Cancer Cell Biology
Elevated levels and distinct patterns of expression of A-type cyclins and their associated cyclin-dependent kinases in male germ cell tumors
Version of Record online: 13 NOV 2003
Copyright © 2003 Wiley-Liss, Inc.
International Journal of Cancer
Volume 108, Issue 5, pages 654–664, 20 February 2004
How to Cite
Liao, C., Li, S. Q., Wang, X., Muhlrad, S., Bjartell, A. and Wolgemuth, D. J. (2004), Elevated levels and distinct patterns of expression of A-type cyclins and their associated cyclin-dependent kinases in male germ cell tumors. Int. J. Cancer, 108: 654–664. doi: 10.1002/ijc.11573
- Issue online: 19 DEC 2003
- Version of Record online: 13 NOV 2003
- Manuscript Accepted: 26 AUG 2003
- Manuscript Revised: 5 AUG 2003
- Manuscript Received: 10 APR 2003
- Lance Armstrong Foundation
- NIH. Grant Number: R01 HD34915
- Wenner-Gren Foundation
- Swedish Foundation for International Cooperation in Research and Higher Education (STINT)
- Swedish Cancer Society
- cyclin A;
- embryonal carcinoma;
- carcinoma in situ;
- yolk-sac tumor;
- Top of page
- MATERIAL AND METHODS
- Supporting Information
Aberrant expression of several key regulators controlling the G1/S phase of the cell cycle has been implicated in human male germ cell tumorigenesis. Given the critical role of cyclin A2 at both the G1/S and G2/M transitions and the essential role for cyclin A1 in male germ cell development, our present study focused on the involvement of the A-type cyclins in the transformation and progression of male germ cell tumors (GCTs). The expression of the A-type cyclins and their catalytic partners Cdk1 and Cdk2 was examined in all types and stages of human male GCTs, including carcinoma in situ(CIS), seminoma and non-seminoma GCTs, along with normal testis samples. Elevated levels of cyclin A2, Cdk1 and Cdk2 were detected in the majority of GCTs and were correlated with the invasiveness of the tumors (p < 0.05). Cyclin A1 expression was virtually undetectable in CIS and seminoma, but was aberrantly expressed in all non-seminomatous GCTs. Cyclin A2 expression was strongly correlated with that of its catalytic partners Cdk1 and Cdk2 in all types of testicular tumors examined (p < 0.05), whereas a strong correlation between cyclin A1 and Cdk1 or Cdk2 was only seen in non-seminomatous GCTs (p < 0.05). Histone kinase activities of cyclin A1/Cdks and cyclin A2/Cdks were found to be elevated in tumors. Our data suggest that aberrant expression of A-type cyclins and their Cdks is a significant factor in male germ cell tumorigenesis. The abundant ectopic expression of cyclin A1 in non-seminomatous GCTs and its absence in CIS and seminomas is likely linked to the tumor transformation and progression and may be relevant to clinical prognosis. Supplementary material for this article can be found on the International Journal of Cancer website at http://www.interscience.wiley.com/jpages/0020-7136/suppmat/index.html.© 2003 Wiley-Liss, Inc.
Human testicular germ cell tumors (GCTs) are the most common malignancies occurring in the young male population and their frequency is increasing.1, 2, 3 The tumors can be divided histologically into 2 primary categories, seminomas and non-seminomas. Seminomas account for half of all GCTs and retain the morphology of undifferentiated germ cells.2 Non-seminomatous GCTs are further subdivided according to the degree of differentiation, with embryonal carcinoma exhibiting the most primitive pattern and teratoma the most mature type of differentiation along diverse lineages.4 This class also includes extra-embryonally differentiated phenotypes such as choriocarcinoma and yolk-sac carcinoma.2 GCTs of all types are frequently associated with carcinoma in situ (CIS), also designated as male germ cell neoplasia, which progresses to invasive lesions.5 Both seminomatous and non-seminomatous GCTs have been suggested to arise from cytologically identical CIS lesions, indicating a common cell of origin of all GCTs.2 Although several risk factors for GCT development have been identified, which include cryptorchidism, spermatogenic or testicular dysgenesis and possible family inheritance, the molecular mechanisms underlying the pathogenesis of GCTs are poorly understood.
