Involvement of the CCND1 gene in hairy cell leukemia

W of Oncology 7: 251-256, 19%. 0 1996 K l v w a Acodrmic Publishers. Printed in the Netherlandr.

Original article Involvement of the CCNDl gene in hairy cell leukemia C. J. de Boer: J. C. Kluin-Neleman~,~ E. Dreef,' M. G. D. Kester? P. M. Kluin,] E. Schuuringl & J. H. J. M. van Kriekenl Departments of 'Pathology and 2Hemarology, University of Lciden, Leiden, The Netherlands

Summary

Introduction

There have been only a few reports on chromosome abnormalities in hairy cell leukemia (HCL) over the years. Various methods, such as the CD40 in vitro proliferation system or B-cell rnitogen stimulated cells, have been described for obtaining good metaphase chromosomes which can be used for cytogenetics. Different anomalies have been described, including clonal aberrations and numeric and structural changes in the chromosomes. However, no specific recurrent chromosomal aberrations were observed [I-91. This paucity of data is explained by the fact that HCL is a rare entity; it is often difficult to obtain good material, and above all, the cells show a very low proliferative capacity. As an alternative, interphase fluorescent in situ hybridization using alphoid and satellite probes specific for 16 human chromosomes (chromosomes 2,4,5,13,14,19 and 21 not included) showed that HCL appeared to be diploid with non-random gain and loss of chromosomes without any specificity regarding the disease [3]. Involvement of chromosomes 5, 7 and 14q has often been observed, but the presence of 11q13 rearrange-

Key words: BCL-1, CCND1, cyclin D l , HCL, imrnunohistochemistry, RNA, t(11; 14)

ments or even t(11;14) are rare [4, 7, 8, 101. We previously reported mRNA overexpression of CCNDl in some cases of HCL Ill], and therefore we analyzed the involvement of the chromosome 11q13 region and in particular the CCND1 gene in HCL. The chromosome l l q l 3 A C L - 1 region is found in approximately 50% and 7O0/0 of mantle cell lymphomas (MCL) at the genomic and cytogenetic level, respectively, translocated to the immunoglobulin heavy chain joining gene complex (Jh) located at chromosome 14q32 (see references [12-141). The gene involved in this translocation is CCND1. The CCNDl gene is overexpressed in almost all cases of MCL, irrespective of the presence of a t(11;14)(q13;q32) rearrangement, at both the RNA and protein levels [ l l , 151. Ln contrast, the CCNDl gene is not, or only weakly, expressed in normal lymphoid cells and most types of non-Hodgkds lymphoma Sporadically, the l l q 1 3 region is rearranged in several other types of B-cell lymphoproliferative disorders such as follicular lymphoma, B-cell prolymphocytic leukemia, chronic lymphocytic leukemia, multiple myeloma (see reference [12])and hairy cell leukemia [4,7,10].

Downloaded from http://annonc.oxfordjournals.org/ by guest on March 25, 2016

Background Previous results suggested increased mRNA expression of CCNDl in hairy cell leukemia (HCL). The CCNDl gene is involved in the t(11;14)(q13;q32) chromosomal rearrangement, a characteristic abnormality in mantle cell lymphoma (MCL). We and others reported that, in contrast to other B-cell lymphomas, almost all MCL have overexpression of the CCNDl gene with a good correlation between RNA and protein analysis. Recent studies showed that overexpression of the cyclin D l protein can be easily detected by imrnunohistochemistry (MC) on formalin-fixed,paraffin embedded tissues. Patients and methods: To investigate whether the CCND1 gene is involved in HCL, we performed M C on a series of 22 cases using formalin-6xed paraffin embedded splenectomy specimens. For IHC the sections were boiled in citrate buffer. The presence of rearrangements within the BCL-1 locus and the CCNDl gene was analyzed in 13 of 22 cases by Southern blot analysis using all available breakpoint probes. Expression of CCNDl was analyzed at the

mRNA level (Northern blot) and protein level (MC). Results: Overexpression of the cyclin D l protein using IHC was observed in all cases, with strong expression in 5 cases. Preexisting B- and T-cell areas of the spleen did not express sigmlicant levels of the cyclin D l protein. Seven of 9 cases analyzed by both M C and Northern blotting showed overexpression of the CCNDl gene with both methods. No genomic abnormalities were observed in any of the 13 cases studied by Southern blot analysis. Additionally, no l l q 1 3 abnormalities were detected by banding analysis of 19 of 22 cases. Conclusions: The elevated levels of CCNDl mRNA and protein in conjunction with the absence of overt rearrangements within the BCL-1 locus distinguish HCL from MCL and other B-cell malignancies. This suggests that activation of the CCND1 gene in HCL is due to mechanisms other than chromosomal rearrangement.

