GENETIC CONTROL
OF TESTICULAR GERM CELL TUMOR SUSCEPTIBILITY IN MICE
By
PHILIP DAVID ANDERSON
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
August 2009
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of
______
candidate for the ______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
ABSTRACT ...... 8
1 CHAPTER 1: BACKGROUND AND SIGNIFICANCE ...... 10
1.1 TESTICULAR CANCER IN HUMANS ...... 11
1.1.1 Effect of age on testicular cancer risk ...... 11
1.1.2 Histopathological subtypes are associated with particular age groups ...... 14
1.1.3 Accelerating rates of TGCT diagnoses throughout the First World ...... 15
1.1.4 Risk factors for TGCTs ...... 16
1.1.5 Genetic contributors to TGCTs in humans ...... 17
1.1.6 Genetics of cancer ...... 19
1.2 TESTICULAR CANCER IN MICE ...... 19
1.2.1 TGCTs in strain 129 mice ...... 20
1.2.2 PGC migration and colonization of the genital ridges ...... 21
1.2.3 The Y chromosome and its role in sex determination...... 22
1.2.4 Biology of TGCT tumorigenesis ...... 23
1.2.5 Defects in key developmental pathways may promote TGCTs ...... 24
1.2.6 TGCT susceptibility genes and modifier genes ...... 26
1.2.7 Mapping TGCT loci ...... 29
1.3 RESEARCH AIMS ...... 38
2 CHAPTER 2: TGCT SURVEY IN A RECIPROCAL B6.129 CSS PANEL AND A PARTIAL 129‐
CHRMOLF CSS PANEL ...... 39
2.1 ABSTRACT ...... 40
2.2 INTRODUCTION ...... 40
2.3 MATERIALS AND METHODS ...... 43
1
2.3.1 Backcrosses ...... 43
2.3.2 Genotyping ...... 44
2.3.3 SNP selection and primer design ...... 45
2.3.4 TGCT survey ...... 45
2.4 RESULTS ...... 46
2.4.1 TGCT occurrence in 129‐ChrMOLF backcross mice ...... 46
2.4.2 Evidence for MOLF modifiers in 129‐Chr 2MOLF and 129‐Chr 18MOLF ...... 46
2.4.3 TGCT occurrence in 129‐ChrB6 and B6‐Chr129 backcross mice ...... 51
2.4.4 The model accurately predicted a modifier(s) on MOLF‐18 ...... 52
2.4.5 High bilateral TGCT prevalence in 129‐Chr XMOLF ...... 52
2.5 DISCUSSION ...... 56
3 CHAPTER 3: THREE TGCT QTLS MAPPED TO CHROMOSOME 18 ...... 61
3.1 ABSTRACT ...... 62
3.2 INTRODUCTION ...... 62
3.3 MATERIALS AND METHODS ...... 66
3.3.1 Construction of 129‐Chr 18MOLF ...... 66
3.3.2 Genotyping ...... 67
3.3.3 Construction of congenic strains from 129‐Chr 18MOLF ...... 67
3.3.4 Identification of novel SSR markers ...... 68
3.3.5 Breakpoint mapping ...... 69
3.3.6 TGCT survey ...... 69
3.3.7 Histology...... 69
3.4 RESULTS ...... 70
3.4.1 Evidence for at least three TGCT QTLs on chromosome 18...... 72
3.4.2 Testes of 129‐Chr18MOLF and congenic mice were affected by calcification ...... 73
3.4.3 Does length of the congenic segments influence TGCT prevalence? ...... 75
2
3.5 DISCUSSION ...... 77
4 CHAPTER 4: THE ROLE OF THE MOUSE Y CHROMOSOME ON TGCT SUSCEPTIBILITY ...... 82
4.1 ABSTRACT ...... 83
4.2 INTRODUCTION ...... 83
4.3 MATERIALS AND METHODS ...... 86
4.3.1 CSSs ...... 86
4.3.2 Sox9Ods crosses...... 87
4.3.3 Genotyping ...... 87
4.3.4 TGCT survey ...... 87
4.3.5 Histology and immunohistochemistry ...... 87
4.4 RESULTS ...... 88
4.4.1 The 129/Sv Y chromosome is not necessary for TGCTs ...... 88
4.4.2 The 129/Sv Y chromosome is not sufficient for TGCTs ...... 89
4.4.3 Adult XX sex‐reversed males do not have TGCTs ...... 91
4.4.4 Neonate XY males, but not XX sex‐reversed males, have germ cells at birth ...... 93
4.4.5 Neonate XY males, but not XX sex‐reversed males, have TGCTs at birth ...... 95
4.5 DISCUSSION ...... 95
5 CHAPTER 5: SUMMARY AND FUTURE DIRECTIONS ...... 100
5.1 SUMMARY ...... 101
5.1.1 Three new CSSs with TGCT QTLs ...... 101
5.1.2 Chromosome 18 has multiple factors influencing TGCT susceptibility ...... 103
5.1.3 At least one factor on the Y chromosome is necessary for TGCTs...... 103
5.2 FUTURE DIRECTIONS ...... 104
5.2.1 Is Apobec1 the TGCT gene on B6‐6? ...... 104
5.2.2 How can the TGCT factors in TQ1 – TQ3 be identified? ...... 105
5.2.3 Do TGCTs and ES cell derivation share genetic determinants? ...... 108 3
5.2.4 Are testis calcifications the remains of aborted TGCTs ...... 109
5.2.5 Is Sry the Y‐linked factor that is necessary for TGCTs? ...... 111
5.3 CONCLUSIONS ...... 111
6 CHAPTER 6: APPENDICES ...... 113
7 CHAPTER 7: REFERENCES ...... 120
4
LIST OF TABLES
TABLE 1‐1: MODIFIER GENES AND CHROMOSOMES THAT AFFECT TGCT SUSCEPTIBILITY IN 129/SV MALES...... 27
MOLF TABLE 2‐1: TGCT DATA FROM BACKCROSS 129‐CHR MALES...... 49
B6 TABLE 2‐2: TGCT DATA FROM BACKCROSS 129‐CHR MALES ...... 54
TABLE 2‐3: THE MODEL ACCURATELY PREDICTED A MOLF MODIFIER ON MOLF‐18...... 55
TABLE 3‐1: RATES OF PARENT STRAINS AND CONGENIC STRAINS AFFECTED WITH TESTICULAR CALCIFICATIONS ...... 74
TABLE 3‐2: LENGTH IN MEGABASES (MB) OF THE CONGENIC SEGMENTS, AND THE PERCENT OF MALE ADULTS AFFECTED WITH
TGCTS...... 76
TABLE 4‐1: TGCTS IN CSS MALES WITH ALTERNATE Y CHROMOSOMES...... 90
TABLE 4‐2: TEST FOR TGCTS IN CONTROL AND SEX‐REVERSED MALES...... 90
5
LIST OF FIGURES
FIGURE 1‐1: RISK OF VARIOUS CANCERS AS A FUNCTION OF AGE ...... 13
FIGURE 1‐2: STEPS TO MAKE A CSS ...... 33
FIGURE 1‐3: STEPS TO MAKE A CONGENIC STRAIN FROM A CSS...... 35
FIGURE 3‐1: TGCT DATA FROM ADULT PARENTAL STRAINS AND CONGENIC STRAINS...... 71
FIGURE 4‐1:CROSSES PERFORMED TO PRODUCE SEX‐REVERSED AND CONTROL N2 MALES...... 92
FIGURE 4‐2: COMPARISON OF TESTES FROM NORMAL MALES AND SEX‐REVERSED MALES ...... 94
6
ACKNOWLEDGEMENTS
I would like to thank Annie Baskin and Stephanie Doerner for giving me their
129-ChrB6 and B6-Chr129 CSS males to screen for TGCTs. Those data are reported in
Chapter 2 and Appendix IV.
I would like to thank Vicki Nelson for helping to survey the congenic mice I report on in Chapter 3.
I would also like to thank Jean Kawasoe, Diane Lui, Elaine Leung, Maritza
Rosales and Vicki Nelson for helping to survey the CSS and N2 males I report on in
Chapter 4.
