0021

doi:1

Corre

Cyclin D3 Expression in Normal Fetal, NormalAdult and Neoplastic Feline Tissue

A. J. Norris, S. M. Griffey*, M. D. Lucroy† and B. R. Madewell‡

Veterinary Medical Teaching Hospital, *Comparative Pathology Laboratory and ‡Department of Surgical andRadiological Sciences, School of Veterinary Medicine, University of California, Davis, CA 95616, and †Department of

Veterinary Clinical Sciences and Cancer Center, Purdue University, West Lafayette, IN 47907, USA

Summary

-99

0

s

Cyclin D3 is a tightly regulated cell cycle protein and member of the cyclin D family—a group of proteinsthat facilitates the progression of a cell through G1 and into the S phase of the cell cycle. All cells use atleast one of the cyclin D proteins for cell cycle regulation. In this study, feline tissues (normal fetal andadult, and neoplastic) were examined immunohistochemically for expression and topographicaldistribution of cyclin D3. Its distribution was similar to that in human tissues in health and neoplasia,and suggested a dual role of cyclin D3 in cell proliferation and differentiation. Immature lymphoid tissueand proliferating epithelial cells in health and neoplasia were immunoreactive for cyclin D3, whereasexpression of the protein in other immunoreactive tissues reflected differentiated cell types.Immunoreactivity for cyclin D3 was particularly striking in germinal centre cells of normal lymphnodes and B-cell lymphomas, and in normal suprabasal epithelial cells of the skin and mucousmembranes of the oropharynx and in squamous cell carcinomas at these sites.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: cat; cyclin D3; tumour

Introduction

The cell cycle consists of four distinct phases,separated by “checkpoints”. The S, G2 and Mphases are generally consistent and uniform induration, and the rate of cell division is governedprimarily by the G1 phase (Andreeff et al., 2000).The coordinated events of the cell cycle areregulated by the activity of cyclins, their catalyticcyclin-dependent kinases, and their inhibitors.Cyclin-dependent kinases form a group of highlyconserved regulatory enzymes that are of criticalimportance in enabling a cell to proceed into itsnext phase of the cell cycle (Harper, 1997). Thereare multiple subsets of cyclin-dependent kinases, ofwhich subsets cdk2, cdk4 and cdk5 requirea conserved group of regulatory proteins knownas the cyclins to bind and stimulate forward cell

75/$ - see front matter

.1016/j.jcpa.2004.12.003

pondence to: B.R. Madewell.

cycle progression. The cdk/cyclin complex has aproregulatory role in the cell cycle, but two familiesof inhibitory proteins can act at the same level ofthe cdk/cyclin complex and restrict the cell fromdividing. These inhibitory proteins include kinaseinhibitory proteins (KIPs) and the inhibitorykinases (INKs) (Reed, 1997; Kato, 1999; Sherrand Roberts, 1999).

Aberrant cell cycle regulation is characteristic ofneoplastic cells—cells that continue to proliferatein the absence of stimulatory signals. Neoplasticcells often have a higher growth fraction thannormal cells and major dysregulation of the cellcycle occurs at the checkpoints (Reed, 1997; Kato,1999; Andreeff et al., 2000 ). Methods and reagentsare now available to facilitate the study of individualcomponents of the cell cycle. The cyclin D familycontains three isoforms—D1, D2 and D3. Cyclin Dcomplexes with cdk 4 or 6 and acts to regulate thecell cycle at G1/S phase. The role of cyclin D

J. Comp. Path. 2005, Vol. 132, 329–339

www.elsevier.com/locate/jcpa

q 2005 Elsevier Ltd. All rights reserved.

A.J. Norris et al.330

expression in both normal and neoplastic humantissue has been explored (Doglioni et al., 1998;Teramoto et al., 1999; Moller et al., 2001). The cyclinD3 isoform facilitates physiological progressionfrom G1 to S after mitogenic stimulation in somecells, whereas its expression in other cells appearsto be related to cellular differentiation. Dysregula-tion of cyclin D3 contributes to the high growthfraction of some murine and human tumours, butlittle is known of its role in tumours of veterinaryimportance.

In the present study, cyclin D3 expression wasexamined immunohistochemically in normalfeline tissue and in a variety of feline neoplasms.

