Oncogene (1997) 14, 415 ± 423  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Induction of Cip/Kip and Ink4 cyclin dependent kinase inhibitors by interferon-a in hematopoietic cell lines

Olle Sangfelt, Sven Erickson, Stefan Einhorn and Dan Grande r

Department of Oncology/Pathology, Radiumhemmet, Building P8, Karolinska Hospital and Institute, S-171 76 Stockholm, Sweden

One prominent e€ect of IFNs is their cell growth exert cell growth inhibitory e€ects. It has since been inhibitory activity. The exact molecular mechanism found that this antiproliferative action of IFNs can be behind this inhibition of proliferation remains to be observed in many cell types from established cell lines, elucidated. Possible e€ectors for IFN-induced growth normal tissues and primary tumor cells (Taylor- inhibition are the recently discovered cyclin-dependent Papadimitriou and Rozengurt, 1985). It has been kinase inhibitors. The e€ect of IFN-a treatment on the postulated that this e€ect may, at least in part, members of the Ink4 and Cip/Kip families of Cdk mediate the antitumor e€ects observed in some inhibitors was investigated in three hematopoietic cell malignancies (Brenning et al., 1985). In vitro culture lines Daudi, U-266 and H9. Two of these cell lines, of susceptible cell types with IFN commonly leads to

Daudi and U-266, respond to IFN-a by G1 arrest, G1 arrest, but sometimes to a lengthening of S-phase or whereas the H9 cell line is not growth arrested by IFN-a. all phases of the (Roos et al., 1984). We show that a p53-independent upregulation of p21 With the discovery of the proteins that normally mRNA occurs following IFN-a treatment in all three cell regulate the cell cycle, some of the molecular e€ectors lines. In Daudi and U-266 cells, the mRNA induction is responsible for the IFN induced cell growth inhibition accompanied by an increase in p21 protein, followed by have been identi®ed. A number of proteins already an increased binding of p21 to Cdk2 and a subsequent identi®ed as key regulators of the cell cycle such as decrease in Cdk2 activity, temporally coinciding with G1 pRb, cyclin A, E2F, Cdk2 and c-myc are downstream arrest. In both these cell lines, there was also an targets in IFNs growth suppressive action (Kimchi, increased binding of p21 to Cdk4. In contrast, p21 1992; Melamed et al., 1993; Zhang and Kumar, 1994). protein was not expressed in H9 cells, despite high levels However, the exact chain of events responsible for the of p21 mRNA following IFN-a treatment. In U-266 antiproliferative e€ect of IFN is still unclear. cells, IFN-a increased not only p21 but also p15 mRNA During the last few years a host of knowledge has and protein levels, followed by an increased association been gained in the ®eld of cell cycle regulation. In all of p15 with Cdk4. Furthermore, IFN-a treatment caused organisms studied so far, passage through the cell a four- to sixfold induction of the p16 E1b transcript in cycle is dependent on the sequential formation and U-266 cells. Expression levels of the other Ink4 and Cip/ activation of complexes consisting of cyclin dependent Kip Cdk inhibitors were not induced by IFN treatment in kinases (Cdks) and their cyclin partners. In mamma- any of the cell lines. We conclude that IFN-a can act as lian cells the complexes active in G1 contain cyclin a potent regulator of Cdk-inhibitor expression, correlat- D-Cdk4/6 and cyclin E-Cdk2 (reviewed in GranÄ a and ing with decreased Cdk activity and cell growth Reddy, 1995). Many proteins have been suggested as inhibition. One mechanism for resistance to IFN may targets of di€erent cyclin/Cdk complexes (Nigg, 1993). be loss of the ability of cells to upregulate these proteins. One of the most studied substrates is the Rb protein, which upon phosphorylation enables the E2F tran- Keywords: interferon; cell cycle; cyclin-dependent scription factor to activate genes necessary for S phase kinase; cyclin-dependent kinase inhibitors progression (Nevins, 1994). The activity of Cdk complexes have also been shown to be regulated by phosphorylation and dephosphorylation on speci®c residues (Morgan, 1995). Recently a new level of Introduction regulation of these complexes has been elucidated by the indenti®cation of low molecular weight cyclin The interferons (IFNs) are a family of proteins which dependent kinase inhibitors (Sherr and Roberts, 1995). are produced by eukaryotic cells when challenged by In the mammalian system, two families of Cdk- viruses and other agents. Following binding to speci®c inhibitors have been identi®ed. The Ink4 family cell surface receptors, IFNs induce transcription of includes the p15, p16, p18 and p19 proteins, which IFN-stimulated genes through activation of the Jak/ all contain several ankyrin repeat like sequences. STAT signaling pathway (Shuai, 1994), whereby IFNs These inhibitors bind speci®cally to Cdk4 and 6, elicit a number of physiological responses in cells, with thereby interfering with cyclin D binding to the the common denominator of inhibiting viral replication kinases. The p16 gene is commonly mutated or (reviewed in De Mayer and De Mayer-Guiginard, deleted in human malignancies (Kamb et al., 1994; 1988). It was shown in the early 60s that IFNs also Nobori et al., 1994). Recently, an alternative ®rst exon (E1b) of the p16 gene has been described. When the E1b exon is spliced onto exon 2 of the p16 gene, the Correspondence: D Grande r original reading frame is shifted, and translation of Received 21 June 1996; revised 16 September 1996; accepted 17 this so called b transcript creates a protein completely September 1996 unrelated to p16. Interestingly this protein seems to Induction of Cdk-inhibitors by interferon O Sangfelt et al 416 have cell growth inhibitory properties, although it has [3H]thymidine incorporation. U-266 as well as Daudi not been shown to be a direct Cdk-inhibitor (Quelle et cells responded to IFN-a with an accumulation of cells

