The human protein p19ARF is not detected in hemopoietic ... - Nature

Oncogene (1997) 15, 2475 ± 2481  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

The human protein p19ARF is not detected in hemopoietic human cell lines that abundantly express the alternative b transcript of the p16INK4a/MTS1 gene VeÂronique Della Valle, Dominique Duro, Olivier Bernard and Christian-Jacques Larsen INSERM U-301, Institute de GeÂneÂtique MoleÂculaire, 27, rue Juliette Dodu, 75010, Paris, France

The p16/MTS1/CDKN2 gene on human chromosome band 9p21 encodes two unrelated proteins: p16INK4a, a speci®c inhibitor of the cyclin D-dependent kinases CKD4 and CDK6, and the structurally unrelated p19ARF protein that arrests cell growth in G1/S and also in G2/M. By use of polyclonal antibodies, the human p19ARF (hp19ARF) protein has been identi®ed in the nucleus of various cells including normal cultured ®broblasts. The level of this protein did not ¯uctuate throughout the cell cycle and was more elevated in ®broblasts with limited or arrested growth, suggesting that p19ARF accumulated in presenescent or senescent cells. Interestingly, hp19ARF was not detected in several hemopoietic tumor cell lines (mainly of B-type lymphoid origin) that expressed abundant amounts of the p16b transcript. This ®nding indicates that in certain tumors, the expression of hp19ARF RNA and protein may be uncoupled. Furthermore, it suggests that disruption of a translational mechanism may be involved in the inactivation of hp19ARF. Keywords: p16INK4a; MTS1; CDKN2; p19ARF

Introduction The human and mouse p16INK4a genes exhibit unusual properties in the sense that they encode two transcripts derived from the two alternative ®rst exons 1a and 1b and they encode two structurally unrelated proteins (Duro et al., 1995; Mao et al., 1995; Stone et al., 1995; Quelle et al., 1995). The a transcript codes for the p16INK4a/MTS1/CDKN2 protein that blocks the G1 to S transition of the cell cycle by speci®cally inhibiting cyclin D1-3-dependent CDK4/6 and the resulting phosphorylation of the pRb protein (for a review, see Sherr, 1996). The b transcript encodes a p19ARF protein of 132aa in man and 156aa in mouse, of which the primary structure is totally unrelated to that of p16INK4a (Larsen, 1996). Strikingly, in spite of this unrelated structure, p19ARF also arrests cells in G1/S and G2/M by a presently unknown mechanism (Quelle et al., 1995; Liggett et al., 1996). Because of its location on human chromosome band 9p21 that is frequently rearranged on both alleles in a variety of solid tumors and acute lymphoblastic leukemias (Olopade et al., 1992; Coleman et al., 1994, and references herein), p16INK4a is considered as a tumor suppressor gene whose inactivation plays a Correspondence: C-J Larsen Received 5 May 1997; revised 11 July 1997; accepted 11 July 1997

major role in the genesis and/or development of tumors. In this respect, the most representative example comes from studies on familial melanomas in which point mutations in the p16INK4a gene have been shown to co-segregate with the occurrence of tumors in patients (for a review, see Flores et al., 1996). Consistently, it has been abundantly documented that these mutations are directly responsible for the incapacity of the mutated p16INK4a protein to bind CDK4/6 and the concomitant loss of its inhibitory function (Koh et al., 1995; Wick et al., 1995; Parry et al., 1996). Altogether, these data strongly point to p16INK4a as the genuine target of alterations that impair the p16 locus. In contrast, the place of p19ARF in tumorigenic processes is still poorly understood. While expression of a wild type copy of human p19ARF in cells devoid of functional Rb protein was shown to arrest cells in G1 (Liggett et al., 1996), a recent report by Sherr's group has shown that point mutations located in the part of exon 2 that is common to p16INK4a and p19ARF did not result in the incapacity of the latter to exert its inhibitory role (Quelle et al., 1997). In addition, the same group also showed that the aminoterminal part of p19ARF encoded by exon 1b is necessary and sucient to abolish the functional activity of the protein. To date, with only one published exception, no mutation in exon 1b has been so far detected in several tumor types (Fitzgerald et al., 1996). For these reasons, p19ARF is not likely to be signi®cantly involved in tumorigenic processes, at least as regards a number of dierent tumor types. In this report we have characterized the human p19ARF protein (hp19ARF) in various cells of hematopoietic origin by use of polyclonal rabbit antibodies to a recombinant hp19ARF protein. A survey of the expression of hp19ARF in these cells indicated that the protein is not expressed in some cell lines in which the b transcript is abundant. The signi®cance of this result is discussed with regard to the tumorigenic process.