Cellular differentiation and proliferation is governed by the cell cycle machinery and involves progression through well-defined transition points: G1, S, M and G2 phases. The key regulators of this machinery are the cyclins, cyclin-dependent kinases (Cdks), CDK inhibitors, (pRb) and E2F.6, 7 The cyclins are usually classified into 2 categories: those that function at G1/S, including the D-type cyclins and cyclin E, and the mitosis-inducing cyclins, primarily the B-type cyclins, that function at the G2/M transition.8, 9 The somatic A-type cyclin, cyclin A2, is unique in that it is believed to function at both G1/S and G2/M.9, 10, 11, 12 Because cyclins and Cdks play critical roles in DNA synthesis and cell division, alteration of their function may lead to the disruption of normal cell division and cell growth and subsequently result in carcinogenesis. Several studies have examined the expression of some cell cycle regulators controlling the G1/S transition in established GCT cell lines and GCT samples from patients. These include cyclins A, B1, D1, D2 and E;13, 14, 15 Cdks 2, 4, 6;15 pRB16 and the Cdk inhibitors p21, p27, p18INK4c and p19INK4d.14, 17 The observation of elevated levels of cyclin D2 was of particular interest given that it maps to 12p13, and a high proportion of testicular tumors show increased copy numbers or duplications of 12p.2, 18
Although the possible role of the A-type cyclins in GCTs has not been considered in detail, there is evidence in support of their function during normal male germ cell development and possibly tumorigenesis. There are 2 mammalian A-type cyclins: the ubiquitously expressed cyclin A2 (referred to as cyclin A in Bartkova et al.14) and the novel cyclin A1.19, 20 In complexes with Cdk2, cyclin A2 is crucial for DNA replication and for the commitment to cell proliferation.21 Cyclin A2 also binds to Cdk1 and plays an essential role in the initiation of mitosis.8 Mammalian cyclin A1 was first identified in the mouse, where its expression is restricted to male germ cells in late meiotic prophase during normal development.19, 22 Human cyclin A1 is also expressed at highest levels in the testis,20 although the cellular specificity of expression in the human testis remains to be determined. The only other sites of very low levels of normal expression of human cyclin A1 that have been reported are brain20 and some population of hematopoietic cells.20, 23 While cyclin A2 appears to be almost ubiquitously expressed, in the mouse male germ line, its expression is restricted to spermatogonia and pre-leptotene spermatocytes and is excluded from later stages of spermatogenic differentiation.22 Both Cdk1 and Cdk2 are expressed in spermatogonia as well, and are more broadly expressed throughout meiotic prophase in spermatocytes in the mouse testis.22 Cyclin A1 and cyclin A2 can both associate with Cdk1 and Cdk2 to form active kinase complexes; however, both A1 and A2 appear to prefer Cdk2 as a partner in testicular cell lysates.19, 22, 24, 25 Finally, gene targeting has revealed an essential role for cyclin A1 in the transition of spermatocytes through G2/M of the first meiotic division.24 Mice lacking functional cyclin A1 are overtly healthy and the females are fertile. The males are sterile, however, exhibiting an arrest at the diplotene stage followed by apoptosis.
Cyclin A2 was one of the first cyclins to be implicated in tumorigenesis: the cyclin A2 gene was found to be a site of integration of hepatitis B virus in a clonal hepatoma.26 The integration resulted in a chimeric protein that lacked the cyclin destruction box and had altered turnover properties and the cells had defects in the G2/M checkpoint.12 Altered expression of cyclin A2 has been reported in a wide range of different types of human tumors, such as astrocytomas,27 colorectal adenocarcinoma,28 soft tissue sarcoma29 and prostate cancer.30 The observation of the expression of cell cycle regulating genes, such as cyclin A2, in tumors may have practical prognostic impact in addition to providing insight into the mechanisms underlying the pathogenesis of the tumor. For example, the correlation of high levels of cyclin A2 and a high level of proliferation can predict a favorable chemotherapy response in patients with soft tissue sarcomas.31 Conversely, expression of cyclin A2 has been correlated with poor patient survival in renal cancers32 and colorectal adenocarcinoma.28
Abnormal expression of cyclin A1 has also been linked to tumorigenesis. Elevated levels of cyclin A1 were detected in hematologic malignancies with myeloid differentiation,23 especially in certain leukemias blocked at the myeloblast and promyelocyte stages of differentiation.33 Further, targeted overexpression of cyclin A1 in early myeloid cells initiated acute myeloid leukemia in transgenic mice.34
Given the absolute requirement for cyclin A1 for normal spermatogenesis and the unique patterns of expression of the A-type cyclins in the mouse and the observation that cyclin A1 is also expressed in the testis of humans, we wished to establish the cellular specificity of expression of the A-type cyclins and their interacting Cdks in normal human testis. Further, the correlation of ectopic cyclin A1 expression and the development of myeloid leukemia20, 23, 33, 34 and the potential utility of A-type cyclins in assessing tumor status and patient outcome prompted examination of A-type cyclin and Cdk expression in tumors originating from male germ cells. Our initial studies focused on a series of male GCTs including seminomas, embryonal carcinomas, teratomas, choriocarcinomas and yolk-sac carcinomas, compared to normal testes, and with the pre-invasive testicular CIS. Our observations provide evidence for a highly conserved cellular specificity of expression, and hence potential function, of the cyclins and Cdks between mouse and human spermatogenic cells. We further observed a high level of expression of cyclin A2 and A-type cyclin associated Cdks in male GCTs, similar to that observed in other tumors, but a striking, ectopic expression of the germ cell-specific cyclin A1 in non-seminomatous GCTs but not in CIS and seminomas. We also found that A-type cyclins/Cdks complexes had increased activities in non-seminomatous GCTs.
MATERIAL AND METHODS
- Top of page
- MATERIAL AND METHODS
- Supporting Information
Tissues and sections
Frozen tumor specimens and paraffin embedded sections of seminoma and embryonal carcinoma were obtained as archival specimens from the Department of Pathology, College of Physicians and Surgeons, Columbia University. The sections of testicular CIS, teratoma, yolk-sac tumor and choriocarcinoma were collected from the Department of Pathology, Lund University, University Hospital, Malmö, Sweden. Two normal testes were obtained from Columbia and 8 from Lund. The surgically resected specimens were fixed with 8% paraformaldehyde in PBS at 4°C overnight. For each sample, all available hematoxylin and eosin stained sections were reviewed by 2 of the authors (S.Q.L. and A.B., pathologists), and a representative block was selected for immunohistochemic analysis. The tissue microarrays (TMAs) were prepared as previously described.35 Briefly, tissue cores with a diameter of 5 mm were punched from a selected area of each “donor” block and brought on to a recipient paraffin array block. Each TMA contains 15–25 tissue core samples. The TMAs were made in multiple copies for immunohistochemic analysis using different antibodies. Tumors were diagnosed and staged by independent pathologists according to the WHO International Histological Classification of tumors36 and American Joint Committee on Cancer (American Joint Committee on Cancer, 1988).