al., 1995). in the G1 phase after 16 ± 24 h of incubation with The Cip/Kip family include the sequence, p21, p27 IFN-a (Figure 1). In contrast, no change in cell cycle and p57 proteins, which share no homology with the distribution was observed in H9 cells after IFN-a Ink4 family (Sherr and Roberts, 1995). The latter treatment (Figure 1). Both U-266 and H9 cells were group of Cdk-inhibitors associate with and inhibit the found to undergo following longer times of cyclin E-Cdk2 and cyclin A-Cdk2 complexes, in exposure to IFN-a (Sangfelt et al., submitted). addition to inhibiting cyclin D-Cdk4 and 6 com- Since the Rb protein is implicated in the control of

plexes. The p21 gene is transcriptionally regulated by G1 progression and has previously been shown to be an the p53 , thereby mediating cell important target for IFN's cell growth inhibitory cycle arrest. However, p53 independent induction of action, we examined the status and e€ects of IFN-a p21 has recently been described (El-Deiry et al., 1993; treatment on this protein in the three cell lines. In Elbendary et al., 1994; Datto et al., 1995; Parker et al., cycling Daudi cells the major portion of the Rb protein 1995; Reynisdo ttir et al., 1995). In addition to appeared as slowly-migrating hyperphosphorylated associating with di€erent cyclin/Cdk complexes, the p21 protein (but not p27 or p57) can bind to proliferating cell nuclear antigen (PCNA), and inhibit DNA replication (Waga et al., 1994). However this activity maps to a di€erent region of the p21 protein than the one interacting with cyclin/Cdk complexes (Chen et al., 1995). Moreover, p21 has also been shown to disrupt the interaction between Cdk2 and the E2F- p107/p130 complexes (Zhu et al., 1995; Shianov et al., 1995). A few previous observations prompted our interest in examining whether Cdk-inhibitors may be targets for regulation by IFNs. Firstly, other extracellular signals for growth arrest such as TGF-b, contact inhibition and serum starvation have been shown to act, at least in part, through the modulation of Cdk- inhibitors. TGF-b induces p15 and p21 at the transcriptional level. It also increases the amount of p27 bound to Cdk2, and reduces Cdk4 levels (Elbendary et al., 1994; Hannon and Beach, 1994; Polyak et al., 1994; Datto et al., 1995., Reynisdo ttir et al., 1995). Other growth inhibitory cytokines may conceivably work in a similar fashion. Secondly, the previously described e€ects on cell cycle regulating proteins by IFN, such as reduced pRb phosphorylation and the inhibition of Cdk2 kinase activity are all compatible with an induction of Cdk-inhibitor activity. We have therefore analysed the e€ects of IFN-a on the expression of the Ink4- and Cip/Kip-families of Cdk- inhibitors in three hematopoietic cell lines.

Results Figure 1 Flow cytometry analysis of asynchronous U-266 (a), As IFN under physiological circumstances is likely to Daudi (b) and H9 cells (c) cultured in the absence or presence of act by arresting actively proliferating cells, we 5000 U/ml of IFN-a for 24 h investigated the e€ects of IFN-a treatment of proliferating cell cultures, rather than on cells released from quiescence. time (hours): 0 4 8 12 24 48 IFN-a e€ects on cell cycle progression in Daudi, U-266 and H9 cells The integrity of the IFN signal transduction pathway Rbphos was shown by demonstrating that all three cell lines were Rb sensitive to IFN-a, as measured by induction of the IFN stimulated gene, 2',5'oligoadenylate synthetase. A 11 ± 40-fold induction of enzyme activity was detected after Figure 2 pRb protein levels in Daudi cells at various timepoints incubation with 5000 U/ml of IFN-a for 24 h. after addition of 5000 U/ml of IFN-a. Positions of the slowly The e€ect of IFN-a on cell cycle distribution and migrating hyperphosphorylated forms (pRbphos) and the rapidly DNA synthesis was examined by ¯ow cytometry and migrating underphosphorylated (pRb) forms are indicated Induction of Cdk-inhibitors by interferon O Sangfelt et al 417 forms (Rbphos). After exposure to IFN-a, a progressive increase in p15 protein expression, which was detected conversion to the fast-migrating underphosphorylated as a double band in this cell line (data not shown). In forms occurred around 8 ± 12 h after addition of IFN-a contrast, no p15 mRNA or protein expression could be (Figure 2). No Rb protein was expressed in U-266 cells detected in cycling Daudi cells and IFN-a treatment (data not shown), possibly due to a genomic deletion did not induce the expression of p15 in these cells (data within the pRb gene previously described in this cell not shown). H9 cells were not examined for p15 or p16 line (Corradini et al., 1994). Thus, the observed G1 expression, as this cell line carries a homozygous arrest in U-266 cells after IFN-a treatment is not deletion of 9p21 sequences including these genes dependent on functional Rb protein. Likewise, there (Borgonovo Brandter et al., 1996). We further was no expression of Rb protein in H9 cells (data not examined the expression of p16 in U-266 and Daudi shown). cells. In none of these cell lines, could expression of the p16 E1a mRNA or p16 protein be detected (data not shown). Interestingly, there was an approximate IFN-a e€ects on the expression of Ink4 inhibitory ®vefold induction of the p16 E1b transcript in U-266 proteins cells, reaching maximum levels at 12 h after IFN-a The e€ects of IFN-a on the expression of the di€erent addition (Figure 3). In contrast, although weakly members of the Ink4 family of Cdk inhibitors were expressed at all time points, the E1b mRNA level examined in all three cell lines. The IFN-a arrested cell was not a€ected by IFN-a treatment in Daudi cells line U-266 expressed two p15 mRNA transcripts (2.2 (Figure 3). and 3.2 kb). IFN-a treatment of these cells resulted in The expression of the newly identi®ed p16 homologue a 4 ± 6-fold induction of both transcripts, reaching a p18 was also examined. All three cell lines expressed two maximum at 8 h after the addition of IFN-a (Figure 3). p18 mRNA transcripts (1.9 and 2.5 kb). In IFN-a The observed increase in p15 mRNA, was paralleled by treated Daudi cells, the ratio was changed between the two transcripts. A decrease in the larger transcript was paralleled by an increase in the expression of the smaller transcript at around 9 to 12 h of IFN-a treatment a (Figure 4). Repeated experiments also showed that this IFN treatment (hours) shift in transcript ratio was accompanied by a transient

0 4 8 12 24 decrease in p18 protein expression levels in Daudi cells ¨ (Figure 4). In U-266 and H9 cells, addition of IFN-a did not change the p18 mRNA or protein expression (data p15 not shown).