Results Preparation and validation of antibodies to the human hp19ARF protein A rabbit antiserum (referred to as serum 73) raised against a recombinant GST-hp19ARF was prepared (see Materials and methods). This serum immunoprecipitated a 35S-methionine-labeled protein with a mobility slightly lower than that of the p16INK4a protein after in vitro transcription and translation of the b and the a transcripts, respectively (Figure 1a and data not shown). This band was not detected by nonimmune

Human p19ARF protein in tumors V Della Valle et al

2476

rabbit serum and was eliminated upon addition of an excess of GST-hp19ARF in the immunoprecipitation reaction mixture. In addition, the serum did not recognize the in vitro p16INK4a product (data not shown). Next, the anti-hp19ARF serum was utilized to immunoprecipitate the protein from extracts of metabolically labeled HeLa cells (the choice of HeLa cells was dictated by the report that the b transcript is expressed in this cell line, see Mao et al., 1995). A protein was readily detected by immunoprecipitation in total cell extracts (Figure 1b, lane 3), but not in extracts of Reh cells chosen as negative control (lane 1), since the p16 locus is homozygously deleted in this cell line (Duro et al., unpublished data, Zhou et al., 1995). The intensity of the band greatly diminished upon mixing of the serum with GST-hp19ARF (lane 4). The same band was also present in cell extracts from Meg-01 human tumor cells with a megakaryoblastic origin (lane 5). Consistent with these data, the protein was also detected by Western blot in immune complexes of HeLa cell and Meg-01 cell extracts but not in those of Reh cells (see Figure 3). In separate experiments with metabolically labeled HeLa cells, the signals obtained by ECL revelation and by autoradiography were superimposable (data not shown). a Clone 13 1

2

3

4

5

Taken together, these data indicate that serum 73 has the ability to speci®cally detect the human p19ARF protein. Indirect immuno¯uorescence studies Serum 73 was also used to perform indirect immuno¯uorescence (IIF) on cells grown on glass coverslips. In the case of HeLa cells, a micropunctuated staining with anti-hp19ARF antibody was observed in the nucleus (Figure 2). Nucleoli were generally not stained. On the other hand, the staining was homogeneous in all cells examined, suggesting that the protein levels do not grossly ¯uctuate during the cell cycle. Depending on the nature of the cells, a more or less intense nuclear staining was observed. For example HS-27 human ®broblasts which were karyotypically and morphologically normal exhibited a much lower ¯uorescence than HeLa cells analysed simultaneously. Nevertheless, the intensity of the staining appeared to be homogeneous in all the nuclei examined, again suggesting that the protein levels did not vary very much throughout the cell cycle. This behaviour did appear not to be entirely consistent with RNA data on human peripheral lymphocytes that indicated a dramatic increase of the b transcript level upon stimulation by phytohemagglutinin plus IL-2 (Stone et al., 1995). This result suggests a parallel ¯uctuation of the protein level. The fact that HeLa cell immuno¯uorescence patterns did not reveal any protein level ¯uctuations may be related to dierent

21.5 KDa —

p19ARF 12.5 KDa —

HeLa

b Reh 1

Hela 2

3

Meg 01 4

5 HS-27

— 21.5 kDa p19ARF Figure 1 A rabbit antiserum to recombinant GST-hp19ARF recognizes the protein in cell lines. (a) Immunoprecipitation of in vitro transcription/translation product of a hp19ARF cDNa clone (clone 13). lane 1: no RNA; lane 2: nonimmune rabbit serum; lane 3: anti-GST-hp19ARF serum; lane 4: same as in lane 3 plus GST protein in excess; lane 5: same as in lane 3 plus GSThp19ARF protein in excess. (b) Immunoprecipitation of extracts of metabolically labeled cell lines: p167/7 Reh cells (lane 1), HeLa cells (lanes 2 ± 4), Meg-01 cells (lane 5). lane 1: anti-hp19ARF serum; lane 2: non immune rabbit serum; lane 3: anti-hp19ARF serum; lane 4: anti-hp19ARF serum plus GST-hp19ARF protein in excess. The faint residual band most likely represents non speci®c material also detected by nonimmune serum in lane 2. lane 5: anti-hp19ARF serum

