human testicular tumors. Mod Pathol 7: 435-439.

FORMAN, D., R. T. OLIVER, A. R. BRETT, S. G. MARSH, J. H. MOSES et al., 1992 Familial testicular cancer: a report of the UK family register, estimation of risk and an HLA class 1 sib-pair analysis. Br J Cancer 65: 255-262.

153 FRIEDMAN, P. N., X. CHEN, J. BARGONETTI and C. PRIVES, 1993 The p53 protein is an unusually shaped tetramer that binds directly to DNA. Proc Natl Acad Sci U S A 90: 3319-3323.

GEBAUER, F., and M. W. HENTZE, 2004 Molecular mechanisms of translational control. Nat Rev Mol Cell Biol 5: 827-835.

GINSBURG, M., M. H. SNOW and A. MCLAREN, 1990 Primordial germ cells in the mouse embryo during gastrulation. Development 110: 521-528.

GORDON, D., C. ABAJIAN and P. GREEN, 1998 Consed: a graphical tool for sequence finishing. Genome Res 8: 195-202.

HARTLEY, A. L., J. M. BIRCH, A. M. KELSEY, H. B. MARSDEN, M. HARRIS et al., 1989 Are germ cell tumors part of the Li-Fraumeni cancer family syndrome? Cancer Genet Cytogenet 42: 221-226.

HARVEY, M., M. J. MCARTHUR, C. A. MONTGOMERY, JR., A. BRADLEY and L. A. DONEHOWER, 1993a Genetic background alters the spectrum of tumors that develop in p53- deficient mice. Faseb J 7: 938-943.

HARVEY, M., A. T. SANDS, R. S. WEISS, M. E. HEGI, R. W. WISEMAN et al., 1993b In vitro growth characteristics of embryo fibroblasts isolated from p53- deficient mice. Oncogene 8: 2457-2467.

HENKE, W., K. HERDEL, K. JUNG, D. SCHNORR and S. A. LOENING, 1997 Betaine improves the PCR amplification of GC-rich DNA sequences. Nucleic Acids Res 25: 3957-3958.

HOUBAVIY, H. B., M. F. MURRAY and P. A. SHARP, 2003 Embryonic stem cell- specific MicroRNAs. Dev Cell 5: 351-358.

HOULSTON, R. S., and J. PETO, 2004 The search for low-penetrance cancer susceptibility alleles. Oncogene 23: 6471-6476.

HUDSON, C., and H. R. WOODLAND, 1998 Xpat, a gene expressed specifically in germ plasm and primordial germ cells of Xenopus laevis. Mech Dev 73: 159-168.

154 JACKS, T., L. REMINGTON, B. O. WILLIAMS, E. M. SCHMITT, S. HALACHMI et al., 1994 Tumor spectrum analysis in p53-mutant mice. Curr Biol 4: 1-7.

JIANG, L. I., and J. H. NADEAU, 2001 129/Sv mice--a model system for studying germ cell biology and testicular cancer. Mamm Genome 12: 89-94.

JORDAN, S. A., R. M. SPEED, F. BERNEX and I. J. JACKSON, 1999 Deficiency of Trp53 rescues the male fertility defects of Kit(W-v) mice but has no effect on the survival of melanocytes and mast cells. Dev Biol 215: 78-90.

KHREBTUKOVA, I., A. KUKLIN, R. P. WOYCHIK and E. J. MICHAUD, 1999 Alternative processing of the human and mouse raly genes(1). Biochim Biophys Acta 1447: 107-112.

KIMURA, T., A. SUZUKI, Y. FUJITA, K. YOMOGIDA, H. LOMELI et al., 2003 Conditional loss of PTEN leads to testicular teratoma and enhances embryonic germ cell production. Development 130: 1691-1700.

KNAUT, H., H. STEINBEISSER, H. SCHWARZ and C. NUSSLEIN-VOLHARD, 2002 An evolutionary conserved region in the vasa 3'UTR targets RNA translation to the germ cells in the zebrafish. Curr Biol 12: 454-466.