Source of antibodies
Antisera specific for cyclin A1 were raised in a rabbit immunized with a GST fusion protein containing amino acids 3–204 of the murine cyclin A1.24 The antisera were affinity-purified by incubation with strips of nitrocellulose membrane blotted with the antigen fusion protein. Monoclonal anti-human cyclin A1 BD Biosciences Pharmingen (San Diego, CA) was also used. Three sources of antibodies against cyclin A2 were used for immunohistochemistry. One polyclonal antibody, which was previously used in our studies on mouse testis,22 was kindly provided by Dr. Mark Carrington.19 One monoclonal antibody was purchased from BD Biosciences Transduction Laboratories (San Diego, CA) and one polyclonal antibody from Upstate USA, Inc. (Lake Placid, NY) against Cdk1 and Cdk2 were from Upstate USA, Inc. (Lake Placid, NY). For immunoblotting analyses, peroxidase-conjugated goat anti-rabbit IgG (Boehringer Mannheim) was used as secondary antibody at a dilution of 1:1500.
Deparaffinization of 7-μm sections of paraffin embedded tissue samples was performed in Histoclear for 5 min, then in ethanol with each concentration of 95%, 85%, 75% and 50% for 2 min. After deparaffinization, antigen retrieval was performed by boiling the slides in 0.01 M citrate buffer, pH 6.0, in a microwave for 10 min. The slides were then treated with 0.03% H2O2 in methanol for 20 min and washed with PBST (1× PBS, 0.1% Triton X-100). Before adding the primary antibody, the slides were blocked for 2 hr with 2.5% goat serum in PBST. Primary antibodies were applied on the slides and the slides were placed in a humid chamber overnight at 4°C. Anti-cyclin A1 polyclonal antibody was diluted 1:100. Anti-cyclin A2 polyclonal or monoclonal antibodies from 3 different sources were diluted 1:250. Anti-Cdk1 and anti-Cdk2 antibodies from Upstate USA, Inc. (Lake Placid, NY) were used at dilution of 1:250. The slides were then washed 3 times with PBST and stained with the Vectastain ABC kit (Vector Laboratories, Burlingame, CA). The slides were then dipped in hematoxylin solution for 1 min, washed with H2O extensively, treated briefly with acid H2O, rinsed with H2O for 10 min, treated with ammonia-H2O briefly, and then washed with H2O for 10 min. The slides were finally washed with 50%, 70%, 85%, 95% and 100% ethanol each for 2 min, dipped into Histoclear and then air dried and mounted in Permount. The specimens were viewed with a Nikon 800 microscope.
The expression of the A-type cyclin and Cdk proteins was evaluated in the entire section. The high level expression of proteins in tumor cells was defined as a stronger staining intensity than that in adjacent normal testis on the same section. The staining signal was then graded according to the proportion of positively-staining tumor cells (%) and staining intensity on a scale of 0 to 4. Expression of the proteins in tumor cells was evaluated as positive if the staining intensity was as strong as that in the normal testis on the same section, and the stained cells were distributed in > 10% of tumor area. At least 1,000 cells were counted and the fraction of positive tumor cells was expressed as a percentage of the total number of cells counted. These percentages and intensities were used in the statistical analysis in Tables I and II. Because of space limitations, the list of all samples examined can be obtained as supplementary information, Tables I to V, from the corresponding author (D.J.W.) (This supplementary material can be found on the International Journal of Cancer website at http://www.interscience.wiley.com/jpages/0020-7136/suppmat/index.html. The subcellular distribution, either nuclear, cytoplasmic or both, was evaluated at 100 X magnification.
|Cyclin A2||–||0.56* (31%)||0.78** (61%)|
|Cdk1||0.56* (31%)||–||0.69* (48%)|
|Cdk2||0.78** (61%)||0.69* (48%)||–|
|Index||Cyclin A1||Cyclin A2||Cdk1||Cdk2|
|Cyclin A1||–||0.418** (17%)||0.475** (23%)||0.314* (9%)|
|Cyclin A2||0.418** (17%)||–||0.643** (41%)||0.529** (28%)|
|Cdk1||0.475** (23%)||0.643** (41%)||–||0.649** (42%)|
|Cdk2||0.314* (9%)||0.529** (28%)||0.649** (42%)||–|
The significance of differences in the levels of expression of proteins among tumors exhibiting different pathologic parameters was analyzed using the Mann-Whitney test. Possible pair-wise correlations between the expression of each of the 4 cell cycle regulators were analyzed using the Spearman Rank Correlation Test, and the linear regression and the estimated curve regression analyses were further performed. The strength of association between the continuous variables is indicated with r in the form of a percentage. Statistical analyses were performed using SPSS version 9.0 (SPSS Inc., Chicago, IL).
