d-type cyclins in adult human testis and testicular cancer: relation to cell type, proliferation,...

J. Pathol. 187: 573–581 (1999)

D-TYPE CYCLINS IN ADULT HUMAN TESTIS ANDTESTICULAR CANCER: RELATION TO CELL TYPE,

PROLIFERATION, DIFFERENTIATION, ANDMALIGNANCY

1, - 2, . Æ2 1*1Department of Cell Cycle and Cancer, Institute of Cancer Biology, Danish Cancer Society, DK-2100 Copenhagen, Denmark

2Copenhagen University, Department of Growth and Reproduction, Rigshospitalet, DK-2100 Copenhagen, Denmark

SUMMARY

D-type cyclins are proto-oncogenic components of the ‘RB pathway’, a G1/S regulatory mechanism centred around the retinoblastomatumour suppressor (pRB) implicated in key cellular decisions that control cell proliferation, cell-cycle arrest, quiescence, anddifferentiation. This study focused on immunohistochemical and immunochemical analysis of human adult testis and 32 testiculartumours to examine the differential expression and abundance of cyclins D1, D2, and D3 in relation to cell type, proliferation,differentiation, and malignancy. In normal testis, the cell type-restricted expression patterns were dominated by high levels of cyclin D3in quiescent Leydig cells and the lack of any D-type cyclin in the germ cells, the latter possibly representing the only example of normalmammalian cells proliferating in the absence of these cyclins. Most carcinoma-in-situ lesions appeared to gain expression of cyclin D2but not D1 or D3, while the invasive testicular tumours showed variable positivity for cyclins D2 and D3, but rarely D1. An unexpectedcorrelation with differentiation rather than proliferation was found particularly for cyclin D3 in teratomas, a conceptually significantobservation confirmed by massive up-regulation of cyclin D3 in the human teratocarcinoma cell line NTera2/D1 induced to differentiatealong the neuronal lineage. These results suggest a possible involvement of cyclin D2 in the early stages of testicular oncogenesis and thestriking examples of proliferation-independent expression point to potential dual or multiple roles of the D-type cyclins, particularly ofcyclin D3. These findings extend current concepts of the biology of the cyclin D subfamily, as well as of the biology and oncopathologyof the human adult testis. Apart from practical implications for the assessment of proliferation and oncogenic aberrations in humantissues and tumours, this study may inspire further research into the emerging role of the cyclin D proteins in the establishment and/ormaintenance of the differentiated phenotypes. Copyright ? 1999 John Wiley & Sons, Ltd.

KEY WORDS—cell cycle; D-type cyclins; human testis; germ cell tumours; differentiation

*Correspondence to: J. Bartek, Institute of Cancer Biology, DanishCancer Society, Strandboulevarden 49, DK-2100 Copenhagen,Denmark. E-mail: [email protected].

Contract/grant sponsor: Danish Medical Research Council;Contract/grant number: 9600821.

Contract/grant sponsor: European Union Biomed 2 Programme;Contract/grant number: BMH4-CT96-0010.

INTRODUCTION

One of the goals of contemporary biomedical researchis to gain a better understanding of the molecular basisof the key cell-fate decisions, including those betweencell-cycle progression versus quiescence and terminaldifferentiation. Diverse lines of evidence point to the lateG1 phase as a critical period of the proliferation–differentiation switch and to the so-called retinoblas-toma protein pathway as a candidate mechanisminvolved in those fundamental cellular processes.1Prominent among the immediate upstream regulators ofpRB are the D-type cyclins, a subfamily of positivecell-cycle regulatory proteins whose expression appearsto follow the availability of mitogens, a property that ledto a concept of D-cyclins as growth factor sensors.2 Thethree D-type cyclin proteins, cyclin D1, D2, and D3,bind and activate their partner cyclin-dependent kinasesCDK4 and CDK6 by mid-to-late G1 and initiate the

CCC 0022–3417/99/050573–09$17.50Copyright ? 1999 John Wiley & Sons, Ltd.

multistep phosphorylation of pRB and the relatedpocket proteins. This modification cancels the pocketproteins’ ability to sequester diverse transcription fac-tors such as E2F, thus allowing for expression ofS-phase genes and the G1/S transition.1,2 Consistentwith the cell cycle-promoting function of the D-typecyclins, cyclin D1 turned out to be a potent proto-oncogene, overexpressed in a wide range of humanmalignancies.3 Cyclin D2 has recently been identified asa candidate proto-oncogene activated in subsets of tes-ticular and ovarian germ cell tumours,4,5 while theproto-oncogenic potential of cyclin D3 remains to bedemonstrated.