U–266 ¨ We further examined the expression and responsive- ness to IFN-a of the p19 Cdk-inhibitor. All cell lines expressed the 1.3 kb p19 transcript and IFN-a

treatment did not a€ect the expression level of this ¨ GAPDH Cdk-inhibitor (data not shown).

b IFN treatment (hours)

03691224 IFN treatment (hours)

¨ 03691224

p16 E1β ¨

U–266

p18

¨

¨ GAPDH

¨ GAPDH ¨ p16 E1β

Daudi IFN treatment (hours)

¨ GAPDH 0 4 8 12 24 30 ¨ p18 Figure 3 E€ects of IFN-a (5000 U/ml) on the steady state mRNA levels of the p15 transcript in U-266 cells and the p16 E1b transcript in U-266 and Daudi cells as determined by Northern Figure 4 E€ects of IFN-a (5000 U/ml) on the steady state blot analysis. Exponentially growing cells were treated with IFN-a mRNA and protein levels of p18 in Daudi cells as determined by for the indicated times. As control for RNA loading, blots were Northern and Western blot analysis. Exponentially growing cells probed with a glyceraldehyde-3-phosphate dehydrogenase were treated with IFN-a for the indicated times. As control for (GAPDH) cDNA probe RNA loading, blots were probed with a GAPDH cDNA probe Induction of Cdk-inhibitors by interferon O Sangfelt et al 418 IFN-a e€ects on the expression of Cip/Kip inhibitory absence of IFN-a, but was highly induced (25-fold) in response to IFN-a treatment (Figure 5). In all three cell proteins lines, p21 induction was obvious already 1 ± 2 h after Expression levels of the Cip/Kip inhibitory proteins addition of IFN-a, with highest expression occuring were examined in all three cell lines. A rapid induction around 12 h of IFN-a treatment (Figure 5). In U-266

of p21 mRNA expression was observed subsequent to and Daudi cells, which are both G1 arrested by IFN-a, IFN-a treatment in all three cell lines (Figure 5). the rise in p21 mRNA levels was paralleled by an Compared to U-266 and H9 cells, Daudi cells showed increase in p21 protein (Figure 6). Interestingly, despite a weaker induction of p21 mRNA (three times), but the observed induction of p21 mRNA in H9 cells, seem to express the highest baseline levels of p21 which are not growth arrested by IFN-a, there was a mRNA (Figure 5). complete lack of p21 protein expression in these cells, U-266 cells expressed moderate levels of p21 mRNA, even in the presence of IFN-a (Figure 6). IFN-a treatment of these cells increased p21 mRNA Since the p21 gene has shown to be positively expression levels more than eight times (Figure 5). In transcriptionally regulated by the tumor suppressor H9 cells the p21 message was barely detectable in the protein p53, we examined whether the observed IFN-a associated p21 induction was dependent on wild type p53. Immunostaining analysis was performed for wild type and mutant p53 expression. Daudi, U-266 as well as H9 cells stained strongly with the monoclonal a IFN treatment (hours) antibody pAb 240 that recognizes mutant p53 protein only, whereas all three cell lines were negative with the 03691224 antibody pAb 1620 that recognizes wild-type p53

¨ p21 protein alone (data not shown). Thus, all three cell lines express mutant p53 protein, consistent with

U–266 previously reported results in these cell lines (Cheng ¨ GAPDH

a IFN treatment (hr) ¨ 0 4 8 12 24 30 p21

Daudi U–266 ¨ p21

¨ GAPDH

Daudi ¨ p21 ¨ IFN treatment (hr) p21 0 4 8 12 24 C

H9

H9 ¨ p21 ¨ GAPDH

b b

Figure 6 E€ects of IFN-a (5000 U/ml) on the steady state Figure 5 E€ects of IFN-a (5000 U/ml) on the steady state protein levels of p21 as determined by Western blot analysis. mRNA levels of p21 as determined by Northern blot analysis. Exponentially growing cells were treated with IFN-a (5000 U/ml) Exponentially growing cells were treated with IFN-a for the for the indicated times (a). Quantitation of p21 protein levels in indicated times (a). Quantitation of p21 mRNA levels in U-266, U-266 and Daudi cells at the indicated timepoints (b). Values Daudi and H9 cells at the indicated timepoints (b). The values represent the relative induction of p21 as normalized to p21 have been normalized for GAPDH mRNA expression and are protein expression levels in untreated cells (timepoint 0). As a reported as the relative induction of normalized p21 mRNA levels control for equal loading and transfer, membranes were stained over untreated cells (timepoint 0) with Ponceau red (Sigma) Induction of Cdk-inhibitors by interferon O Sangfelt et al 419 and Haas, 1990; Mazars et al., 1990; Wiman et al., cells, was followed by an increase in the amount of 1991), suggesting that the p21 induction was not p53 p15 bound to Cdk4 (Figure 8). Also after dependent. immunoprecipitation, p15 protein was detected as a We further examined the expression and responsive- double band (15 kD and 17 kD). Since U-226 cells ness to IFN-a of the p27 and p57 Cdk-inhibitors. As do not express the p16 protein (examined using two measured by Western blotting, the p27 protein was di€erent p16 antiobodies), the upper band is not due expressed in U-266 and Daudi cells. However, IFN-a to crosshybridization to p16. Neither was it due to treatment did not alter the expression of p27 in these crosshybridization with p18 protein, as p18 is cells (data not shown), In constrast, no p27 expression detected at a di€erent position (data not shown). could be detected in H9 cells. p57 mRNA expression could not be detected in any of the three cell lines (data not shown). Discussion