Patients and methods Patient materials As a referral hospital, the Department of Hematology received during the late 1970s and the early 1980s many spleen samples from Dutch HCL patients. The diagnosis of HCL was established on the basis of the clinical picture, cytomorphology of neoplastic cells, histology of the spleen and bone marrow, cytochemical staining (tartrate-resistant acid phosphatase positive), and immunophenotyping (reactivity with monoclonal antibodies against CDllc, CD19, CD25, CD103 and expression of monotypic immunoglobulin). Single-cell suspensions were obtained from b l d and HCL spleens by gentle mechanical disruption and isolation by Ficoll-lsopaque density gradient centrifugation and frozen in liquid nitrogen. Immunophenotypically, all of the suspensions contained more than 80% hairy cells. The samples analyzed in this study are derived from untreated patients. For Southern and Northern blot analyses, viable frozen cells were used as described, derived either from spleen samples (18 samples) or peripheral b l d samples (2 samples) [I 11. For W C of the cyclin D l protein, formalin-fixed, paraffin embedded spleen tissue was used as described 1151. In this study 13, 15, and 22 cases were analyzed with Southern blot, Northern blot and irnmunohistochemistry, respectively, as summarized in Table 1. Of the total of 29 different HCL patients analyzed in this study, 26 were described previously for cytogenetic analysis 171. In brief, viably frozen cells were cultured by stimulation of the CD40 antigen using anti-CD40

Tablel. Summary of the analysis of chromosome l l q 1 3 and CCNDl involvement in HCL. No. analyzed Southern blot

13

l lq13 rearrangement

0

Northern blot

15

strong expression (*) moderate expression (+) undetectable (-)

5 7 3

IHC

22

strong expression (tt) moderate expression (+) undetectable (-)

5 17 0

monoclonal antibodies and analyzed by cytogenetics. By this technique, three cases showed clonal aberrations, four showed nonclonal aberrations, and normal karyotypes were observed in 20

cases. Nucleic acid analysis Southern blot analysis was performed as described 1121. Briefly, high molecular weight DNA was isolated by proteinase K treatment and phenol-chloroform extraction. 10 yg of DNA was digested with at least three different restriction enzymes (BarnHI, EcoRI, HindIII or Pstl; h h r i n g e r Mannheim Germany) according to the recommendations of the supplier, blotted onto nylon filters (Hybond N+, Amersham, U.K.) and serially hybridized with various probes (see below). Cases were included only when a rearrangement was observed with the immunoglobulin heavy chain joining gene complex (IgH) probe Jh, indicating a clonal tumor population. The following probes were used for analysis of rearrangements within the l l q 1 3 region: BCL-lb/MTC, p94F5, p l l E H , PRADI-pDY1 (representing the 5' region of CCNDl), U21B31 (representing the 3' nontranslated region of CCNDl), 3128-cDNA (representing the cDNA of CCNDl) Ill]. Comigration with chromosome 14q32 was assessed by hybridization with a Jh probe. A total of 13 patients were analyzed. Northern blot analysis was performed as described Ill].In brief, the single cell suspension was washed with phosphate buffered saline (PBS) and RNA was subsequently isolated using ureuml lithiurnchloride and phenol-chloroform extraction. 10 pg of RNA was size fractionated on a 1% formaldehyde agarose gel, blotted onto nitrocellulose filters (Schleicher & Schull, Keene, NH,U.S.A.) and serially hybridized with U21B31 and 3128-CCNDI. The amount of RNA loaded on the gel was analyzed with probes for GAPDH or 28s (obtained from Dr. M. Quax-Jeuken and Dr. J. Schalken, respectively, University of Nijmegen). The cell lines UMSCC-1 and Jurkat were used as positive and negative controls, respectively. A total of 15 cases were analyzed. Three categories were used: -: no detectable expression, +: expression equal to UMSCC-1 and normal lymphoid tissues, ++: expression stronger than UMSCC-1 and normal lymphoid tissues. Hybridization conditions were as described Ill]. Briefly, filters were pre-hybridized for 2 hours at 65 'C in hybridization mix (6 X SSC, 5 X Denhardt's solution, 0.5OA SDS, 50 pg/ml salm sperm DNA and 100/0 polyethyleenglycol (DNA filters) or 10% dextransulphate (RNA filters)). Purified DNA probes were radiolabeled with 20 pCi of [aJ2P]dCTPusing a random-primed labeling kit (Pharmacia, Uppsala, Sweden). Filters were washed for 30 minutes in 1 X SSC/O.l0A SDS and 30 minutes 0.1 X SSC/O.l0/0 SDS at 65 'C after 16 hours of hybridization. Filters were exposed to Kodak X-Omat AR films with intensifying screens at -70 'C.