Tissue samples were homogenized by sonication in ice-cold lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH 8.0, 1% NP-40, 2 mM EDTA, 10 mM NaF, 1 mM NaVO4, 1 mM DTT) containing 10 μg/ml aprotinin, 25 μg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF) and 5 μg/ml pepstatin. The extracts were centrifuged at 12,000g for 5 min at 4°C. The protein concentration of the supernatant fraction was determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Richmond, CA). Protein (100 μg) was subjected to 7.5% SDS-PAGE. The subsequent blots were probed with primary antibodies. Cyclin A1 and cyclin A2 were detected using 1:500 dilutions24 and Cdk1 and Cdk2 at 1:1,000 dilutions. The blots were then washed and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Roche-Boehringer Mannheim, Indianapolis, IN) and visualized using the Enhanced Chemi-Luminescence (ECL) detection system (Amersham, Biosciences, Piscataway, NJ) and Kodak X-OMAT films.
Co-immunoprecipitation and kinase assay
Protein lysates were prepared as described above. Monoclonal anti-cyclin A1 or anti-cyclin A2 antibodies (2–3 μg) together with 30 μl protein G-sepharose beads were added to 1 mg lysates. After rocking the samples for 2 hr at 4°C, the immune complexes were washed 3 times with lysis buffer as described above and eluted by adding 2× sample buffer and boiling for 10 min. The samples were then subjected to 12% SDS-PAGE. For kinase assays, cyclin A1 or cyclin A2 associated complexes were first co-immunopurified with monoclonal antibodies against these proteins. Beads containing the immunocomplexes were washed extensively 3 times in lysis buffer and twice in EB kinase buffer (50 mM Hepes, pH 7.5, 10 mM MgCl2, 80 mM B-glycerophosphate, 20 mM EGTA, 1 mM DTT). The kinase reactions were performed by adding 40 μl kinase buffer (10 μM ATP, 5 μM cAMP-protein kinase inhibitor (Sigma-Aldrich, St. Louis, MO), 50 μg/ml calf thymus DNA), histone H1 (Upstate USA, Inc.) and 2.5 μCi of γ32P ATP to the beads. The samples were then incubated at 30°C for 20 min. The reactions were stopped by adding 15 μl of 2× SDS sample buffer followed by boiling for 5 min. The eluted proteins from beads were separated by SDS-PAGE followed by drying the gels and exposure to autoradiography.
- Top of page
- MATERIAL AND METHODS
- Supporting Information
Expression of A-type cyclins and their known Cdk partners in normal human testes
Cyclin A1 was reported to be expressed in the human testis;20 however, the cellular and subcellular distribution of the A-type cyclins and their associated cyclin-dependent kinases in human testis has not been determined. We examined the expression of these proteins in human testes from 10 adult males without history of testicular tumors, using immunohistochemistry, in order to accurately evaluate their role in tumorigenesis of male GCTs. The general morphology of human adult testis is similar to that of the mouse, and all of the stages of spermatogenesis are similarly represented. However, the cellular associations that characterize the rodent testicular epithelium are not found in the human; it is therefore not possible to define the stages of meiotic cell cycle progression as precisely.37 In general, spermatogonial stem cells that divide mitotically distribute in the more peripheral zone of the seminiferous tubules cut in a cross-section, and primary spermatocytes, which undergo a prolonged prophase, are localized in the next layer toward the lumen. Spermatocytes undergoing the first and second meiotic division are distributed in the next layer, and the spermatids that are eventually produced from the spermatocytes occupy the most central region of the tubules.
The first cells in the human spermatogenic lineage to express cyclin A1 were the pachytene spermatocytes (Fig. 1a and 1b), a localization that was very similar to that observed in adult mouse.19, 22 No cyclin A1 protein was detected in spermatogonia or in the somatic compartment of the testis. Low levels of cyclin A1 protein appeared to remain in the early round spermatids of the human testis (Fig. 1b) but have not been observed in the mouse.22 Similar to the mouse spermatocytes, the localization was predominantly nuclear (Fig. 1b), with only weak cytoplasmic staining noted. In contrast to cyclin A1, cyclin A2 protein was readily detected in spermatogonia (Fig. 1c and 1d) and in occasional somatic cells in the interstitial region (Fig. 1d). There was some weak staining in what appear to be early pre-leptotene spermatocytes (data not shown); however, cyclin A2 was not detected in obvious pachytene spermatocytes (Fig. 1d).
