|Home | About | Journals | Submit | Contact Us | Français|
Germ cell tumors (GCTs) affect infants, children, and adults and are the most common cancer type in young men. Progress in understanding the molecular basis of GCTs has been hampered by a lack of suitable animal models. Here we report the identification of a zebrafish model of highly penetrant, heritable testicular GCT isolated as part of a forward genetic screen for cancer susceptibility genes. The mutant line develops spontaneous testicular tumors at a median age of 7 months, and pedigree analysis indicates dominant inheritance of the GCT susceptibility trait. The zebrafish model exhibits disruption of testicular tissue architecture and the accumulation of primitive, spermatogonial-like cells with loss of spermatocytic differentiation. Radiation treatment leads to apoptosis of the tumor cells and tumor regression. The GCT-susceptible line can serve as a model for understanding the mechanisms regulating germ cells in normal development and disease and as a platform investigating new therapeutic approaches for GCTs.
Proper development of the germline is critical to ensure fertility and, ultimately, survival of the species. In humans, aberrations in germ cell development are linked to gonadal dysgenesis, infertility, and germ cell tumors (GCTs). GCTs are malignant neoplasms of germ cells that occur in neonates, infants, children, adolescents, and adults. In children the tumors are relatively rare, occurring at a rate of about 2 per million.1 The incidence in adults is 70–80 per million, and it has been rising steadily for unknown reasons.2 Testicular GCT is in fact the most common cancer in young men aged 15–40. GCTs are a heterogeneous group of neoplasms that includes yolk sac tumors (also called endodermal sinus tumors), immature and mature teratomas, choriocarcinoma, embryonal carcinoma, and germinomas (referred to as dysgerminomas in females and seminomas in males).3 Therapy for GCTs has been essentially static since the discovery that these tumors are exquisitely sensitive to cisplatin-based regimens.1,4 Although effective, these chemotherapy regimens cause hearing loss, infertility, lung damage, and kidney failure. Moreover, emerging data on long-term survivors of testicular cancer show a doubling in risk of second malignancy and of early onset of cardiovascular disease.1 Prospects for improved, targeted therapy of GCTs are poor, however, because of the lack of molecular detail about the causes of these tumors. No somatic or inherited mutations causing GCTs have been definitively identified.
Although there is overlap in tumor spectrum, it is generally helpful to consider GCTs of children and young adults as clinicopathologically distinct entities4 (Fig. 1). Type I tumors occur in neonates, infants, and prepubertal children, and consist of teratomas (benign) and yolk sac tumors (malignant). Characteristic chromosomal aberrations include losses of chromosome 1p and 6q, and gain of 1q and 20q. In postpubertal adolescents and adults (type II tumors), a different spectrum is seen, with the appearance of germinomas (called seminomas in males and dysgerminomas in females), which are tumors of primitive, undifferentiated germ cells. The other malignant GCTs in this age group are collectively called nongerminomas, and consist of embryonal carcinoma, yolk sac tumor, teratoma, and choriocarcinoma. Germinomas and nongerminomas appear to arise from a common precursor, the carcinoma in situ cell,5,6 share common chromosomal aberrations including isochromosome 12p, and are frequently found together in mixed tumors. Finally, type III GCTs are the spermatocytic seminomas occurring in older males, which are thought to arise from spermatogonia or spermatocytes and characteristically exhibit gain of chromosome 9.7,8 These tumors are confined to the testis and show a mixture of small, intermediate, and large cells.