Recently, the seemingly exclusive mitogenic role of theD-type cyclins has been challenged by unexpected find-ings of dramatic up-regulation of cyclin D3 in quiescentdifferentiating myotubes6 and increased abundance ofcyclin D2 in diverse conditions associated with growtharrest.7 These intriguing data indicate that the biologicalfunctions of the D-cyclins are more complex than appre-ciated so far, and possibly include some role(s) in celldifferentiation of at least some lineages. To gain furtherinsight into the emerging multiple roles of the individualD-type cyclins, and to elucidate their unique versusredundant expression patterns in relation to cell type,proliferation, differentiation, and tumourigenesis, we
decided to examine some of these issues in the naturally
Received 30 June 1998Accepted 23 November 1998

574 J. BARTKOVA ET AL.

occurring ‘model’ of human testis and testiculartumours. The unique properties which make this fasci-nating biological system well suited for such studiesinclude the presence in the same tissue of two types ofcell cycles, mitotic and meiotic; the co-existence ofdefined cell types such as the germ cells, Sertoli andLeydig cells; the availability of a non-invasive commonprecursor lesion as a well-defined stage of early onco-genic transformation; and several tumour types, includ-ing teratomas with their display of differentiationcomponents with features of diverse cellular lineages.

MATERIALS AND METHODS

described previously.

Tissue samples

Tumour and adjacent tissue specimens were obtainedduring orchidectomy of patients with testicular cancer.Testicular tissues, used here as normal controls, weresurgically removed for the following conditions: contral-ateral testicular tumour (n=2), prostate cancer (n=1),contralateral Leydig tumour (n=1), and diagnosticbiopsy due to oligospermia (n=1). Carcinoma in situ(n=25) was diagnosed on morphological criteria,8 sup-ported in some cases by immunohistochemical stainingfor placental-like alkaline phosphatase. The tumourspectrum analysed included the Leydig cell tumour(n=1) and diverse types of testicular germ cell tumoursfrom a total of 31 patients (Table I), classified asspermatocytic seminoma (n=1), pure seminoma (n=12),pure embryonal carcinoma (n=2), pure teratoma(n=2), and combined tumours of more than onehistological type (n=14). The samples were fixed ineither buffered formalin or modified Stieve’s fixative (7·6per cent glutaraldehyde, 4 per cent acetic acid, inphosphate-buffered saline, pH 7·4), and embedded inparaffin wax.

Antibodies and immunohistochemistry

The production and characterization of mouse mono-clonal antibodies DCS-6 to cyclin D1,9 DCS-22 to cyclinD3,10 and MO-1 to CDK-711 have been reported pre-viously and the antibodies were used as either neathybridoma tissue culture supernatant or ascitic fluiddiluted 1:500 for immunoblots and immunostaining.Further antibodies included polyclonal rabbit antiseraagainst cyclin D2 (C-17, Santa Cruz Biotechnology,Santa Cruz, California, U.S.A., used at 1/800 dilution)and against the Ki67 proliferation marker (Dako,Glostrup, Denmark, diluted 1/75 as recommended bythe manufacturer), as well as normal mouse and rabbitsera used as negative controls. For immunohistochem-istry, the deparaffinized sections were microwave-treated (10 min in citrate buffer, pH 6·0) for antigenunmasking before incubation with the primary anti-bodies, followed by detection using the sensitiveimmunoperoxidase-based Vectastain Elite kit (VectorLaboratories, Glostrup, Denmark) and metal enhance-ment as described previously.12 Nuclear counterstainingwas omitted for photography. The Vectastain kit was

Copyright ? 1999 John Wiley & Sons, Ltd.

also employed for immunocytochemical staining ofthe NTera 2/clone D1 (NT2/D1) cells13 cultured andfixed (10 min in cold methanol–acetone, 1:1) on glasscoverslips.

Cell culture

The human teratocarcinoma cell line NT2/D1 wascultured essentially as described in refs 13 and 14, withminor modifications. The cells were grown in Dulbecco’smodified Eagle’s medium (DMEM) with 10 per centfetal bovine serum and antibiotics and passaged twiceweekly in a ratio 1:4. To induce differentiation alongthe neuronal lineage, exponentially growing cells weretreated with retinoic acid (RA, 10 m, Sigma, SaintLouis, MO, U.S.A.) in the above culture medium andfresh RA-containing medium was exchanged every 3–4days for 4 weeks, followed by further cultivation in theinitial medium without RA until the emerging neuronalcells formed a network and ganglia-like structures.