Materials and Methods

Tissues

Paraffin wax blocks selected from the pathologyarchive of the University of California VeterinaryMedical Teaching Hospital (1987–2002) includedan entire midgestation feline fetus, normal felinetissue and various feline tumours. Neoplastictissues selected were oral squamous cell carcinoma(nZ22), cutaneous squamous cell carcinoma (16),basal cell tumour (23), lymphoma (33), intestinalor colonic adenocarcinoma (13), mast cell tumour(5), thyroid adenoma and carcinoma (5) andmeningioma (5). Normal feline tissues includedstomach (2), intestine (2), oral/pharyngealmucosa and larynx (2), lymph node (4), pancreas(2), thyroid gland (2), skeletal muscle (2), tongue(2), skin (4), urinary bladder (2) and singlespecimens of lung, uterus, ovary, liver, salivarygland, gall bladder, spleen, kidney, mammarygland and adrenal gland. Hyperplastic lymphnodes (2) and lymphoid depleted lymph nodes(3) were also examined. For each specimen,sections (5 mm) were stained with haematoxylinand eosin (HE) and examined by light microscopy.Tumours were first examined for confirmation ofprevious histological classification. The tissuesamples selected had been fixed in formalin forno more than a limited period (!48 h) to preventprotein denaturation that might otherwise haveinterfered with appropriate immunohistochemicallabelling.

Because preliminary immunolabelling showedparticularly strong reactivity in skin and squamouscell carcinomas, and in lymphoid organs andlymphomas, further histological categorization ofsquamous cell carcinomas and malignant lympho-mas was undertaken. The squamous cell carci-nomas were categorized on the basis of

characteristics considered to indicate aggressivetumour behaviour. The samples were given a scorefor degree of invasion and keratinization. Invasionwas scored as 1 if the neoplastic cells remained insitu or showed only mild invasion of the superficialdermis, 2 if the neoplastic cells invaded the mid-dermis, and 3 if the neoplastic cells were veryinvasive and extended to the surgical margins. Thedegree of keratinization was scored as 1 if mostneoplastic cells were basal epithelial with littledifferentiation toward keratinization, 2 if theneoplastic nests had a thin layer of basal epitheliumthat matured centrally to prominent keratinization,and 3 if the majority of neoplastic nests andindividual neoplastic cells were keratinized. Thetumours were also categorized on the basis ofhistological patterns–invasive nests, solid sheets,cords or central comedomes with sloughed neo-plastic keratinized cells. For malignant lymphomas,the tumours were further categorized into T- andB- cell immunophenotypes on the basis of labellingreactions described below.

Immunohistochemistry

Cyclin D3 expression was detected with an avidin–biotin technique and commercial mouse mono-clonal antibody specific for cyclin D3. All tissueswere immunolabelled with the cyclin D3 antibody,and the lymphomas, squamous cell carcinomas andbasal cell tumours were also immunolabelled forthe proliferation protein Ki67. Sections (5 mm)were dewaxed in xylene and in 100% ethanol.Endogenous peroxidase was blocked with hydro-gen peroxide 3% in methanol for 20 min. Sectionswere immersed for 2 min in 95% ethanol, for 2 minin 75% ethanol, and for 5 min in tap water. Antigenretrieval was accomplished by microwaving(100 8C) in citrate-buffer for two 1-min cycles, thesections then being allowed to cool to roomtemperature before placing in a humidifiedchamber. A solution of 10% normal horse serumwas applied to the sections for 20 min. Afterblotting, the primary antibody (NCL-Cyclin D3:Novocastra Laboratories, Newcastle-upon-Tyne,UK) diluted 1 in 20 was applied and sections wereincubated overnight at 4 8C. The sections wererinsed twice (5 min each) in phosphate-bufferedsaline (PBS) and the secondary antibody (biotiny-lated horse anti-mouse; Vector Laboratories, Bur-lingame, CA, USA) diluted 1 in 800 was applied atroom temperature for 60 min. The sections wererinsed twice in PBS (5 min each) followed bya 30-min incubation with avidin–biotin complex(ABC Elite; Vector Laboratories) diluted 1 in 50,