KRISTENSEN, A. T., J. BJORHEIM, J. WIIG, K. E. GIERCKSKY and P. O. EKSTROM, 2004 DNA variants in the ATM gene are not associated with sporadic rectal cancer in a Norwegian population-based study. Int J Colorectal Dis 19: 49-54.

LAM, M. Y., K. K. YOUNGREN and J. H. NADEAU, 2004 Enhancers and suppressors of testicular cancer susceptibility in single- and double- mutant mice. Genetics 166: 925-933.

LEAHY, M. G., S. TONKS, J. H. MOSES, A. R. BRETT, R. HUDDART et al., 1995 Candidate regions for a testicular cancer susceptibility gene. Hum Mol Genet 4: 1551-1555.

LEE, J. M., and A. BERNSTEIN, 1993 p53 mutations increase resistance to ionizing radiation. Proc Natl Acad Sci U S A 90: 5742-5746.

155 LEVINE, A. J., J. MOMAND and C. A. FINLAY, 1991 The p53 tumour suppressor gene. Nature 351: 453-456.

LIN, H., 1997 The tao of stem cells in the germline. Annu Rev Genet 31: 455- 491.

LINDELOF, B., and G. EKLUND, 2001 Analysis of hereditary component of cancer by use of a familial index by site. Lancet 358: 1696-1698.

LOOIJENGA, L. H., A. J. VERKERK, M. C. DEKKER, R. J. VAN GURP, A. J. GILLIS et al., 1998 Genomic imprinting in testicular germ cell tumours. Apmis 106: 187-195; discussion 196-187.

LU, D., D. WILLARD, I. R. PATEL, S. KADWELL, L. OVERTON et al., 1994 Agouti protein is an antagonist of the melanocyte-stimulating-hormone receptor. Nature 371: 799-802.

MACPHERSON, G., C. S. HEALEY, M. D. TEARE, S. P. BALASUBRAMANIAN, M. W. REED et al., 2004 Association of a common variant of the CASP8 gene with reduced risk of breast cancer. J Natl Cancer Inst 96: 1866-1869.

MALKIN, D., F. P. LI, L. C. STRONG, J. F. FRAUMENI, JR., C. E. NELSON et al., 1990 Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250: 1233-1238.

MANOVA, K., and R. F. BACHVAROVA, 1991 Expression of c-kit encoded at the W locus of mice in developing embryonic germ cells and presumptive melanoblasts. Dev Biol 146: 312-324.

MARTINHO, R. G., P. S. KUNWAR, J. CASANOVA and R. LEHMANN, 2004 A noncoding RNA is required for the repression of RNApolII-dependent transcription in primordial germ cells. Curr Biol 14: 159-165.

MATIN, A., G. B. COLLIN, Y. ASADA, D. VARNUM and J. H. NADEAU, 1999 Susceptibility to testicular germ-cell tumours in a 129.MOLF-Chr 19 chromosome substitution strain. Nat Genet 23: 237-240.

MATIN, A., and J. H. NADEAU, 2001 Sensitized polygenic trait analysis. Trends Genet 17: 727-731.

156 MATSUI, Y., D. TOKSOZ, S. NISHIKAWA, D. WILLIAMS, K. ZSEBO et al., 1991 Effect of Steel factor and leukaemia inhibitory factor on murine primordial germ cells in culture. Nature 353: 750-752.

MCWHINNEY, S. R., G. BORU, P. K. BINKLEY, M. PECZKOWSKA, A. A. JANUSZEWICZ et al., 2003 Intronic single nucleotide polymorphisms in the RET protooncogene are associated with a subset of apparently sporadic pheochromocytoma and may modulate age of onset. J Clin Endocrinol Metab 88: 4911-4916.

MEHTA, A., M. T. KINTER, N. E. SHERMAN and D. M. DRISCOLL, 2000 Molecular cloning of apobec-1 complementation factor, a novel RNA-binding protein involved in the editing of apolipoprotein B mRNA. Mol Cell Biol 20: 1846-1854.