Extracellular signals are important in the regulation of Changes in binding of Cdk-inhibitors to Cdk2 and Cdk4 cellular proliferation. A large number of growth factors and alterations in Cdk2 activity following IFN-a have been found to date. In contrast only a small treatment number of cytokines which induce cell growth The cell lines were examined for the expression of inhibition have been described. Among the latter are mRNA and protein for Cdk2, 4 and 6. Cdk2 and the TGF-b and IFN families of proteins (Taylor- Cdk4 mRNA and protein was expressed at similar Papadimitriou and Rozengurt, 1985; Massague , 1990). levels in all three cell lines (data not shown). Only The recent elucidation of how the cell cycle is U-266 cells expressed signi®cant levels of Cdk6 regulated, including the realization of the importance mRNA (data not shown). IFN-a treatment had no of cyclin-Cdk complexes and their newly discovered e€ect on mRNA or proteins levels of any of these inhibitors, has made it possible to study the e€ects of Cdks (data not shown). In both U-266 and Daudi growth inhibitory cytokines on these proteins. TGF-b cells, the increase in p21 protein expression was has been found to upregulate the activity of at least followed by an increase in the amount of p21 bound three Cdk-inhibitors, p15, p21 and p27, in addition to to Cdk2, as analysed by immunoprecipitation of decreasing the expression of Cdk4 (Elbendary et al., Cdk2 complexes followed by Western blotting with 1994; Hannon and Beach, 1994; Polyak et al., 1994; p21 antibodies (Figure 7). Furthermore, immunopre- Datto et al., 1995; Reynisdo ttir et al., 1995). The cipitated Cdk2 complexes were analysed for kinase results from these studies also indicate that the e€ect of activity using histone H1 as substrate. There was TGF-b on cell cycle regulating proteins can be cell type 60 ± 70% decrease in Cdk2 phosphorylating activity speci®c (Polyak et al., 1994). Other extracellular in IFN-a treated U-266 and Daudi cells after 16 ± growth inhibitory signals, such as contact inhibition 24 h of treatment (Figure 7). Similarly, we also and cAMP induced cell cycle arrest, have also been observed an increased binding of p21 to Cdk4 shown to be associated with an increased expression of following IFN-a treatment in both U-266 and Daudi Cdk-inhibitors (Kato et al., 1994; Polyak et al., 1994). cells (Figure 8). Additionally, the observed IFN-a Less is known about the mechanism behind the induced increase in p15 expression levels in U-266 antiproliferative action of IFN. Such knowledge seems

a IFN treatment (hours)

0824 c ¨ Cdk2

U–266 ¨ p21

b IFN treatment (hours)

0824 ¨ Cdk2

Daudi ¨ p21

Figure 7 E€ects of IFN-a (5000 U/ml) on association of p21 with Cdk2 in U-266 (a) and Daudi cells (b). Coupled immunoprecipitation with Cdk2 antibodies and immunoblotting with p21 antibodies was performed on cell lysates from cultures treated with IFN-a for the indicated times. The same cellular extracts that are described for a and b were analysed for Cdk2 associated histon H1 kinase activity (c). The values are reported as the relative percentage of Cdk2 kinase activity as compared to untreated cells (timepoint 0) Induction of Cdk-inhibitors by interferon O Sangfelt et al 420 to inhibition of translation or a truncated protein that a IFN treatment (hours) is not detected by the monoclonal antibody used in

0 8 16 24 these studies. The lack of p21 protein expression may ¨ theoretically be the reason for the lack of cell growth Cdk4 arrest by IFN-a in cells as discussed below. The fact that the induction of p21 occurs within 1 ±

2 h after addition of IFN-a, preceding G1 arrest,

indicates that the e€ect on p21 expression is not

¨ secondary to IFN-a induced cell growth inhibition. ¨ p15 U–266 This conclusion is also supported by the ®nding of p21 mRNA induction in H9 cells, which are not growth arrested by IFN-a. Taken together, the results suggest that the rapid increase in p21 after IFN-a treatment

may play a causative role in IFN-a mediated growth ¨ p21 arrest. It also supports a model in which a single Cdk inhibitor can serve as an e€ector of diverse signals for cell cycle arrest, including signals from within the cell such as DNA damage, as well as extracellular b IFN treatment (hours) signalling mediating by TGF-b and IFN-a. The data obtained with these growth inhibitory cytokines are 0 8 16 24

somewhat paradoxical in comparison with the data ¨ Cdk4 indicating induction of p21 as an immediate-early gene following mitogenic stimulation. This may be explained Daudi by the fact that p21 at low concentrations enhances Cdk complex kinase activity whereas high levels lead to an inhibition of kinase activity (Zhang et al., 1994;