Cyclin D3 Expression in Feline Tissue 331

followed by two 5-min PBS rinses. After rinsing,diaminobenzidene-peroxidase (Sky-Tech Labora-tories, Logan, UT, USA) was applied for 3–5 minto produce a permanent (brown) colour change inthe reactive cells. The sections were immersed intap water for 5 min and stained with Mayer’shaematoxylin for 30s. They were then dehydrated,cleared and mounted under a coverslip. Positive(control) tissues for cyclin D3 expression includednormal human skin and lymph node. In negativecontrols, the primary mouse cyclin D3 antibody wasreplaced by purified rabbit serum IgG (Dako,Carpinteria, CA, USA) diluted as for the primaryantibody. For Ki67 immunolabelling, the primaryantibody (MIB1; Dako) diluted 1 in 80 was used;the procedures were similar to those describedabove except that a red chromogen (Fast Red;BioGenex, San Ramon, CA, USA) was used toobtain a permanent colour change in the reactivecells. The T-cell and B-cell lymphomas werepreviously immunophenotyped for CD3 (mono-clonal rat antibody for CD3 epsilon kindly providedby Dr P.F. Moore) and CD79a (HM57 clone; Dako)expression.

Tissue specimens labelled with cyclin D3 anti-body were examined by light microscopy at !10magnification. Tumours containing O5% of nucleiimmunolabelled for cyclin D3 were consideredpositive, as previously described (Moller et al.,2001). The specificity of the human monoclonalantibody for feline cyclin D3 was validated bycomparing the immunolabelling of the felinetissues with that of human control tissues, as wellas the use of the strikingly strong tissue immuno-labelling indicated on the preliminary feline studyto explore those tissues in greater depth (Doglioniet al., 1998).

For quantitation of cyclin D3 expression, fiveadjacent, non-overlapping fields were selected atrandom and examined with a 36-square (6!6)ocular grid and !400 magnification. One hundredtumour cells were counted per field for a total of500 cells. The positive cells were expressed as apercentage of the total cells counted.

The same procedure was used for quantitation ofKi67 expression.

Fig. 1. Histology (HE; top) and immunolabelling (ABC;bottom) of feline fetal tissue, showing cyclin D3expression in various neuronal, mesenchymal andepithelial tissues (!2). Inset: immunoreactivity of theintervertebral nerve ganglion nuclei. !200.

Statistical Analysis

Quantitative data were compared between groupswith the unpaired Student’s t-test and one-wayanalysis of variance (ANOVA). Correlationsbetween quantitative variables were evaluated withsimple linear regression. Statistical analyses weredone with computer software (SigmaStat 3.0; SPSS,

Inc., Chicago, IL, USA). For all tests, a value ofP!0.05 was considered statistically significant. Dataare reported as meanGSD, unless otherwise stated.

Results

Fetal Tissue

Cyclin D3 immunoreactivity was present in cellnuclei in some mesenchymal tissues (cartilage,skeletal muscle, smooth muscle, brown fat sur-rounding the pancreas) and some stromal cells ofthe dermis. There was intense cyclin D3 labelling insome nervous tissues such as the neurons in theintercostal ganglion (Fig. 1), the neurons in theplexuses throughout the intestine, and someneuronal or glial cells in the cerebral hemispheres.The oronasal pharyngeal epithelium, renal tubularepithelium, and intestinal crypt and villous epi-thelial cells were also variably immunoreactive, aswere the sinusoidal Kupffer cells in the liver.

Normal Adult Tissue

In mucous membranes of the oral cavity (Figs. 2and 3) and epithelium of the skin (Figs. 4 and 5),there was nuclear immunoreactivity of the supra-basilar cell layers. In the stratified squamous tissue,immunoreactive cells were particularly obvious inthe stratum spinosum and stratum granulosumlayers. The most intense immunolabelling wasobserved in the oral mucosa, including tongue.In lymphoid tissue, the germinal centre cells oflymph nodes (Figs 6 and 7) and Peyer’s patches

Figs 2 and 3. Histology (Fig. 2; HE) and cyclin D3 immunolabelling (Fig. 3; ABC) of normal feline keratinized stratified squamousepithelium from tongue, showing expression of cyclin D3 in suprabasal epithelium. !200.