Gel electrophoresis and immunoblotting

The total cellular protein lysates were prepared byharvesting the subconfluent NT2/D1 cells before (day 0)and at different times after exposure to RA in electro-phoresis sample buffer. Cell extracts balanced forprotein amount were separated by polyacrylamide gelelectrophoresis in the presence of sodium dodecylsulphate (SDS PAGE) on a 10 per cent gel, withprestained molecular weight markers run in parallel. Theseparated proteins were blotted onto nitrocellulosemembrane and visualized by the ECL method as

9

RESULTS

Normal adult testes

Immunohistochemical analysis of normal humanadult testicular tissues (n=5) and morphologically nor-mal tissues surrounding the tumours (n=31) revealeddistinct, cell-type-restricted immunostaining patterns foreach of the three cyclin D proteins (Fig. 1 and Table I).Whereas cyclin D1 was detectable in Sertoli cells of somecases (Fig. 1A), cyclin D3 showed a strong and consist-ent staining signal in Leydig cells of all cases examined(Fig. 1B and inset in Fig. 1B) and both cyclins D1 andD3 were detected in some scattered mesenchymal ele-ments. Localization of cyclins D1 and D3 was largelynuclear, with a concomitant weaker cytoplasmic signalin some cells. Interestingly, the germ cells of non-neoplastic seminiferous tubules appeared devoid of anyof the three D-type cyclin proteins and cyclin D2 wasgenerally negative in the cell types present in the normaladult testis (see Table I for the summary of the stainingresults). There was no obvious correlation between thecyclin D staining and proliferation as detected by theproliferation marker Ki67, in that many proliferatingcells were cyclin D-negative, and the strongest positivityfor cyclin D3 was found in the quiescent Leydig cells.

J. Pathol. 187: 573–581 (1999)

g patterns*

Cyclin D3

eterogeneous Negative Sparse Heterogeneous

5 (+/+ +)55

31 (+/+ +)3118

9 (&/ + ) 251 (+/+ +)

1 ( + ) 1 (+)5 ( + ) 6 7 (&/+) 3 (+)3 ( + ) 3 7 (&/+) 1 (+)

3 (&/ + ) 2 5 (&/+) 18 (+/+ +)1 ( + ) 1 3 (&/+) 7 (+/+ +)

1 1 (&) 3 (+)3 (+/+ +)2 (+/+ +)

1 (+) 1 (+ +)1 ( + ) 1 (+)1 (&) 1 (+)

ning of up to 10 per cent; heterogeneous=positive staining of

S or CIS and some atrophic tubules.

575D

-CY

CL

INS

INT

ES

TIC

UL

AR

CA

NC

ER

Copyright

?1999

JohnW

iley&

Sons,L

td.J.

Pathol.

187:573–581

(1999)

Table I—Immunohistochemistry of D-type cyclin proteins in testicular tissues and tumours

Tissues and tumours

Immunohistochemical stainin

Cyclin D1 Cyclin D2

Negative Sparse Heterogeneous Negative Sparse H

Normal tests (n=5)Leydig cells 5 5Sertoli cells 4 1 (&) 5Germ cells 5 5

Normal tissue adjacent to tumours (n=31)Leydig cells 31 31Sertoli cells 23 8 (&) 31Germ cells† 18 18

Carcinoma in situ (n=25) 25 10 6 (&/ + )Leydig tumour (n=1) 1 1Spermatocytic seminoma (n=1) 1 ( + )Seminoma (n=16‡) 15 1 (&) 3 8 (&/ + )Embryonal carcinoma (n=11)§ 8 3 ( + ) 2 6 ( + )Teratoma (n=12)QDifferentiation component

Total 7 15 (&/ + ) 3 (+) 11 11 ( + )Glandular 1 8 (&/ + ) 2 (+) 5 5 ( + )Stratified squamous nonkeratinizing 2 3 (&/ + ) 1 4 ( + )Stratified squamous keratinizing 1 1 ( + ) 1 (+) 3Cartilage 1 1 ( + ) 2Myotubes 1 1 (&) 2 ( + )Smooth muscle 1 (&)Peripheral nerve 1

*The results are expressed as the number of cases in one of three categories: negative=no detectable staining; sparse=positive stai11–70 per cent cells. The intensity of staining is indicated in parentheses: (&) weak, (+) moderate, (+ +) strong.

†Morphologically normal seminiferous tubules and germ cells were found in only 18 of these 31 cases, the rest containing either CI‡Twelve pure seminomas and four seminomatous components of combined tumours.§Two pure embryonal carcinomas and nine embryonal carcinoma components of combined tumours.QTwo pure teratomas and ten teratomatous components of combined tumours.