7
Genetic Control of Testicular Germ Cell Tumor Susceptibility in Mice
Abstract
By
PHILIP DAVID ANDERSON
Testicular germ cell tumors (TGCTs) are the most common types of testicular cancer as well as the most common tumor in men aged 15 – 34. Despite their prevalence, the genetic control of susceptibility is poorly understood. Family history suggests a strong genetic component to TGCT susceptibility, but studies have found only one rare, weakly associated Y-linked locus. The 129 family of inbred mouse strains is an established animal model of TGCTs. Linkage studies in mice have been largely unsuccessful, and no genes controlling susceptibility have been found.
The goal of my research was to identify chromosomes and chromosomal regions containing TGCT QTLs (quantitative trait loci) using chromosome substitution strains
(CSSs). My analysis showed that three CSSs have TGCT QTLs. To map the QTLs on chromosome 18, I made a panel of eight congenic strains from 129-Chr 18MOLF. I surveyed congenic males for TGCTs and found evidence for three previously unknown
TGCT QTLs on chromosome 18.
Because studies have shown conflicting evidence for a Y-linked factor(s) promoting TGCTs in humans, I tested the role of this chromosome on TGCT development in mice. Sry, the master sex-determining gene in mammals, is on the Y chromosome and is normally required for testis formation and male development. In the
8 absence of the Y chromosome, mice develop as fertile females. To bypass the requirement for Sry in sexual development, I took advantage of the Odd Sex mutation that causes XX animals to develop as males without the Y chromosome. Sex-reversed XX males develop testes with PGCs, which enabled tests of whether the Y chromosome was required for TGCT tumorigenesis. Although normal males developed TGCTs at an appreciable rate, no TGCTs were found in sex-reversed males. I concluded that at least one factor on the Y chromosome was required for TGCT initiation.
The results of my research support the theory that TGCTs are a highly complex trait, and that CSSs are an effective tool for mapping QTLs with weak effects. A more thorough understanding of the heritable component of TGCTs in mice will improve our understanding of the genetic control of susceptibility to complex diseases.
9
1 CHAPTER 1: Background and Significance
10
1.1 TESTICULAR CANCER IN HUMANS
Testicular cancer is the most common malignancy affecting young men aged 15 -
34, accounting for over 1% of all male cancers (KRAIN 1973; JEMAL et al. 2006). Over
90% of testicular cancers are testicular germ cell tumors (TGCTs) (TALERMAN 1985) and
are derived from germ cells, the pluripotent precursors of gametes. Somatic cell-derived
tumors of the testis (e.g. Leydig cell tumors, Sertoli cell tumors) are rare in comparison to
TGCTs, and approximately 90% of somatic testicular cancers are benign (KUMAR et al.
2005).
1.1.1 Effect of age on testicular cancer risk
Testicular cancer is unusual because the risk for the disease peaks at three different times throughout life, instead of generally increasing with age (NICHOLSON and
HARLAND 1995; FALLON et al. 2006) (Figure 1-1). The first peak occurs in infants less
than two years of age. Risk for TGCTs drops after infancy until approximately the age of
ten. After ten, risk for TGCTs increases, and reaches a second peak at approximately age
thirty. After thirty, risk for TGCTs decreases until approximately age seventy. Starting at age seventy, risk again increases, reaching a third peak at approximately age eighty.
After eighty, risk again decreases.
The presence of distinct peaks suggests the importance of developmental pathways on TGCT risk. For example, the TGCT risk for newborns is higher than the
TGCT risk for five-year-olds. These data may indicate that the TGCTs of early childhood are antagonized by maternal factors such as pregnancy hormones, and after birth TGCT risk decreases because the infant is removed from the source of the hormones. There is some evidence to suggest that dosage of placental hormones affects
11
TGCT risk: dizygotic (DZ) twins are at higher risk than monozygotic (MZ) twins for
TGCTs, perhaps reflecting the presence of two hormone-secreting placentas in DZ pregnancies compared to the usual one placenta in MZ pregnancies (BRAUN et al. 1995;
SWERDLOW et al. 1997).
Risk for TGCTs dramatically increases between ages ten and twenty, which is
correlated with the onset of puberty. These data may suggest the influence of pubertal
hormones on TGCT risk. Although other cancer types are thought to arise mostly
because of DNA mutations in key genes, the presence of three distinct peaks and general
decrease in risk after age thirty suggest TGCTs may be a special class of cancers that are
sensitive to developmental milestones.
12
SEER data (F cancer peaks three times throughout type are indicated by the “all Figure 1-1: Risk of various cancers ALLON et al. 2006). s” suffix. Unlike other cancer t as a function of age. Lines life. This diagram was graphed by the Candidate based on representing more ypes, the risk for testicular than one tumor
13
1.1.2 Histopathological subtypes are associated with particular age groups
Each of the three peaks in TGCT prevalence is associated with particular histopathological subtypes. In infants, the tumors are mostly yolk sac tumors, teratomas, and teratocarcinomas (NICHOLSON and HARLAND 1995; KUMAR et al. 2005). All three
subtypes are thought to develop directly from primordial germ cells (PGCs) or via an
embryonal carcinoma (EC) cell intermediate (EC cells are the malignant counterparts of
PGCs) (DAMJANOV 1984). Whereas yolk sac tumors arise by differentiation along
extraembryonic lines, teratomas result from differentiation along the lines of embryonic
germ layers- endoderm, mesoderm and ectoderm. Teratocarcinomas are a mixed tumor
type consisting of teratomas plus undifferentiated EC cells. The continued presence of
EC cells in teratocarcinomas and other TGCTs is associated with the potential for distant metastasis (MARTIN and EVANS 1974; ANDREWS 2002). Teratomas, on the other hand,
are considered benign and are associated with good prognosis in this age group (MARTIN
and EVANS 1974; ISAACS 2004).
The testicular cancers that occur in the second peak are mostly seminomas, and less commonly, mixed teratomas and yolk sac tumors with embryonal carcinoma
(NICHOLSON and HARLAND 1995; KUMAR et al. 2005). Approximately 50% of all
TGCTs are seminomas, which are the dominant TGCT type in young adults but are very
uncommon in infants (KUMAR et al. 2005). Seminomas result from differentiation of transformed germ cells into sheets of large, uniform “seminoma cells” (KUMAR et al.
2005). Teratomas have a far greater risk of harboring small EC foci in adults than in
infants (KUMAR et al. 2005). Therefore clinicians typically treat mature teratomas in the
adult as if they were malignant, reflecting the increased risk. In contrast to TGCTs in
14 infants, TGCTs in young adults are associated with carcinoma in situ (intratubular germ cell neoplasia), a transformed pre-invasive lesion. Tumor karyotype studies have shown that both carcinoma in situ and TGCTs in young adults are usually aneuploid
(OOSTERHUIS et al. 1989), commonly with deletions in 1p and 6q (BEHRMAN et al. 2004).
Approximately 80% of TGCTs in young adults exhibit isochromosome 12p (SANDBERG
et al. 1996). Gain of 12p material has not been found in carcinoma in situ (ROSENBERG et al. 2000; SUMMERSGILL et al. 2001), suggesting that gain of 12p (or loss of 12q) is
associated with progression from carcinoma in situ to invasive disease.
The testicular cancers that occur in older adult men are mostly spermatocytic
seminomas and lymphomas of the testis (NICHOLSON and HARLAND 1995; KUMAR et al.
2005). Although both spermatocytic seminomas and seminomas are derived from germ cells, spermatocytic seminomas have little in common clinically or histologically with seminomas, and represent a distinct tumor type. Unlike seminomas, spermatocytic seminomas are slow-growing, not associated with carcinoma in situ, and seldom metastasize (KUMAR et al. 2005). All of these characteristics contribute to good
prognosis.