Despite the differences noted above, there is striking evidence that both childhood and adolescent/adult GCTs arise due to defects in the totipotent early germline progenitors known as primordial germ cells (PGCs). First, the heterogeneous phenotype of GCTs indicates a totipotent cell of origin; GCTs can be germinomas (which retain primitive germ cell characteristics); tumors of extraembryonic tissues (yolk sac, trophoblast); and tumors differentiated to endoderm, mesoderm, and ectoderm (teratomas). Second, GCTs show persistent or ectopic expression of genes normally enriched in PGCs, such as Oct-4, NANOG, and STELLA.9,10 Third, disorders of male genitourinary development such as cryptorchidism and spermatogenic or testicular dysgenesis are strong risk factors for later developing testicular GCT,11 again supporting the idea that GCTs have their origin during embryonic and fetal development. Finally, the high frequency of extragonadal GCTs in the midline likely reflects the migratory path taken by PGCs during development, after they arise on the yolk sac.12
The development of the germline in zebrafish includes many features that are conserved from fish to mammals.13–16 Zebrafish PGCs are specified beginning with localization of specific transcripts such as vasa to the germ plasm at the four-cell stage. An initial population of four to five germ cells undergoes several rounds of division as the cells migrate to their eventual position in the presomitic mesoderm. The factors controlling PGC specification and migration are broadly conserved from flies to fish to mammals (reviewed in Refs.15,17,18). In both sexes, the mature gonad develops between 14 and 28 days postfertilization through a primitive ovary-like stage. In presumptive males, this tissue undergoes apoptosis as the male germline emerges. Zebrafish do not have identified sex chromosomes, thus sexual development is somewhat plastic and subject to environmental influences.19–23
Previous studies have described the spontaneous development of gonadal neoplasms in male zebrafish >2 years of age. Moore et al. described the tumor spectrum at 30–34 months of age in wild-type zebrafish and in the carriers of the gin genomic instability phenotype.24 Testicular hyperplasias (enlarged testes containing all stages of spermatogenesis) were found in 48% of wild types and 25% of gin heterozygotes. Benign seminomas, which they defined as tumors of predominantly one cell type derived from an early stage of spermatogenesis, were seen in 17% of wild-type fish at 30–34 months of age; the incidence was 53% in gin carriers. Amsterdam et al. documented the tumor spectrum in a survey of nearly 10,000 two-year-old zebrafish. Overall, 473 tumors were found, of which approximately 40% were described as seminomas of the testis.25 The testis has also been identified as a target of carcinogens in multiple fish species, including rainbow trout, medaka, and zebrafish.26,27 Spitsbergen et al. reported testicular neoplasms of 5 of the 68 juvenile fish treated with the carcinogens N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) and in 1 of the 99 juveniles treated with 7,12-dimethylbenz[a]anthracene (DMBA).28,29 Bauer and Goetz29a carried out a histology-based genetic screen in zebrafish and identified 11 mutations that produced visible gonadal phenotypes in only one sex per family. To date, however, no line of zebrafish that consistently gives rise to GCTs of the ovary or testis has been isolated and propagated. Here we describe the identification, through an ethylnitrosourea (ENU)-based forward genetic approach, of a mutant zebrafish line that spontaneously develops highly penetrant testicular GCTs at an average age of 7 months. The GCT susceptibility trait was identified in the carriers of the shortstop cell proliferation mutation, but we show that the trait is not linked to shortstop. The trait is dominantly inherited, and the GCTs display disordered testicular tissue architecture and the accumulation of primitive spermatogonial-like cells, with impaired germ cell differentiation. Similar to human seminomas, the tumors are sensitive to radiation therapy. This novel model can serve as a basis for understanding the development of GCTs in humans.
Zebrafish were maintained according to the standard procedures30 on a 14h light/10h dark cycle in a recirculating system (Aquaneering, San Diego, CA). Zebrafish embryos were obtained by natural spawning of adults and staged according to Kimmel et al.31 Tricaine methanesulfonate was used as an anesthetic for all procedures and for euthanasia. All work with zebrafish was carried out under protocols approved by the Institutional Animal Care and Use Committees at Children's Hospital Boston and UT Southwestern Medical Center, both American Association for Laboratory Animal Care (AALAC)-accredited institutions.
The forward genetic screen for cell proliferation and cancer-susceptibility mutants has been described elsewhere.32 Briefly, male zebrafish were mutagenized with ethlynitrosourea and mated to unmutagenized females. Eggs were squeezed from individual F1 females and in vitro fertilized with UV-inactivated sperm. At 24h postfertilization, embryos were fixed and subjected to whole-mount immunohistochemistry with anti-phosphohistone H3 antibody. F1 females whose progeny showed alterations in cell proliferation were outcrossed to establish F2 families, and mutant phenotypes were reidentified via random crosses of F2 siblings. All of the identified mutations were homozygous embryonic lethal, and mutant lines were maintained as heterozygotes.