Both Cdk1 and Cdk2 have been shown to be associated with cyclin A1 in mouse testis by immunoprecipitation,19, 25 while cyclin A2 appears to associate preferentially with Cdk2 as compared to Cdk1 in mouse testis.22 To confirm that the putative partners for the A-type cyclins are indeed expressed in the appropriate testicular cell types, we examined the cellular and subcellular localization of Cdk1 and Cdk2 by immunohistochemistry. Cdk1 was expressed in all the recognizable stages of spermatogenesis, including spermatogonia, with the intensity of staining notably stronger in round spermatids (Fig. 1e) and pachytene spermatocytes (Fig. 1f). Cdk1 appeared to be present in both the nuclear and cytoplasmic compartments in the spermatocytes (Fig. 1f). Cdk2 expression was also detected in spermatogonia and spermatocytes and exhibited a predominantly nuclear distribution (Fig. 1g and 1h). The overall distribution of Cdk1 and Cdk2 in human male germ cells was similar and was comparable to that was observed in the mouse testis.22
Expression of A-type cyclins and Cdks in CIS lesions and seminomas
Testicular CIS is believed to be the common precursor lesion of the seminoma and non-seminomatous GCTs. To study whether there are any alterations of the A-type cyclins and Cdks in male GCTs and whether these alterations are related to the tumor formation and progression, we first investigated the correlation of expression of the A-type cyclins expression with CIS (n = 15), a noninvasive lesion, and seminoma in which tumor cells are invasive and aggressive (n = 17). Seminoma are more like germ cells in appearance and display little evidence of differentiation (reviewed previously2, 38). Immunohistochemic analysis of 15 CIS revealed that cyclin A1 is absent in the majority of CIS lesions. Only very weak expression of cyclin A1 was seen in occasional cells in only 2 out of 15 (13%) of CIS specimens (Fig. 2a). The majority of the seminomas also showed no staining with the cyclin A1 antibody (Fig. 3a), similar to the pattern in CIS, with only 2 of 17 (8%) showing very weak cyclin A1 expression. Cyclin A1 protein was detected in pachytene spermatocytes of the paired adjacent normal testes, indicating that the absence of cyclin A1 in the majority of tumor cells in the majority of the seminomas was not due to experimental artifacts. In contrast to cyclin A1 protein, cyclin A2 was readily detected in the majority (75%) of the CIS samples (Fig. 2b). Elevated levels of cyclin A2 were detected in tumor cells (Fig. 3b) and in almost all (88%) of the seminomas examined. One of the specimens was subsequently identified as containing an atypical, very poorly differentiated seminoma. Comparison of the relative intensity of cyclin A2 staining with the adjacent normal testis and CIS cells revealed that the overall intensity of cyclin A2 staining is much stronger in seminomas. High levels of cyclin A2 expression were found in 100% of the tumors with lymph node and distant metastases. The subcellular localization of cyclin A2 in the tumor cells was predominantly nuclear but was also detected in the cytoplasm in CIS (Fig. 2b) and seminomas (Fig. 3b).
We next examined the expression of the catalytic partners of the A-type cyclins, Cdk1 and Cdk2, in CIS lesions and seminomas. Cdk1 and Cdk2 were detected in the majority of the CIS lesions (Fig. 2c and 2d) and seminomas (Fig. 3a and 3b). As observed for cyclin A2, the tumor cells in seminomas stained more intensely for Cdk1 and for Cdk2 than the spermatogenic cells in the adjacent normal tubules and CIS cells (Fig. 4). The expression of these 2 proteins was also found in all tumors with lymph node and distant metastases. Cdk1 was localized to both cytoplasm and nucleus, but was predominant in cytoplasm, with only weak nuclear staining seen (Fig. 4a). In contrast, Cdk2 protein was predominantly localized in the nuclei of the tumor cells (Fig. 4a).
Expression of A-type cyclins and Cdks, and the altered activities of cyclin A1/Cdk and cyclin A2/Cdk complexes in non-seminomatous GCTs
In contrast to seminomas, which retain a somewhat germ cell-like appearance, embryonal carcinomas exhibit morphologies resembling mitotic cells in stages of primitive postzygotic differentiation. It has been postulated that embryonal carcinoma cells undergo somatic and extrasomatic differentiation and thus form teratomas.5 In normal spermatogenesis, cyclin A1 is required for the first meiotic division,24 while cyclin A2 is thought to play a role in the mitotic spermatogonia and in pre-leptotene spermatocytes.22 We thus predicted that the “embryonic-appearing” embryonal carcinomas would express cyclin A2, possibly at higher levels than in seminomas, and would not express the meiosis-specific cyclin A1. To test this hypothesis and to search for the tumor-associated elevated expression of A-type cyclins and their associated Cdks in non-seminomatous tumors, 13 embryonal carcinomas, 21 teratomas, 16 yolk-sac carcinomas and 7 choriocarcinomas were examined for cyclin A1, cyclin A2, Cdk1 and Cdk2 expression.
As expected, cyclin A2 protein was strongly detected in all the samples of teratomas (Fig. 3e), yolk sac carcinomas (Fig. 3g) and choriocarcinomas (Fig. 3i) and the majority of embryonal carcinomas (Fig. 3c) (10 out of 13). Surprisingly, however, elevated expression of cyclin A1 protein was also detected in these tumors (12 of the 13 embryonal carcinomas, 100% of other tumors). Using TMA approaches, tumor samples and the paired adjacent normal testicular tubules from different types and stages were assembled on the same slide and the intensity of the immunostaining of different antibodies was evaluated. Cyclin A1 protein was localized in the nuclei and cytoplasm of tumor cells in embryonal carcinoma (Fig. 3c), while it was predominantly cytoplasmic in the tumor cells from the other types of tumors (Fig. 3e, 3g, 3i). Cyclin A2 was detected in both the nuclear and cytoplasmic compartments of all tumors (Fig. 3b, 3d, 3f, 3h, 3j).
Immunohistochemic analysis of the non-seminomatous carcinomas revealed the elevated expression of Cdk1 protein in 12 of the 13 embryonal carcinomas, 20 of 21 teratomas and in all of the yolk sac carcinomas and choriocarcinomas examined, while increased Cdk2 was present in all these tumor samples. The subcellular distribution of Cdk1 was predominantly cytoplasmic (Fig. 4, left panels), while in contrast, the intensely localized staining of Cdk2 was predominantly nuclear and the cytoplasm was weakly stained (Fig. 4, right panels). The strong expression of cyclin A1, cyclin A2, Cdk1 and Cdk2 was detected in 100% of the tumor samples in which lymph node and distant metastases had been noted (data not shown).