A.J. Norris et al.332

(Figs 8 and 9) were immunoreactive. There wasscattered immunoreactivity in the interfollicularcells, and strong reactivity in the neurons of themyenteric plexuses throughout the intestine,nerves in the lungs and salivary gland, andparaganglion cells of the adrenal gland. Therewas variable epithelial mucosal labelling of theurinary bladder, stomach, intestine and gall blad-der. Some cells in the dermis (interpreted to be

Figs 4 and 5. Histology (Fig. 4; HE) and cyclin D3 immunolabellinepithelium from haired skin, showing expression of

macrophages, based on histological appearance)were immunolabelled, as well as some interstitialcells in the lungs.

Hyperplastic Lymph Node

The immunoreactivity for cyclin D3 was O5% forthe lymphocytes within the germinal follicularcentres and for some parafollicular cells.

g (Fig. 5; ABC) of normal feline keratinized stratified squamouscyclin D3 in suprabasal epithelium. !200.

Figs 6 and 7. Histology (Fig. 6; HE; !200) and cyclin D3 immunolabelling (Fig. 7; ABC; !600) of normal feline peripheral lymphnode, showing expression of cyclin D3 in the germinal centres with scant immunoreactivity in the paracortex.

Cyclin D3 Expression in Feline Tissue 333

Lymphoid-depleted Lymph Node

The immunoreactivity was weak, but in one of threelymph nodes O5% of follicular cells wereimmunolabelled.

Lymphomas

Thirty-three lymphoma specimens were evaluated.Lymphomas were categorized by immunophenotypeand anatomical site. There were 18 B-cell lymphomas(Figs 10 and 11) and 15 T-cell lymphomas (Figs 12and 13). The mean cyclin D3 immunoreactivity wassignificantly (P!0.001) higher for the B-cell lym-phomas (22.3G12.85%) than for T-cell lymphomas(1.2G3.22%). The mean Ki67 immunoreactivity was

Figs 8 and 9. Histology (Fig. 8; HE) and cyclin D3 immunolabellin(GALT). Arrow in Fig. 9 depicts the prominent imm

significantly (P!0.001) higher for the B-cell lym-phomas (61.9G17.84%) than for T-cell lymphomas(31.2G26.72%).

The lymphomas were categorized anatomicallyas alimentary (nZ20), multicentric (3), extranodal(8) or cutaneous (2). There were no significantdifferences in cyclin D3 immunoreactivity or Ki67immunoreactivity between the various anatomicallocations (PO0.07 for all comparisons).

The cats with lymphoma consisted of 19 malesand 14 females. The mean cyclin D3 immunor-eactivity was not significantly (PZ0.209) differentfor the male cats (16.7G17.54%) as compared withthe female cats (9.6G13.09%). Likewise, the meanKi67 immunoreactivity did not differ significantly

g (Fig. 9; ABC) of normal feline gut associated lymphoid tissueunoreactivity of the central region !100.

Figs 10 and 11. Cyclin D3 immunolabelling of a gastrointestinal B-cell lymphoma. Neoplastic B lymphocytes have invaded the wall ofthe intestine and are strongly immunoreactive. ABC. Fig. 10 !100. Fig. 11 !600.

A.J. Norris et al.334

(PZ0.351) between males (52.4G25.57%) andfemales (43.3G29.36%).

The median age of the cats with lymphoma was11 years (nZ33; range, 3 to 18 years). The meancyclin D3 immunoreactivity did not differ signifi-cantly (PZ0.762) between cats aged R11 years(13.1G16.51%; nZ17) and those aged !11 years(11.5G11.80%; nZ15). Likewise, the mean Ki67immunoreactivity did not differ significantly(PZ0.096) between cats aged R11 years (40.1G29.82%) and those aged !11 years (56.0G21.33%). There was a significant although weakcorrelation between cyclin D3 immunoreactivity

Figs 12 and 13. Cyclin D3 immunolabelling of a peripheral lymph ndepicts scant immunolabelling of reactive lymphoc

and Ki67 immunoreactivity (P!0.001; Fig. 14) inthe lymphoma specimens.

Squamous Cell Carcinoma

Thirty-eight squamous cell carcinomas (SCCs) wereevaluated. They were categorized by location as oralcavity (nZ22) (Figs 15 and 16), nasal plane (7),other facial site (7) (Figs 17 and 18), and non-facial(2). The mean cyclin D3 immunoreactivity fornasal plane SCCs (42.1G6.49%) was significantlyhigher than that for other facial sites (22.6G13.45%) and for oral SCCs (28.1G9.83%)

ode expanded by neoplastic T lymphocytes. The arrow in Fig. 13ytes. ABC. Fig. 12 !100. Fig. 13 !600.