1.1.3 Accelerating rates of TGCT diagnoses throughout the First World
Studies have shown a pattern of increasing TGCT prevalence in Europe and North
America over the last thirty years. In Denmark and Norway, which have the highest rates
of TGCT in the world, men have almost a 1% risk of developing a TGCT in their
lifetimes (SKAKKEBÆK et al. 2007). During the 1990’s, TGCT diagnoses accelerated
2.4% - 4.2% per year in every country studied in Northern Europe except Denmark
(RICHIARDI et al. 2004). In England and Wales, the age-standardized incidence rate grew
15 from 2.9 per 100,000 in 1971 to 5.4 per 100,000 in 1997 (POWER et al. 2001). In three
provinces of Canada, the incidence of seminomas and other TGCTs increased 53% and
91% respectively in the 26 years from 1970 to 1995 (LIU et al. 2000).
Increasing prevalence of TGCTs is mirrored in the USA. In the USA, the
incidence of the disease has been increasing in European-American men for at least the
past fifty years (MCKIERNAN et al. 1999; SHAH et al. 2007), and African-American and
Asian-American men for at least the past thirty five years (MCGLYNN et al. 2005; SHAH
et al. 2007). European-American men are at highest risk of the disease, with 6.36 cases
per 100,000 per year developing a tumor in 2003, a rise of 61% from 1978 (SHAH et al.
2007). Although Asian-American men have a reduced risk (2.23 affected per 100,000 per year in 2003) relative to European-American men, their risk has increased at nearly the same rate, 62.8%, since 1978 (SHAH et al. 2007). African-American men have the
lowest risk (1.30 per 100,000 per year in 2003), although their risk has increased 39.8% since 1978 (SHAH et al. 2007).
1.1.4 Risk factors for TGCTs
In addition to race, there are several other known risk factors for TGCTs. The
greatest risk factors are family history (FORMAN et al. 1992), cryptorchidism (MØLLER et
al. 1996), testicular dysgenesis (KAPLAN et al. 1981; YOUNG and SCULLY 1990), and
prior diagnosis of TGCT (COLLS et al. 1996; HOFF WANDERÅS et al. 1997; RÜTHER et al.
2000). Infertility (JACOBSEN et al. 2000), dizygotic twinship (BRAUN et al. 1995;
SWERDLOW et al. 1997), testicular atrophy (HAUGHEY et al. 1989; MØLLER et al. 1996)
and unusual sex chromosome karyotypes (YOUNG and SCULLY 1990) are also associated
with the development of TGCTs (DIECKMANN and PICHLMEIER 2004). All of these
16 factors relate to abnormal testicular development and prompted Skakkabæk and colleagues to propose the existence of a “testicular dysgenesis syndrome” that includes
TGCTs and is caused by genetic and environmental factors (SKAKKEBÆK et al. 2001).
Environmental and occupational factors have also been proposed to play a role in
TGCT risk. Much of the research in environmental risk factors has focused on the
contribution of estrogenic and anti-androgenic pollutants (endocrine disruptors) to TGCT
risk. Men engaged in the mining, chimney sweeping and food and beverage industries
are at higher risk for nonseminoma, suggesting exposure to occupational antagonists
contributes to TGCT risk (KNIGHT et al. 1996). Although there is debate about the relative importance of environmental risk factors versus genetic risk factors to TGCTs, there is little doubt that both environmental and genetic factors contribute to the disease
(FORMAN et al. 1990; HARDELL et al. 1998; RICHIARDI et al. 2002).
1.1.5 Genetic contributors to TGCTs in humans
Family history of TGCTs is among the most significant risk factors for TGCTs:
sons of affected fathers have a 4- to 6-fold increased risk of developing the disease, and
brothers of affected sibs have an 8- to 10-fold increased risk (FORMAN et al. 1992;
LINDELÖF and EKLUND 2001). Despite the strong family history, identification of
susceptibility genes has been exceedingly difficult in humans. Traditional linkage studies
have been largely unsuccessful because of the limited number of multigenerational
pedigrees containing several affected cases, the sterility that often results from treatment,
and the genetic complexity of the disease. Mapping experiments conducted on families
collected by the International Testicular Cancer Linkage Consortium have revealed only tenuous linkages. Although numerous weak linkages were reported by Bishop (BISHOP
17
1998), only one significant linkage has been reported to Xq27 (RAPLEY et al. 2000;
LUTKE HOLZIK et al. 2006). Unfortunately, later studies failed to confirm the linkage to
Xq27 (CROCKFORD et al. 2006). To date, only one factor involved in susceptibility
(gr/gr) has been described (NATHANSON et al. 2005), and it was found through a
candidate gene approach and not by unbiased linkage. Because of the difficulties in
human linkage studies, mouse models are an attractive alternative and could be used to test individual autosomes or Y-linked loci for TGCT genes.
The gr/gr TGCT factor is a 1.6 megabase deletion within the AZFc region of
Yq11 that had been associated with subfertility in hemizygous men (REPPING et al. 2003;
LYNCH et al. 2005). Because infertility is a risk factor for testicular cancer (JACOBSEN et
al. 2000), Nathanson et al. tested the prevalence of this deletion in a large, international
cohort of men with and without TGCT (NATHANSON et al. 2005). The deletion is present
in 1.3% of unaffected males. However it is present in 2% of males with TGCT who lack
a family history and 3% of males with a family history (NATHANSON et al. 2005). For
that reason, gr/gr is considered a rare, low-penetrance TGCT factor. Interestingly, men with gr/gr and a maternal family history of TGCTs are at over 4-fold greater risk than men with gr/gr and a paternal family history (NATHANSON et al. 2005). Furthermore,
gr/gr does not entirely delete any genes. Rather, it reduces the copy number of three
genes that are duplicated in multiple palindromic repeats on the Y chromosome. gr/gr
deletes two of the four DAZ genes, one of the three BPY2 genes, one of the two CDY1
genes, and several additional transcriptional elements (NATHANSON et al. 2005). All
three genes are expressed in male germ cells with functions in germ cell survival,
development and differentiation (REIJO et al. 1995; KLEIMAN et al. 2003; GINALSKI et al.
18
2004), suggesting that perturbations of these functions in germ cells promote TGCTs. Y chromosome deletions in mouse models could be used to help elucidate the role of the Y in TGCT susceptibility.
1.1.6 Genetics of cancer
Cancer is a heterogeneous disease characterized by inappropriate cell growth and proliferation. Cancers in somatic cells are partly or mostly caused by the inheritance or accumulation of mutations in two types of genes, proto-oncogenes and tumor suppressor genes. Proto-oncogenes promote cancer when inappropriately activated. Genes promoting proliferation and inhibiting apoptosis are affected by oncogenic mutations.
Mutations in proto-oncogenes usually resulting in constitutively active alleles. In contrast, tumor suppressors normally inhibit cell growth and proliferation. Mutations in tumor suppressors are usually involve loss of function of the protein. Therefore, a single functional gene copy is usually sufficient to prevent cancer initiation. For example, p53 normally promotes cell cycle arrest and apoptosis in response to DNA damage.
Inactivation of p53 prevents a damaged cell from inducing apoptosis and simultaneously allows unregulated growth, eliminating a key checkpoint for preventing the accumulation of mutations (MENDELSOHN et al., 2001).
1.2 TESTICULAR CANCER IN MICE
Mice have a closer phylogenetic relationship to humans compared to other
common model organisms (e.g. fruit flies, zebrafish, chick), which is why mice are used
to model human diseases, including testicular cancer. Mice present several advantages
over other mammalian models. Compared to other mammals, mice are physically small,
have large numbers of offspring in frequent litters, and a short generation time. A large
19 number of phenotypically and genotypically-diverse inbred strains are available that have unique sets of characteristics. These phenotypic and genotypic differences are exploited in mapping studies in which strains that vary in a particular phenotype are crossed to map genes involved in susceptibility.
1.2.1 TGCTs in strain 129 mice
Mice from the 129 family of inbred strains spontaneously develop TGCTs and 1-
10% of males have at least one TGCT by 3 weeks of age (STEVENS and HUMMEL 1957).
Males from other inbred strains are rarely affected (STEVENS and HUMMEL 1957;
STEVENS 1967a). Strain 129 mice are grouped into one of several substrains to reflect the
genetic variability within the family (SIMPSON et al. 1997; THREADGILL et al. 1997).