To further verify the apparent high levels of expression of the A-type cyclins, Cdk1 and Cdk2, immunoblotting was performed in selected tumor samples (Fig. 5). Densitometric quantification of the bands on the immunoblot revealed that the tumor tissues displayed higher levels of expression of these proteins than in the adjacent normal testis sample, consistent with the observations by immunohistochemic analysis. The immunoblot analysis also confirmed the absence of detectable cyclin A1 expression in seminomas (Fig. 5).
To evaluate whether A-type cyclins were able to form complexes with their Cdk partners in tumors and whether the elevated expression of the A-type cyclins resulted in a loss of the association with Cdks, co-immunoprecipitation was further performed using tumor samples and the normal adjacent testis from a patient with embryonal carcinoma. Cdk2 was detected in cyclin A1 associated immunocomplexes from both normal (Fig. 6a) and tumor samples (Fig. 6b). The presence of Cdk2 in cyclin A2-associated complexes from normal (Fig. 6c) and tumor samples (Fig. 6d) was also confirmed. This result was reproduced using both normal and tumor samples from the additional 2 patients (data not shown) and suggested that complexes between cyclin A1 and Cdks or cyclin A2 and Cdks are not disrupted in non-seminomas. To investigate whether the activities of A-type cyclins/Cdks complexes were altered in tumors, histone kinase assays were performed using tumor and the paired adjacent normal samples from 3 patients with embryonal carcinoma. Cyclin/Cdk immunocomplexes were co-precipitated from the same amount of tumors and normal samples; one portion of the immunoprecipitates was subjected to immunoblotting analysis using antibodies against A-type cyclins, Cdk1 and Cdk2. The bands indicated that these proteins from the tumors and normal samples showed similar intensity, suggesting that similar amounts of cyclin/Cdk immunocomplexes were immunoprecipitated from the tumor and normal samples. However, histone H1 was more heavily phosphorylated in the tumor samples (Fig. 7), suggesting the presence of an increased kinase activity of cyclin A1/Cdks and cyclin A2/Cdks in tumors.
Distinct expression of A-type cyclins and Cdks in mixed germ cell tumors
To further verify that A-type cyclins and their associated Cdks exhibit distinct expression patterns in preinvasive CIS and invasive seminomas or non-seminomatous GCTs, we examined 5 samples containing mixed GCTs. On the same section where CIS, seminoma, embryonal carcinoma, teratoma and yolk sac carcinoma were all found, cyclin A1 showed no staining for CIS and seminoma but gave strong staining for embryonal carcinoma, teratocarcinoma and yolk sac carcinoma (data not shown). The staining intensity was much higher than that was shown in the adjacent normal testis. In contrast to cyclin A1, cyclin A2, Cdk1 and Cdk2 were detected in CIS and were consistently overexpressed in seminomas or non-seminomatous GCTs.
Correlations between expression of the A-type cyclins and their kinase partners, Cdk1 and Cdk2, and tumor pathology
To determine whether the levels of expression of cyclin A1 and cyclin A2 were correlated with the stages and severity of the tumor, we subdivided the seminomas and embryonal carcinomas into 2 groups according to tumor pathology, characterized by (i) invasion of the rete testis; (ii) invasion of the tunica albuginea; and (iii) invasion into the vascular space in veins and lymphatics (data not shown). We then performed the Mann-Whitney Test to analyze if there were significant differences in the expression of those proteins between the highly invasive and less invasive seminomas and embryonal carcinomas. Statistical analysis was performed to compare the differences of the proportion of cells expressing the 4 cell cycle proteins between the tumors with or without the characteristics mentioned above (data not shown). Significant differences were observed in the expression of cyclin A2 (Mann-Whitney U [U] = 5, p = 0.02) and Cdk2 (U = 9, p = 0.04) between the tumors with or without tunica albuginea invasion. Even higher significant differences were observed in the expression of the 4 cell cycle proteins between the invasive tumors characterized with vascular space invasion present in veins and lymphatics and tumors without metastasis (cyclin A1, U = 0, p = 0.01; cyclin A2, U = 0.5, p = 0.02; Cdk1, U = 0.5, p = 0.02; Cdk2, U = 0, p = 0.02). However, there were no significant differences in the protein expression between the tumors with and without the characteristics of rete testis invasion for cyclin A1 (U = 18, p = 0.72) and Cdk1 (U = 19, p = 0.87). These results suggested that the expression of the A-type cyclins and Cdks is significantly higher in tumors with tunica albuginea invasion and the presence of vascular space invasion in veins and lymphatics than tumors without those characteristics.