¨ p21 Macleod et al., 1995). p21 transcriptional activation has previously been Figure 8 E€ects of IFN-a (5000 U/ml) on association of p15 and reported in response to p53 induction as well as by p53 p21 with Cdk4 in U-266 (a) and Daudi cells (b). Coupled independent pathways (El-Deiry et al., 1993; Elbendary immunoprecipitation with Cdk4 antibodies and immunoblotting with p21 and p15 antibodies was performed on cell lysates from et al., 1994; Datto et al., 1995 Macleod et al., 1995; cultures treated with IFN-a for the indicated times Parker et al., 1995). The potential role of p53 in IFN-a mediated induction of p21 was therefore determined by examination of the p53 status in these cell lines. All three cell lines expressed p53 protein that was detected by especially warranted, since IFNs are widely used in the antibodies recognizing mutant p53, in accord with treatment of malignant disease, with the antiprolifera- previous studies in these cell lines (Cheng and Haas, tive e€ect as one possible e€ector mechanism. IFNs 1990; Mazars et al., 1990; Wiman et al., 1991). This have been shown, in analogy with TGF-b, to prevent suggests that cell growth inhibition by IFN-a is not the phosphorylation of pRb, decrease E2F DNA dependent on the presence of wild type p53 and that binding activity, reduce c-myc and cyclin A levels, as IFN-a, similar to TGF-b, can induce p21 in a p53 well as to decrease the activity of Cdk2 complexes independent manner. We are currently investigating (Kimchi, 1992; Melamed et al., 1993, Zhang and whether the p21 promotor is directly transcriptionally Kumar, 1994). One mechanism behind alterations in regulated by IFN-a. The published p21 promotor the above mentioned proteins could be an e€ect of sequence does not display any obvious interferon- IFN on Cdk-inhibitory proteins. stimulated-response-elements (ISREs). IFN-a thus acts In the present study we therefore examined the either indirectly, through the induction of some other e€ects of IFN-a on the expression of Cdk-inhibitors, as protein(s), alternatively IFN-a induced transcription can well as Cdk2, 4 and 6, in three hematopoietic cell lines. be mediated by elements other than the classical ISRE. Two of the cell lines, Daudi and U-266, respond to Interestingly, it has recently been shown that the p21

IFN-a by a G1 growth arrest, whereas the H9 cell line gene is transcriptionally activated by IFN-g through a is not growth arrested by IFN-a. Our results STAT1-dependent pathway (Chin et al., 1996). demonstrate that IFN-a causes a rapid induction of Additionally, we observed an induction of p15 mRNA for the p21 Cdk inhibitor in all three cell lines. mRNA and protein levels rising within a few hours

In the G1 arrested cell lines, Daudi and U-266, this is of IFN-a treatment in U-266 cells, whereas no such followed by an increase in p21 protein levels. Analysis induction was found in Daudi cells. The observed of Cdk4 and Cdk2 complexes showed a strong increase increase in p15 expression in U-266 cells was followed in the amount of p21 bound, and a subsequent by an increased binding to Cdk4. Whether this leads to decrease in kinase activity of Cdk2 in IFN-a treated an inhibition of Cdk4/6 activity is currently investi-

cells, coinciding with G1 arrest in these cell lines. gated, as well as whether increased binding of p15 Interestingly, in H9 cells, which are not growth displaces p27 associated with Cdk4/6 making it arrested by IFN-a, there is a complete lack of p21 available for Cdk2 binding. Theoretically this could protein expression in spite of a 25-fold increase in p21 lead to additional inhibition of Cdk2 activity, as has mRNA levels. The reason for the lack of p21 protein been suggested in TGF-b induced cell cycle arrest expression is unknown. It is possible that this cell line (Polyak et al., 1994; Reynisdo ttir et al., 1995). The lack carries an intragenic mutation in the p21 gene, leading of p15 expession and IFN-inducability in Daudi cells Induction of Cdk-inhibitors by interferon O Sangfelt et al 421 could be due to methylation of the p15 gene, as has however, not exclude the possible importance of other been found in this and other Burkitt lymphoma cell factors, such as regulation of cyclins, Cdc25 or CAK lines (Herman et al., 1996; and U Klangby Karolinska activity, not investigated in the present study. Another Institute, personal communication). possibility could be that IFN-a alters the amount of In U-266 cells we observed a rapid ®vefold induction p27 bound to Cdk2, as has been described for TGF-b. of the newly discovered p16 E1b transcript. It has These issues are currently under investigation. Resis- previously been disputed whether this transcript tance to IFN induced cell growth inhibition may be translates into a functional protein, as the original due to an inability of cells to induce Cdk-inhibitors, as reading frame is shifted when the E1b exon is spliced in the case of H9. Whether similar defects exist in other onto exon 2 of the p16 gene. Recently a translation cell lines not undergoing cell cycle arrest subsequent to product has been found in rodent cells and, interest- IFN treatment, remains to be studied. Since the cell ingly, this protein has cell growth inhibitory activity growth inhibitory e€ect of IFN may play an important (Quelle et al., 1995). The ®nding that the expression of role in IFN's clinical antitumor action, a better this gene can be regulated by IFN-a adds support to understanding of the molecular mechanisms behind the hypothesis that the resulting protein has a function this activity as well as causes for resistance, may be of in cellular physiology. help in improving tumor therapy with IFN. In neither Daudi nor U-266 cells did we observe any expression or induction of p16 E1a transcript. The undetectable p16 expression in U-266 cells is Materials and methods somewhat intriguing since these cells carry a pRb gene mutation (Corradini et al., 1994), which in other cell Cell lines, culture conditions and interferon preparation lines has been shown to correlate with high levels of Three malignant lymphoid cell lines were used in this p16 protein (Parry et al., 1995). pRb has been study; the H9 cell line, derived from the continuous T-cell suggested to be an important downstream e€ector of lymphoma cell line Hut-78 (Mann, 1989), the myeloma cell IFN induced cell growth inhibition (Kimchi, 1992). line U-266, kindly provided by Dr Kenneth Nilsson,

The fact that U-266 cells are blocked in G1 indicates Uppsala Sweden, and the Burkitt lymphoma (BL) cell that the presence of wild type pRb is not mandatory line Daudi (ATCC, Rockville, MD). All cell lines were grown in RPMI (GIBCO, Berlin, to obtain G1 arrest by IFN. It can of course not be excluded that other members of the pocket protein Germany) supplemented with 10% heat-inactivated fetal calf family may be active in this cell line, and that the serum (Seromed; Biochrom KG, Berlin, Germany), 2 mM decrease in Cdk2 activity may presumably lead to a glutamine, 50 mg/ml of streptomycin and 50 mg/ml of penicillin and were kept at 378C in a humid incubator with decreased phosphorylation of these or other unknown 5% CO2. The cell lines were cultured in the presence and target proteins. Since it has been suggested that p16 absence of IFN-a for up to 48 h. RNA and protein was and p18 are dependent on functional pRb protein for prepared from cells harvested at di€erent time points after their action (Kun-Liang et al., 1994; Lukas et al., addition of IFN-a.