Fig. 14. A scatter plot depicting significant (P!0.001)correlation (r2Z0.351) between Ki67 immunoreac-tivity and cyclin D3 immunoreactivity in felinelymphoma specimens. The dashed lines representthe 95% confidence interval.

Cyclin D3 Expression in Feline Tissue 335

(PZ0.004 and PZ0.013, respectively). The meanKi67 immunoreactivity for nasal plane SCCs(57.9G6.31%) was not significantly different fromthat for other facial sites (46.1G13.45%) or for oralSCCs (49.4G15.22%; PZ0.431).

SCC specimens were assigned a score (1, 2, or 3)for degree of invasion and keratinization. Therewere no significant differences in mean cyclin D3immunoreactivity among SCC specimens when

Figs. 15 and 16. Cyclin D3 immunolabelling of an oral squamousuprabasal epithelium in addition to that of thunderlying stroma (Fig. 15, arrows). ABC. Fig. 15

grouped according to invasion score (PZ0.604)or keratinization score (PZ0.705). Likewise, therewere no significant differences in mean Ki67immunoreactivity among SCC specimens whengrouped according to invasion score (PZ0.183)or keratinization score (PZ0.850).

SCC specimens were classified histopathologi-cally as organized into islands (nZ9), possessingmixed features (i.e., islands and cords; 17), orpossessing other features (e.g., nests, solid, cords,comedome; 12). Among the histological groupsthere were no significant differences in mean cyclinD3 immunoreactivity (PZ0.700) or mean Ki67immunoreactivity (PZ0.633).

The gender of 36 cats with SCC was known (17males and 19 females). The mean cyclin D3immunoreactivity did not differ significantly(PZ0.379) between the males (28.7G10.94%)and females (32.0G11.20%). Likewise, the meanKi67 immunoreactivity did not differ significantly(PZ0.925) between the males (50.4G16.43%) andfemales (49.9G13.02%).

The coat colour was known in 36 cats with SCC andincluded 15 white or partly white cats and 21 cats withno white in their coats. There was no significantdifference (PZ0.142) in mean cyclin D3 immunor-eactivity between cats with white in their coats (26.4G13.48%) and non-white cats (32.1G9.36%). Like-wise, there was no significant difference (PZ0.312)in mean Ki67 immunoreactivity between cats withwhite in their coats (53.5G18.33%) and non-whitecats (48.31G12.40%).

The median age of the cats with SCC was 14 years(nZ38; range, 6 to 26 years). The mean cyclin D3

s cell carcinoma, showing immunoreactivity of non-neoplastice neoplastic keratinocytes that are forming nests within the!200. Fig. 16 !600.

Figs 17 and 18. Cyclin D3 immunolabelling of a nasal plane squamous cell carcinoma, showing immunoreactivity of non-neoplasticsuprabasal epithelium in addition to the neoplastic keratinocytes forming nests within the underlying stroma(arrows). ABC. Fig. 17 !200. Fig. 18 !600.

A.J. Norris et al.336

immunoreactivity was significantly (PZ0.025)lower in cats aged R14 years (26.8G10.53%;nZ20) than in those aged !14 years (35.0G10.22%; nZ16). However, the mean Ki67 immu-noreactivity was not significantly (PZ0.868) differ-ent for cats aged R14 years (49.8G16.44%) andthose aged !14 years (50.60G12.20%). There wasno significant correlation between cyclin D3immunoreactivity and Ki67 immunoreactivity(PZ0.437; Fig. 19) in the SCC specimens.

Fig. 19. A scatter plot depicting a lack of correlation(r2Z0.0178) between Ki67 immunoreactivity andcyclin D3 immunoreactivity in feline squamous cellcarcinoma specimens (PZ0.437). The dashed linesrepresent the 95% confidence interval.