Most of the published experiments pertaining to mouse TGCTs were performed with
mice from the 129/Sv subline, in particular the 129S1/Sv +p +Tyr-c +Kitl-SlJ inbred strain
[129S1/SvImJ, formerly 129S3/SvImJ] (hereafter 129/Sv).
129/Sv males are affected by two different kinds of TGCTs: testicular teratomas
and teratocarcinomas. Teratomas are tumors that arise from complete differentiation of
PGCs or EC cells into cells or tissue derived from ectoderm, mesoderm or endoderm
(STEVENS and HUMMEL 1957). Teratocarcinomas are teratomas that retain some
undifferentiated PGCs or EC cells. The teratomas and teratocarcinomas of mice closely resemble the teratomas and teratocarcinomas of newborn humans (STEVENS and HUMMEL
1957; YOUNGREN et al. 2003): both originate embryonically, both lack carcinoma in situ, and both lack isochromosome 12p. In most cases, the arrangement of tissue types in
TGCTs is random. However, some normal tissue co-associations have been reported,
20 such as glandular epithelium with ducts (STEVENS 1967a), and muscle attachment to
cartilage and bone (STEVENS 1973).
1.2.2 PGC migration and colonization of the genital ridges
TGCTs in mice arise directly or indirectly from primordial germ cells (PGCs), the embryonic precursors to gametes (STEVENS 1967b). PGCs are the first embryonic cell
type to become lineage-restricted, which occurs just prior to gastrulation (embryonic day
[E]6.5 – E8.5). At E6.25, the PGC precursors are localized to the proximal posterior
epiblast as a population of approximately 6 Blimp1-positive cells adjacent to the extra- embryonic ectoderm (OHINATA et al. 2005). Blimp1 expression is important for the survival of the early PGC lineage, as few Blimp1-deficient PGCs survive past E7
(OHINATA et al. 2005). By labeling individual epiblast cells, Lawson and Hage
established that PGCs are derived from the proximal epiblast cells closest to the extra-
embryonic ectoderm (LAWSON and HAGE 1994). Reciprocal transplantations of distal
epiblast and proximal epiblast just prior to gastrulation showed that the position of the
cells within the epiblast is important for PGC determination (TAM and ZHOU 1996), and
not segregation of a maternally-inherited cytoplasmic determinant (e.g. pole plasm in
Drosophila) in the proximal epiblast. Between E7 – E7.5, PGCs reduce their rate of cell
division from once every 7 hours to once every 16-17 hours (LAWSON and HAGE 1994).
In addition, the PGCs up-regulate Stella (SATO et al. 2002) and maintain high levels of
TNAP (GINSBURG et al. 1990), Nanog (CHAMBERS et al. 2003), and Oct3/4 (SCHOLER et al. 1990; NICHOLS et al. 1998), among other genes.
By late gastrulation (E8.5), PGCs are localized to the ventral portion of the
hindgut. From there they migrate to the dorsal side of the hindgut, then through the
21 dorsal mesentery in the direction of the developing genital ridges. The genital ridge is the embryonic precursor to the gonads, kidneys, and adrenals. Throughout their migration the PGCs are dependent on the c-kit signal transduction pathway for their continued proliferation and migration: homozygous mutants in either kit receptor (Kitw/w) or kit
Sl/Sl ligand (Kitl ) have reduced germ cell numbers (COULOMBRE and RUSSEL 1954;
BENNETT 1956), and few PGCs in Kit-null homozygotes arrive at the genital ridges
(BUEHR et al. 1993). Normally, approximately 20,000 PGCs colonize each genital ridge
between E10 and E11 (TAM and SNOW 1981), while the ridges are still forming. It is
important that the PGCs arrive during this narrow window: cells in the midline down-
regulate Kitl at E10.5, and any PGCs still in the midline undergo BAX-dependent
programmed cell death (RUNYAN et al. 2006).
1.2.3 The Y chromosome and its role in sex determination
The Y chromosome is a small, gene-poor sex chromosome on which Sry, the
master sex determining gene in mammals, is located. Between E11 and E11.25, the
epithelial cell derivatives (and only epithelial cell derivatives) in the indifferent gonad express Sry (SINCLAIR et al. 1990, HIRAMATSU et al. 2009). Expression of Sry causes masculinization of the indifferent gonad and the differentiation of the supporting epithelial derivatives into Sertoli cells. In the absence of Sry, the supporting epithelial derivatives differentiate into granulosa cells. Several genes, particularly Sox9, play a crucial downstream role in the sex determination pathway (CHABOISSIER et al. 2004).
Ectopic expression of a Sox9 transgene in XX gonads is sufficient to elicit testis formation in transgenic XX mice lacking Sry and the Y chromosome (VIDAL et al. 2001),
demonstrating that neither Sry nor the Y chromosome are required for sex determination.
22
Odd Sex (Ods) is a dominant insertional mutation that causes complete female to male sex reversal in XX mice lacking the Y chromosome and Sry (BISHOP et al. 2000).
The mutation was created in the TGCT-resistant FVB inbred strain by the integration of a
Tyrosinase minigene ~1 Mb upstream of Sox9 on chromosome 11, which resulted in inappropriate Sox9 expression in XX mice (BISHOP et al. 2000). Adult sex-reversed XX
Sox9Ods/+ mice are morphologically identical to their XY littermates except in the testes,
in which the seminiferous tubules of sex-reversed mice are devoid of germ cells,
precluding spermatogenesis. Although sex-reversed mice have some germ cells in the
gonad until at least birth, complete germ cell death occurs before sexual maturation,
causing sterility. Sterility in the XX Sox9Ods/+ males is consistent with the sterility
observed with other female-to-male sex-reversing mutations (SUTCLIFFE and BURGOYNE
1989).
1.2.4 Biology of TGCT tumorigenesis
The remarkable diversity of cell and tissue types in TGCTs suggested a
pluripotent stem cell origin for the tumors (FRIEDMAN and MOORE 1946; DIXON and
MOORE 1952; MELICOW 1955). In 1967, Leroy Stevens directly tested whether PGCs
were the cells of origin for TGCTs in mice. Working in TGCT-susceptible strain 129
mice, he grafted germ cell-deficient KitlSl/Sl genital ridges and normal Kitl+/+ genital
ridges into normal Kitl+/+ adult testes. He found that wild-type grafts formed TGCTs at a
high rate, whereas germ cell-deficient grafts formed few tumors (STEVENS 1967b). In a
separate experiment, Leroy Stevens used grafting experiments to refine the timing of
TGCT tumorigenesis (STEVENS 1964). Again working in 129/Sv, he transplanted intact genital ridges from E12.5 and E13.5 fetuses into adult testes. Although grafts from E12.5
23 genital ridges formed teratomas in 82% of cases, only 8% of E13.5 grafts formed teratomas. Taken together, these experiments provided strong evidence that PGCs are the cells of origin of TGCTs, and embryonic genital ridges are competent to form teratomas by E12.5.
1.2.5 Defects in key developmental pathways may promote TGCTs
Defects in several cellular processes may explain the origin of TGCTs from
PGCs. In particular, the switch from mitotic to meiotic cell division has generated considerable interest. In C. elegans, deletion of the Notch family member GLP-1 causes premature entry of PGCs into meiosis (BERRY et al. 1997). When a dominant hyper-
morphic allele of GLP-1 was microinjected into adult hermaphrodites, PGCs failed to
enter meiosis and continued to proliferate mitotically, causing tumor-like growths
(BERRY et al. 1997). Mammals have four orthologues of GLP-1, Notch1 – Notch4.
Gene expression studies showed that Notch2 and Notch4 are consistently up-regulated in human seminomas and carcinoma in situ (ADAMAH et al. 2006). In addition to Notch family genes, members of the retinoic acid pathway are also of interest because retinoic acid promotes entry into meiosis in both males and females (CHUNG and WOLGEMUTH
2004; VERNET et al. 2006; BOWLES and KOOPMAN 2007). Retinoic acid stimulates Stra8,
whose expression is correlated with initiation of meiosis (OULAD-ABDELGHANI et al.