Carcinoma in situ

A characteristic profile of D-type cyclin expression,quite distinct from that seen in histologically normaltesticular tissues, was found among the 25 cases ofcarcinoma in situ (CIS) of the testis (Table I). Thus, incontrast to the lack of detectable levels in normal andnon-tumourous adult testes, cyclin D2 protein was aber-rantly expressed in significant subsets of the CIS cells in60 per cent of the cases. An example of the immuno-histochemical staining for cyclin D2 is shown in Fig. 1C.On the other hand, parallel examinations with anti-bodies to cyclins D1 and D3 gave consistently negativeresults in all 25 cases of CIS, as well as in a case ofintratubular seminoma (Table I and Fig. 1D). Stronglypositive non-tumour cells on the same sections, such asLeydig cells in staining for cyclin D3 (Fig. 1B and 1D)served as positive internal controls, documenting thatboth the antibodies used and the tissue processingprocedure were suitable for this type of analysis. Theabundance of cyclin D2 appeared ‘case-specific’ ratherthan reflecting the degree of proliferation in CIS, sincethere was no direct correlation with the proportion ofKi67-positive cells.

Fig. 1—Immunohistochemical detection of D-type cyclins on paraffin sections of normal human testis and testicular in situtumours. (A) Cyclin D1 in Sertoli cells of normal testis (antibody DCS-6). (B) Cyclin D3 is undetectable in CIS cells, incontrast to positive adjacent Leydig cells. The inset shows a detail of cyclin D3 nuclear staining of Leydig cells in normal testis(antibody DCS-22). (C) CIS cells positive for cyclin D2 (antibody C17). (D) Intratubular seminoma is negative, whilesurrounding Leydig cells stain strongly for cyclin D3 (antibody DCS-22)

Testicular tumours

The cyclin D expression patterns in this group ofmainly germ cell tumours of diverse histological typeswere generally more complex and more variable thanthose seen in either normal testes or CIS (Table I), yet


several conclusions about the expression of the individ-ual D-type cyclins in relation to tumour type and majorbiological parameters can be reached. In seminomas,cyclin D1 was absent, in contrast to elevated expressionof cyclin D2 in the vast majority of cases and thevariable immunostaining signal for cyclin D3 (Table I).Somewhat analogous patterns were found amongembryonal carcinomas, in that most specimens showedvariable proportions of cells expressing cyclins D2(Fig. 2A) and D3 (Fig. 2B), while only sparse cyclinD1-positive tumour cells were seen in a small subset ofcases (Table I). The general lack of a positive correlationbetween cyclin D expression and cell proliferationamong the tumours was reminiscent of the normal testesand CIS, and one example of strong cyclin D3 stainingin a tumour with only rare proliferating cells is shown inFig. 2C and 2D.

Distinct, unexpected cyclin D expression patternswere found in the group of teratomas and teratomatouscomponents of combined tumours of more than onehistological type (Table I). Among the characteristicfeatures of teratomas were moderate-to-low expressionof cyclins D1 and D2 and variable, strong expression ofcyclin D3 in the majority of cases. As can be seen fromexamples of various differentiation components includ-ing keratinizing (Fig. 3A and 3B) and non-keratinizing(Fig. 3C and 3D) stratified epithelia, glandular epithelial(Fig. 3E) or peripheral nerve (Fig. 3F) structures fromseveral teratoma cases, cyclin D3 was often abundant incells with highly differentiated features (see also Table I).Parallel examination of the Ki67 marker confirmed

J. Pathol. 187: 573–581 (1999)

577D-CYCLINS IN TESTICULAR CANCER

the non-proliferative status of the cells most stronglyaccumulating cyclin D3. Taken together, these resultsindicated a positive correlation of cyclin D3 proteinlevels with differentiation.

Cyclin D3 and differentiation in cell culture

To elucidate further the emerging relationshipbetween abundance of cyclin D3 and differentiation intesticular tumours, we employed a widely used exper-imental model system of a teratocarcinoma cell lineNT2/D1, inducible to differentiate along the neuronallineage in vitro.13,14 Consistent with the previously pub-lished data,13,14 exposure of the exponentially growingNT2/D1 cells to RA resulted in characteristic morpho-logical and cytoskeletal changes over a period of severalweeks, eventually giving rise to quiescent cell popu-lations of which a significant subset showed typicalfeatures of neuronal differentiation, including extendedneurite networks and the formation of ganglia-likemulticellular structures (data not shown). Time courseimmunocytochemical and immunoblotting analyses ofthe three D-type cyclin proteins in the NT2/D1 cellsprior to addition of RA, and during the process ofdifferentiation, showed fast and dramatic accumulationof cyclin D3 upon induction (Fig. 4). At the single-celllevel, cyclin D3 protein was only detectable in nucleiof a small subset of the exponentially growing cells(Fig. 4A), followed by a substantial increase in terms ofboth the percentage of positive cells and the signalintensity upon induction of differentiation (Fig. 4B). Inrepeated experiments, the reproducible increase of cyclin