Basal Cell Tumours

Twenty-three basal cell tumours were studied(Figs 20 and 21). The mean cyclin D3 immunor-eactivity was 5.8G9.14%, whereas the mean Ki67immunoreactivity was 17.4G13.28%. The sex of 22cats with basal cell tumours was known, 13 beingmale and nine female. The mean cyclin D3immunoreactivity was not significantly (PZ0.784)different in male (4.2G6.76%) and femalecats (5.10G8.02%). Likewise, the mean Ki67immunoreactivity was not significantly (PZ0.505)different in the males (16.3G9.81%) and females(13.8G6.63%).

The age of 21 cats with basal cell tumours wasknown, the median age being 14 (range, 7 to 18years). The mean cyclin D3 immunoreactivity didnot differ significantly (PZ0.569) in cats aged R14years (5.1G8.02%; nZ12) and those aged !14years (3.2G6.12%; nZ9). Similarly, the mean Ki67immunoreactivity was not significantly (PZ0.269)different in cats aged R14 years (13.3G8.59%)and those aged !14 years (17.7G8.66%). There

was a significant although weak correlationbetween cyclin D3 immunoreactivity and Ki67immunoreactivity (P!0.001; Fig. 22) in the basalcell tumour specimens.

Other Tumours

Of 13 adenocarcinomas derived from the intestinalepithelium, six showed cyclin D3 immunolabelling

Figs 20 and 21. Cyclin D3 immunolabelling of a cutaneous basal cell tumour, showing expression in a single neoplastic basal cell(arrow). ABC. Fig. 20 !200. Fig. 21 !600.

Cyclin D3 Expression in Feline Tissue 337

of O5% of cells. Of five thyroid tumours (fouradenomas and one carcinoma), none was immuno-reactive for cyclin D3. Little or no immunoreactivityfor cyclin D3 was observed in other tumours(thyroid, 0/5; mammary, 0/5; mast cell, 1/5;meningioma, 1/5).

Fig. 22. A scatter plot depicting a significant (P!0.001)correlation (r2Z0.519) between Ki67 immunoreac-tivity and cyclin D3 immunoreactivity in feline basalcell tumour specimens. The dashed lines representthe 95% confidence interval.

Discussion

The cyclin D family includes isoforms D1, D2, andD3. Cyclin D3 in normal and neoplastic felinetissues was examined in this study. The expressionof cyclin D is cell-specific and all cells are thought toexpress at least one isoform of cyclin D. In humantissue, cyclin D3 has dual patterns of expression. Innormal human tissue, cyclin D3 is expressed inactively dividing areas of lymphoid and suprabasalepithelial cells, and in well-differentiated mesench-ymal cells of some other tissues (Doglioni et al.,1998). This pattern of cyclin D3 expression wasfound in the present study in feline fetal and adulttissues, with greatest reactivity in lymphoid, muco-sal and cutaneous epithelial cells and in someneural cells (peripheral nerve, myenteric plexusand adrenal paraganglion cells).

In human lymphomas, cyclin D3 expression iscorrelated with tumour proliferation and pro-gression, whereas in other tumours, the expressionis heterogeneous and inconsistent (Doglioni et al.,1998). In the feline lymphomas studied here,tumours with high cellular proliferation (demon-strated by Ki67 immunoreactivity) contained agreater proportion of cells expressing cyclin D3than did tumours with low cellular proliferation. Inone study of cyclin D2 and D3 in human lymphoid

tissues, there was high expression in the germinalcentres of normal lymph nodes, and in follicular B-cell lymphomas (Teramoto et al., 1999). Moller et al.(2001) demonstrated high expression of cyclin D3in 43 of 198 non-Hodgkin’s lymphomas, but T-celllymphomas were more likely than B-cell lympho-mas to express cyclin D3. Furthermore, in high-grade lymphomas in which cyclin D3 wasexpressed, in contrast to lymphomas in which itwas not expressed, there was direct correlation

A.J. Norris et al.338

between more rapid tumour growth rates, shorterdisease-free intervals, and shorter survival timesafter treatment (Moller et al., 2001). In the catsstudied here, cyclin D3 expression was greater inlymphomas with B-cell immunophenotype than inlymphomas derived from T cells. Alimentarylymphomas had significantly lower cyclin D3expression than did tumours derived from othersites presumably–related to the predominant T-cellimmunophenotype of the alimentary lymphomas.