1996). The enzyme that degrades retinoic acid, CYP26B1, is down-regulated in PGCs of normal females and up-regulated in PGCs of normal males around E12.5. High levels of
CYP26B1 in male PGCs prevent expression of “meiosis factor” Stra8 and preclude entry into meiosis at E13.5. As a result, perturbations in the expression of Cyp26B1 in male
24
PGCs may play a role in PGC transformation by prolonging mitosis in a manner similar to hypermorphic GLP-1 in C. elegans.
TGCTs have also been hypothesized to arise from mutations in genes that are expressed during meiosis. Attention has been focused the Sycp3 gene in particular, because it maps close to an ‘insignificant’ linkage on human chromosome 12q (RAPLEY
et al. 2003; CROCKFORD et al. 2006), and because 12q is often lost in TGCT patients
(SANDBERG et al. 1996; ROSENBERG et al. 2000; SUMMERSGILL et al. 2001). SYCP3 is a
crucial component of the synaptonemal complex, a protein-DNA structure that forms
between all four sister chromatids during meiotic prophase I. Sycp3 is necessary for
-/- meiosis: Sycp3 males are sterile because of a meiotic defect during zygotene (YUAN et al. 2000). Sycp3 is normally expressed exclusively in meiotic cells. However, it is also expressed in human EC cells, carcinoma in situ, and seminomas (ADAMAH et al. 2006).
In addition, the cohesin Smc1β has attracted attention. Deletion of this gene in mice
results in severe mutant phenotypes including failure to complete meiotic recombination,
aneuploidies and aberrant telomeres (REVENKOVA et al. 2004; HODGES et al. 2005).
Aneuploidies are associated with certain human TGCT subtypes (OOSTERHUIS et al.
1989).
Anomalies in the switch from mitosis to meiosis have been postulated as a potential cause for neoplastic conversion of PGCs (ADAMAH et al. 2006). In male mice,
PGCs enter mitotic arrest at E13.5 and stay arrested until shortly after birth, when they resume proliferation (MCLAREN 1984). Instead of arresting in mitosis, PGCs in females
enter into meiosis at E13.5 (MCLAREN 1984). Co-expression of both mitosis and meiosis
genes may provide the instability that could initiate tumor formation. Because the
25 decision to enter meiosis is sex-dependent, aberrations in the sex determination pathway and Sry may be involved. The timing of Sry expression is very important for normal testis development: delayed Sry expression (E11.3) results in development of intersex gonads (ovotestes) (HIRAMATSU et al. 2009). A delay in testis specification could alter sex-specific gene expression patterns. However, the effect of timing or level of expression of Sry on TGCT initiation has not been evaluated in humans or mice.
1.2.6 TGCT susceptibility genes and modifier genes
Susceptibility genes and modifier genes are distinct types of genes that influence
complex traits, like TGCTs. Susceptibility genes are defined as genes that are necessary
for a disease (HASTON and HUDSON 2005). Modifier genes are not necessary for disease,
but rather influence various aspects of the presentation of disease (HASTON and HUDSON
2005). For example, modifier genes can influence the severity or rate of progression of
disease by affecting the penetrance, dominance, expressivity or pleiotropy of
susceptibility genes (NADEAU 2001). Although no TGCT susceptibility genes have been
discovered in mice, several TGCT modifier genes have been found that affect the
penetrance (the proportion of individuals with the disease genotype that also have the
disease phenotype) of their target susceptibility genes. Modifier genes that increase the
penetrance of susceptibility genes are called ‘enhancers’, and modifier genes that reduce
the penetrance of susceptibility genes are called ‘suppressors’. Table 1-1 lists some
known TGCT modifier genes identified to date. Characterization of TGCT modifier
genes of both types continues to provide insight into the developmental pathways leading
to disease initiation and progression.
26
Process Modifier Type of modifier Nature of mutation (of normal allele) Pten-/- Enhancer Knock-out Regulation of cell cycle Trp53-/- Enhancer Knock-out Regulation of cell cycle Apobec1-/- Enhancer and Knock-out RNA editing suppressor Ay Suppressor Deletion Translation KitlSl Enhancer Deletion PGC migration Kitlgb Enhancer Deletion PGC migration Dnd1Ter Enhancer Opal mutation PGC survival MOLF-19 Enhancer CSS Unknown
Table 1-1: Modifier genes and chromosomes that affect TGCT susceptibility in 129/Sv males.
27
Several TGCT modifier genes enhance the frequency of male 129/Sv mice affected with TGCTs. Mutants in Kitl (Steel) and Kit (White-spotting, [W]) were both first described as dominant coat color mutants with defects in melanogenesis, hematopoiesis and PGC development (DURHAM 1908; SARVELLA and RUSSELL 1956).
Kitl encodes the ligand of kit receptor (Kit). Positional cloning of Kitl in 1990 and
subsequent interval mapping showed that the Steel allele encompasses a 973,366
nucleotide deletion on mouse chromosome 10 that entirely removes Kitl and as many as
six flanking genes (COPELAND et al. 1990; RUNYAN et al. 2006; HEANEY et al. 2008).
SlJ Slgb Sld Other alleles of Kitl (Kitl , Kitl , Kitl ) entail smaller deletions (HEANEY et al. 2008).
KitlSl heterozygotes on the 129/Sv background develop TGCTs at more than double the
+ rate of normal Kitl 129/Sv males (14% vs. 6%, respectively (STEVENS 1967a)) and have
Sl reduced numbers of PGCs (BUEHR et al. 1993; BEDELL et al. 1995). Kitl homozygotes
are almost devoid of PGCs by E12.5 (BENNETT 1956) and show late embryonic lethality
because of severe anemia (SARVELLA and RUSSELL 1956), so the effect of two alleles of
Steel on TGCT prevalence cannot be measured in adult mice.
Teratoma (Ter) is a semi-dominant allele of Dnd1 that enhances the frequency of
129/Sv males with TGCTs. The Ter mutant arose spontaneously while Leroy Stevens
was backcrossing KitW from C57BL6/J onto the 129/Sv inbred background to evaluate its
effect on tumorigenicity (STEVENS 1973). Tumor prevalence is 17% and 94% in Ter
heterozygotes and homozygotes, respectively (NOGUCHI and NOGUCHI 1985). Dnd1 is
important for PGC survival: Dnd1Ter/Ter mice have approximately ten PGCs at E11.5 and
are infertile on all inbred backgrounds tested (BHATTACHARYA et al. 2007).
28
Dnd1 is the mouse orthologue of the zebrafish dead end (dnd) gene which functions in PGC survival and migration (WEIDINGER et al. 2003). Positional cloning of
Ter revealed a single CÆT transition in the Dnd1 ORF on mouse chromosome 18
(YOUNGREN et al. 2005). This point mutation introduces a premature stop codon at amino acids 178 and 190 in protein isoforms DND1-α and DND1-β respectively. DND1 is known to localize to the 3’ end of certain mRNA transcripts to prevent miRNA- mediated binding and repression (KEDDE et al. 2007). That fact suggests Dnd1 promotes
expression of genes that are key to normal germ cell development (KEDDE et al. 2007).
The ways in which Dnd1 blocks miRNA binding and exacerbates TGCT risk remain to be elucidated.
Interestingly, many known modifiers that increase TGCT prevalence also decrease PGC numbers, associating reduced PGC numbers with increased tumorigenesis.
However, PGC-deficient KitW mutants fail to develop TGCTs at greater rates than control
129/Sv males (STEVENS 1967a; STEVENS 1973), demonstrating that reduction of PGCs is
not sufficient for increased rates of tumorigenesis.
1.2.7 Mapping TGCT loci
The number of genes influencing TGCT susceptibility can be estimated based on
backcrosses and intercrosses of susceptible and resistant inbred strains. Backcrosses of
strain 129 mice to seven TGCT-resistant inbred strains produced only one affected mouse
in 11,291 unaffected mice (STEVENS and MACKENSEN 1961). If all susceptibility genes
are assumed to be recessive, equally penetrant, and additive, approximately 23 genes are
required for TGCTs.