Figure 2 Indirect immuno¯uorescence. Top: HeLa cells. Bottom: HS-27 human ®broblasts. For each cell type, the picture on the right results from competition of anti-hp19ARF antibodies with GST-hp19ARF in excess (1 mg/ml of serum). Notice that the micropunctuated pattern of HeLa cell nuclei is less obvious than on original pictures


cell status (in this case normal primary cells versus tumor cell line) and dierent experimental conditions. Finally, Western blotting of immune complexes con®rmed that in normal ®broblasts, hp19ARF was expressed at low levels (Table 1; see also Figure 5). Characterization of hp19ARF in various cells Hemopoietic cells of dierent sources (cell lines or fresh cells) were analysed by immunoprecipitation followed by Western blot for their p16INK4a and hp19ARF contents. A set of these analyses is presented in Figure 3 and summarized in Table 1. While some cells (HeLa, Meg-01, Nalm-6) displayed signi®cant levels of the protein, or only low levels (K562), others were totally negative (Reh, Raji, Daudi, HL-60, Jurkat), although the reasons for this negative

expression were dierent. Reh cells have a homozygous deletion of the p16 locus (Duro et al., unpublished data and Zhou et al., 1995). But some of the other cell lines, particularly those with a mature B-type phenotype (Raji, RPMI 8226, Daudi) were veri®ed by Southern blot analysis to bear the whole p16 locus. Strikingly, these latter cells exhibited abundant levels of the b transcript which were readily detected by Northern blot analysis (see Table 1). These results that were reproduced in independent experiments and by dierent procedures (IP, IP/WB, immmuno¯uorescence) made it impossible to correlate positively the expression of RNA and protein. Since p16INK4a and hp19ARF are encoded by the same gene, the p16INK4a protein expression in the dierent cell lines was also examined (Figure 3). In general, our results were in agreement with those already published

Table 1 IP

Protein hp19 IP/WB

IIF

Methylation

Comments

++ +++ +++ ++ 7 ++ + + ND ++ ND ND

+ 7 ND ND 7 7 ND ND ND ND ND +/7

+ 7 7 7 7 7 ND ND 7 + + +

+ 7 ND ND 7 7 7 ND ND + ++ +

7 + + + ND + ND ND ND + ND ND

a and b-transcripts detected only by PCR Burkitt cells. No a-transcript Myeloma cells Burkitt cells. No a-transcript Homozygous deletion of the p16/p15 locus

ND ND ND ND

ND 7 ND

+ +/7

ND + ND ++

ND ND ND ND

Cells

b-Transcript North. PCR

HeLa Raji RPMI 8226 DAUDI Reh HL-60 PBL T Lymphoc. Jurkat Nalm-6 DAMI HS27

7 +++ +++ ND 7 + ND 7 7 7 ND ND ND ND ND ND

PFF K-562 MEL MEG-01

ARF

+

Puri®ed by rosetting Pre-B ALL cells human megakaryoblastic cell line Human diploid ®broblasts senescent or presenescent cells primary foreskin ®broblasts murine erythroleukemic cells human megakaryoblastic cell line

21 kD —

14 kD —

Hela (RS)

Hela

K562

HL60

Jurkatt

Daudi

Meg 01

Nalm6

Raji

Reh

p 16INK4a

Hela

North.: Northern Blot; PCR: Polymerase Chain reaction analysis. In most of the cases, the assays were repeated. IP: immunoprecipitation following metabolic cell labeling; IP/WB: immunoprecipitation and analysis of the immune complexes by immunoblotting; IIF: indirect immuno¯uorescence

—

—

p19ARF

21 kD —

14 kD —

—

—

Figure 3 Expression of the hp19ARF protein in hematopoietic cell lines. Immune complexes of dierent cell lines extracts were electrophoresed and hp19ARF was revealed by Western blot analysis. The same cell extracts were also treated for p16INK4a. HeLa cells and Reh cells were included as positive and negative controls, respectively. RS: nonimmune rabbit serum