G1–S transition defects in cell cycle control areconsidered to be critical in tumour development.Low expression of the kinase inhibitory proteinp27kip-1 was previously identified in feline lympho-mas, allowing cells to pass through the G1–Scheckpoint and to undergo uncontrolled prolifer-ation (Madewell et al., 2002). The results reportedhere suggest that cyclin D3, one of the majorregulatory proteins in the cell cycle checkpointbetween G1 and S, may also be overexpressed insome cells, further facilitating their passagethrough this checkpoint and favouring prolifer-ation. In some human lymphomas, high cyclin D3expression is directly related to patient survivaltimes (Moller et al., 2001; Fillipits et al., 2002).Whether high cyclin D3 expression in feline B-celllymphomas is also related to outcome has not beendetermined. It is conceivable that therapy targetedat reducing cyclin D3 expression within lymphoidtumours would delay tumour progression andextend survival times.

In this study, there was also a direct relationbetween cyclin D3 and Ki67 immunoreactivity inthe basal cell tumours, reflecting the patterns ofexpression of these proteins in the normal proto-type tissues. In several of the analyses for squamouscell carcinomas, however, there was no suchrelation. These findings are consistent with theheterogeneous expression of cyclin D3 in somehuman tumours and underscore the complexregulation of cellular proliferation (Doglioniet al., 1998).

There are limited data on the expression ofcyclin D3 in skin tumours of either human beingsor other mammals. Zhan et al. (1997) examined theexpression of cyclin D3 in chemically induced skintumours, which was found to be higher insquamous cell carcinomas than in papillomas,and higher in metastases than in the primarytumours; there was a direct relation between cyclinD3 expression and tumour progression. A study ofmurine mammary squamous cell carcinomas(Pirkmaier et al., 2003) suggested that cyclin D3was implicated in the pathogenesis of squamousmetaplasia and carcinoma in multiparous animals.

Similarly, in the feline tissues studied here, normaland neoplastic squamous epithelium were highlyimmunoreactive for cyclin D3. A study by Quonet al. (2001) suggested that cyclin D1 influenced theoutcome of squamous cell carcinomas of the headand neck in human patients.

In the present study, cutaneous and oralsquamous cell tumours were further categorizedby location, degree of keratinization, and invasion.However, aside from high expression of cyclin D3in tumours derived from the nasal plane, therewere no significant differences in the expression ofeither cyclin D3 or Ki67 between anatomical sites.In a previous study of cats with squamous cellcarcinoma of the nasal plane, tumour volume andproliferation index were found to be predictive oftumour response to radiation therapy (Theon et al.,1995). Whether cyclin D3 expression also serves topredict the radioresponsiveness of SCC of the nasalplane remains unknown.

Cyclin D3 in human and murine thyroiddisorders was examined by Motti et al. (2003),who reported overexpression in hyperplasias andadenomas. In the five thyroid adenomas examinedhere, no cyclin D3 expression was observed.

In the gastrointestinal tract, cyclin D3 expressionwas observed in fetal and adult epithelium and insix of 13 adenocarcinomas. In the human gut,cyclin D3 is implicated in the pathogenesis ofstromal tumours by its inhibitory effect on p27(Pruneri et al., 2003). In the feline fetal tissues,some immunoreactive stromal cells were detected,but whether this reactivity was localized to theprimitive cells of Cajal or spindle cells from whichthe gastrointestinal stromal tumours are derived isunknown.

The results reported here show that cyclin D3 isexpressed in some feline tissues and is conserved insome tumours. The absence of expression in othertissues and neoplasms reflects cell specificity for thedifferent cyclin D proteins. Further studies arerequired to define the role of cyclin D3 in theaetiopathogenesis of feline neoplasms, and toassess its possible value in predicting the biologicalbehaviour of tumours or their response to therapy.

Acknowledgments

The authors acknowledge the Center for Compa-nion Animal Health, University of California Davis,School of Veterinary Medicine for financial sup-port, and the immunohistochemical assistance ofDiane Naydan and Judy Wall, University of Califor-nia. Thanks are also due to Dr John Peauroi for the

Cyclin D3 Expression in Feline Tissue 339

laboratory samples contributed from VDx (Veter-inary Diagnostic Services).