1995; Parry et al., 1995), the data indicate that IFN Recombinant human IFN-a2b (from Schering-Plough, induced G1 arrest can occur independently of these Kenilworth, NJ) was used for the in vitro experiments. This Cdk inhibitors. IFN preparation had a speci®c activity of 2.06108 units/mg In addition to the above ®ndings we also observed of protein and the purity was 499%. an IFN-a induced shift in p18 transcripts, accompanied by a slight transient decrease in p18 protein levels in DNA labeling and ¯ow cytometric analysis Daudi cells. The signi®cance of this ®nding is unclear, For DNA histograms, cells were ®xed in 4% formaldehyde but one possibility may be that this occurs secondary in phosphate bu€er during 18 h followed by 95% ethanol to pRb dephosphorylation, which has been shown to treated with subtilisin Carlsberg solution (0.1% Sigma in¯uence the expression of p16 (Li et al., 1994). protease XXIV, 0.1 M Tris-HCl, 0.07 M NaCl, pH 7.2) and The reason for resistance to IFN-a induced growth stained with DAPI-SR101 solution as previously described arrest in the H9 cell line has yet to be determined. It (Castro et al., 1993). The samples were analysed using a might depend on several factors, such as deletion of the PAS II ¯ow cytometer (Partec, MuÈ nster, BRD), equipped p15 and p16 genes, lack of pRb expression as well as the with a 100 W mercury arc lamp HBO 100. DAPI inability to produce p21 protein. The fact that Daudi ¯uorescence was measured above 435 nm. Flow cytometric cells arrest in the absence of p15/p16 expression/ data were analysed using Multicycle program (Phoenix Flow Systems, San Diego, CA). induction and that U-266 cells are blocked in the absence of functional pRb argues for the lack of p21 protein as the most important candidate for resistance to Western blot and immunocytochemistry analysis IFN-a induced cell growth inhibition in the H9 cell line. Western blot analysis was performed essentially as In summary, IFN-a regulates the levels of Ink4 and described (Finke et al., 1987). Whole cell extracts were Cip/Kip Cdk-inhibitors in analogy with TGF-b and prepared by lysis in equal volumes of Nonidet P-40 (NP- other extracellular signals for growth arrest. Interest- 40) bu€er (100 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1.0% ingly, we also observed an induction of the newly NP-40%, 15 mM aprotinin, 50 mM leupeptin, 1 mM phenyl- described p16 E1b transcript in one of the cell lines. As methyl-sulfonyl ¯uoride (PMSF), 1 mM sodium vanadate) in the case of TGF-b, the response to IFN-a with and 46Laemmli bu€er (125 mM Tris-HCl, pH 6.8, 4% sodiumdodecyl-sulfate (SDS), 20% glycerol, 200 mM regard to Cdk-inhibitors seems to be cell line dithiothreitol (DTT)). Protein concentration in cell dependent. A common denominator in the two G1 extracts was quantitated spectrophotometrically with a arrested cell lines was the induction of p21. This may protein assay kit, according to the instructions of the imply that the increase in p21 levels is causative for the manufacturer (Bio-Rad, Hercules, CA). Proteins were cell growth inhibitory activity of IFN-a. This does, resolved on a 7% (for Rb protein) or a 13% (for all Induction of Cdk-inhibitors by interferon O Sangfelt et al 422 other proteins) SDS-polyacrylamide gel, electroblotted to bu€er and boiling for 5 min. An aliquot of the reaction was membranes (Hybond C-extra, Amersham Life Sciences) by loaded on a 13% SDS ± PAGE and separated. The gel was semidry transfer. As a control for equal loading and dried and exposed to X-ray ®lm (Amersham Life Sciences). transfer, membranes were stained with Ponceau red (Sigma). For protein detection, immunoblots were probed Northern blot analysis with the following monoclonal antibodies; anti-Rb (1:1000), anti-p16 (1:500), anti-p21 (1:500), anti-p27 Total cellular RNA isolation and Northern blot analysis (1:1000) and the following polyclonal antibodies; anti-p15 were carried out as previously described using 20 mgof (1:50), anti-p16 (1:1000), anti-p18 (1:3000), anti-Cdk2 RNA per lane (Heyman et al., 1994). Northern blot ®lters (1:500) and anti-Cdk4 (1:1000). Antibody-antigen interac- (Hybond C-extra, Amersham Life Sciences) were hybri- tions were detected by incubation with