2477


2478

in the literature. Again, the reasons for the negative results varied according to the cell type. In addition to the presence of homozygous deletions of the p16INK4a gene in some cell lines (Reh), for others (Raji, Daudi, RPMI 8226, HL-60) the p16INK4a promoter has been shown to be hypermethylated, thus resulting in total absence of expression (Duro et al. 1996). To our knowledge, no methylation of the 1b promoter has been reported so far (see also Swaord et al., 1997). Transient transfection experiments In an attempt to get insight into one mechanism responsible for the absence of the hp19ARF, the translation capacity of the b transcript cDNA isolated from one hp19ARF protein negative cell line, Raji, was tested. By using transcription/translation in vitro assay, the cDNA appeared to generate a product that was immunoprecipitated by serum 73 and that migrated at the expected position for hp19ARF (data not shown). The ability of b cDNAs to encode the hp19ARF protein was further tested in transfection experiments (Figure 4). The hp19ARF-encoding sequence of the clone 13 cDNA (isolated from RPMI 8226 cells, Duro et al., 1995) was inserted in frame with hemagglutinin (HA) into the DEB vector (Boulukpoz et al., 1989), and the recombinant vector was used to transfect Cos cells and HeLa cells (see Materials and methods). After 24 h, the cells were metabolically labeled for 4 h and immune complexes from cell extracts were analysed by autoradiography. A strong band running at the expected potision for the HA-hp19ARF protein was detected by serum 73 (Figure 4, lane 2) as well as by anti-HA antibodies (lane 4). As expected, this band was completely abolished upon addition of an excess of GST-hp19ARF protein (lane 3).

A similar result was obtained by transfecting cells (C-33A cells and U2-0S cells) with another construct comprised of a hp19ARF cDNA isolated from Raji cells and placed under control of a cytomegalovirus (CMV) promoter. A band running at the position of hp19ARF was detected on blots resulting from IP of transfected cell extracts (data not shown). Indirect immuno¯uorescence analysis of HeLa cells recovered 48 h after transfection generated a nuclear pattern (Figure 4b). Similar patterns were observed with anti-HA antibodies. By taking into account the mean of several experiments, 10 ± 15% of the cells appeared to be positively stained. Because of the high intensity of the positively transfected cells, the micropunctuated staining pattern was not obvious. It also diered from the pattern displayed by murine NIH3T3 cells infected with p19ARF, which mainly consisted of speckles (Quelle et al., 1997). The hp19ARF protein exhibits a moderate stability HeLa cells were labeled for 6 h with [35S]-methionine+cysteine, then submitted to chase for 19 h in normal medium, the analysis of the hp19ARF content from identical numbers of cells indicated that some signi®cant labeling of the protein was still present at the 8th hour of chase but not at the 19th hour (Figure 5a). These data are consistent with a moderate stability of the protein exceeding at least 8 h in proliferating HeLa cells. However, it cannot be excluded that the cell line status of HeLa cells introduces a bias as it has been recently mentioned that p19ARF levels tend to accumulate in cell strains passaged in culture (Quelle et al., 1997). Accordingly, the relative stability of p19ARF (as well as that of p16INK4a) may rely on the particular status of tumoral cells inde®nitely maintained in culture.

b

a

hp19ARF

C 1

2

3

4 — 30 kDa

hp 19ARF

— 21.5 kDa

Figure 4 Transfection of a hp19ARF-encoding cDNA vector in HeLa cells. (a) Immunoprecipitation of a cell extract from metabolically labeled cells (4 h of labeling). lane 1: negative control (C); lane 2: anti-GST-hp19ARF serum; lane 3: same as in lane 2 plus GST-hp19ARF protein in excess; lane 4: anti-HA antibodies. The intensity of the band migrating at the position expected for the endogenous p19ARF varied widely in dierent experiments. (b) Indirect immuno¯uorescence. Top: anti-hp19ARF serum. Staining with FITC anti-rabbit antibodies. Bottom: staining of nuclei with DAPI. Because of the high ¯uorescence levels it was not possible in these experiments to decide whether nucleoli were not interested by the staining


The hp19 protein may accumulate in presenescent or senescent cells ARF

In experiments performed with the normal human ®broblast HS-27 cell line, we noticed that the growth rate of these cells decreased during the time they were studied, suggesting that they were evolving toward senescence. Consistent with this possibility, the morphology of the cells became ¯at and enlarged while proliferation ®nally stopped at subcon¯uent densities. This was in contrast with cultures of primary foreskin ®broblasts (PFF cells) that grew steadily to con¯uence and adopted the typical morphology of normal ®broblasts. In order to compare the expression of hp19ARF in both cell types, immunoprecipitation plus Western blotting were performed on extracts prepared

a 1

2

3

4

5

p19ARF—

1

2

3

4

p16INK4a—

HS-27

PFF

RPMI 8226

HeLa

Reh

b

— p19ARF

Figure 5 hp19ARF protein is metabolically stable and accumulates in senescent cells. (a) Pulse-chase experiment. A HeLa cell culture was labeled for 6 h with [35S]-methionine+cysteine, then chased in the presence of cold medium. Identical numbers of cells were processed at the indicated times. Immune complexes were electrophoresed and blotted. The blots were autoradiographed. The protein total amounts in each lane were estimated by immunoblot and chemoluminescence (data not shown). lane 1: 6 h pulse labeling; lane 2: 6 h labeling+8 h chase; lane 3: 6 h labeling+19 h chase; lanes 4 and 5: negative controls with non immune rabbit serum. The band above the p16INK4a band is non speci®c. (b) Western blot analysis of immune complexes from dierent cell extracts. Cell extracts corresponding to identical cell numbers were processed as described in Figure 4. Following electrophoresis of the immune complexes, revelation of the p19ARF signal was carried out by ECL (serum dilution: 1 : 500). The prominent band at the top of each lane corresponds to the non speci®c Ig light chain signal

from equal numbers of HS-27 and PFF cells (Figure 5b). While the protein was detected in both cells, the intensity of the signal was signi®cantly higher in HS-27 cells than in PFF cells. These data suggested that the hp19ARF protein tends to accumulate in cells that acquire a senescent phenotype. In this respect, hp19ARF may be similar to p16INK4a which has been reported to be present at elevated levels in primary ®broblasts at senescence (Hara et al., 1996; Alcorta et al., 1996), as well as in uroepithelial cell cultures undergoing senescence (Rezniko et al., 1996). Discussion The data presented in this paper on the human p19ARF (hp19ARF) protein are in general agreement with results already reported for the murine p19ARF protein (Quelle et al., 1995). Both murine and human molecular species are found in the nucleus and appear to be expressed at low levels in normal cells. In contrast, we have detected abundant amounts of the human protein in some cultured tumor cells (HeLa cells, Meg-01 cells). In view of the inhibitory cell cycle activity of p19ARF, this ®nding suggests that an inactive form of the protein is present in these tumor cells. Alternatively, it may re¯ect a defect in the biochemical pathway in which hp19ARF is involved. Presently, no partner of hp19ARF has been identi®ed. It is, therefore, interesting that Quelle et al. recently mentioned the association of mp19ARF with several proteins by virtue of their speci®c coimmunoprecipitation from metabolically labeled murine cells infected with a retrovirus encoding the wild type p19ARF protein (Quelle et al., 1997). Importantly, these proteins were not detected when a construct lacking the N-terminal moiety only encoded by exon 1b was utilized to transfect recipient cells. The other interesting result presented in this work is the absence of the protein in cells that abundantly express the b transcript (Duro et al., 1995). Noticeably, in all the experiments presented here, the same cell lines were used to analyse in parallel RNA and protein expression. Several hypotheses may be formulated to explain this discrepancy. Either the protein is immediately degraded by proteolytic activity immediately following its translation. In this respect, a p27kip1 proteasome-mediated degradative activity has been recently reported in extracts of colon carcinoma tissues that contained no p27 protein or only low levels (Loda et al., 1997). In contrast, carcinomas expressing high p27 protein levels appeared to be devoid of this proteolytic activity. While a similar situation may exist for hp19ARF, we do not presently favor it. First, to be active, the putative protease activity should be immediately available. Although this condition is likely to occur in cell extracts where all cell components have been disrupted during the extraction procedure, it appears most unlikely that protease activity is at work based on the results provided by IIF studies involving immediate ®xation and therefore stabilization of the cell structure (see Table 1). Furthermore, by analogy with the p27kip1 data (Loda et al., 1997), some residual hp19ARF could be expected to be present in the cells exhibiting the proteolytic activity. Finally, it is striking that no lysine

2479


2480

residue that is required for ubiquitination, the ®rst step of degradation, is present in the primary structure of hp19ARF. However, we cannot rule out the possibility that a protease activity is present in those tumor cells in which the b transcript and hp19ARF expression are not correlated. A second possibility is that the hp19ARF mRNA is actually not translated in the protein-negative cell line analysed in the present study. In this respect, a particular structure of the 5' part of the native b transcript synthesized in these cell lines might prevent its ecient translation. However, all our eorts to isolate more 5' transcripts were unsuccessful. Other groups have also failed to isolate some longer transcripts (Stone et al., 1995). These data indirectly support the idea of an anomalous 5' mRNA structure that inhibits its translation. Interestingly, a sequence analysis of the 5' genomic sequence surrounding exon 1b has shown the presence of a CpG-rich region and a GT-rich region suggestive of a GT box promoter element (Mao et al., 1995). In spite of numerous CpG dinucleotides, this region has not been reported to be methylated (Swaord et al., 1997). Another point to consider is the presence of a speci®c factor inhibiting the translation of the b transcript in tumor cells that express it at high levels. In view of the cell cycle inhibitory function of hp19ARF, this hypothesis is attractive as it is conceivable that such a putative inhibitor would have been selected by cancer cells in order to relieve them from growth inhibition. One evident prediction of this model is that hp19ARF is involved in tumorigenic processes. However, the most recent data on this point do not suggest that this is the case (see Quelle et al., 1997 and references herein). Finally, a completely dierent possibility is that the hp19ARF protein present in some human tumor cell lines is not recognized by the antibodies we used. The recent suggestion that a truncated protein, comprised of the only exon 1b-encoded moiety, might be expressed in tumor cells is plausible as such a truncated species might not contain the epitopes reacting with our antibodies (Quelle et al., 1997). We do not have currently any indication of the epitopic speci®city of the serum used throughout this study. In conclusion, similar to the uncoupling of hp19ARF RNA and protein expression described herein, it has been reported that, in mouse skin carcinoma cell lines and human hepatocarcinomas (cell lines and tumors), the presence of the a transcript was not accompanied by p16INK4a protein expression (Linardopoulos et al., 1995; Hui et al., 1996). Since no mutation was apparently detected, these results as well as ours suggest that p16INK4a expression and hp19ARF expression may be inactivated by posttranscriptional regulation rather than by genomic alterations. Moreover, the fact that the hp19ARF protein is frequently absent in tumor cells that express its mRNA, suggests that the protein might play a still unknown function in tumorigenic processes despite the absence of any point mutation in the coding sequence. The presence of increased protein levels in certain tumor cells grown in culture (for example HeLa cells) does not contradict this notion as, in these cells other partners, (or eectors) of hp19ARF may be impaired or missing.

Materials and methods Cell and cell culture conditions All cells growing in suspension (HL60, Daudi, Meg01, Nalm-6, Raji, Reh) were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS), 5 mM glutamine and antibiotics (penicillin and streptomycine). HeLa cells, Cos cells and HS-27 cells were grown in DMEM medium supplemented with 10% FCS, 5 mM glutamine and antibiotics. HS-27 cell line of human normal ®broblast was provided by Dr Berthold Henglein (Institut Curie, Paris). The Reh cell line was derived from an acute lymphoblastic leukemia (Zhou et al., 1995). The PFF cell line (provided by Dr P Jansen-DuÈrr, DKFZ, Heidelberg) was established from primary foreskin fibroblasts of a 6 year old child. It is currently at early passage numbers. C-33-A cells (derived from a cervical tumor) and U-2 OS cells (derived from an osteosarcoma) came from the American Type Culture Collection. Metabolic cell labeling was carried out in methionine+cysteine-free medium (ICN, Biomedicals, Inc. Costa Mesa, California) plus 10% non dialyzed CFS and in the presence of 150 mCI/ml of 35S-methionine-cysteine for 4 h (New England Nuclear). Recombinant GST-hp19ARF protein and preparation of rabbit antiserum The EcoRI fragment from a bluescript SK-hp19ARF recombinant vector (clone 13, see Duro et al., 1995) was subcloned in frame with glutathione S-transferase (GST) into the EcoRI site of the pGEX 3X vector. Puri®cation of the GST fusion protein was performed as previously described (Coligan et al., 1995). Brie¯y, upon IPTG induction, bacterial cells were lysed in NTE buer (50 mM Tris-HCl pH 8.0, 100 m M NaCl, 50 mM EDTA, 0.5% Tween, 0.5% Triton, 1 mM PMSF, 0.1% Aprotinine, 2 mg/ ml Leupeptine, 2 mg/ml Pepstatine) and further disrupted by sonication. Then, fusion proteins were puri®ed on gluthatione-sepharose columns (Pharmacia LKB) and analysed by electrophoresis in denaturing 15% polyacrylamide gels containing sodium dodecylsulphate (SDS). The anti-hp19ARF polyclonal serum used throughout this study was prepared by immunizing a rabbit with 750 mg of puri®ed GST-hp19ARF recombinant protein. The whole immunization procedure was done according to the standard protocol of Eurogentec (Belgium). Immunoprecipitation and Western blot analysis In vitro protein translation 2 mg of supercoiled DNA template were submitted to in vitro transcription/translation assays using the TNTT7-coupled reticulocyte lysate system (Promega Corporation, Madison, Wisconsin, USA) according to the manufacturer's instructions. Immunoprecipitation of the labeled reaction products, gel electrophoresis were performed as previously described (Soulard et al., 1993). Cell extracts Cells were washed in saline phosphate buer, then harvested in SDN buer (100 mM Tris HCl pH 8, 150 mM NaCl, 5 mM EDTA, 0.5% SDS, 0.5% NP40, 0.5% sodium deoxycholate, 1 mM PMSF, 0.1% Aprotinine, 2 mg/ ml Leupeptine, 2 mg/ml Pepstatine) and sonicated. Upon incubation on ice for 30 min, the samples were centrifugated for 20 min at 14 000 r.p.m. and supernatants were used immediately or stored at 7808C. Protein amounts corresponding to 56106 cells were immunoprecipitated for 45 mn at 48C in SDN buer with protein A sepharose 6MB (Pharmacia) and nonimmune rabbit serum. 5 ml of antihp19ARF rabbit polyclonal antiserum or 3 ml of anti-p16INK4 antibody (sc-468, Santa Cruz Biotechnology, California)


were added to the supernatants recovered by centrifugation at 12 000 r.p.m. for 1 min. Incubation was continued overnight at 48C, and protein A sepharose was added for 45 min at 48C. The immune complexes were processed as previously reported (Soulard et al., 1993) and half of the reaction mixtures were run on 16.5% polyacrylamide gel using Tricine buer (SchaÈgger et al., 1987). The proteins were then transferred to nitrocellulose membrane which were probed for 4 h with 1/500 dilution of anti-hp19ARF or anti-p16INK4 rabbit sera, then incubated with 1/2000 dilution of anti-rabbit Ig coupled to horseradish peroxidase (Amersham) for 20 min at 48C. Finally, bound antibody was detected by chemiluminescence (ECL, Amersham) according to the manufacturer's instructions.

Indirect immuno¯uorescence microscopy Cells were ®xed with 3% paraformaldehyde in PBS pH 7.4 for 10 min at room temperature, washed in PBS, permeabilized by 0.2% Triton X-100 in PBS and washed three more times with PBS. The slides were incubated with 1/200 dilution of anti-hp19ARF polyclonal rabbit sera or 1/50 dilution of anti-HA mouse monoclonal antibody (Boehringer Mannheim). Finally, the slides were washed three times in PBS and incubated for 20 min at room temperature with

1/200 dilution of ¯uorescein-conjugated, donkey and sheep respectively, anti-rabbit or anti-mouse Ig (Amersham). Transfection procedures Because of the presence of a SacI site inside the hp19ARF insert, clone 13 DNA was digested at the unique EcoRI cloning site of the vector, and the resulting fragment was subcloned into bluescript KS plasmid. Upon ampli®cation, the hp19ARF insert of the recombinant subclone DNA was recovered by SacI digestion and integrated in frame with hemagglutinin (HA) into DEB vector (Boulukoz et al., 1989, a gift of Dr J Ghysdael, Institut Curie, Orsay, France). Hela and Cos cells were transfected by using the standard calcium phosphate technique (Sambrook et al., 1989). Hela cells were also transfected using the DMRIE-C reagent according to the manufacturer's instructions (Gibco-BRL). Acknowledgements This work was supported by grants from the `Association pour la Recherche sur le Cancer (ARC), Villejuif' and the `Fondation contre la LeuceÂmie'. We are grateful to P Jansen-DuÈrr for providing a sample of PFF cells. We also greatly thank Dr A Bloom for improving the English text.

References Alcorta DA, Xiong D, Phelps D, Hannon G, Beach D and Barrett JC. (1996). Proc. Natl. Acad. Sci. USA, 93, 13742 ± 13747. Boulukoz KE, Pognonec P, Rabault B, Begue A and Ghysdael J. (1989). Mol. Cell. Biol., 9, 995 ± 1008. Coleman A, Fountain JW, Nobori T, Olopade OI, Robertson G, Housman DE and Lugo TG. (1994). Cancer Res., 54, 344 ± 348. Coligan JE, Dunn BM, Ploegh HL, Speicher DW and Wing®eld PT. (1995). Current protocols in Protein Science, Vol 1, J. Wiley and Sons, Inc. Duro D, Bernard O, Della Valle V, Berger R and Larsen C-J. (1995). Oncogene, 11, 21 ± 29. Duro D, Bernard O, Della Valle V, Leblanc T, Berger R and Larsen C-J. (1996). Cancer Res., 56, 848 ± 854. Fitzgerald MG, Harkin DP, Siva-Arrieta S, MacDonald DJ, Isselbacher KJ, Sober AJ and Haber DA. (1996). Proc. Natl. Acad. Sci. USA, 93, 8541 ± 8545. Flores JF, Walker GJ, Glendening JM, Haluska FG, Castresana JS, Rubio M-P, Pastor®de GC, Boyer LA, Kao WH, Bulyk M, Barnhill RL, Hayward NK, Housman DE and Fountain JW. (1996). Cancer Res., 56, 5023 ± 5032. Hara E, Smith R, Parry D, Tahara H, Stone S and Peters G. (1996). Mol. Cell. Biol., 16, 859 ± 867. Hui AM, Sakamoto M, Kanai Y, Ino Y, Gotoh M, Yokota J and Hirohashi S. (1996). Hepatology, 24, 575 ± 579. Koh J, Enders GH, Dynlacht BD and Harlow E. (1995). Nature, 375, 506 ± 510. Larsen C-J. (1996). Oncogene, 12, 2041 ± 2044. Liggett WH Jr, Sewell DA, Rocco J, Ahrendt SA, Koch W and Sidransky D. (1996). Cancer Res., 56, 4119 ± 4123. Linardopoulos S, Street AJ, Quelle DE, Parry D, Peters G and Sherr CJ. (1995). Cancer Res., 55, 5168 ± 5172. Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta GF, Jessup JM and Pagano M. (1997). Nature Med., 3, 231 ± 234.

Mao L, Merlo A, Bedi G, Shapiro GI, Edwards CD, Rollins BJ and Sidransky D. (1995). Cancer Res., 55, 2995 ± 2997. Olopade OI, Bohlander SK, Pomykala H, Maltepe E, Van Melle E, Le Beau MM and Diaz MO. (1992). Genomics, 14, 437 ± 443. Parry D and Peters G. (1996). Mol. Cell. Biol., 16, 3844 ± 3852. Quelle DE, Zindy F, Ashmun RA and Sherr CJ. (1995). Cell, 83, 993 ± 1000. Quelle DE, Cheng M, Ashmun R and Sherr CJ. (1997). Proc. Natl. Acad. Sci. USA, 94, 669 ± 673. Rezniko CA, Yeager TR, Belair CD, Savelieva E, Puthenveettil JA and Stadler WM. (1996). Cancer Res., 56, 2886 ± 2890. Sambrook J, Fritsch EF and Maniatis T. (1989). In Molecular cloning. Second edition, Cold Spring Harbor Laboratory Press. SchaÈgger H and von Jagow G. (1987). Analyt. Biochem., 166, 368 ± 379. Sherr CJ. (1996). Science, 274, 1672 ± 1677. Soulard M, Della Valle V, Siomi MC, Pinol-Roma S, Codogno P, Bauvy C, Bellini M, Monod G, Dreyfuss G and Larsen C-J. (1993). Nucleic Acids Res., 21, 4210 ± 4218. Stone S, Jiang P, Dayananth P, Tavtigian SV, Katcher H, Parry D, Peters G and Kamb A. (1995). Cancer Res., 55, 2988 ± 2994. Swaord DS, Middleton SK, Palmisano WA, Nikula KJ, Tesfaigzi J, Baylin SB, Herman JG and Belinsky SA. (1997). Mol. Cell. Biol., 17, 1366 ± 1374. Wick ST, Dubay MM, Imanil I and Brizuela L. (1995). Oncogene, 11, 2013 ± 2019. Zhou M, Gu L, James CD, He J, Yeager AM, Smith SD and Findley HW. (1995). Leukemia, 9, 1159 ± 1161.

2481