References

Andreeff, M., Goorich, D. W. and Pardee, A. B. (2000).Cell proliferation, differentiation, and apoptosis. In:Cancer Medicine, 5th Edit., R. C. Bast Jr., D. W. Kufe,R. E. Pollock, R. R. Weichselbaum, J. F. Holland and E.Frei III, Eds, BC Decker Inc, London, pp. 17–32.

Doglioni, C., Chiarelli, C., Macri, E., Dei Tos, A. P.,Meggiolaro, E., Palma, P. A. and Barbareschi, M.(1998). Cyclin D3 expression in normal, reactive andneoplastic cells. Journal of Pathology, 185, 159–166.

Fillipits, M., Jaeger, U., Pohl, G., Stranzl, T., Simonitsch,I., Kaider, A., Skrabs, C. and Pirker, R. (2002). CyclinD3 is a predictive and prognostic factor in diffuselarge B-cell lymphoma. Clinical Cancer Research, 8,729–733.

Harper, J. (1997). Cyclin dependent kinase inhibitors.In: Cancer Surveys, Vol. 29, Checkpoint Controls andCancer, M. B. Kastan, Ed., Cold Spring HarborLaboratory Press, New York, pp. 91–107.

Kato, J. (1999). Induction of S phase by G1 regulatoryfactors. Frontiers in Bioscience, 4, d787–d792.

Madewell, B. R., Griffey, S. M., Wall, J. and Gandour-Edwards, R. (2002). Reduced expression of cyclin-dependent kinase inhibitor p27 kip-1 in felinelymphoma. Veterinary Pathology, 38, 698–702.

Moller, M. B., Nielsen, O. and Pedersen, N. T. (2001).Cyclin D3 expression in non-Hodgkin lymphoma.American Journal of Clinical Pathology, 115, 404–412.

Motti, M. L., Boccia, A., Belletti, B., Bruni, P., Troncone,G., Cito, L., Monaco, M., Chiappetta, G., Baldassarre,G., Palombini, L., Fusco, A. and Viglietto, G. (2003).Critical role of cyclin D3 in TSH-dependent growth ofthyrocytes and in hyperproliferative diseases of thethyroid gland. Oncogene, 23, 7576–7586.

Pirkmaier, A., Dow, R., Ganiatsas, S., Waring, P., Warren,K., Thompson, A., Hendley, A. and Germain, D.

(2003). Alternative mammary oncogenic pathways areinduced by D-type cyclins: MMTV-cyclin D3 trans-genic mice develop squamous cell carcinoma. Onco-gene, 22, 4425–4433.

Pruneri, G., Mazzorol, G., Fabris, S., Del Curto, B.,Bertolini, F., Neri, A. and Viale, G. (2003). Cyclin D3immunoreactivity in gastrointestinal stromal tumoursis independent of cyclin D3 gene amplification and isassociated with nuclear p27 accumulation. ModernPathology, 16, 886–892.

Quon, H., Liu, F. F. and Cummings, B. J. (2001).Potential molecular prognostic markers in head andneck squamous cell carcinomas. Head and Neck, 23,147–159.

Reed, S. I. (1997). Stereotypical control of the G1/Stransition. In: Cancer Surveys, vol. 29, CheckpointControls and Cancer, M. B. Kastan, Ed., Cold SpringHarbor Laboratory Press, New York, pp. 7–23.

Sherr, C. J. and Roberts, J. M. (1999). CDK inhibitors:positive and negative regulators of G1-phase pro-gression. Genes and Development, 13, 1501–1512.

Teramoto, N., Pokrovskaja, K., Szekely, L., Polack, A.,Yoshino, T., Akagi, T. and Klein, G. (1999). Expressionof cyclin D2 and D3 in lymphoid lesions. InternationalJournal of Cancer, 81, 543–550.

Theon, A. P., Madewell, B. R., Shearn, V. I. and Moulton,J. E. (1995). Prognostic factors associated with radio-therapy of squamous cell carcinoma of the nasalplane in cats. Journal of the American Veterinary MedicalAssociation, 206, 991–996.

Zhan, S. Y., Liu, S. C., Goodrow, T., Morris, R. and Klein-Szanto, A. J. P. (1997). Increased expression of G1cyclins and cyclin-dependent kinases during tumourprogression of chemically induced mouse skin neo-plasms. Molecular Carcinogenesis, 18, 142–152.

Received; July 27th; 2004

Accepted;December 13th; 2004

� �