horseradish- dized with the following cDNA probes, p21 p16, p18, p19, peroxidase-conjugated anti-mouse and anti-rabbit IgG Cdk2, 4 and 6.Ap16E1b speci®c probe was generated by antibodies followed by enhanced chemiluminescence PCR ampli®cation and restriction cleavage of the PCR (Amersham, Aylesbury, UK). All antibodies used for product. The following primer pair was used: 5'- Western blot analysis were obtained from PharMingen TCCCAGTCTGCAGTTAAGG-3' and 5'-GTCTAAGT- (San Diego, CA), except for p18 antibodies (kindly CGTTGTAACCCG-3'. The resulting product was then provided by Dr Y Xiong, University of North Carolina, cleaved with SacII. Chapel Hill, NC) and p15 antibodies (Santa Cruz Poly(A)+ RNA was isolated using Dynabeads mRNA Biotechnology, CA). p53 expression in individual cells DIRECT kit (Dynal, Oslo, Norway) according to the was analysed in cytospinic preparations, using two di€erent protocol provided by the supplier. Poly(A)+RNA ®lters monoclonal anti-p53 antibodies, pAb 240 and 1620 were hybridized with cDNA probes for p15 and p57. The (Oncogene Science, Uniondale, NY), as previously de- amount of RNA in each sample was normalized by scribed (Sangfelt et al., in press). The pAb 240 reacts only hybridizing the ®lters with a glyceraldehyde-3-phosphate with mutant human p53. The pAb 1620 reacts with wild dehydrogenase (GAPDH) RNA probe. All probes were type human p53 (although some mutant forms may react). labeled by an oligonucleotide labeling kit, using the Cells from a follicular lymphoma patient were used as a protocols provided by the supplier (Pharmacia, Uppsala, negative control for p53 expression. Cells from a BL cell line, Sweden). BL41, transfected with a temperature sensitive p53 construct (Ramqvist et al., 1993), were used as controls for mutant and Quantitation of band intensity wild type p53 expression. BL41 cells were also used as negative and positive controls for p21 expression in Western Hybridization signals were quantitated by scanning analysis. Cells from the H9 cell line and the SaOS-2 densitometry using an Ultroscan XL (Pharmacia LKB osteosarcoma cell lines, were used as negative and positive Biotechnology, Uppsala, Sweden). controls for p16 expression. [3H]thymidine incorporation Immunoprecipitation and histone H1 kinase assay For measurement of DNA synthesis 0.2 ml triplicate For immunoprecipitation cells were harvested at di€erent cultures were set up in roundbottomed 96-well microtiter timepoints, washed once with PBS and cell pellets were plates. One hour before harvesting, 1.0 mCi of lysedinLSLDbu€er(50mM HEPES at pH 7.4, 50 mM [3H]thymidine with the speci®c activity of 5 Ci/mmol NaC1, 10% Glycerol, 0.1% Tween 20, 0.3 mM Na- (Amersham, England) was added to each well. Harvesting orthovanadate, 50 mM NaF, 80 mM b-glycerophosphate, was performed with a Skatron harvesting machine 20 mM Na-pyrosphophate, 1 mM DTT and 1 mM PMSF, (Lierbyen, Norway) and radioactivity was measured with 10 mg/ml of leupeptin, 10 mg/ml of antipain, 100 mg/ml of a scintillation counter (Packard Instruments, Downers benzamidine hydrochloride, 5 mg/ml of aprotinin and Grove, Ill). 100 mg/ml of soybean trypsin inhibitor). Cell lysates were precleared by incubation with protein A-sepharose over 2',5'oligoadenylate synthetase (2',5'-A) assay nightat48C, with gentle agitation. Lysates were incubated with anti-Cdk2 or Cdk4 antibodies (kindly provided by Dr The cytoplasmic levels of 2',5'-A were determined in NP- D Beach, Cold Spring Harbor; USA) for 1 ± 2 h at 48C. 40 lysed cells following incubation with poly(rI) poly(rC) Immunocomplexes were then bound to protein A-Sephar- agarose beads as previously described (Grande r et al., ose and washed several times with LSLD bu€er. 1990). Immunoprecipitated proteins were resolved by separation on SDS ± PAGE, electroblotted to ®lters and hybridized with di€erent antibodies as described above. Acknowledgements For kinase assay, immunoprecipitation was performed as The excellent technical assistance of Ms Ann-Charlotte described above. Immunocomplexes were subsequently BjoÈ rklund and Elisabet Anderbring is gratefully acknowl- washed three times in kinase bu€er (50 mM Tris-HCl at pH edged. Drs Olga Aprelikova, David Beach, Ulf Klangby,