D3 occurred already at day 1 of the treatment and thehigh levels of the protein persisted, or even moderatelyincreased, in the advanced stages of the 7-weekfollow-up period (see Fig. 4C for examples of theimmunoblotting results from the first 8 days of treat-ment). In sharp contrast, the levels of both cyclins D1and D2 were very low before or after induction ofdifferentiation, either protein being only weakly detect-able in a small subset of NT2/D1 cells throughout theexperiment (not shown). Furthermore, by monitoringbromodeoxyuridine incorporation into newly replicatedDNA, we could confirm the previously reported obser-vation that the NT2/D1 cells continue to proliferate forthe initial period of approximately 2 weeks of the RAtreatment, with gradual slowing of the proliferation rate,before becoming quiescent during the third week oftreatment (ref. 13 and our unpublished data). Thus,reminiscent of the expression patterns of cyclin D3observed in testicular tissues and tumour specimens (seeabove), the elevated levels of cyclin D3 correlate posi-tively with the induction and maintenance of differentia-tion, rather than with proliferation of the NT2/D1teratocarcinoma cell model.

Fig. 2—Heterogeneous immunohistochemical staining patterns of cyclin D2 (A, antibody C17) and cyclin D3 (B, antibodyDCS-22) in embryonal carcinoma and almost homogeneous staining of cyclin D3 in a Leydig cell tumour (C, antibodyDCS-22). (D) Low proliferation rate in the same Leydig cell tumour as shown in C is demonstrated by only rare Ki67-positivecells

DISCUSSION

Based on the initial mRNA analyses of tissue extractsand limited immunohistochemical studies of some tis-sues, the D-type cyclins appear to be expressed in a celltype-restricted and partly overlapping manner.12,15–17

Our present data support and extend this concept by

J. Pathol. 187: 573–581 (1999)


Fig. 3—Immunohistochemical detection of D-type cyclins on paraffin sections of human testicular teratomas. A and B showadjacent serial sections of a keratinizing stratified squamous epithelial cyst stained for cyclin D1 (antibody DCS-6) and cyclin D3(antibody DCS-22) respectively. C and D show adjacent serial sections of non-keratinizing stratified squamous epithelium stainedfor cyclin D2 (D, antibody C17) and cyclin D3 (C, antibody DCS-22) respectively. Cyclin D3 is expressed in glandular epithelium(E) and peripheral nerve (F) components of teratomas (both antibody DCS-22)

demonstrating that among the characteristic cell types inadult human testis, the Leydig cells express high levels ofcyclin D3 but not cyclins D1 or D2, while Sertoli cellsshow variable expression of cyclin D1 but not D2 or D3.The apparent lack of any of the three D-type cyclins inadult germ cells is conceptually interesting, since thismay represent the only example of normal mammaliancells proliferating in the absence of this class ofG1-phase cyclins. Consistent with our data, absence ofcyclin D1 and D2 proteins in adult human male germcells has also been noticed by Chaganti and co-workers.4It is tempting to speculate that this unique situationreflects the lack of the critical substrate of the cyclinD-dependent kinases, the retinoblastoma tumour sup-pressor protein, previously reported for human malegerm cells.18 This hypothesis is also supported by ourprevious data that in human tumour cells with mutantpRB, and in embryo fibroblasts from RB gene-knockoutmice, the function of the cyclin D-dependent kinases


becomes dispensable.19,20 The in situ scenario that wereport here for the adult human germ cells is alsoreminiscent of the pRB/cyclin D-independent prolifer-ation of the mouse embryonal stem (ES) cell line, inwhich the G1 control mechanism dependent on the pRBpathway becomes functional only during the induceddifferentiation of these pluripotent cells.21

Studies of the molecular pathogenesis of the malegerm cell tumours have largely been guided by thestriking phenomenon of aberrant extra copies of chro-mosome 12p in a major fraction of testicular cancers.22