To determine whether there is a correlation of high levels of expression of the A-type cyclins and their Cdk partners in GCTs in general, thus suggesting a possible role in the pathogenesis of this class of tumors, rank correlation tests were performed on the 17 seminomas and non-seminomatous GCTs (13 embryonal carcinomas, 21 teratomas, 16 yolk sac carcinomas and 7 choriocarcinomas) listed in Table I and Table II. The results revealed that there were strong correlations of expression between cyclin A2 with Cdk1 (rs = 0.56), between cyclin A2 and Cdk2 (rs = 0.78) and between Cdk1 and Cdk2 (rs = 0.69) in seminomas (Table I). There was also significant correlation between A-type cyclins and Cdks expression in non-seminomatous GCTs (p < 0.05 in Table II). Both linear and curve regression analyses were further performed to predict whether the expression of any of the 4 cell cycle proteins are dependent on each other. There was a linear correlation between A-type cyclins and Cdks shown in the regression graphs (data not shown). These results indicate that the simultaneous strong expression of cyclin A2, Cdk1 and Cdk2 might be characteristic of seminomas, while a simultaneous expression of these 3 proteins plus cyclin A1 is characteristic of non-seminomatous GCTs in human testis.
- Top of page
- MATERIAL AND METHODS
- Supporting Information
In our study, we have defined the differential pattern of expression and their subcellular localization of the cyclin A1, cyclin A2, Cdk1 and Cdk2 proteins in normal human adult testis, providing insight into their possible function in spermatogenesis. We have further shown the abnormal expression of these proteins in male testicular tumors of different stages and types including CIS, seminomas and several non-seminomatous carcinomas. Our analysis revealed the remarkable observation that the novel A-type cyclin, cyclin A1, which is normally meiosis-restricted, is aberrantly expressed in all types of non-seminoma GCTs, which are the aggressive tumors that are resistant to radiotherapy. However, cyclin A1 is absent in nonaggressive CIS and seminomatous GCTs, both of which are sensitive to the treatment of radiotherapy. An additional new finding shown in our results is that although cyclin A1 and cyclin A2 are able to interact with Cdk2 in non-seminomatous carcinomas as they do in normal testes, the activities of cyclin A1/Cdks and cyclin A2/Cdks are altered in tumors.
The cellular specificity of expression of cyclin A1, cyclin A2, Cdk1 and Cdk2 proteins in normal mouse testis has been determined.22 The abundant expression of cyclin A1 is restricted to late pachytene and diplotene spermatocytes and is predominantly present in the nucleus. In our present study, we now extend this remarkably restricted expression pattern to the human testis as well. Our previous molecular genetic studies have further demonstrated an absolute requirement for cyclin A1 in the progression of spermatocytes into the first meiotic division.24 Mice lacking cyclin A1 exhibit a block in spermatogenesis, resulting in male-specific infertility, but are otherwise healthy. Whether the absence of cyclin A1 function would also result in human male-specific sterility is of considerable interest, as is the question of whether cyclin A1 also functions in the first meiotic division in the human. In both mouse and human, the presence of cyclin A1 only in cells that have undergone DNA synthesis and are about to undergo meiotic divisions indicate a meiotic, G2/M-specific and no G1/S function for this distinct cyclin in the normal male germ line. We were thus surprised to observe the high levels of cyclin A1 expression in the embryonal carcinomas, teratomas, yolk sac carcinomas and choriocarcinomas examined in our present study (discussed further below).
In contrast, in both mouse22 and human (our present study), expression of cyclin A2 appears to be limited to mitotic spermatogonia and possibly early (pre-leptotene) spermatocytes. The function of cyclin A2 in spermatogenesis is not known, as the null mutation of Ccna2 resulted in early embryonic lethality, the basis of which remains to be elucidated.39 However, the presence of cyclin A2 in specific and nonoverlapping (with cyclin A1) cell types in the testis may indicate unique and evolutionarily conserved functions for these 2 distinct A-type cyclins in male germ cell development. Their differential expression patterns may correlate with their specific roles in the pathways of spermatogenesis. The distribution of the A-type cyclins' catalytic partners Cdk1 and Cdk2 in the adjacent normal human testicular tubules resembled the pattern of expression in the mouse, being present in both mitotic and meiotic germ cells.
In human male GCTs, the normal pathway of spermatogenesis has been disrupted, and germ cells that would have normally entered a terminal meiotic differentiation now yield tumor cells of unlimited proliferative capacity. It has been postulated that the etiology of testicular germ cell tumors includes several discrete steps as follows.5 An “initiation stage,” equivalent to parthenogenetic activation of germ cells is followed by a “promotion stage,” which results in recognizable intracellular carcinoma cells but not obvious invasiveness, and is termed as CIS. Next is a “progression stage,” in which CIS develops into an invasive carcinomas. The absence or downregulation of cyclin A1 in CIS and seminoma indicates that these 2 types of tumor may be independent of pathways involving cyclin A1. However, in non-seminomatous GCTs and mixed GCTs that included non-seminomatous GCTs, cyclin A1 exhibited elevated levels, suggesting that cyclin A1 may be involved in tumor progression and invasion of these types of tumors. Although the biologic and genetic mechanisms underlying the development of germ cell tumors remain enigmatic, it is believed that both seminoma and non-seminomatous carcinoma are directly derived from CIS.5 It is also postulated that seminoma is a prerequisite stage for formation of other non-seminomatous carcinomas.
Given the unique function of A-type cyclins in spermatogenesis and in G1/S and G2/M cell cycle transition, our results suggest that the abnormal expression of A-type cyclins and their associated Cdks might also be a critical event of male GCTs oncogenesis. Results from other research groups have suggested that aberrant expression of cyclin D2 resulting from gene amplification is an early event in human male germ cell tumorigenesis.13 Overexpression of cyclin D2 and Cdk4 at both mRNA and protein levels were found in seminomas and non-seminomas from a large numbers of patients, implying that deregulation of the G1-S checkpoint may be an important event for the development of GCTs.15 Cyclin D2 expression in normal human testis was occasionally found in spermatogonia2 and was also reported to be present in spermatids.15 The expression pattern of cyclin D2 is quite different from that observed for the A-type cyclins, indicating that they may be involved with specific stages of spermatogenesis and possibly tumor formation as well. Our data on the high levels of expression of cyclin A2 in seminomas are consistent with a previous observation,14 but also reveal that cyclin A2 is expressed in CIS cells and in non-seminomatous GCTs. Its expression appeared to be more abundant in aggressive tumors, pointing to its role in tumor progression. Cyclin A2 may contribute to tumor cell differentiation, because cyclin A2 expression decreased in the human NTera2/D1 teratoma cell line when the cells were induced to differentiate by the addition of retinoic acid.14 These data further suggest that aberrant expression of cell cycle regulators in male GCTs is likely correlated with the disruption of germ cell differentiation.
In parallel with the high levels of expression of A-type cyclins in male GCTs, we observed correspondingly high levels of expression of the Cdk proteins. Our observation on the localization of Cdk2 in normal testis and the protein level of Cdk2 in different types of GCTs is in contrast to the report of Schmidt et al.15 Schmidt et al. reported that Cdk2 protein was mainly present in Sertoli cells in normal testis, cells that are not dividing. Further, this observation differs from previous results in the mouse model, where Cdk2 is clearly detected at both the mRNA40 and protein levels.22 Cdk2 was down-regulated at the mRNA level in certain numbers of samples from different types of GCTs compared to the normal adjacent tissues.15 However, it is not clear whether Schmidt et al.15 examined Cdk2 at the protein level in large numbers of tumor samples and whether their results on Cdk2 at protein level in tumor cells are representative. Among all types of GCTs in a large number of patient samples examined, we found that Cdk2 protein was absent in occasional tumors, although it is highly expressed in the majority of tumors.
We also observed an alteration of the intracellular distribution of the A-type cyclins and Cdks in tumor cells, as compared to adjacent normal testicular cells. In seminomas and non-seminomatous carcinomas, the tumor cells exhibited a predominantly cytoplasmic localization of Cdk1, and both nuclear and cytoplasmic localization of cyclin A2, while in normal testis, the same proteins were mainly nuclear. In embryonal carcinomas, the subcellular distribution of cyclin A2 and Cdk1 was also altered, in that they were predominantly cytoplasmic, as compared to their nuclear localization in the normal testis. A predominantly cytoplasmic localization of Cdk1 has also been observed in oral carcinoma cells in which high levels of expression of cyclin A2 and cyclin B1 proteins were noted.41 In normal oral epithelium, Cdk1 is mainly nuclear. It was suggested that the cytoplasmic Cdk1 might be nonfunctional or incapable of binding to cyclin B1.41 We show that both cyclin A1 and cyclin A2 were able to bind to Cdk2 in samples from several patients with non-seminomatous GCTs; however, the activities of cyclin A1/Cdks and cyclin A2/Cdks were increased in tumors. Because an abnormal subcellular localization of these proteins was detected in the tumors, this may affect the relative kinase activities of the complexes as well as their downstream targets.
In conclusion, the expression of cyclin A1, cyclin A2, Cdk1 and Cdk2 was closely related to malignant cellular features in male GCTs. One of the striking observations of our study was the characteristic difference in the expression pattern of cyclin A1 between seminomas and non-seminomatous GCTs. The level and timing of cyclin A1 expression in normal human testis is consistent with its documented role in normal mouse spermatogenesis: progression into the first meiotic division. Its aberrant, robust expression in non-seminomatous GCTs and its down-regulation in the majority of CIS and seminomas may be either a cause or the consequence of tumorigenesis in GCTs. Furthermore, its expression may contribute to the more aggressive nature and more poorly differentiated properties of embryonal carcinomas. Our understanding of the initiation and progression of male GCTs may greatly benefit from systematic studies on the key cell cycle regulators in male germ cells and in subtypes of male GCTs. Elucidation of the alterations of cell cycle regulators, particularly proteins involved in both G1/S and G2/M transitions, will provide insight into the pathogenesis of male germ cell tumorigenesis.
- Top of page
- MATERIAL AND METHODS
- Supporting Information
We thank Ms. E. Laurion and Ms. R. Arbing for help in preparing the figures, Dr. B. Persson for help in the statistical evaluations, Ms. E. Nilsson for technical help and Dr. P. Mikulowski and Dr. G. Landberg for valuable suggestions. This work was supported in part by a grant from The Lance Armstrong Foundation to D.J.W., a grant from NIH (R01 HD34915) to D.J.W. and by grants from the Wenner-Gren Foundation, The Swedish Foundation for International Cooperation in Research and Higher Education (STINT), NFR and The Swedish Cancer Society to C.L.
- Top of page
- MATERIAL AND METHODS
- Supporting Information
- 5Pathobiology of human germ cell neoplasia. In: DamjanovI, ed. Recent results in cancer research. vol. 123. Berlin Heidelberg: Springer-Verlag, 1991., .
- 36World Health Organization international histological classification of germ cell tumors of the testes. vol II. Oxford: Pergamon, 1986., , .
- Top of page
- MATERIAL AND METHODS
- Supporting Information
|suppmat_page.pdf||486K||Supporting Information file suppmat_page.pdf|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.