MICHAUD, E. J., S. J. BULTMAN, M. L. KLEBIG, M. J. VAN VUGT, L. J. STUBBS et al., 1994 A molecular model for the genetic and phenotypic characteristics of the mouse lethal yellow (Ay) mutation. Proc Natl Acad Sci U S A 91: 2562-2566.

MILLER, M. W., D. M. DUHL, H. VRIELING, S. P. CORDES, M. M. OLLMANN et al., 1993 Cloning of the mouse agouti gene predicts a secreted protein ubiquitously expressed in mice carrying the lethal yellow mutation. Genes Dev 7: 454-467.

MINTZ, B., and E. S. RUSSELL, 1957 Gene-induced embryological modifications of primordial germ cells in the mouse. J Exp Zool 134: 207-237.

MOLYNEAUX, K. A., J. STALLOCK, K. SCHAIBLE and C. WYLIE, 2001 Time-lapse analysis of living mouse germ cell migration. Dev Biol 240: 488-498.

MONTES DE OCA LUNA, R., D. S. WAGNER and G. LOZANO, 1995 Rescue of early embryonic lethality in mdm2-deficient mice by deletion of p53. Nature 378: 203-206.

MOORE, F. L., J. JARUZELSKA, M. S. FOX, J. URANO, M. T. FIRPO et al., 2003 Human Pumilio-2 is expressed in embryonic stem cells and germ cells and interacts with DAZ (Deleted in AZoospermia) and DAZ-like proteins. Proc Natl Acad Sci U S A 100: 538-543.

157 MOORE, K. J., D. A. SWING, N. G. COPELAND and N. A. JENKINS, 1990 Interaction of the murine dilute suppressor gene (dsu) with fourteen coat color mutations. Genetics 125: 421-430.

MORGAN, H. D., H. G. SUTHERLAND, D. I. MARTIN and E. WHITELAW, 1999 Epigenetic inheritance at the agouti locus in the mouse. Nat Genet 23: 314-318.

MULLER, A. J., A. K. TERESKY and A. J. LEVINE, 2000 A male germ cell tumor- susceptibility-determining locus, pgct1, identified on murine chromosome 13. Proc Natl Acad Sci U S A 97: 8421-8426.

NICHOLSON, P. W., and S. J. HARLAND, 1995 Inheritance and testicular cancer. Br J Cancer 71: 421-426.

NIGRO, J. M., S. J. BAKER, A. C. PREISINGER, J. M. JESSUP, R. HOSTETTER et al., 1989 Mutations in the p53 gene occur in diverse human tumour types. Nature 342: 705-708.

NOCE, T., S. OKAMOTO-ITO and N. TSUNEKAWA, 2001 Vasa homolog genes in mammalian germ cell development. Cell Struct Funct 26: 131-136.

NOGUCHI, T., and M. NOGUCHI, 1985 A recessive mutation (ter) causing germ cell deficiency and a high incidence of congenital testicular teratomas in 129/Sv-ter mice. J Natl Cancer Inst 75: 385-392.

OLIVER, R. T., 1990 Atrophy, hormones, genes and viruses in aetiology germ cell tumours. Cancer Surv 9: 263-286.

OREN, M., and V. ROTTER, 1999 Introduction: p53--the first twenty years. Cell Mol Life Sci 55: 9-11.

PARANT, J., A. CHAVEZ-REYES, N. A. LITTLE, W. YAN, V. REINKE et al., 2001 Rescue of embryonic lethality in Mdm4-null mice by loss of Trp53 suggests a nonoverlapping pathway with MDM2 to regulate p53. Nat Genet 29: 92-95.

158 PHAROAH, P. D., A. ANTONIOU, M. BOBROW, R. L. ZIMMERN, D. F. EASTON et al., 2002 Polygenic susceptibility to breast cancer and implications for prevention. Nat Genet 31: 33-36.

PLOCHER, T. A., and T. L. POWLEY, 1976 Effect of hypophysectomy on weight gain and body composition in the genetically obese yellow (Ay/a) mouse. Metabolism 25: 593-602.

RAJPERT-DE MEYTS, E., M. KVIST and N. E. SKAKKEBAEK, 1996 Heterogeneity of expression of immunohistochemical tumour markers in testicular carcinoma in situ: pathogenetic relevance. Virchows Arch 428: 133-139.