7.5, 10 mM MgCl2 and 1 mM DTT). Washed complexes were Katja Pokrovskaja, Charles Sherr, Klas Wiman and Yue resuspended in kinase bu€er and the activity was assayed by Xiong are thanked for kindly providing probes and addition of 5 mg histone H1 (Sigma), 1 mM ATP and 5 mCi antibodies. Dr Yisong Wang, Dr Ismail Okan and Dr Ulf [g-32P]ATP (5000 Ci/mmole) (Amersham Life Sciences) to a Klangby are thanked for helpful discussions. This study ®nal volume of 50 ml and incubated for 30 min at 308C. The was supported by grants from the Swedish Cancer Society reaction was stopped by addition of one vol 26SDS sample and the Cancer Society of Stockholm.

References

Borgonovo Brandter L, Heyman M, Rasool O, Liu Y, Brenning G, AÊ hre A and Nilsson K. (1985). Scand. J. Grande r D and Einhorn S. (1996). Eur. J. Haematol., 56, Haematol., 35, 543 ± 549. 313 ± 318. Induction of Cdk-inhibitors by interferon O Sangfelt et al 423 Castro J, Heiden T, Wang N and Tribukait B. (1993). Mann D. (1989). Aids Res. Human Retrovir., 5, 253 ± 255. Cytometry, 14, 793 ± 804. Massague J. (1990). Ann. Rev. Cell. Biol., 6, 597 ± 641. Chen J, Jackson PK, Kirschner MW and Dutta A. (1995). Mazars G-R, Portier M, Zhang X-G, Jourdan M, Bataille R, Nature, 374, 386 ± 388. Theillet C and Klein B. (1990). Oncogene, 7, 1015 ± 1018. Cheng J and Haas M. (1990). Mol. Cell. Biol., 10, 5502 ± Melamed D, Tiefenbrun N, Yarden A and Kimchi A. (1993). 5509. Mol. Cell. Biol., 13, 5255 ± 5265. Chin YE, Kitagawa M, Su W-CS, You Z-H, Iwamoto Y and Morgan DO. (1995). Nature, 374, 131 ± 134. Fu X-Y. (1996). Science, 272, 719 ± 722. Nevins R. (1994). Curr. Opinion Gen. Devel., 4, 130 ± 134. Corradini P, Inghiarami G, Astol® M, Ladetto M, Voena C, Nigg EA. (1993). Curr. Opinion Cell. Biol., 5, 187 ± 193. Ballerini P, Gu W, Nilsson K, Knowles DM, Boccadoro Nobori T, Miura K, Wu DJ, Lois A, Takabyashi K and M, Pileri A and Della-Favera R. (1994). Leukemia, 8, Carson DA. (1994). Nature, 368, 753 ± 756. 758 ± 767. Parker SB, Eichele G, Zhang P, Rawls A, Sands AT, Bradley Datto MB, Li Y, Panus JF, Howe DJ, Xiong Y and Wang X- A, Olson EN, Harper JW and Elledge SJ. (1995). Science, F. (1995). Proc. Natl. Acad. Sci. USA, 92, 5545 ± 5549. 267, 1024 ± 1027. De Mayer E and De Mayer-Guiginard J. (1988). Interferons Parry D, Bates S, Mann DJ and Peters G. (1995). EMBO J., and other regulatory cytokines John Wiley: New York. 14, 503 ± 511. El-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons PolyakKJ,KatoJ-Y,SolomonMJ,SherrCJ,Massague J, R, Trent JM, Lin D, Mercer E, Kinzler KW and Roberts JM and Ko€ A. (1994). Genes & Dev., 8, 9±22. Vogelstein B. (1993). Cell, 75, 817 ± 825. Quelle DE, Zindy F, Ashmun RA and Sherr CJ. (1995). Cell, Elbendary A, Berchuck A, Davis P, Havrielsky L, Bast RC 83, 993 ± 1000. Jr, Iglehart JD and Marks JR. (1994). Cell. Growth. Ramqvist T, Magnusson KP, Wang Y, Szekely L, Klein G Di€er., 5, 1301 ± 1307. and Wiman KG. (1993). Oncogene, 8, 1495 ± 1500. Finke J, Rowe M, Kallin B, Ernberg I, Rose n A, Dillner J Reynisdo ttir I, Polyak K, Iavarone A and Massague J. and Klein G. (1987). J. Virol., 61, 3870 ± 3878. (1995). Genes & Dev., 9, 1831 ± 1845. GranÄ a X and Reddy EP. (1995). Oncogene, 11, 211 ± 219. Roos G, Leandersson T and Lundgren E. (1984). Cancer Grande rD,OÈberg K, Lundqvist M-L, Eriksson B, Tiensuu Res., 44, 5358 ± 5362. Janson E and Einhorn S. (1990). Lancet, 336, 337 ± 339. Sangfelt O, Einhorn S, BjoÈ rklund A-C, Wiman KG, Okan I Hannon GJ and Beach D. (1994). Nature, 371, 257 ± 261. and Grande r D. (1996). Int. J. Cancer, 67, 106 ± 102. Herman JG, Jen J, Merlo A and Baylin SB. (1996). Cancer Shiyanov P, Bagchi S, Adami G, Kokontis J, Hay N, Arroyo Res., 56, 722 ± 727. M, Morozov A and Raychaudhuri P. (1996). Mol. Cell Heyman M, Grande rD,BroÈndum-Nielsen K, Cederblad B, Biol., 16, 734 ± 744. Liu Y, Xu B and Einhorn S. (1994). Leukemia, 8, 425 ± Sherr CJ and Roberts JM. (1995). Genes & Dev., 9, 1149 ± 434. 1163. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman Shuai K. (1994). Curr. Opinion in Cell Biol., 6, 253 ± 259. K, Tavtigian SV, Stockert E, Day III RS, Johnson BE and Taylor-Papadimitriou J and Rozengurt E. (1985). Interfer- Skolnick MH. (1994). Science, 264, 436 ± 440. ons, their impact in biology and medicine, J Taylor- Kato J Matsuoka M, Polyak K, Massague J and Sherr CJ. Papadimitriou J (ed). Oxford Medical Publications: (1994). Cell, 79, 487 ± 496. Oxford p. 81. Kimchi A. (1992). J. Cell. Biochem., 50, 1±9. Waga S, Hannon GJ, Beach D and Stillman B. (1994). Kun-LiangG,JenkinsCW,LiY,NicholsMA,WuX, Nature, 369, 574 ± 578. O'Keefe CL, Matera AG and Xiong Y. (1994). Genes & Wiman KG, Magnusson KP, Ramqvist T and Klein G. Dev., 8, 2939 ± 2952. (1991). Oncogene, 6, 1633 ± 1639. Li Y, Nichols M, Shay J and Xiong Y. (1994). Cancer Res., Zhang H, Hannon GJ and Beach D. (1994). Genes Dev., 8, 54, 6078 ± 6082. 1750 ± 1758. Lukas J, Parry D, Aagard L, Mann DJ, Bartkova J, Strauss Zhang K and Kumar R. (1994). Biochem. Biophys. Res. M, Peters G and Bartek J. (1995). Nature, 375, 503 ± 506. Comm., 200, 522 ± 528. Macleod KF, Sherry N, Hannon G, Beach D, Tokino T, Zhu L, Harlow E and Dynlacht BH. (1995). Genes Dev., 9, Kinzler K, Vogelstein B and Jacks T. (1995). Genes & Dev., 1740 ± 1752. 9, 935 ± 944.