Twenty-eight-day-old fry from backcrosses of mutant heterozygotes were exposed to dimethyl sulfoxide or ethanol vehicle control, 5ppm DMBA or 1.5ppm MNNG for 24h.28,29 At 3, 6, or 12 months after treatment, fish were euthanized with tricaine, fixed 48h in 4% paraformaldehyde (PFA)/1× phosphate-buffered saline, decalcified 5 days in 0.25M ethylenediamine tetraacetic acid, dehydrated, and paraffin embedded. Three hematoxylin and eosin–stained sagittal step sections from each fish were examined for tumors. Prospective tumors were reviewed and validated by two independent pathologists.
Adult male zebrafish were subjected to whole-body irradiation from a 137Cs source. About 24Gy total dose was administered in six daily fractions of 4Gy. Treated animals were allowed to recover and observed for signs of tumor regression. TUNEL staining was carried out on 4μm unstained paraffin sections from testes dissected from the animals 1 week after irradiation, using the ApopTag Red In Situ Apoptosis detection kit (Chemicon; Millipore, Billerica, MA). Sections were counterstained with Hoechst 33342 and imaged on a Nikon Eclipse 80i microscope using Nikon Elements imaging software (Melville, NY).
Zebrafish were anesthetized with 0.04mg/mL tricaine and immobilized ventral side up in a clay mold immersed in aquarium water with 0.04mg/mL tricaine. Ultrasound images were acquired with the Vevo770 imaging system (Visual Sonics, Toronto, Canada) using a 40MHz ultrasound probe. About 40-μm-thick images were obtained in B mode at 30Hz.
Previously, we described a genetic screen to identify mutations that altered embryonic cell proliferation in the progeny of ENU-mutagenized zebrafish.32,33 Using whole-mount immunohistochemistry with anti-phosphohistone H3 to evaluate cell proliferation, seven mutant lines were identified including lamc1cz61, sfdcz213, mybl2cz226, espl1cz280, llgcz3322, sdscz319, and slhcz333 (reviewed in Ref.32). One of the goals of the screen was to identify mutations that increased cancer susceptibility in adult zebrafish. As all the lines were homozygous embryonic lethal, we investigated the incidence of cancer in heterozygote carriers of the cell proliferation mutations. With the exception of lamc1cz61, described below, the incidence of spontaneous tumors in these lines was low (13/3597; 0.36%). This was likely due to the requirement for multiple genetic lesions in neoplastic transformation. Therefore, outcross progeny of heterozygous carriers for all lines were subjected to a single dose of carcinogen at 28 days postfertilization. The carcinogens used were DMBA and MNNG, both of which had been shown to potently induce tumors in zebrafish.28,29 Animals were sacrificed at 3, 6, or 12 months following treatment, and histologic step sections were examined for the presence of tumors. Data for treatment of the mybl2cz226 and espl1cz280 lines with MNNG have previously been reported.32,33 Table 1 shows the tumor spectrum after DMBA and MNNG treatment of all of the lines (n=12,471 animals total). Fish developing tumors were not genotyped in this analysis, and heterozygous carriers of the mutations and wild-type siblings are considered together. Although there is variability between lines, some general observations can be made. DMBA is metabolized in the liver, and hepatobiliary tumors predominate in the DMBA-treated zebrafish; also seen are epithelial neoplasms of the gills. After exposure to MNNG, a different spectrum is seen (Table 1). Mesenchymal tumors such as spindle-cell sarcomas and invasive, neoplastic vascular tumors are the most common tumor type (43%). These results are similar to those reported by Spitsbergen et al.28,29
We noted variability in the incidence of GCTs among the different mutant lines, after either DMBA or MNNG treatment. Two lines, lamc1cz61 and mybl2cz226, had notably elevated incidence of GCT after carcinogen treatment (Table 1). (A third line, sdscz319, exhibited 40% GCTs after MNNG treatment; however, the small number of only 32 tumors makes the significance of this observation unclear.) The GCTs that develop in these lines after DMBA or MNNG treatment typically consist of enlarged, bilateral testes in which primitive appearing, spermatogonial-like cells accumulate, with impairment of spermatogenesis. The increased GCT incidence in these two independent lines, compared with the other lines derived in our screen, suggests that multiple loci can be mutated to cause increased susceptibility of the testis to carcinogens. The mybl2cz226 line carries a loss-of-function mutation in the mybl2 gene. We genotyped 20 fish that developed GCTs from this line and found that 10 carried the mybl2cz226 mutation and 10 were wild type at the mybl2 locus. Therefore, the GCT susceptibility in this line is not linked to the mybl2cz226 cell proliferation mutation, and the observed increase in GCT incidence was likely due to a different mutation that was induced by the initial ENU treatment. We did not find a significant incidence of spontaneous GCTs in the mybl2cz226 line (see below), and we did not recover a derivative of this line with enhanced GCT susceptibility.