29
The advent of genetic markers in the early 1990’s facilitated efforts to distinguish the strain derivation of alleles distributed over most of the mouse genome (DIETRICH et al. 1992). In turn, these markers enabled whole genome linkage studies designed to map susceptibility genes and cancer modifier genes. Whole genome linkage studies are based on a cross between a tumor-susceptible and tumor-resistant inbred strain, followed by intercrossing or backcrossing of F1s to promote segregation of strain-specific alleles influencing tumor susceptibility. Phenotyping and genotyping of F2s or N2s enable mapping of susceptibility loci and cancer modifier loci.
The first whole genome linkage study aimed at mapping TGCT genes was performed by Gayle Collin, who crossed 129T1/Sv-+Oca2 Tyrc-ch Dnd1Ter/+ (hereafter,
129T1/Sv) females to MOLF/EiJ (hereafter MOLF) males and then backcrossed F1
progeny to 129T1/Sv (COLLIN 1994). MOLF is a wild-derived inbred strain of Mus
musculus molossinus and was used as the outcross strain to take advantage of the large
number of polymorphic markers between 129/Sv and MOLF (MATIN et al. 1998).
MOLF also has better intercross reproductive performance than strains derived from Mus
spretus (SILVER 1995) and males of that strain are TGCT-resistant (MATIN et al. 1998).
Arranging the cross in this way facilitated mapping of dominant MOLF enhancers and
recessive 129/Sv susceptibility genes. The cross was sensitized with Dnd1Ter/+ in order to boost the number of affected N2s, thereby increasing the statistical power of the linkage
analysis. Collin found no significant linkages when all tumor-bearing N2s were
considered together, or when the unilateral cases were considered separately from the
bilateral cases (COLLIN 1994; COLLIN et al. 1996). However, the greatest ‘insignificant’
linkage was to an excess of 129-derived markers on chromosome 19. (COLLIN 1994).
30
When bilateral TGCTs were considered separately from unilateral TGCTs, a weak linkage was detected with an excess of MOLF-derived markers on chromosome 19.
These results suggested the presence of at least one TGCT-promoting factor on mouse chromosome 19 (COLLIN 1994). Unfortunately hundreds more (estimated n > 400)
affected N2s would be required to validate these weak linkages. For this reason, TGCT
susceptibility was considered too complex for conventional genetic approaches.
Chromosome substitution strains (CSSs) provide a powerful alternative resource
to segregating crosses for discovering genes involved in TGCT susceptibility.
Chromosome substitution strains, also known as consomic strains, are inbred strains in
which one entire chromosome has been replaced by its homologue from a different strain
of mouse. The steps to make a CSS are illustrated in Figure 1-2, using 129/Sv as the host
strain and MOLF/Ei as the donor strain. CSSs have several advantages that facilitate
studies of complex traits such as TGCTs. For example, to detect chromosomes with
quantitative trait loci (QTLs) affecting a trait, the phenotypes of the CSS and host strain
are compared. If the trait value of the CSS is significantly different from the trait value
of the host strain, then the chromosome must have at least one QTL influencing the trait.
Fine-mapping of QTLs is accomplished by making panels of congenic strains in which
the donor chromosome is partitioned into segments of various lengths on the inbred host
background. The steps to make a congenic strain from a CSS are illustrated in Figure 1-
3. Congenic strains have all the same advantages of CSSs and can be constructed in as
few as three generations, compared to the ten or more generations that are needed to
make traditional congenic strains from a two-strain cross. CSSs accelerate QTL
discovery by removing several time consuming steps inherent to linkage crosses. Unlike
31
N2s or F2s, mice from the same CSS are genetically identical, making genotyping unnecessary. Since genotyping is not required, large numbers of CSS mice can be quickly bred and screened. In addition, tests of dominance are easier in CSSs: Trait values can be measured in both homosomic (two copies of the donor chromosome) and heterosomic (one copy of the donor chromosome, resulting from a backcross of the CSS to its host strain) animals to infer dominance of factors involved in the trait. CSSs also have statistical advantages. Since the entire genome is not segregating in CSS mice, the background “noise” from QTLs on other chromosomes is diminished, and the proportion of the trait variance from QTLs on the donor chromosome is enriched. For that reason, at least 37% fewer animals are needed to detect QTLs in CSSs compared to an F2 intercross
(BELKNAP 2003).
32
Figure 1-2: Steps to make a CSS (NADEAU et al. 2000; SINGER et al. 2004). 129/Sv are mated to MOLF and the N1F1 progeny backcrossed to 129/Sv. N2 mice with non- recombinant MOLF chromosomes are backcrossed to 129/Sv. Selection and backcrossing are repeated until the N10 generation. To identify non-recombinant MOLFchromosomes, the mice are genotyped for at least four loci on each chromosome. As the backcrosses proceed, the genetic background becomes enriched in 129/Sv sequences. At the N10 generation, males and females with non-recombinant MOLF chromosomes are brother-sister mated to make the substituted MOLF chromosome homozygous on the 129/Sv background.
Continued on next page:
33
34
Figure 1-3: Steps to make a congenic strain from a CSS. Heterosomic CSS mice are mated to 129/Sv and progeny are typed at six equally-spaced markers on the substituted chromosome (shown as dashes in the second cross only). Recombinants with congenic segments are backcrossed to 129/Sv and progeny re-genotyped with the six markers to identify mice with the same congenic segments. Males and females with the same congenic segments are brother sister mated. Offspring homozygous for the congenic segments are identified using the same six markers. Finally, mice homozygous for the congenic segments are brother-sister mated to perpetuate the congenic strain.
35
A CSS takes several years and hundreds of mice to construct. Substitution of the donor chromosome is achieved by repeated backcrossing to the host strain with selection for the non-recombinant donor chromosome. After at least ten generations of backcross, mice carrying one copy of the substituted chromosome are brother-sister mated and progeny selected with two copies of the substituted chromosome. So-called ‘homosomic’ mice are then brother-sister mated to perpetuate the CSS. Slight modifications of this process are necessary to substitute the mitochondria and sex chromosomes. Using this technique, entire CSS panels (a set of twenty-two inbred strains in which every chromosome from one inbred strain has been substituted onto the genetic background of a second inbred strain) have been constructed in mice (SINGER et al. 2004) and rats
(MALEK et al. 2006).
Matin made the first mouse autosomal chromosome substitution strain to test whether MOLF chromosome 19 had TGCT QTLs (MATIN et al. 1999), as predicted by
Collin (COLLIN 1994). Chromosome 19 from the MOLF inbred strain was substituted
onto the 129/Sv inbred background and males of the new CSS, 129-Chr 19MOLF, were tested for TGCTs. 80% of the 129-Chr 19MOLF males developed at least one TGCT, but only 5% of the control 129/Sv males developed at least one TGCT (MATIN et al. 1999).
Furthermore, 56% of the 129-Chr19MOLF males developed bilateral disease as opposed to
0% of the 129/Sv control males. These results confirmed Collin’s (1994) linkage to a
QTL on chromosome 19 promoting TGCTs.
A panel of congenic strains was made to map the QTL(s) (YOUNGREN et al.
2003). Interestingly, the position of the MOLF-derived congenic segments on
chromosome 19 did not predict TGCT risk. Ordinarily, trait-promoting genes are located
36 at discrete sites on chromosomes. For that reason, congenic strains of different lengths tend to have similar trait values because they contain the same trait-promoting genes.
With two exceptions (both congenic segments in the sub-telomere), a direct correlation was observed between the length of the congenic segments and increased TGCT susceptibility. That unexpected discovery could be explained by approximately five equally-penetrant enhancers on MOLF chromosome 19 or an epigenetic modification affecting expression of genes on the centromeric and middle regions of chromosome 19.
Misregulation of large numbers of genes could explain the existence of so many apparent modifiers. Similar “length effects” have not been previously described, and future experiments with 129-Chr 19MOLF and other 129.MOLF CSS-derived congenic strains may provide interesting new insights into the epigenetic changes associated with chromosome substitution.