RAPLEY, E. A., G. P. CROCKFORD, D. TEARE, P. BIGGS, S. SEAL et al., 2000 Localization to Xq27 of a susceptibility gene for testicular germ-cell tumours. Nat Genet 24: 197-200.

RAY, B., S. I. HAJDU and W. F. WHITMORE, JR., 1974 Proceedings: Distribution of retroperitoneal lymph node metastases in testicular germinal tumors. Cancer 33: 340-348.

RODRIGUEZ, E., S. MATHEW, V. REUTER, D. H. ILSON, G. J. BOSL et al., 1992 Cytogenetic analysis of 124 prospectively ascertained male germ cell tumors. Cancer Res 52: 2285-2291.

ROOSEN-RUNGE, E. C., and J. LEIK, 1968 Gonocyte degeneration in the postnatal male rat. Am J Anat 122: 275-299.

RUCKER, E. B., 3RD, P. DIERISSEAU, K. U. WAGNER, L. GARRETT, A. WYNSHAW- BORIS et al., 2000 Bcl-x and Bax regulate mouse primordial germ cell survival and apoptosis during embryogenesis. Mol Endocrinol 14: 1038- 1052.

SAKURAI, T., H. KATOH, K. MORIWAKI, T. NOGUCHI and M. NOGUCHI, 1994 The ter primordial germ cell deficiency mutation maps near Grl-1 on mouse chromosome 18. Mamm Genome 5: 333-336.

SCHOTTENFELD, D., M. E. WARSHAUER, S. SHERLOCK, A. G. ZAUBER, M. LEDER et al., 1980 The epidemiology of testicular cancer in young adults. Am J Epidemiol 112: 232-246.

159 SILVERS, W. K., 1979 The Coat Colors of Mice.

SINGER, J. B., A. E. HILL, L. C. BURRAGE, K. R. OLSZENS, J. SONG et al., 2004 Genetic dissection of complex traits with chromosome substitution strains of mice. Science 304: 445-448.

SKAKKEBAEK, N. E., E. RAJPERT-DE MEYTS, N. JORGENSEN, E. CARLSEN, P. M. PETERSEN et al., 1998 Germ cell cancer and disorders of spermatogenesis: an environmental connection? Apmis 106: 3-11; discussion 12.

SRIVASTAVA, S., Z. Q. ZOU, K. PIROLLO, W. BLATTNER and E. H. CHANG, 1990 Germ-line transmission of a mutated p53 gene in a cancer-prone family with Li-Fraumeni syndrome. Nature 348: 747-749.

STALLOCK, J., K. MOLYNEAUX, K. SCHAIBLE, C. M. KNUDSON and C. WYLIE, 2003 The pro-apoptotic gene Bax is required for the death of ectopic primordial germ cells during their migration in the mouse embryo. Development 130: 6589-6597.

STEVENS, L. C., 1966 Development of resistance to teratocarcinogenesis by primordial germ cells in mice. J Natl Cancer Inst 37: 859-867.

STEVENS, L. C., 1967a The biology of teratomas. Adv Morphog 6: 1-31.

STEVENS, L. C., 1967b Origin of testicular teratomas from primordial germ cells in mice. J Natl Cancer Inst 38: 549-552.

STEVENS, L. C., 1973 A new inbred subline of mice (129-terSv) with a high incidence of spontaneous congenital testicular teratomas. J Natl Cancer Inst 50: 235-242.

STEVENS, L. C., and K. P. HUMMEL, 1957 A description of spontaneous congenital testicular teratomas in strain 129 mice. J Natl Cancer Inst 18: 719-731.

STEVENS, L. C., and J. A. MACKENSEN, 1961 Genetic and Environmental Influences on Teratocarcinogenesis in Mice. J Natl Cancer Inst 27: 443- 453.

160 STOKOE, D., 2001 Pten. Curr Biol 11: R502.

TAM, P. P., and M. H. SNOW, 1981 Proliferation and migration of primordial germ cells during compensatory growth in mouse embryos. J Embryol Exp Morphol 64: 133-147.