The identification of the gene(s) whose activity mightexplain the proto-oncogenic driving force of this cyto-genetic abnormality is an urgent priority. The geneencoding cyclin D2 resides on chromosome 12p15,16 andoverexpression of cyclin D2 in a large subset of testiculargerm cell tumours,4 and cell lines derived from suchtumours,5 raised the hope that cyclin D2 could representthe elusive target.4 Most recently, however, a detailed

J. Pathol. 187: 573–581 (1999)


Fig. 4—Abundance of cyclin D3 in human teratocarcinoma cells NT2/D1 before and duringRA-induced differentiation. Immunoperoxidase staining with antibody DCS-22 shows low levels ofcyclin D3 in undifferentiated cells (A, day 0), in contrast to the high abundance of cyclin D3 in cells atday 8 of RA treatment (B). (C) Immunoblotting analysis of cyclin D3 levels in NT2/D1 cells before (day0) and during the initial 8 days of RA treatment (upper panel, antibody DCS-22), compared withCDK-7 (lower panel, antibody MO-1), a positive control protein known to be expressed in bothproliferating and differentiated cells.11 The low level of cyclin D3 at day 0 can be visualized upon longerexposure. The molecular weight of the markers is given in kilodaltons

genetic analysis of the minimal part of chromosome 12preproducibly overrepresented in human male germ celltumours indicates that the cyclin D2 gene is localizedoutside the critical region and thus seems an unlikelycandidate to explain the cytogenetic findings.23 One ofthe results of our present comparative analysis of allthree D-type cyclin proteins was confirmation of theoverabundance of cyclin D2, recently reported byHouldsworth et al.,4 in our study associated with themajority of the early lesions (60 per cent of the CIScases), and with a significant fraction of the invasivemale germ cell tumours. Thus, regardless of whether thisresult directly reflects the chromosome 12p abnormalityor not, and given the plausible proto-oncogenic role ofthe D-type cyclins,1–3 the potential contributionof cyclin D2 to the pathogenesis of testicular cancerwarrants further investigation. Our findings of nosignificant expression of cyclin D1 and D3 in CISstrongly suggest that neither of these two cyclins isinvolved in the early stages of testicular tumourdevelopment. Cyclin D1 was also absent in the vastmajority of the invasive tumours, while the variableexpression of cyclin D3 may be explainable by itsunexpected association with differentiation (see below),


rather than as a sign of its proliferation-promotingactivity.

Possibly the most significant among the novel obser-vations made in this study is the expression of theD-type cyclins, particularly of cyclin D3, in a range ofdifferentiated quiescent cell types. The striking examplesof this phenomenon described here in the normal/non-malignant testicular tissues include cyclin D1 in Sertolicells or cyclin D3 in Leydig cells, and in the tumours,especially cyclin D3 in differentiated epithelial, muscle,and nerve cells of the teratomas or teratomatous com-ponents of the combined tumours. The well-definedNT2/D1 teratocarcinoma cell culture model allowed usto verify the correlation of cyclin D3 abundance withdifferentiation in a controlled time course experiment.The dramatic increase of the cyclin D3 protein alreadyat day 1 suggests that the cyclin D3 level could serve asa sensitive marker of very early stages of differentiation,at least in the neuronal lineage. Furthermore, our exper-iments with bromodeoxyuridine incorporation to assessDNA replication showed that the accumulation of cyclinD3 persists, and even increases, at least several weeksinto post-replicative, quiescent stages of differentiationin this model. Considering the strong expression of

J. Pathol. 187: 573–581 (1999)


cyclin D3 in the peripheral nerve structures in teratomas(this study) and analogous results for peripheral nervesin normal human tissues (J. Bartkova, unpublishedresults), these data indicate that cyclin D3 may beinvolved in initiation as well as maintenance of thedifferentiated phenotype. Taken together with thereported accumulation of cyclin D3 in differentiatingmyotubes in vitro,6 the high levels of this cyclin inquiescent Leydig cells and diverse differentiated compo-nents of teratomas also support the notion that cyclinD3 may be involved in the differentiation process of arange of cell lineages in many tissues, an unexpected rolefor a protein so far regarded mainly as a positivecell-cycle regulator.1,2,24