Immunohistochemical analysis M C for the cyclin D l protein was performed on formalin-fixed, paraffin embedded tissue from splenectomy specimens as recently described 115). Ln brief, tissue sections were adhered to 3-aminopropyltriethoxysilane (APES) coated slides and boiled for 25 minutes in 10 mM citrate buffer pH 6.0 and subsequently incubated with the primary antibdy. We used the cyclin D l monoclonal antibody DCS-6 culture supernatant in two dilutions (1:500 and 1 :2500). Staining was assessed according to staining of endothelial cells and fibroblasts in the same tissue section. Hence, four categories were used: -: weaker than endothelial cells and fibroblasts; -: equally strong as endothelial cells and fibroblasts; + and ++: stronger and much stronger than endothelial cells and fibroblasts. A total of 22 cases were analyzed.

Downloaded from http://annonc.oxfordjournals.org/ by guest on March 25, 2016

In this study we describe the analysis of rearrangements within the llql3ACL-1 region and expression of the CCNDl gene at the protein level using irnrnunohistochemistry in HCL. This study includes 22 cases analyzed by MC, but also a compilation of previously published banding analyses of HCL using the CD40 culturing system [7], Southern blot analysis for rearrangements within the BCL-1 locus and previously reported Northern blot data [ll]. Methods for detecting overexpression of cyclin D l protein by M C on routinely processed biopsies of MCL and other B-cell neoplasias, including HCL, have been recently published [15-181. In our series of HCL, all 22 cases studied by M C showed overexpression of the cyclin D l protein, but no breakpoints within the BCL-1 locus could be detected by any of the above-mentioned techniques. This strongly suggests that mechanisms other than translocations are implicated in the deregulation of the CCNDl gene in HCL.

Results

-

t Cyclin D l

Figure 1. RNA analysis of CCNDl expression in HCL. 10 pg of RNA was size fractionated on a 1% agarose, formaldehyde gel, transferred to a filter and serially hybridized with both CCNDl probes and a 2 8 s probe as control for the amount of RNA loaded. RNA of the cell line UMSCC-1 was used as a reference for CCND1 expression. Cases of HCL indicated at the top, the molecular weight marker at the left, the transcripts o n the right.

MCL (Figure 2b), but stronger than in all other B-cell lymphomas and leukemias (n-48) and normal lymphoid tissues ( n = 10) tested previously [15J.In some HCL, residual foci of the splenic white pulp were detected. These areas did not stain for cyclin D l (Figure 2a).

Table 2. Immunohistochemistry for cyclin D l in hairy cell leukemia No.

Case'

Sex/ age'

Cytogenetics'

Southern blot analysis Jhb MTC

p94PS p l l E H

RNAC M C 5'Dl

3'Dl

cDNADl

Cclsur analymi for mRNA and IHC Gy-W Ra S-J Rie Jo Ka Hof M ~ Y LY Cam am&& Mu Ma W-J G roe wo

J-B

Ek Gra Za

KO Hoo wou vDa

normal RR normal RR normal RR abnormal, nonclonal RR 46, XY, add (14) (q32), del (6) (p21) [ I ] normal RR abnormal, nonclonal RR 46, XY, del (7) (q32) llV46, XY [291 normal na normal na normal na wirh IHC only

normal RR na normal R R abnormal, clonal R R 46,XX [21p45,X,-X,add (4)(p15),del (lO)(q24) Ill 46, XX, der (18) t(5;18)(ql l;q23) 129V46, XX (171 abnormal, clonal R R 47, XY, +5 [181/46, XY [I 1 na na na na na na na na na na na na normal na na na normal na na na normal na na na normal na na na normal na na na normal n a n a na

' Adapted from 171; RR: J h rearrangement present; 'adapted from [ll]; sample left overnight in culture medium at 37'C. na not analyzed; nb sample could not be scored. Scoring: - negative; + moderate expression; strong expression.

- -

--

*-

1:500

1:2500

Downloaded from http://annonc.oxfordjournals.org/ by guest on March 25, 2016

Our previously published Northern blot results [11] suggested overexpression of the CCNDl gene in at least one-third of HCL. In 12 of 15 cases expression was observed, with high expression in 5 cases and moderate levels of expression in 7 cases (Figure 1). Expression of the CCNDl gene was unlikely to be due to an admixture of accessory cells in the HCL samples since all samples contained more than 80°h hairy cells. Based on these observations and previous reports that M C on formalin-fixed, paraffin embedded tissues is a reliable and sensitive method for detecting (over) expression of the cyclin D l protein in B-cell neoplasia [15-181, we analyzed a large series of HCL spleens by IHC (Table 1and 2). Using the well-defined cyclin D1speclfic monoclonal antibody DCS-6 [19],we observed overexpression in tumor cell nuclei in all cases of HCL (n 22). Five cases showed a very strong staining, and the remaining 17 cases showed a moderate level of overexpression. Overexpression was defined by a stronger staining in tumor cells than in fibroblasts and endothelial cells within the same section. Variation in staining intensity among individual tumor cells within the same section was observed in all cases (Figure 2a). In general, the staining intensity was weaker than in

turing in the CD40 system [7].Two cases showed clonal and two non-clonal chromosomal abnormalities, and 15 cases showed a normal karyotype. None of these cases had a t(11;14)(q13;q32) or other abnormalities involving llq13. Discussion

-

Nine of 22 HCL were tested by both M C and Northern blot analysis (Table 2). In 7/9 cases increased expression was observed with both Northern blot and irnrnunohistochernistry, although no clear correlation was found in the level of overexpression. In 2/9 cases no overexpression was detected by Northern blot analysis, whereas irnrnunohistochemistry showed moderate expression. All cases of HCL showed the major transcript of 4.5 kb, and no abnormally sized transcripts were detected. Nine of 22 cases analyzed by M C were also studied by Southern blot for translocations within the BCL-1 locus. By none of the 6 available breakpoint probes did any of these cases show rearrangements indicative of a t(11;14)(q13;q32). The probes included a 3' probe for CCNDl by which translocations and deletions within the 3' untranslated region of CCNDl can be detected [Ill. The overall results of our analysis are shown in Tables 1 and 2. Nineteen of 22 cases analyzed by IHC had been studied before by banding analysis after cul-

Downloaded from http://annonc.oxfordjournals.org/ by guest on March 25, 2016

Figure 2 Lmmunohistochemistry of cyclin D l in HCL and normal spleen. Sections were stained for cyclin D l with the DCS-6 antibody (1:2500 dilution of culture supernatant; origud magnification 2 0 0 ~ ) A: . HCL, showing inliltration of the red pulp by HCL with variable nuclear staining for cyclin D l . Note the absence of staining wthm the preexisting lymphoid cells of the white pulp (asterisk). B: MCL, showing a mantle zone growth pattern. Note the very strong expression in the neoplastic mantle zone cells and its absence in the pre-existing foUicle center cells (asterisk).

Recent reports from our group and others show that M C is a reliable method to detect (over-) expression of the CCNDl gene in B-cell neoplasia [15-181. Using this method, we observed that in a series of 22 HCL, all cases showed overexpression of the cyclin D l protein. Five cases showed very strong expression and the remaining 17 showed moderate levels. These results corroborate the Northern blot results indicatmg that the majority of HCL have a relatively high expression of the cyclin D l protein. The high variability observed in expression levels, and even the lack of expression in 2 of 9 cases with Northern blot analysis is perhaps attributable to the use of different materials used for the two methods. For M C freshly fixed tissues were used but for mRNA analysis splenic cell suspensions were used. CCNDl mRNA has a short half-life (around 30 minutes [20, 211). We speculate that during the timeconsuming procedure of isolation of the HCL cells using FACS-sorting, and subsequent freezing and thawing of the cells, the CCND1 gene product might be partially degraded in some cases. Another possibility could be the presence of an alternatively spliced CCNDl transcript, as has recently been described [22]. Alternative splicing is, however, not likely to have happened in our HCL samples, since in our study CCND1 mRNA levels were analyzed with two different CCNDl probes, a cDNA probe (3128-CCND1) which also hybridizes with the alternatively spliced transcript, and a probe in the 3'-nontranslated region (U21B31), and both probes showed identical results. Therefore, detection of cyclin D l protein overexpression using M C seems to be a more reliable method than Northern blotting. In contrast to the present results, Zuckerberg et al. [17] detected overexpression of the cyclin D l protein in only 1/15 cases of HCL. This could be due to the use of another less-sensitive cyclin D l antibody, unable to detect the moderate levels of overexpression. In MCL cyclin D l overexpression is observed in almost all cases [ l l , 151. Depending on the method used, a t(11;14)(q13;q32) can be detected in 50°/0-750/~ of the cases [12-141. In the 22 HCL cases with cyclin D l overexpression using MC, we did not detect chromosomal or genomic 11q13 rearrangements by any of the used methods. Clonal and non-clonalllq13 abnormalities, especially del (ll)(q13q2l), have been described by Haglund et al. in 8 cases of 36 mitogenstimulated HCL [4]. Terminal deletions at l l q are frequent in chronic lymphocytic leukemia (CLL) as well, and may occur in tumor subclones. In our recent study, Coignet et al. analyzed 5 cases of CLL with sirni-

--

--

-

-

volvement of the CCNDl gene in HCL. Our results show that CCNDl is indeed involved in HCL. At present, we do not know whether this high expression reflects a structural oncogenic event in the genome of the tumor cells undetectable by current methods, deregulation at the posttranscriptional level, or whether in hairy cells it is physiologic. The latter hypothesis can not be totally excluded, since the normal counterpart of hairy cells is not known and since very few cells in normal lymphoid tissues show a considerable level of expression of the cyclin D 1 protein [IS]. Acknowledgements .-

-~

-

. -. -

We t h a n k ~ rJ.. Bartek of the Danish Cancer Society (Copenhagen, Denmark) for providing us with the DCS-6 cyclin D l specific monoclonal antibody, and Dr. A. Arnold, Dr. T. Carey, Prof. Dr. D. Catovsky, Dr. T. Meeker, Dr. G. Peters, Dr. M. Quax-Jeuken,Prof. Dr. T. Rabbits, Dr. J. Schalken, Prof. Dr. Y. Tsujimoto and Dr. M. Williams for providing probes, and Karin Kleiverda for performing the FISH analysis.

References Chang KL,Stroup R Weiss LM.Hairy cell leukemia. Current status. Anatomic Pathol 1992; 97: 719-38. Ueshima Y, Alimena G, Rowley JD et al. Cytogenetic studies in patients with hairy cell leukemia Hematol Oncol 1983; 1: 215-26. Lewis JP, Tanke HJ, Raap AK et al. Hairy cell leukemia: An interphase cytogenetic study. Leukemia 1993; 7: 1334-8. Haglund U, Juliusson G, Stellan B et al. Hairy cell leukemia is characterized by clonal chromosome abnormalities clustered in specific regions. Blood 1994; 83: 2637-45. Han T, Sadamori N, Block AMW et al. Cytogenetic studia in chronic lymphocytic leukemia, prolymphocytic leukemia and hairy cell leukemia: A progress report. Nouvelle Revue Franqaise d'Htmatologie 1988; 30: 393-5. Golomb HM,Lindgren V Rowley JD. Hairy cell leukemia: An analysis of the chromosomes of 26 patients. Virchows Archives 1978; 29: 113-20. Kluin-Nelemans JC, Beverstock GC, Mollevanger P et al. Proliferation and cytogenetic analysis of hairy cell leukemia upon stimulation via the CD40 antigen. B l d 1994; 84: 313441. Britc-Babapulle V, Pittman S, Melo JV et al. The 14q+ marker in hairy cell leukemia A cytogenetic study of 15 cases. Leuk Res 1986; 10: 131-8. Ohyashiki K, Ohyashiki JH, Takeuchi J et al. Cytogenetic studies in hairy cell leukemia Cancer Genet Cytogenet 1987; 24: 109-17. Nacheva E, Fischer P, O'Connor S et al. Complex chromcsoma1 rearrangements in and unusual variant of hairy cell leukemia Cancer Genet Cytogenet 1992; 62: 186-90. De Boer U, Van Kriekcn JHJM, Kluin-Nelemans JC et al. Cyclin D l messenger RNA overexpression as a marker for mantle cell lymphoma Oncogene 1995; 10: 1833-40. De Boer CI,Loyson S, Kluin PM et al. Multiple breakpoints within the BCL-1 locus in B-cell lymphoma: Rearrangement of the cyclin D l gene. Cancer Res 1993; 53: 4148-52. Raffeld M, Jaffe ES. BCL-1, t(11;14), and mantle cellderived lymphomas. B l d 1991; 78: 259-63.

Downloaded from http://annonc.oxfordjournals.org/ by guest on March 25, 2016

lar terminal deletions seeking BCL-1 rearrangements using interphase and metaphase FISH with cosmids spanning a 750 kb region around the MTC of the BCL-1 locus [23]. In none of these cases could a BCL-1 breakpoint be detected. Recently, we used the same method in 2 cases of HCL with cytogenetically documented 11q13 abnormalities [7]. No BCL-1 breakpoints were detected (unpublished data). Unfortunately, both the 5 CLL cases with terminal deletions [23] and the present 2 HCL cases with llq13 translocations could not be studied for cyclin D l expression since no paraffin embedded material was available. Our compiled data on expression of the CCNDl gene and llq13 rearrangements in HCL suggest that either HCL harbor .relevant 11q13 breakpoints I+. detectable by the current methods, or, more likely, that HCL cells have other mechanisms leading to deregulation of the CCNDl gene. Of note, our study made deletions and translocations immediately 3' of the CCNDl gene very unlikely. Such events have been described in a number of cell lines and tumors [ l l , 24271 [own unpublished data], and cause the accumulation of mainly the 1.5 kb or abnormally sized CCNDl transcripts by removing mRNA destabilizing sequences. No gross abnormalities were detected in HCL. Furthermore, genomic alterations of the CCND1 gene itself are unlikely to be involved in HCL because we did not observe genomic aberrations with either the 5' CCNDl probe or a CCNDl cDNA probe. Overexpression of the CCNDl gene is suggested to play a subtle role in tumor development. CCNDl has an essential function in the cell cycle, especially in the GI-S phase transition (reviewed in [28, 291). Overexpression induced by transfected CCND1 constructs resulted in a shorter G1 phase and total time required for one cell cycle in cell lines. Transgenic mice with an Ep-CCND1 construct show subtle changes in B- and T-cell composition and only develop B-cell lymphomas after deregulation of other genes such as myc [30, 311. The role of the CCNDl gene in HCL is remarkable, especially since the tumor cells have a very low growth fraction. Blocking of CCND1 expression by antisense or antibody treatment of cell lines results in a cell cycle block and G1 phase-arrest. The cyclin D l protein has, furthermore, been shown to be associated with other cell cycle-related proteins such as cyclindependent kinase 4 (CDK4), CDK6, p16 and the retinoblastoma gene product pRb. Cyclin Dl-CDK complexes can phosphorylate pRb which thereby releases transcription factors. Cyclin Dl-CDK complexes are inactive when complexed with p16, an inhibitor of O K . p16 is therefore believed to be a tumor suppressor gene: absence of p16 could result in the constant activation of cyclin Dl-CDK4 complexes. In two reports on p16 in HCL, no obvious deletions or point mutations were found [32, 331. Constant activation of CCNDl due to loss of p16 is therefore not likely to occur in HCL. To our knowledge this is the first report on the in-