Following the study of Collin (1994), a second whole genome linkage study was performed by Muller and colleagues using C57BL/6 instead of MOLF as the outcross
+/- Ter/+ strain and the Jacks Trp53 mutant instead of Dnd1 as the sensitizer (MULLER et al.
2000). When bilateral and unilateral TGCTs were considered together, one significant linkage was reported in a region of 129-derived markers on chromosome 13. That region was not significant in the cross by Collin (COLLIN 1994). Muller and colleagues failed to
find any significant linkages to chromosome 19. These contrasting results could be
explained by the use of different sensitizers: Trp53+/- acts additively with Dnd1Ter/+ in
interaction tests, suggesting at least two parallel pathways contribute to TGCT
susceptibility (LAM et al. 2007). The results could also be explained by the use of
different outcross strains, or too few affected N2 males.
37
1.3 RESEARCH AIMS
Although TGCTs have been studied in strain 129 mice for at least 55 years, no susceptibility genes have been discovered. Our understanding of the genetic control of
TGCT formation and progression is therefore poor, and many susceptibility genes and modifier genes remain undiscovered. Identifying these genes would improve our understanding of the etiology of TGCTs in mice and may provide insight into the genetic control of TGCTs in humans.
Matin demonstrated that QTLs controlling highly complex traits like TGCTs can be localized to individual chromosomes using CSSs (MATIN et al. 1999). I therefore
constructed a partial CSS panel in which selected MOLF donor chromosomes were substituted onto the 129/Sv host background. In Chapter 2, I describe how these CSSs were built and surveyed for QTLs affecting TGCTs. Although ten CSSs were constructed, only two (129-Chr 2MOLF and 129-Chr 18MOLF) had dominant QTLs affecting
TGCTs. I therefore prioritized development of these two CSSs and made congenic
strains from one of them (129-Chr 18MOLF) to map the QTLs involved in TGCTs. I
describe how the congenic strains were made and surveyed in Chapter 3.
In Chapter 4, I evaluated the role of the Y chromosome on TGCT susceptibility. I tested two hypotheses: first, whether the 129/Sv Y chromosome is necessary or sufficient
for TGCTs, and second, whether any Y chromosome is necessary for TGCTs. I
determined that although the 129 Y chromosome is not required for TGCTs, at least one
Y chromosome factor is required for TGCTs.
Chapter 5 contains my closing thoughts and ideas for future experiments.
38
2 CHAPTER 2: TGCT survey in a reciprocal B6.129 CSS panel and a partial
129-ChrMOLF CSS panel
39
2.1 ABSTRACT
Despite a strong heritable component, little is known about the genetic control of susceptibility to testicular germ cell tumors (TGCTs) in humans or mice. Although the mouse model of TGCT teratocarcinogenesis has been studied intensely for at least 55 years, TGCT genes remain elusive, a fact in part attributable to the genetic complexity of the disease. Evidence for five QTLs containing TGCT enhancers on chromosome 19 was found using the 129-Chr 19MOLF chromosome substitution strain (CSS), demonstrating
the power of CSSs to map TGCT loci to individual chromosomes. In this Chapter, I
tested whether ten new incipient 129-ChrMOLF CSSs had dominant QTLs affecting TGCT
prevalence at 3 – 5 weeks of age. I surveyed backcross males for TGCTs from the 129-
ChrB6, B6-Chr129 and 129-ChrMOLF panels. The results of this survey indicate the
presence of dominant TGCT QTLs on one B6-derived and two MOLF-derived chromosomes.
2.2 INTRODUCTION
Testicular germ cell tumors (TGCTs) are the most common solid tumors in men aged 20 – 35 years (DIXON and MOORE 1952; BISHOP 1998). Family history is a
significant risk factor, as the risk of developing TGCTs is 8-10 times higher among brothers and four times higher among sons of affected individuals (FORMAN et al. 1992;
LINDELÖF and EKLUND 2001). Unfortunately, traditional linkage studies to identify
genes that control susceptibility have been difficult to perform for at least three reasons: a
shortage of multigenerational pedigrees with sufficient numbers of affected individuals,
the sterility that often results from treatment, and the genetic complexity of the disease.
40
As a result, the genetic factors that contribute to the disease are still poorly understood, and only one low-penetrance susceptibility locus has been identified.
The teratomas and teratocarcinomas of strain 129 mice are an established model of the teratomas and teratocarcinomas of human infants. 129 males spontaneously develop TGCTs at a rate of 1 – 8%, depending on the substrain (STEVENS and HUMMEL
1957). TGCTs in mice are composed of a number of adult and embryonic tissue types,
implying a stem cell origin of the tumors. Indeed, TGCTs fail to form from grafts
derived from mice lacking germ cells, showing that primordial germ cells (PGCs) are the
cells of origin for TGCTs (STEVENS 1967b). Grafts from E10.5 to E12.5 genital ridges
were sufficient to induce TGCTs when introduced into adult testes, whereas ridges from
later developmental timepoints were not sufficient to induce TGCTs, demonstrating that the “critical period” of TGCT formation is between this two day period (STEVENS 1964).
As in humans, control of TGCT formation in mice is complex: segregating crosses
between 129/Sv and other strains yielded one affected mouse in 11,291 mice surveyed,
strongly suggesting multigenic control (STEVENS and MACKENSEN 1961).
A whole genome linkage study was performed to map QTLs containing TGCT
genes (COLLIN 1994; COLLIN et al. 1996). Although the strongest association was to chromosome 19, the study failed to discover statistically significant linkages on any
chromosome (COLLIN 1994). Because traditional genetic approaches lack the power to
locate TGCT QTLs, our lab made the first mouse autosomal chromosome substitution
strain (CSS) to test for TGCT genes on chromosome 19 (MATIN et al. 1999). A CSS is
an inbred strain with an entire chromosome replaced by its homologue from a different
strain of mouse, and is made by repeated backcrossing with selection for the chromosome
41 being substituted. After ten generations of backcross, the CSS is brother-sister mated to make the substituted chromosome homozygous. CSSs that have been backcrossed less than ten generations are called ‘incipient CSSs.’ Several logistical and statistical advantages make CSSs preferable to linkage crosses for finding susceptibility genes for complex diseases such as TGCTs: genotyping is unnecessary because CSS mice are genetically identical, strain background effects are uniform and fixed, CSS mice are quick and easy to breed relative to N2s or F2s used in linkage, there are no dominance effects, and up to 10 times fewer animals are required to achieve significant results
(NADEAU et al. 2000; BELKNAP 2003).
129-Chr 19MOLF was made by substituting chromosome 19 from MOLF/Ei
(MOLF) onto the 129S1/SvImJ (129/Sv) host inbred background. MOLF is a wild-
derived inbred strain of the Japanese species Mus molossinus. Mus molossinus is
genetically distant from Mus domesticus and Mus musculus, the principal ancestral
species of laboratory inbred strains like 129/Sv (YANG et al. 2007). As a result,
MOLF/Ei carries DNA sequence variants that can manifest as modifier alleles on other
inbred backgrounds, making it a good candidate for donor chromosomes in CSS studies
MOLF (MATIN et al. 1998). 82% of 129-Chr 19 males developed at least one TGCT
(MATIN et al. 1999). These results indicate the presence of at least one QTL that
promotes TGCTs on chromosome 19.
Despite the absence of TGCTs in MOLF, at least one factor from MOLF
chromosome 19 enhances TGCTs on the 129/Sv background. That fact, together with
weak linkages to several other MOLF chromosomes (COLLIN 1994), suggest that TGCT
modifiers exist on other MOLF chromosomes. To test for novel TGCT modifiers, I
42 constructed and surveyed ten new incipient 129.MOLF CSSs. The ten MOLF/Ei chromosomes substituted were 2, 3, 7, 11, 13, 16, 18, X, Y and the mitochondrial chromosome. Two of the chromosomes, 13 and X, were selected because QTLs influencing TGCT susceptibility on those chromosomes had been documented previously in mice (MULLER et al. 2000) or humans (RAPLEY et al. 2000), respectively.