When considered in a broader perspective of cyclin Dbiology, including their emerging proto-oncogenicpotential,1–3 our present results may have several con-ceptual as well as practical implications. Of practicalimportance is the observed lack of correlation betweencyclin D expression and proliferation, reflecting theabsence of any D-cyclins in proliferating germ cells, aswell as detectable expression in quiescent Sertoli andLeydig cells, and variable levels in tumours of both lowand high fraction of Ki67-positive cells. Consequently,in contrast to cyclins A and B, the abundance of theD-cyclin proteins should not be used as a marker ofproliferation, perhaps with the exception of cyclin D1 insome normal tissues, in which it correlates with theproliferating compartments.12 Conceptually, overabun-dance of cyclin D3 in human malignancies cannot besimply regarded as a proto-oncogenic event. Caution inthis respect should also be exercised when judging thelevels of cyclins D2 and D1 in some tumour types. Theaccumulating evidence now indicates a likely dual rolefor cyclin D3, in promoting G1/S phase transition and inthe initiation and/or maintenance of differentiation. Thesignificance of this concept should become more appar-ent when the phenotype of the cyclin D3 gene knock-outmice is assessed. Furthermore, crossing such mice, ifviable, with the available cyclin D1- and D2-gene knock-out mice should allow the degree of redundancy withinthe D-cyclin subfamily to be evaluated in vivo. The factthat the most striking defect of the male mice lacking thecyclin D2 gene is testicular hypoplasia5 strongly pointsto a critical role of this cyclin in testicular development.Since cyclin D2 protein is not detectable in the adulttestis (ref. 4 and this study), we will analyse fetal andearly postnatal testes in animal models and the limitedrelevant human material available to us to find outwhether the expression of cyclin D2 in the carcinomain situ cells (ref. 4 and this study) represents a reappear-ance of an early developmental feature, rather thanreflecting the extra copies of chromosome 12p. Theformer scenario would be consistent with the dynamicsof cyclin D2 mRNA during development of the murinetestis25 and cyclin D2 could thus become the nextaddition to the growing list of onco-developmentalproteins whose expression is shared by the gonocytes inearly testicular development, and by the cells of CIS.26,27

Although speculative at present, such a mechanismwould agree with our hypothesis that testicular carci-noma in situ may represent a developmental defect


potentially caused by multiple factors,27,28 a conceptsupported also by the exceptionally high incidence oftesticular neoplasia in patients with a range of sex-differentiation and gonadal development disorders.29,30

It is also possible that both the stage-related reappear-ance and the overrepresented chromosome 12p contrib-ute to the aberrant expression of cyclin D2 in CIS. Asatisfactory resolution of this issue must probably awaitthe elucidation of transcriptional control over the cyclinD2 gene promoter and the identification of the precisetiming of chromosome 12p abnormalities relative to CISdevelopment.

ACKNOWLEDGEMENTS

We thank Dr Peter Andrews for kindly donating theNT2/D1 cell line. This work was supported by grantsfrom the Danish Medical Research Council (No.9600821), the Danish Cancer Society, and the EuropeanUnion Biomed 2 Programme (grant No. BMH4-CT96-0010).

REFERENCES1. Bartek J, Bartkova J, Lukas J. The retinoblastoma protein pathway and the

restriction point. Curr Opin Cell Biol 1996; 8: 805–814.2. Sherr CJ. G1 phase progression: cycling on cue. Cell 1994; 79: 551–555.3. Hall M, Peters G. Genetic alterations of cyclins, cyclin-dependent kinases,

and Cdk inhibitors in human cancer. Adv Cancer Res 1996; 68: 67–108.4. Houldsworth J, Reuter V, Bosl GJ, Chaganti RSK. Aberrant expression of

cyclin D2 is an early event in human male germ cell tumorigenesis. CellGrowth Differ 1997; 8: 293–299.

5. Sicinski P, Donaher JL, Geng Y, et al. Cyclin D2 is an FSH-responsive geneinvolved in gonadal cell proliferation and oncogenesis. Nature 1996; 384:470–474.

6. Kiess M, Gill RM, Hamel PA. Expression of the positive regulator of cellcycle progression, cyclin D3, is induced during differentiation of myoblastsinto quiescent myotubes. Oncogene 1995; 10: 159–166.

7. Meyyappan M, Wong H, Hull C, Riabowol KT. Increased expression ofcyclin D2 during multiple states of growth arrest in primary and establishedcells. Mol Cell Biol 1998; 18: 3163–3172.

8. Skakkebæk NE. Possible carcinoma-in-situ of the testis. Lancet 1972; ii:516–517.

9. Lukas J, Pagano M, Staskova Z, Draetta G, Bartek J. Cyclin D1 proteinoscillates and is essential for cell cycle progression in human tumour celllines. Oncogene 1994; 9: 707–718.

10. Bartkova J, Zemanova M, Bartek J. Abundance and subcellular localisationof cyclin D3 in human tumors. Int J Cancer 1996; 65: 323–327.

11. Bartkova J, Zemanova M, Bartek J. Expression of CDK7/CAK in normaland tumour cells of diverse histogenesis, cell-cycle position and differentia-tion. Int J Cancer 1996; 66: 732–737.