We noted an increased incidence of both spontaneous and carcinogen-induced GCTs in fish derived from the shortstop line (originally designated as shpCZ61 [reviewed in Ref.32] and renamed lamc1cz61, see below) (Table 1). The shortstop mutation causes a characteristic morphologic defect in homozygous mutant progeny,32 allowing the identification of heterozygote carrier parents after genetic incrosses. As with other lines, heterozygous carriers were outcrossed to wild type, and the resultant progeny were exposed to carcinogen at day 28. Although liver, pancreas, and biliary tumors still predominate in the DMBA-treated animals, the proportion of fish developing GCTs in this background increased to 26%. The increased incidence of GCT was even more striking after MNNG treatment, representing 57% of all tumors in this group.
The shortstop/lamc1cz61 background was also notable for an increased incidence of spontaneous GCTs. We raised control vehicle-treated fish for 12 months, sacrificed the fish, and examined them for spontaneous GCTs (n=1061 total). No spontaneous GCTs were identified in the sfdcz213, espl1cz280, llgcz3322, or slhcz333 lines. The prevalence of spontaneous GCTs in the mybl2cz226 line was 2/164 (1.2%); in the sdscz319 line the prevalence was 1/141 (0.7%). In contrast, the prevalence of spontaneous GCT in the lamc1cz61 line was significantly increased at 18/259 (7%) (p=0.008 for the comparison of mybl2cz226 vs. lamc1cz61; p=0.005 for the comparison of sdscz319 vs. lamc1cz61; both by two-tailed Fisher's exact test).
The occurrence of spontaneous GCTs in the lamc1cz61 background and the heightened susceptibility to develop GCTs after carcinogen treatment are suggestive of an inherited tumor predisposition syndrome. We noted that the progeny of different heterozygous lamc1cz61 parents exhibited different prevalence of GCTs. Therefore, we investigated whether the GCT susceptibility exhibited founder-specific effects. We assembled a detailed pedigree of fish descended from the original shortstop female to document the GCT incidence in specific families (Fig. 2). The GCT trait was inherited by the progeny of multiple different unrelated males crossed to the shortstop female, indicating that the trait was derived from this female. The trait was dominantly inherited and could be transmitted by both males and females, though only males exhibited a phenotype. After two generations of outcrossing, five different carriers of the shortstop/lamc1cz61 cell proliferation mutation (identified via phenotyping of incross embryos) were tested for ability to transmit the TGCT trait in outcrosses to a wild-type strain. Although all carriers generated outcross progeny, only one female (Fig. 2, arrowhead) transmitted the trait, suggesting that the GCT trait was not linked to the shortstop mutation (Table 2).
To directly test the possible linkage, we identified the mutated locus in shortstop. Initial possible linkage to four different chromosomes was obtained by carrying out bulk segregant mapping with a pool of 20 phenotypically wild-type and 20 shortstop mutant embryos against a panel of 250 microsatellite markers, as described.34 Using a further panel of 88 mutant embryos, we assigned shortstop to zebrafish chromosome 2 near the microsatellite markers Z8448 and Z21490 (18/88 and 12/88 recombinants, respectively). This location was close to the reported position of the sleepy mutants, which carry inactivating mutations in the laminin c1 gene.35 The phenotype of shortstop was nearly identical to that of sleepy, including altered morphology, notochord defects, and defective laminin assembly (not shown). Therefore, we performed complementation analysis with sleepy/lamc1m86 (generously provided by Derek Stemple; Wellcome Trust Sanger Institute, Cambridge, UK). All the five matings between shortstop/lamc1cz61 and sleepy/lamc1m86 heterozygotes produced embryos with the sleepy phenotype (an average of 22±2% mutant embryos). Therefore, shortstop is an allele of zebrafish lamc1. We prepared and sequenced lamc1 cDNA from shortstop mutant embryos and found that shortstop lamc1 contains an 11bp insertion in exon 27 of the coding sequence, leading to the aberrant insertion of the sequence Phe-Phe-Ser-STOP in the predicted lamc1 protein sequence. Further analysis showed that shortstop mutants contain a T>A mutation at nucleotide 80 of the 91bp intron 26–27, creating a cryptic splice acceptor site and leading to the 11bp insertion. Therefore, shortstop is an allele of laminin c1, similar to sleepy, and the line is thus designated lamc1cz61.
Having identified the mutation in shortstop we genotyped animals that developed GCTs from the lamc1cz61 line and found that the GCT susceptibility trait is not linked to the laminin c1 mutation; for the 15 affected males in family IV-1 (Fig. 2), 7 carried the lamc1cz61 allele and 8 did not. As further confirmation, we examined the spontaneous and DMBA-induced tumor incidence in outcrosses of sleepy/lamc1m86 heterozygotes. There were no spontaneous GCTs observed, and no GCTs developed in 327 DMBA-treated outcross progeny from lamc1m86 (data not shown). Thus, the GCT susceptibility trait is not linked to the laminin c1 gene and likely represents a second-site mutation that was induced during the original ENU mutagenesis.
Figure 3 shows the histologic appearance of normal testis and spontaneous testicular GCTs from the GCT-susceptible line and a rare ovarian GCT from carcinogenesis studies. The normal testis is composed of cysts or lobules, surrounded by a basement membrane and by Sertoli cells and other supporting somatic cells.16,36 As in other metazoans, spermatogenesis in zebrafish depends on the presence of self-renewing spermatogonial stem cells, which are normally found in small clusters adjacent to the basement membrane (Fig. 3A, B, arrowheads). Within the cyst, gametogenesis proceeds in an orderly fashion as type A spermatogonia undergo mitotic and then meiotic division, differentiating successively to type B spermatogonia, primary spermatocytes, secondary spermatocytes, and spermatozoa. In the GCT-susceptible line, this orderly architecture and differentiation sequence is disrupted (Fig. 3C–F). As early as 3 months of age, ectopic spermatogonial-like cells can be found distant from the basement membrane (Fig. 3C, D, arrows). Mature sperms are still present, and males with early tumors can successfully mate, though with reduced fertility owing to a general impairment in spermatogenesis. In more advanced tumors (Fig. 3E, F), there is a severe reduction or complete loss of differentiation. The architecture of the cyst is disrupted as the spermatogonial cells fill the entire space of the cyst. Marked, bilateral testicular enlargement is a consistent finding. Occasionally, immature oocytes can be seen within the cyst (Fig. 3G, arrows). This is also sometimes seen in wild-type fish after carcinogen exposure.
Heterozygous female carriers of the GCT susceptibility trait are fertile and phenotypically normal. Likewise, germ cell neoplasms of the ovary are extremely rare after carcinogen treatment. One example, shown in Figure 3H, exhibits primitive-appearing germ cells within a cellular stroma, with resemblance to a human ovarian dysgerminoma.
In humans, seminomas of the testis are radiosensitive, and radiation therapy is part of the standard treatment in many cases of testicular cancer.1 The high response rate of human GCTs to radiation and chemotherapy is thought to be related to the presence of a robust DNA damage response, leading to tumor cell apoptosis in response to DNA damage.37–39 We tested whether zebrafish testicular GCTs are susceptible to radiation treatment. Male carriers of the GCT-susceptibility trait were allowed to develop tumors, as evidenced by visible abdominal distension. Animals with tumors were then subjected to 24Gy of total body radiation from a 137Cs source, in six daily fractions of 4Gy each. After radiation therapy, the animals were allowed to recover and were examined for evidence of tumor response (Fig. 4). Figure 4A shows an affected male in which reduction of the abdominal mass, suggestive of tumor shrinkage, is evident 18 days after radiation. A different fish is shown in Figure 4B at 36 days after treatment. A third male with a large tumor did not have a detectable response (Fig. 4C). Altogether, six of seven treated animals had a response, with one animal having a partial response followed by tumor recurrence. To understand the mechanism of tumor response, we carried out TUNEL staining of the testis in a treated animal, 1 week after a completion of radiation exposure (Fig. 4D, E). Histologic examination showed the typical pattern of excessive primitive germ cells with impaired spermatocytic differentiation (Fig. 4D). TUNEL staining showed abundant apoptosis in the spermatogonial-like tumor cells, which spared the more differentiated spermatocytes (Fig. 4E).
It is possible that the failure of radiation therapy in the tumor shown in Figure 4C was due to the advanced stage of the tumor at the time of treatment. For better understanding of the natural history of tumorigenesis, including early detection and more accurate quantification of tumor response, a noninvasive imaging approach would likely prove useful. The use of microultrasound to image tumors in the zebrafish was recently described.40 We therefore asked whether this technique could be used to detect GCTs in living fish from the GCT-susceptible line. Figure 5 shows a male with a subclinical testicular mass. The tumor, which measured 2.1×6.3mm, can clearly be seen on the grayscale ultrasound image.
The zebrafish has recently emerged as a powerful system for cancer gene discovery.41 Long in use as a carcinogenesis model, zebrafish have proven amenable to forward and reverse genetic approaches to modeling human neoplasms. Previously, we reported a forward genetic screen for mutations affecting embryonic cell proliferation.32 In mice, mutations in oncogenes and tumor-suppressor genes often exhibit effects on embryonic cell proliferation. Therefore, one of the goals of the zebrafish screen was to determine whether adult zebrafish carrying cell proliferation mutations (including lamc1cz61, sfdcz213, mybl2cz226, espl1cz280, llgcz3322, sdscz319, and slhcz333) showed increased cancer susceptibility. We observed adult heterozygotes for spontaneous tumor incidence and also subjected all lines of fish to carcinogen exposure. Zebrafish heterozygous for loss-of-function mutations in mybl2 and espl1 were more susceptible to neoplasia than their wild-type siblings, identifying these genes as haploinsufficient tumor suppressors.32,33
Here we report that a high incidence of testicular GCTs was observed in progeny from the shortstop/lamc1cz61 line. The increased incidence was observed after both DMBA and MNNG exposure. There was also a high incidence of spontaneous testicular GCT in control vehicle-treated fish. There was a strong variation among shortstop/lamc1cz61 heterozygotes in the ability to transmit the testicular tumor trait, leading us to suspect that the testicular tumor trait might not be linked to the mutation causing the shortstop phenotype. Using microsatellite mapping and a candidate gene approach, we identified a splice-site mutation in lamc1 in shortstop. shortstop is thus allelic to sleepy/lamc1.35 We then showed directly that the GCT trait was not linked to shortstop/lamc1cz61. The F1 progeny of ENU-mutagenized males typically contains 100–200 different mutations,42,43 making it very likely that the GCT trait arose due to a second-site mutation that was transmitted to the original lamc1cz61 female.
The overall incidence of spontaneous testicular neoplasms in wild-type fish has been reported as <1%.25,44 In our analysis of all lines other than lamc1cz61, the incidence of spontaneous testicular GCT at 12 months was 0.37%. The finding that fish from the lamc1cz61 background had an incidence of spontaneous testicular GCT of 7% is thus significant. However, this analysis of the lamc1cz61 line includes both carriers and noncarriers of the GCT susceptibility trait. Through pedigree analysis we could identify both male and female carriers of the trait and find that the incidence of testicular GCT among outcross male progeny of carriers is at least 30–50% (Fig. 3). Because only one-half of these fish would have inherited the mutation causing the trait, we estimate the penetrance of the trait as at least 60%. Determination of the true penetrance will require identification of the mutant locus to allow unambiguous genotyping of affected and unaffected males. Preliminary evidence indicates that zebrafish homozygous for the trait are viable and exhibit increased penetrance and shorter latency of tumor onset (J.C. Neumann and J.F. Amatruda, in preparation).
The zebrafish GCT-susceptible line thus represents a novel, highly penetrant, heritable model of testicular GCT. Grossly, the lesions are nearly always bilateral, and the gonad is enlarged; the tumors can occupy the entire body cavity leading to visible abdominal distension, but do not invade other organs. At the histologic level, the tumors are notable for the loss of orderly cyst architecture and for partial or complete impairment of spermatocytic differentiation. The tumor cells resemble spermatogonia; by fluorescence-activated cell sorting analysis (not shown) the cells have 2n or 4n DNA content, suggesting they are premeiotic. The lack of spermatogenesis and the universal finding of gonad enlargement strongly suggest that the balance between germ cell self-renewal and germ cell differentiation is pathologically altered in the tumors. Other factors such as impaired apoptosis may play a role as well. Additionally, it is not known whether the primary defect in the GCT-susceptible line is germ cell autonomous or not. Reciprocal germline transplants between wild-type and GCT-susceptible fish, now ongoing, may help to clarify whether the product of the mutated gene is required in germ cells, somatic cells, or both.
Testicular neoplasms that occur spontaneously in fish >18 months of age or those occurring after carcinogenesis appear similar to the genetic model described here, with the typical findings of enlarged gonads and impaired spermatogenesis. The increased incidence of testicular GCTs in older fish may result from the accumulation of mutations, or alternatively from age-related changes in steroid hormones or other cell signaling pathways. The frequency of testicular GCTs after carcinogenesis suggests that multiple targets exist in which somatic mutations lead to a GCT phenotype. Similar sensitivity has been seen in other fish species such as rainbow trout and medaka.26,27 In rodents, carcinogen exposure frequently causes testicular atrophy, but testicular neoplasms are rare. Testicular neoplasms have, however, been reported in reverse genetic models in which HPV E6 and E7 or the growth factor glial cell-derived neurotrophic factor (GDNF) are overexpressed in the testis.45,46 These interspecies differences may reflect differences in carcinogen metabolism or other features of the testicular microenvironment, or may instead be due to particular features of the germ cells of teleosts that make them vulnerable to chemical mutagenesis.
No animal model that fully replicates the distinctive features of human GCTs currently exists. The tumors in the zebrafish GCT-susceptible mutant share important features with human testicular GCT, particularly seminomas and spermatocytic seminomas, including the accumulation of primitive, undifferentiated germ cells, the profound defect in spermatocytic differentiation, and (in the case of seminomas) the sensitivity to radiation treatment. However, important differences remain. The zebrafish tumors are not invasive and metastatic, and do not appear to differentiate into other tumor histologies such as embryonal carcinoma and teratoma, as do human seminomas. Regardless of these differences, the availability of a heritable, highly penetrant model of germ cell neoplasia may provide valuable insight into the origins of human GCTs. It is noteworthy, for example, that exposure of the lamc1cz61 line to either DMBA or MNNG resulted in enhanced germ cell tumorigenesis, indicating that the model could be used to understand gene–environment interactions in GCTs. The sensitivity of the zebrafish GCTs to radiation treatment appears to depend at least in part on an apoptotic response to DNA damage, consistent with models of human GCT treatment sensitivity. This idea can now be tested by crosses to lines deficient in TP5347 or other components of the DNA damage response. Additionally, the line could serve as a platform for forward genetic or chemical genetic strategies to identify interacting genes and potential new therapies. Ultimately, the zebrafish model described here may provide valuable insight into the molecular pathways that regulate the balance between self-renewal and differentiation in germ cells, both in normal development and disease.
We thank James Ziai, Kathryn Finley, and Michelle Wells for technical assistance; Jeffrey Kutok and Jonathan Glickman (Brigham and Women's Hospital, Boston) for advice on pathology; Derek Stemple (Wellcome Trust Sanger Institute) for sharing data before publication and for providing the lamc1 line; and Leonard Zon (Children's Hospital, Boston) for generously supporting the early phases of this project. This research was supported by a grant from the Lance Armstrong Foundation and by grant 1R01CA135731 from NIH/NCI.
No competing financial interests exist.