VAUGHN, D. J., G. A. GIGNAC and A. T. MEADOWS, 2002 Long-term medical care of testicular cancer survivors. Ann Intern Med 136: 463-470.

VENKATACHALAM, S., Y. P. SHI, S. N. JONES, H. VOGEL, A. BRADLEY et al., 1998 Retention of wild-type p53 in tumors from p53 heterozygous mice: reduction of p53 dosage can promote cancer formation. Embo J 17: 4657-4667.

VOS, A., J. W. OOSTERHUIS, B. DE JONG, J. BUIST and H. SCHRAFFORDT KOOPS, 1990 Cytogenetics of carcinoma in situ of the testis. Cancer Genet Cytogenet 46: 75-81.

WATERSTON, R. H., K. LINDBLAD-TOH, E. BIRNEY, J. ROGERS, J. F. ABRIL et al., 2002 Initial sequencing and comparative analysis of the mouse genome. Nature 420: 520-562.

WEIDINGER, G., J. STEBLER, K. SLANCHEV, K. DUMSTREI, C. WISE et al., 2003 dead end, a novel vertebrate germ plasm component, is required for zebrafish primordial germ cell migration and survival. Curr Biol 13: 1429- 1434.

WEIR, H. K., L. D. MARRETT, N. KREIGER, G. A. DARLINGTON and L. SUGAR, 2000 Pre-natal and peri-natal exposures and risk of testicular germ-cell cancer. Int J Cancer 87: 438-443.

WOLFF, G. L., 1987 Body weight and cancer. Am J Clin Nutr 45: 168-180.

WOLFF, G. L., D. W. ROBERTS and D. B. GALBRAITH, 1986 Prenatal determination of obesity, tumor susceptibility, and coat color pattern in viable yellow (Avy/a) mice. The yellow mouse syndrome. J Hered 77: 151-158.

YAMADA, K. M., and M. ARAKI, 2001 Tumor suppressor PTEN: modulator of cell signaling, growth, migration and apoptosis. J Cell Sci 114: 2375-2382.

161 YEN, T. T., A. M. GILL, L. G. FRIGERI, G. S. BARSH and G. L. WOLFF, 1994 Obesity, diabetes, and neoplasia in yellow A(vy)/- mice: ectopic expression of the agouti gene. Faseb J 8: 479-488.

YOON, H., S. LIYANARACHCHI, F. A. WRIGHT, R. DAVULURI, J. C. LOCKMAN et al., 2002 Gene expression profiling of isogenic cells with different TP53 gene dosage reveals numerous genes that are affected by TP53 dosage and identifies CSPG2 as a direct target of p53. Proc Natl Acad Sci U S A 99: 15632-15637.

YOUNGREN, K. K., J. H. NADEAU and A. MATIN, 2003 Testicular cancer susceptibility in the 129.MOLF-Chr19 mouse strain: additive effects, gene interactions and epigenetic modifications. Hum Mol Genet 12: 389- 398.

ZAPATA-VELANDIA, A., S. S. NG, R. F. BRENNAN, N. R. SIMONSEN, M. GASTANADUY et al., 2004 Association of the T allele of an intronic single nucleotide polymorphism in the colony stimulating factor 1 receptor with Crohn's disease: a case-control study. J Immune Based Ther Vaccines 2: 6.

ZHENG, S. L., B. L. CHANG, D. A. FAITH, J. R. JOHNSON, S. D. ISAACS et al., 2002 Sequence variants of alpha-methylacyl-CoA racemase are associated with prostate cancer risk. Cancer Res 62: 6485-6488.

ZHIVOTOVSKY, B., and G. KROEMER, 2004 Apoptosis and genomic instability. Nat Rev Mol Cell Biol 5: 752-762.

ZHOU, Y., and M. L. KING, 2004 Sending RNAs into the future: RNA localization and germ cell fate. IUBMB Life 56: 19-27.

ZUR HAUSEN, H., 2001 Oncogenic DNA viruses. Oncogene 20: 7820-7823.

162