12. Bartkova J, Lukas J, Strauss M, Bartek J. Cell cycle-related variation andtissue-restricted expression of human cyclin D1 protein. J Pathol 1994; 172:237–245.

13. Andrews PW. Retinoic acid induces neuronal differentiation of a clonedhuman embryonal carcinoma cell line in vitro. Dev Biol 1984; 103: 285–293.

14. Pleasure SJ, Lee VM-Y. NTera 2 cells: a human cell line which displayscharacteristics expected of a human committed neuronal progenitor cell.J Neurosci Res 1993; 35: 585–602.

15. Inaba T, Matsushime H, Valentine M, Roussel MF, Sherr CJ, Look AT.Genomic organization, chromosomal localization, and independent expres-sion of human cyclin D genes. Genomics 1992; 13: 565–574.

16. Xiong Y, Menninger J, Beach D, Ward D. Molecular cloning and chromo-somal mapping of CCND genes encoding human D-type cyclins. Genomics1992; 13: 575–584.

17. Usuda H, Saito T, Emura I, Naito M. Immunohistochemistry of cyclin D3in pulmonary carcinomas. Virchows Arch 1996; 428: 159–163.

18. Cordon-Cardo C, Richon VM. Expression of the retinoblastoma protein isregulated in normal human tissue. Am J Pathol 1994; 144: 500–510.

19. Lukas J, Parry D, Aagaard L, et al. Retinoblastoma-protein-dependentcell-cycle inhibition by the tumour suppressor p16. Nature 1995; 375:503–506.

20. Lukas J, Bartkova J, Rohde M, Straus M, Bartek J. Cyclin D1 is dispensablefor G1 control in retinoblastoma gene-deficient cells independently of cdk4activity. Mol Cell Biol 1995; 15: 2600–2611.

J. Pathol. 187: 573–581 (1999)


21. Savatier P, Lapillonne H, van Grunsven LA, Rudkin BB, Samarut J.Withdrawal of differentiation inhibitory activity/leukemia inhibitory factorup-regulates D-type cyclins and cyclin-dependent kinase inhibitors in mouseembryonic stem cells. Oncogene 1995; 12: 309–322.

22. Chaganti RSK, Murty VVVS, Bosl GJ. Molecular genetics of male germ celltumors. In: Vogelzang NJ, Shipley WU, Scardino PT, Coffey DS, eds.Comprehensive Textbook of Genitourinary Oncology. Baltimore: Williamsand Wilkins, 1996: 932–940.

23. Mostert MC, Verkerk AJMH, van de Pol M, et al. Identification of thecritical region of 12p over-representation in testicular germ cell tumors ofadolescents and adults. Oncogene 1998; 16: 2617–2627.

24. Ando K, Ajchenbaum-Cymbalista F, Griffin JD. Regulation of G1/Stransition by cyclins D2 and D3 in hematopietic cells. Proc Natl Acad SciU S A 1993; 90: 9571–9575.

25. Nakayama H, Nishiyama H, Higuchi T, Kaneko Y, Fukumoto M, Fujita J.Change of cyclin D2 mRNA expression during murine testis developmentdetected by fragmented cDNA subtraction method. Dev Growth Differ 1996;
38: 141–151.

26. Jørgensen N, Rajpert-De Meyts E, Græm N, Müller J, Giwercman A,Skakkebæk NE. Expression of immunohistochemical markers for testicularcarcinoma in situ by normal human fetal germ cells. Lab Invest 1995; 72:223–231.

27. Skakkebæk NE, Berthelsen JG, Giwercman A, Müller J. Carcinoma-in-situof the testis: possible origin from gonocytes and precursor of all types ofgerm cell tumours except spermatocytoma. Int J Androl 1987; 10: 19–28.

28. Rajpert-De Meyts E, Jørgensen N, Brøndum-Nielsen K, Müller J,Skakkebæk NE. Developmental arrest of germ cells in the pathogenesis ofgerm cell neoplasia. APMIS 1998; 106: 198–206.

29. Skakkebæk NE, Rajpert-De Meyts E, Jørgensen N, et al. Germ cell cancerand disorders of spermatogenesis. An environmental connection? APMIS1998; 106: 3–12.

30. Savage MO, Lowe DG. Gonadal neoplasia and abnormal sexual differen-tiation. Clin Endocrinol 1990; 32: 519–533.

J. Pathol. 187: 573–581 (1999)

d-type cyclins in adult human testis and testicular cancer: relation to cell type, proliferation,...

Documents