Chromosome 7 was selected based on evidence for a recessive 129/Sv-derived TGCT
factor (COLLIN 1994; COLLIN et al. 1996). In the absence of direct evidence for genes
affecting TGCTs, the other chromosomes were selected for diversity in chromosome size
and gene content.
In addition to surveying ten 129-ChrMOLF CSSs, I surveyed nineteen B6-Chr129 and twenty-one 129-ChrB6 CSSs that were being constructed as part of another project in
the Nadeau Laboratory. Although I surveyed males from a total of fifty CSSs for
TCGTs, I found only one CSS (129-Chr 6B6) with dominant TGCT-enhancing QTLs, and
two CSSs (129-Chr 2MOLF and 129-Chr 18MOLF) with dominant TGCT-suppressing QTLs.
2.3 MATERIALS AND METHODS
2.3.1 Backcrosses
MOLF/Ei and 129S1/SvImJ mice were purchased from the Jackson Laboratory
(Stock numbers 000550 and 002448, respectively; Bar Harbor, Maine). MOLF/Ei mice
were maintained on LabDiets 7960 rodent chow (PMI Nutrition International; Saint
Louis, Missouri) and 129/Sv mice were maintained on 5010 rodent chow (PMI Nutrition
International; Saint Louis, Missouri). MOLF/Ei were mated to 129/Sv and the progeny
were backcrossed to 129/Sv. N2 mice were genotyped and mice with non-recombinant
chromosomes of interest were backcrossed to 129/Sv. This step was repeated until at
43 least the N10 generation. Non-recombinant N10+ mice were brother-sister mated to make the substituted chromosome homozygous. This research was approved by the
CWRU Institutional Animal Care and Use Committee.
2.3.2 Genotyping
Mice were genotyped with either the SNaPshot single base extension (SBE) protocol (Applied Biosystems; Foster City, California) or microsatellite markers. For
SBE, SNaPshot was performed as described by the manufacturer’s protocol with the following exceptions. 1 unit of shrimp alkaline phosphatase (USB Corporation;
Cleveland, Ohio) and 0.2 units of Exonuclease I (USB Corporation; Cleveland, Ohio) each were added to 15 ul of PCR product to remove unused dNTPs and PCR primers.
The custom SBE oligonucleotide, which anneals immediately upstream of the SNP, was used at a working concentration of 0.2 uM to 1.0 uM, depending on the efficiency of each primer. Reactions were multiplexed at least four and at most six times. After primer extension, the reactions were subjected to capillary electrophoresis on ABI DNA
Analyzers (Applied Biosystems; Foster City, California) in the CWRU Genomics Core.
The identities of the SNPs and the primers used for genotyping are listed in Appendix I.
For SNP genotyping, at least four and at most six SNPs were typed per chromosome.
Two SNPs were as close to the ends of each chromosome as possible, and two to four
SNPs were evenly spaced in between. There were no gaps >23 cM.
For microsatellite marker genotyping, polymorphic MIT markers were identified
(http://www.cidr.jhmi.edu/mouse/mult_inf.html) and PCR products were resolved in agarose or polyacrylamide gels. Genotypes were inferred by length polymorphism.
Appendix II contains the list of microsatellite markers used for genotyping each
44 chromosome. For microsatellite marker genotyping, four MIT markers were typed per chromosome. Two markers were as close to the ends of each chromosome as possible, and two were evenly spaced in between. There were no gaps >34 cM.
2.3.3 SNP selection and primer design
Polymorphic SNPs for 129/Sv and MOLF were selected from the preliminary release of the Inbred Laboratory Mouse Haplotype Map
(http://www.broad.mit.edu/personal/claire/MouseHapMap/Inbred.htm) and the
Wellcome-CTC Mouse Strain SNP Genotype Set
(http://www.well.ox.ac.uk/mouse/INBREDS/). Genetic map positions of the SNPs were inferred by MIT microsatellite markers surrounding the SNPs. PCR primers were selected and the NCBI BLAST program was used to ensure the PCR primers were specific to the target sequence.
2.3.4 TGCT survey
3-5 week old males (a minority were >5 weeks) were euthanized by CO2 inhalation or cervical dislocation. Incisions were made in the abdomen to expose the testes. TGCTs are identifiable by macroscopic inspection (STEVENS and LITTLE 1954),
by abnormalities in size, color, shape or texture, or a combination of one of more of these
traits. Mice with TGCTs were scored as positive for unilateral or bilateral TGCTs. CSS
TGCT prevalences were analyzed using the chi-square goodness-of-fit test and compared
to the control 129/Sv TGCT prevalence. The significance threshold for all calculations
was set at p < 0.05. The p values reported are after Bonferroni correction for multiple
testing. Female mice lack testes and TGCTs and were not included in these analyses.
45
2.4 RESULTS
2.4.1 TGCT occurrence in 129-ChrMOLF backcross mice
In total, I examined 140 control 129/Sv males and 1,920 post-N2 129-ChrMOLF backcross males for TGCTs. The males were examined at 3 – 5 weeks of age, which is the developmental timepoint used in prior TGCT surveys (STEVENS and LITTLE 1954;
STEVENS and HUMMEL 1957). In Table 2-1, I show the backcross generations sampled
for each CSS, the total number of backcross males screened, the number and percent
affected with unilateral and bilateral TGCTs, and the number and percent of TGCTs
attributable to a segregating modifier (explained in detail below). The percent of affected
control 129/Sv males, 5%, agrees closely with recently published estimates for that strain
(YOUNGREN et al. 2003; LAM et al. 2004; LAM et al. 2007). All backcross progeny in
Table 2-1 were either heterozygous for the substituted chromosome, recombinant for the
substituted chromosome, or non-recombinant for the 129/Sv chromosome. A
disproportionate number of males were screened for the 129-Chr 7MOLF, 129-Chr YMOLF,
and 129-Chr MitoMOLF CSSs either because more cages of those strains were kept or
because they had superior breeding performance. In contrast, 129-Chr 3MOLF, 129-Chr
XMOLF and 129-Chr 13MOLF were remarkably poor breeders and comparatively few males
from these CSSs were surveyed. The poor breeding performance in these CSSs may be
due to MOLF-derived factors on the donor chromosomes that affect fertility.
2.4.2 Evidence for MOLF modifiers in 129-Chr 2MOLF and 129-Chr 18MOLF
The donor chromosomes may have MOLF modifiers affecting TGCT prevalences. To predict which donor chromosomes have factors influencing TGCT prevalence, I tested what affect a dominant MOLF modifier on each donor chromosome
46 would have on TGCT occurrence in the backcross CSSs. Analysis of genotyping data showed that the backcross mice consisted mostly of males that were recombinant with respect to their donor chromosomes. If segregation were normal, each recombinant mouse had a 50% chance of having the MOLF modifier, and each non-recombinant mouse had a 0% or 100% chance of having the modifier, depending on whether the mouse was non-recombinant with respect to the MOLF/Ei chromosome or the 129/Sv chromosome. Both types of non-recombinants have an equal chance of being produced.
Because one type of non-recombinant always has the dominant modifier and the other always lacks it, their effects on the TGCT prevalence statistically cancel one another out.
Finally, both types of non-recombinant are uncommon (approximately 20% of backcross mice, each) relative to the recombinants (approximately 60% of backcross mice)
(NADEAU et al. 2000). For these reasons, estimates of modifier effects take into consideration only the recombinants.
Recombinants that lack the putative MOLF modifier are expected to have TGCTs at the control 129/Sv rate (5%). However, recombinants that have the dominant MOLF modifier are affected with TGCTs at varying rates, depending on the strength and nature of the modifier. For each CSS, I applied the following model, in which the observed frequency of affected males is described as a composite of two different frequencies; the frequency of affected males with the MOLF modifier, and the frequency of affected males without the MOLF modifier. In the equation, m represents the fraction of mice that inherit the MOLF allele. Instead of assuming normal segregation (m = 0.5), we calculated the values of mA by averaging the proportion of mice inheriting the MOLF
47 allele for all microsatellite markers on donor chromosome A. When the MOLF allele was inherited preferentially, m was greater than 0.5: