proposed to act as a rheostat, regulating the ..... immediately in liquid nitrogen and stored at 7708C until ... removal of as much liquid as possible, pellets were.
Oncogene (1997) 14, 2111 ± 2122 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
The role of p27Kip1 in gamma interferon-mediated growth arrest of mammary epithelial cells and related defects in mammary carcinoma cells Beth L Harvat1, Prem Seth2 and AM Jetten1 1
Cell Biology Section, Laboratory of Pulmonary Pathobiology, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, North Carolina 27709; 2Medical Breast Cancer Section, Medicine Branch, National Cancer Institute, NIH, Bethesda, Maryland 20892, USA
Interferon gamma (IFNg) induces growth arrest in normal human mammary epithelial cells by establishing a block during mid-G1 corresponding to the time when the retinoblastoma protein (Rb) would normally be inactivated by hyperphosphorylation. IFNg inhibits the kinase activities of cdk2, cdk4 and cdk6 within 24 h of treatment. Protein levels of the cdks and G1 cyclins do not change within this time period, although cdk4 levels are signi®cantly reduced by 48 h. IFNg treatment induces p27Kip1 protein levels, presumably by a posttranscriptional mechanism as no change was observed in the mRNA levels. In addition, IFNg-induced inhibition of cdk2 and cyclin E-associated kinase activities is accompanied by a 4.5-fold or greater increase of p27Kip1 in cdk2 complexes. p27 may also have a role in the inhibition of cdk4/6 kinase activities, as p27 protein associated with these complexes was increase by 55 ± 70% after IFNg. In mammary carcinoma cell lines which are resistant to growth inhibition by IFNg, p27 levels are not induced by IFNg nor is cdk2 kinase activity inhibited, despite high baseline levels of p27 in cdk2 complexes. However, exogenous expression of p27 in these cells induces growth arrest. In addition, puri®ed p27 protein added to cdk2 complexes immunoprecipitated from carcinoma cells is able to inhibit the kinase activity in a dose dependent manner. Our results suggest that p27Kip1 has a role in mediating IFNg-induced terminal growth arrest. Resistance of mammary carcinomas to growth inhibition by IFNg does not appear to involve resistance of cdk2 complexes to the action of p27, but rather an inability to appropriately regulate the balance of cdk2, cyclin E and p27 levels. Keywords: p27Kip1; growth inhibition; mammary epithelial cells; cdk2; cyclin E; mammary carcinoma
Introduction Cell cycle arrest in the G1 phase of the cell cycle has been associated with checkpoints regulated by cyclindependent kinase (cdk) complexes and their inhibitors (cki's). One of the critical events in progression through G1 is the hyperphosphorylation and inactivation of the growth inhibitory function of the retinoblastoma protein, Rb (Sherr and Roberts, 1995 for review). The primary cdk complexes involved in initiating Rb phosphorylation are thought to contain Correspondence: AM Jetten Received 18 November 1996; accepted 21 January 1997
cdk4 or cdk6 and a member of cyclic D family, because the timing of Rb hyperphosphorylation has been demonstrated to coincide most closely with the onset of cyclin D-associated activity (Resnitzky and Reed, 1995). Which cyclin D isoform is the most active and/ or abundant appears to be dependent in part on cell type speci®city (Ewen et al., 1993a and references therein; Sherr, 1995). Cdk2 in complex with cyclin E, and later cyclin A, may also participate in maintaining the hyperphosphorylation of Rb in later stages of G1 and S phase when the complexes are active, as Rb is a substrate for these complexes both in vitro (Akiyama et al., 1992) and in vivo (Hinds et al., 1992; Dowdy et al., 1993; Resnitzky et al., 1995). Evidence has been presented indicating that cdk2/cyclin E-speci®c phosphorylation of Rb, apart from cdk4/6/cyclin D activity, may not merely maintain Rb hyperphosphorylation but is essential for entry into S phase (Hatakeyama et al., 1994). The activity of the kinases that phosphorylate Rb are negatively regulated by two families of cki's (Elledge and Harper, 1994; Hunter and Pines, 1994; Peter and Herskowitz, 1994; Sherr and Roberts, 1995 for reviews). The Ink4 family, including p16Ink4, p15Ink4b, p18 and p19, speci®cally bind to and inhibit cdk 4 and 6 (Hall et al., 1995), either by disrupting the association of the cdk with its D cyclin partner (Serrano et al., 1993; Xiong et al., 1993b; Parry et al., 1995) or preventing the activation of the cdk by cdk-activating kinase (CAK) (Aprelikova et al., 1995; Hirai et al., 1995). p15Ink4b has been shown to be upregulated during growth arrest by TGFb (Hannon and Beach, 1994), while p18 has been shown to primarily associated with cdk6 (Guan et al., 1994). The second family consists at present of p21Cip1, p27Kip1 and p57Kip2. These inhibitors have inhibitory activity toward all G1 cdks (Xiong et al., 1993a) and bind to cdk/cyclin complexes with much higher anity than to the cdk alone (Harper et al., 1995). Their bindings appears both to block the activation of the cdk by CAK (Ko et al., 1993; Kato et al., 1994; Polyak et al., 1994a; Aprelikova et al., 1995; Reynisdottir et al., 1995) and to inactivate preformed, active complexes (Firpo et al., 1994; Kato et al., 1994; Nourse et al., 1994; Polyak et al., 1994b; Toyoshima and Hunter, 1994; Eblen et al., 1995). p21Cip1 is found in cdk/cyclin complexes throughout G1 and has been described as a potential modulator of normal cell cycle traverse (Firpo et al., 1994; Li et al., 1994; Noda et al., 1994; Nourse et al., 1994; Sheikh et al., 1994) as well as an eector of p53-induced growth arrest in response to DNA damage (El-Deiry et al., 1993). p27 is the primary inhibitor of cdk2/cyclin E activity in cells
p27Kip1 and MEC growth arrest BL Harvat et al
2112
arrested by TGFb (Polyak et al., 1994a; Slingerland et al., 1994) and increased p27 protein levels have been associated with density arrest and growth factor deprivation (Hengst et al., 1994; Polyak et al., 1994b; Slingerland et al., 1994). While the target of cyclin D-associated kinases appears to be primarily limited to Rb, the role of the cdk2/cyclin E complex in enabling cell cycle progression is less well understood. Importantly, at least some events regulated by cdk2/cyclin E are independent of the presence of Rb, because cyclin E, but not D1, is required for entry into S phase in cells lacking functional Rb (Lukas et al., 1995; Ohtsubo et al., 1995). One potential target of cyclin E-associated kinase activity which is not shared by cyclin D complexes is histone H1, an in vitro substrate of cyclin E complexes, suggesting that cyclin E may have a role in altering chromatin structure in preparation for DNA replication. Adachi and Laemmli (1994) demonstrated in Xenopus embryo extracts that cdk2 activity is involved in the unwinding of DNA in replication foci forming on post-mitotic, decondensing chromosomes. Cdk2/cyclin E has also been implicated in having a direct role in the initiation of replication (Jackson et al., 1995; Strausfeld et al., 1994; Duronio and O'Farrell, 1995; Heichman and Roberts, 1994 for review). Interferon gamma (IFNg) has been reported to induce growth arrest in a number of cell types. We have previously shown that IFNg causes irreversible growth arrest in human keratinocytes (Saunders and Jetten, 1994) and mammary epithelial cells (MECs) (Harvat and Jetten, 1996). This growth arrest is complete within 24 h of treatment and is accompanied by a shift from the inactive, hyperphosphorylated to the active, hypophosphorylated forms of both Rb and another Rb family member, p130, that has been associated with regulation of the G0 to G1 transition (Herzinger et al., 1995; Sardet et al., 1995; Vairo et al., 1995). IFNg-induced growth arrest in MECs is associated with the down-regulation of the E2Fregulated genes cyclin A, cdc2 and c-myc, consistent with the role of Rb and p130 in repressing E2F activity and thus synthesis of proteins needed for DNA replication (reviews by MuÈller, 1995; Adams and Kaelin, 1995; Johnson, 1995). IFNg prevents Rb hyperphosphorylation in quiescent, EGF-deprived MECs that have been allowed to re-enter the cell cycle after the addition of EGF. The establishment of the block in DNA synthesis, Rb phosphorylation and cyclin A expression all coincide within a small window of time in mid-G1. Our current studies attempt to identify the mechanism(s) by which IFNg mediates growth arrest and speci®cally, how it aects the cyclin dependent kinase complexes that are responsible for the phosphorylation of Rb. We have discovered that IFNg causes the inhibition of cdk2, cdk4 and cdk6 kinases. IFNg induces p27Kip1 protein and increases the association of p27 in cdk2 complexes, suggesting that p27 plays a signi®cant role in the kinase inhibition and therefore in IFNg-mediated growth arrest. IFNg does not induce expression of other Kip family members, p21Cip1/Waf1 or p57Kip1. p15Ink4b protein levels increased by threefold at 16 h of treatment, while p16Ink4 mRNA levels did not change and the protein was undetectable. In addition,
we show in mammary carcinoma cell lines that IFNg treatment is unable to produce a decrease in cdk2 activity or induce p27 above already high basal levels, consistent with our previous ®nding that IFNg was not able to inhibit growth or block the hyperphosphorylation of Rb in these cell lines.
Results Eects of IFNg treatment of MECs on G1 cdk kinase activity The hyperphosphorylation of Rb has proven to be a critical checkpoint in growth control involved in regulating progression through late G1 and into S phase. Previously ®ndings in our laboratory have indicated that the mid-G1 growth arrest induced in MECs by IFNg treatment correlates with the appearance of the hypophosphorylated form of Rb and occurs during the same time interval in G1 in which Rb would normally be hyperphosphorylated in proliferating cells. (Harvat and Jetten, 1996). The growth arrest is demonstrated by a decrease in [3H]methyl-thymidine incorporation within 4 h, declining to 50% by 12 h and to 6% of controls within 24 h. In this study, we investigate the mechanisms by which IFNg could prevent the hyperphosphorylation of Rb, beginning with an analysis of the G1 cyclin dependent kinases which are known to be able to phosphorylate Rb both in vivo and in vitro. Therefore, we ®rst measured the eect of IFNg treatment on cdk2, cdk4 and cdk6 kinase activity. Cell lysates were prepared under non-denaturing conditions from proliferating cells or cells treated for 24 h with IFNg and immunoprecipitated with antibodies to speci®c cdks (see Materials and methods). Immunoprecipitated cdk2 complexes from proliferating MECs demonstrated signi®cant activity toward a histone H1 substrate, and that activity was reduced to background levels by IFNg treatment (Figure 1a). Similarly, cyclin Eassociated kinase activity, presumably due to cdk2, was likewise down-regulated. Antibodies which were neutralized by their cognate peptide prior to addition of lysate were unable to immunoprecipitate kinase activity, indicating the speci®city of the measured activity found in the extracts. The ability of both cdk2 (Figure 1b) and cyclin E-associated (data not shown) protein complexes to phosphorylate a GSTRb-C-terminal fusion protein (Guan et al., 1994) in vitro was equally down-regulated. In a similar manner, kinase activity toward the GST-Rb substrate that could be immunoprecipitated from proliferating cells by antibodies to cdk4 or cdk6 was also abolished after 24 h of treatment with IFNg, as was the activity in cyclin D1 or D2 immune complexes (Figure 1b). Presumably, the kinase activity associated with cyclins D1 and D2 is due primarily to cdk4 and/or cdk6, although cdk2-immunoprecipitated complexes contain some cyclin D1 (Figure 4), implying that a small portion of total cdk2 activity could be present in cyclin D1 complexes. These results show that IFNg treatment was able to eciently inhibit all of the G1 cyclin dependent kinases that are predominantly associated with the hyperphosphorylation of Rb.
p27Kip1 and MEC growth arrest BL Harvat et al
2113
a
a
cdk2
cdk4 Histone H1➝
cdk6 8 16 24 48 Time (hrs)
0
4
0
4
0
4
0
3 10 30 100 500 IFNγ(IU/ml)
0
3 10 30 100 500 IFNγ(IU/ml)
b b cyclin D1
cyclin E GST-Rb ➝
8 16 24 48 Time (hrs)
c p27KIP1 0
8 16 24 48
Time (hrs)
3 10 30 100 500 IFNγ(IU/ml)
d Figure 1 Inhibition of G1 cdk kinase activity by IFNg. Kinase activity toward (a) histone H1 or (b) GST-Rb-C-terminal was measured in complexes immunoprecipitated from proliferating MECs or MECs treated for 24 h with 200 U IFNg. The immunoprecipitating antibody is indicated at the top of each set of lanes. In lanes marked as containing peptide, extracts from proliferating cells were mixed with antibodies which had been neutralized by prior incubation with their cognate peptide as a control for the speci®city of the detected kinase activity in the precipitated complexes. Quantitation was by densitometry
Eects of IFNg treatment of MECs on levels of G1 cdks, cyclins and the inhibitors of their kinase activity Cdk kinase activity, and consequently the state of Rb phosphorylation, has been shown to be aected by relative levels of the G1 cdks, their partner cyclins, and/ or inhibitors of their kinase activity. We investigated whether the inhibitory eect of IFNg on G1 kinase activity was related to decreases in either the cdks themselves or their activating cyclins. Cellular proteins harvested from proliferating MECs or cells treated with IFNg were analysed by immunoblot for cdk levels over a time course or dose of IFNg treatment. Cdk4 levels were down-regulated starting 24 h after addition of IFNg to the media and were signi®cantly reduced at 48 h. Neither cdk2 nor cdk6 levels were signi®cantly aected by IFNg treatment during the period or dose that inhibited DNA synthesis (Figure 2a). The downregulation of cdk4 occurred late in the process of growth arrest, after the incorporation of [3H]-thymidine had reached its nadir (Harvat and Jetten, 1996). We also analysed the eect of IFNg on the levels of the cyclin partners of the G1 cdk's and found that IFNg treatment caused no changes in the levels of cyclins D1 or E (Figure 2b). The predominant isoform of cyclin D in these cells is D1. Cyclin D2 and D3 were detectable, but also appeared unchanged by IFNg treatment (data not shown). These ®ndings suggest that, although decreased cdk4 levels may have a role in maintaining the hypophosphorylated state of Rb, the initial stages of the block to Rb phosphorylation could not be attributed to changes
— p15Ink4b 0
4
8
16
24
48
Figure 2 Time course and dose eects of IFNg treatment on G1 cdk, cyclin and cki proteins. Proliferating MEC were treated with 200 U/ml IFNg and harvested at the times indicated or treated with varying doses as indicated and harvested at 24 h (cdk2, cdk6, cyclin D1, cyclin D2, p27) or 48 h (cdk4). 20 mg of extract protein from each lysate (30 mg for p27 and 100 mg for p15) was subjected to immunoblot analysis with detection by ECL for the speci®c proteins
in levels of G1 cyclins or cdks. Suspecting that one or more of the cdk inhibitors might therefore be involved in the IFNg-induced block to Rb hyperphosphorylation, we analysed the eects of IFNg on the expression of several members of the Kip and Ink families. We have previously shown that p21Cip1/Waf1 was easily detectable in cycling cells and was down-regulated after 24 h of IFNg treatment (Harvat and Jetten, 1996), a delayed response which followed the inhibition of DNA synthesis similar to that of cdk4. Down-regulation of p21 has also been reported in the irreversible growth arrest that occurs as cells become senescent (Alcorta et al., 1996). Conversely, levels of p27Kip1 protein increased by 16 h and in multiple independent experiments, reached a stable elevated plateau 6 ± 40-fold above untreated cells by 24 h (Figure 2c). The variability of the fold induction was not due to signi®cant dierences in p27 levels in treated cells, but rather to variable basal levels which appeared to increase with the passage number of the cells and correlated with the percentage of cells which were no longer cycling (personal observation). Elevation of p27 protein was maintained for at least 72 h after cells had been exposed to IFNg, even if cells were refed media without IFNg after 24 h (data not shown). Interestingly, p27 mRNA levels did not change signi®cantly over a time course of 48 h of IFNg treatment (Figure 3a), indicating that the increase that is seen at the protein level must be regulated posttranscriptionally. p57Kip2, another member of this family of kinase inhibitors, was not detectable by the antibody available to us (data not shown).
p27Kip1 and MEC growth arrest BL Harvat et al
2114
IFNg treatment of MECs induced p15Ink4b protein levels approximately threefold after 16 h of treatment, which by 48 h returned essentially to levels in untreated cells (Figure 2d). Using three dierent antibodies, we were unable to detect p16Ink4 protein in MECs, in agreement with the ®ndings of others (Sandhu et al., submitted), even though p16 mRNA can be detected (Figure 3b) and is unaected by IFNg treatment. While the role of the small increases seen in p15 protein levels in IFNg-mediated growth arrest needs to be explored further (see Discussion), these studies indicated that the primary eect of IFNg treatment on the ckis that might negatively regulate the hyperphosphorylation of Rb is the induction of p27 protein levels. p27Kip1 is able to bind universally to the various G1 cdk/cyclin complexes and could account for the inhibition of kinase activity which we detect in IFNg-treated cells. The presence of p27 protein increases in cdk complexes immunoprecipitated from IFNg-treated cells p27 has been shown to inhibit cdk activity based on a tightly regulated threshold in which as little as threefold increase in protein levels is sucient to induce kinase inactivation and growth arrest (Kato et al., 1994). These minimal increases in p27 correlate with the detection of increases of p27 in cdk complexes. Therefore, we predicted that the increased levels of p27 seen in IFNg-treated MECs should be re¯ected in changes in the amount of p27 associated with the cdks. As expected, we found that antibodies to cdk2 immunoprecipitated signi®cantly higher levels of associated p27 protein from extracts of cells treated with IFNg for 24 h than from extracts of proliferating cells (Figure 4). In untreated MECs, cdk4 and cdk6 had considerably levels of p27 protein associated with them, consistent with the proposed model that cdk4/6/ cyclin D complexes can sequester p27 in proliferating cells (Toyoshima and Hunter, 1994), permitting cdk2 to be active (Polyak et al., 1994a). IFNg treatment induced a moderate 55 ± 70% (5twofold) increase of
a p27
— 2.5 kb
GADPH
— 1.6 kb 0
2
4
8
16
24
48
b p16ink4a
— 1.0 kb
GADPH
— 1.6 kb 0
4
8
16
24
48
Figure 3 Eects of IFNg on mRNA levels from speci®c cdk inhibitor genes. The levels of cki mRNAs over a time course of IFNg treatment were investigated by (a) Northern analysis of total cellular mRNA (30 mg) or (b) Northern analysis of poly(A)+-mRNA (approximately 4 mg). The level of GADPH mRNA was used as a control for equal loading of RNA
p27 in these complexes, in contrast to the greater than 20-fold increase of p27 seen in cdk2 complexes in the same experiment (Figure 4b). Upon overexposure, small levels of p27 could also be detected in cdk2immunoprecipitated complexes from proliferating cells. We interpreted this to be a re¯ection of the heterogenous nature of this primary culture which could contain cells that have permanently withdrawn from the cell cycle and have elevated levels of p27. We cannot completely rule out, however, that small amounts of p27 could be present in cdk2 complexes in actively cycling cells or at a given stage of the cell cycle. Reprobing the same blots with antibodies to the G1 cyclins demonstrated that cyclin D1 could be found predominantly in complexes with cdk4 and cdk6, but a fraction was also in complex with cdk2, as reported by others (Xiong et al., 1992; Bates et al., 1994). Cyclin E was associated solely with cdk2. Furthermore, after normalization of cyclin levels with the relative amount of cdk immunoprecipitated in each complex, we demonstrated that the amounts of cyclin D1 that were associated with each of the cdks and the amount of cyclin E that was associated with cdk2 were not aected by IFNg treatment (Figure 4b). We also probed immunoblots of cdk immunoprecipitates from two independent experiments with antibodies to p21 and consistently found that the amounts of p21 associated with cdk2 and cdk4 were minimally increased (*25%) with IFNg treatment (Figure 4, note that since IFNg does not aect the levels of cdks at 24 h, the data in Figure 4b has been normalized to the amount of cdk that was immunoprecipitated), while the amount of p21 that was associated with cdk6 decreased. These results suggest that p21 may participate in inhibiting cdk2 and cdk4, but not cdk6, activity in an additive fashion with p27 by binding multiple molecules in a single complex, demonstrated by Zhang et al. (1994) and Harper et al. (1995) to be necessary for kinase inhibition. These ®ndings provide additional evidence that, in normal cells, p27 has an important role in IFNg-mediated growth arrest, although they indicate other ckis could contribute to the kinase inhibition as well. p27 association with cdk2 is a rapid event following IFNg treatment The inhibition of cdk2 activity by IFNg appeared to be primarily the result of increased availability of p27 to associate with the complex. We next investigated if the IFNg-induced association of p27 with cdk2 agreed with the time course in which growth was arrested and Rb hyperphosphorylation was blocked in MECs. We immunoprecipitated cdk2 from extracts of asynchronous, proliferating cells that had been treated for dierent periods of time with IFNg and analysed the immunoprecipitates for the appearance of p27 in the complexes. Cdk2 complexes accumulated increasing levels of p27 protein (Figure 5a) during a time course that paralleled the previously reported decrease in DNA synthesis. Interestingly, an increase in p27 was ®rst detected in cdk2 complexes within 2 h of treatment, even before the induction in p27 protein levels was discernible. This detectable increase in p27 association with cdk2 coordinated with the down
p27Kip1 and MEC growth arrest BL Harvat et al
2115
a
b
— p27
— p21
— cdk2
— cdk4
— cdk6
— cyclinD1
— cyclin E
Figure 4 Eects of IFNg on G1 cdk complexes. Antibodies against the G1 cdks were used to immunoprecipitate associated complexes from extracts of proliferating MECs and cells treated for 24 h with IFNg. The immunoprecipitates were then analysed by immunoblot for the presence of cyclins and the cki's, p21 and p27, by serial probing of the same blots with the relevant antibodies. Graphs represent changes in the quantities of cyclins or cki's found in the complex relative to the amount of cdk that was immunoprecipitated, with the amount found in the IFNg-treated complex given an arbitrary value of 100%
regulation of in vitro kinase activity (Figure 5b). As with total p27 levels in the cells, the relative elevation in p27 seen in the complexes after 24 h of treatment was dependent on the levels seen in complexes from untreated cells and varied from 4.5-fold upwards after being adjusted for the amount of cdk2 that was immunoprecipitated. Consistent with the ®nding that cdk2 levels do not change with IFNg treatment (Figure 2a), reprobing the same blot with anti-cdk2 con®rmed that approximately equal amounts of cdk2 complexes were immunoprecipitated and loaded on the gel for each time point. Mammary carcinoma cell lines that are resistant to growth arrest by IFNg are unable to down-regulate cdk2 kinase activity or up-regulate p27 protein levels We have previously shown that IFNg was unable to induce growth arrest or block Rb phosphorylation in a number of mammary carcinoma cell lines (Harvat and Jetten, 1996). We analysed these cell lines, all of which contain wild type Rb except for MDA-MB-549 cells, for the eect of IFNg on cdk2 activity and p27 inducibility to test if there was a correlation with their resistance to IFNg-induced growth arrest. IFNg had little or no eect on cdk2 kinase activity in any of the mammary carcinoma cell lines that we tested, except that the very high levels of kinase activity in the MDAMB-549 cell line are somewhat reduced (Figure 6).
Furthermore, the carcinoma cell lines had 1.5- to 15fold higher levels of cdk2 activity than did MECs. The elevated kinase activity generally correlated with 10- to 38-fold higher levels of cdk2 protein in the carcinoma cell lines compared to normal MECs (Figure 7a). In addition, carcinoma cells contained higher levels (1.5to 4.2-fold) of cyclin E, as well as the truncated forms of cyclin E which have also been shown to have kinase activity (Keyomarsi et al., 1995). Surprisingly, all cell lines demonstrated substantial levels of p27 protein in proliferating cells, nine- to 40fold higher than that found in proliferating MECs (Figure 7a). Although in some cases, the level of p27 protein appeared to be inversely related to the relative amount of cdk2 kinase activity (MDA-MB-231, MDAMB-549, ZR 75-1) as might be expected, other cell lines (e.g. MCF7) contained high levels of both cdk2 activity and p27 protein. Furthermore, p27 levels were not inducible by IFNg in any of the cell lines, suggesting that IFNg was not able to modulate p27 levels above the threshold required for kinase inhibition in the resistant mammary carcinoma cells (see Discussion). The lack of inhibition of cdk2 kinase activity in the presence of high levels of p27 in carcinoma cells could be explained if the p27 protein was unable to bind to cdk2/cyclin E complexes. However, immunoprecipitation of cdk2 complexes followed by immunoblot detection of p27 revealed that p27 protein was detectable in the cdk2 complexes from all proliferating
p27Kip1 and MEC growth arrest BL Harvat et al
2116
a — p27
— cdk2 P
0
2
4
8
16
24
b — Histone H1
0
2
4
8
12 16 24
Figure 5 Recruitment of p27 to cdk2 complexes with IFNg treatment. (a) Immunoprecipitation studies using anti-cdk2 were conducted on extracts from MECs treated for various times with 200 U/ml IFNg as indicated and the pellets were analysed on the same immunoblot for the presence of p27 and cdk2. P indicates pretreatment of the cdk2 antibody with a neutralizing peptide. The graph represents the amount of p27 protein found in each complex relative to that found in the complex from untreated cells (time 0) after normalization to the amount of cdk2 that was immunoprecipitated. (b) Histone H1 kinase activity of immunoprecipitated cdk2 complexes during a similar time course of IFNg treatment
cell lines (Figure 7b). As suggested by the inability of IFNg to cause an increase in the total level of p27 protein, the levels of p27 that were associated with cdk2 in the carcinoma cells did not change with IFNg treatment. Furthermore in some cases, the p27:cdk2 ratio in the complexes appeared even higher than that seen in the IFNg-treated MECs (Figure 7b, compare IFNg-treated MECs with untreated MCF-7s), yet cdk2 kinase activity was unaected. To further investigate the role of p27 in growth arrest in mammary epithelial cells, we introduced exogenous expression of p27 in both normal MEC and two carcinoma cell lines, using an adenovirus carrying the p27Kip1 cDNA (Adp27). We chose the MCF-7 line because of its apparent lack of relationship between levels of p27 in cdk2 complexes and cdk2 activity and the MDA-MB-549 cell line because of indications that it was slightly responsive to endogenous p27 levels. Exogenous expression of p27 was sucient to inhibit [3H]thymidine incorporation in the MECs in a dose dependent manner (Figure 8a). Interestingly, both carcinoma cell lines were also sensitive to growth inhibition by Adp27. The higher levels of Adp27 needed to inhibit DNA synthesis in MCF-7 cells produced p27 levels lower than that expressed in the other cell lines (Figure 8b) and therefore may re¯ect infection or transcription eciency rather than resistance to p27 eects. In agreement with Figure 7a, longer exposure of the ECL reaction for p27 demonstrated the presence of endogenous levels of p27 in cycling and Adnull cells (not shown). Levels of exogenous p27 which inhibited DNA synthesis eectively blocked Rb hyperphosphor-
a
ZR- MDAMDAMEC T47D MB-231 MCF-7 75-1 MB-549 – + – + – + – + – + – + IFNγ — p27
ZR- MDAMDAMEC T47D MB-231MCF-7 75-1 MB-549 – + – + – + – + – + – +
— cdk2
Histone H1 — — cyclin E
b
MDAZR- MDAMEC T47D MB-231 MCF-7 75-1 MB-549 – + – + – + – + – + – + — p27
Figure 6 The regulation of cdk2 kinase activity in human mammary carcinoma cells by IFNg. Appropriate lysates from proliferating cells and cells treated for 24 h with IFNg were subjected to immunoprecipitation with cdk2 antibodies and the resultant complexes were analysed with kinase assays for activity toward histone H1. Quantitation was by densitometry and fold activity was calculated relative to that found in proliferating MECs
— cdk2 Figure 7 Eects of IFNg on components of the cdk2 complexes in human mammary carcinoma cell lines. (a) 20 mg of protein from total cell lysates were analysed by immunoblot for the indicated proteins. (b) Immunoblot analysis of cdk2 immunoprecipitates as in Figure 5 for p27 and cdk2 protein levels done on the same blot
p27Kip1 and MEC growth arrest BL Harvat et al
2117
a
a
3H-thymidine
incorporation% of control
100
– MEC 549 MCF–7
80
MEC MEC-MB-549
60 MCF-7 40 b
20
IP: 0 0
100
200
300
pfu/cell b
MEC –
n
549 p
–
n
p27
cdk2
MEC 549 MCF7 MEC 549 MCF7 p27 protein – + – – + – + – + – + – + – + – – + – – – – – – + + – – – – IFNγ
MCF7 p
–
n
p
— Adenovirus — p27
c
+ IFNγ – p27 protein
549
MCF7
– n p
– n p
— Adenovirus — Rb
— cyclin A Figure 8 Eects of over-expression of p27 protein on DNA synthesis in MEC and two carcinoma cell lines. (a) Cells exposed to various doses as indicated of Adnull or Adp27 for 48 h were then pulse-labeled with [3H]thymidine and the incorporation determined. Results are plotted as % inhibition of thymidine incorporation in the cells exposed to Adp27 compared to that measured in cells exposed to Adnull. (b) Western blots of extracts from cells infected at the dose level which inhibited DNA synthesis, showing levels of p27 protein expressed after 48 h. (c) Protein extracts used in (b) were analysed by immunoblot for the eect of Adp27 on Rb phosphorylation and cyclin A expression
ylation in the Rb-positive MCF-7 cells and cyclin A expression in both cell lines (Figure 8c). We have also found that Adp27 increases the levels of hypophosphorylated p130 in both MEC and carcinoma cells (data not shown). In a second approach, cdk2 complexes were immunoprecipitated from proliferating normal MECs and the two carcinoma cell lines. Immediately prior to initiating kinase assays on these complexes, the pellets were incubated with various levels of puri®ed p27 protein for 30 min. As seen in Figure 9a, p27 protein was able to inhibit the observed kinase activity in a dose dependent manner similar to its eects in vivo. The relative amounts of p27 required for kinase inhibition were essentially equivalent in MECs and MCF7 cells, while a tenfold higher dose was required to completely inhibit complexes from MDA-MB-549 cells. These results indicate that cdk2 complexes in carcinoma cells are not resistant to p27-mediated inhibition, although there may be dierences in anity or activity of p27 for the complex. In support
Figure 9 Eects of exogenous puri®ed recombinant p27 protein on cdk2 kinase activity from proliferating normal MEC and two human carcinoma cell lines. (a) Complexes immunoprecipitated by antibodies to cdk2 from 100 mg protein from extracts were tested for kinase activity toward histone H1 after incubation with increasing amounts of bacterially expressed, puri®ed his6-p27 fusion protein. The range of levels of p27 protein used for the carcinoma cells was tenfold greater than the range used for MECs. Endogenous p27 and cdk2 content of similar complexes is characterized in Figure 7b. (b) histone H1 kinase activities associated with complexes immunoprecipitated by antibodies to p27 are shown. 250 mg of protein from extracts of MECs or proliferating carcinoma cells were used in the immunoprecipitation. Where indicated, puri®ed p27 protein was added at the highest dose shown in (a) for the respective cell type. Supernates from the p27 IPs were divided equally and reimmunoprecipitated with cdk2 antibodies. Kinase activity of the non-p27-associated cdk2 in the pellet was measured. Autorad exposure of assays from p27 immunoprecipitations was 6 times longer than that of the cdk2 immunoprecipitations in both (a) and (b)
of this interpretation, we found that complexes from untreated or IFNg-treated MECs or carcinoma cells which were immunoprecipitated by an antibody against p27 contained only small amounts of kinase activity (Figure 9b), suggesting that most of the kinase associated with p27 was inactive. The remaining kinase activity was inhibited by the addition of recombinant p27 protein, indicating that changes in p27/cdk2 stoichiometry in the active p27 complexes were sucient for inhibition. The p27 immunoprecipitates from the carcinoma cells contained more than half of the total cdk2 and 495% of the p27 in the extract (data not shown). Finally, we immunoprecipitated cdk2 from the supernatants of the p27 immunoprecipitations to determine if residual cdk2 activity remained. Supernatants from untreated MECs contained cdk2 kinase activity nearly equivalent to that found in complete extracts (compare lane 8 of Figure 9b with lane 1 of Figure 9a), as might be predicted by the lack of p27 association with cdk2 in these extracts, while supernatants from IFNg-treated MECs contained virtually no kinase activity. Supernatants from both the carcinoma cell lines contained large amounts of cdk2 activity, which again was inhibited by the addition of recombinant p27 to the reaction. Thus, we were unable to detect any complex containing cdk2 in carcinoma cells which was resistant to inhibition by p27. Taken as
p27Kip1 and MEC growth arrest BL Harvat et al
2118
a whole, these results imply that elevated levels of p27, such as are induced by IFNg treatment, are sucient to induce growth arrest in normal mammary epithelial cells and that there is a general inability of the carcinoma cells to aect the p27:cdk2 ratio, even with an exogenous stimulus, to levels sucient to reach the threshold level necessary to overcome their high expression of cdk2 and cyclin E. Discussion IFNg induces an irreversible growth arrest in MECs that correlates with the point in G1 of the cell cycle at which Rb would normally be hyperphosphorylated. In this study, we have attempted to elucidate the mechanisms by which IFNg prevents the phosphorylation of Rb. We have shown that IFNg treatment results in the inhibition of the activity of all three G1 cdks known to be primarily responsible for the hyperphosphorylation of Rb, either in complexes immunoprecipitated by antibodies to the kinase itself or to the G1 cyclin partners, cyclins D1, D2 or E. Furthermore, IFNg treatment of MECs increases the levels of the cdk inhibitor p27Kip1 over a 24 h time course that correlates with the previously reported IFNg-induced decrease in thymidine incorporation and disappearance of hyperphosphorylated forms of Rb (Harvat and Jetten, 1996). p27 mRNA levels do not change signi®cantly during the time course in which protein levels increase, lending support to the growing body of evidence that post-transcriptional regulation of p27 is a key mechanism in growth arrest (Polyak et al., 1994b; Halevy et al., 1995; Pagano et al., 1995; Hengst and Reed, 1996). Levels of both the G1 cyclins and cdk's were essentially unchanged within these 24 h, although cdk4 levels began to decrease about 24 h post-treatment and were signi®cantly reduced by 48 h. Similar results have been reported in other systems (Ewen et al., 1993b; Slingerland et al., 1994), implicating down-regulation of cdk4 as a component for the maintenance of long-term growth arrest. Coimmunoprecipitation studies indicated that the amount of p27 associated with cdk2 increased signi®cantly with IFNg treatment, while p27 levels in cdk4 and cdk6 complexes increased only modestly above that found in proliferating cells. Furthermore, at 24 h of treatment, both cdk2 and cdk4, but not cdk6, complexes consistently contained small increases of p21 protein relative to that in complexes from proliferating cells, indicating there may be some additive eect of inhibition from multiple Kip family molecules. Our results suggest that IFNg-induced irreversible growth arrest is a substantially dierent process than that described for TFGb-mediated reversible growth arrest (Massague and Polyak, 1995; Alexandrow and Moses, 1995 for reviews). In the proposed model for TGFbmediated growth arrest of epithelial cells, rapid 15 ± 30fold increases in p15Ink4b levels, induced at the transcriptional level (Hannon and Beach, 1994), cause the release of p27 from sequestration by cdk4/cyclin D complexes. p27 in turn binds and inactivates cdk2/ cyclin E complexes (Reynisdottir et al., 1995; Polyak et al., 1994a). In MECs, TGFb similarly up-regulates p15 mRNA and protein and thereby causes the dissociation of cdk4 and 6 from cyclin D (Sandhu et al., submitted).
IFNg treatment of MECs, on the other hand, saw not only an increase in p27 association with cdk4 or cdk6, albeit small, but cyclin D1 remained associated with cdk4 and cdk6 and a threefold induction of p15 protein was observed after a lag time of at least 8 h. The increases in p15 protein, which appear to be posttranscriptionally induced (unpublished observations), may contribute to inhibition of cdk4 and 6 as well as a redistribution of existing p27 into cdk2 complexes as in the TGFb model. However, the dramatic increase in protein levels of p27 is more likely to in¯uence the continued increased association of p27 and cdk2. Three additional ®ndings support this interpretation - the increase in p27 correlates with the appearance of early markers of squamous dierentiation, the levels of p27 remain elevated even if IFNg is removed from the media, and MECs arrested by TGFb treatment do not demonstrate such a sustained, large increase in p27 levels (unpublished results). These dierences in p27 regulation and the composition of cdk complexes, as well as the results which demonstrate that increased p27 levels alone are sucient to arrest MEC growth, support the hypothesis that the induced levels of p27 have a signi®cant role in IFNg-mediated irreversible growth arrest. Still at issue is whether the relatively small increases in p27 levels found upon IFNg treatment in cdk4 and cdk6 complexes contribute to the observed dramatic inhibition of their kinase activities or are solely an indication that the levels of p27 are sucient to saturate cdk4/6 complexes. Cdk4 and cdk6 have been proposed to act as a rheostat, regulating the availability of p27 for its primary physiological role, cdk2 inhibition (Soos et al., 1996). As such, the presence of p27 in cdk2 complexes from IFNg-treated cells without a reduction in the amount of p27 in cdk4/ 6 complexes argues that saturation of available sites in cdk4/6 complexes has occurred. The eect of p27 saturation is particularly signi®cant in light of studies which have shown that Rb kinase activity in cycling hematopoietic B cells (MANCA) cells can be immunoprecipitated by antibodies to p27 and that the kinase responsible for the kinase activity in the p27 complexes is cdk6 (Soos et al., 1996). IFNg treatment of MECs, however, clearly causes inhibition of cdk6 activity. p15 levels increase at approximately the same time as p27 levels and therefore, the data generated to date does not distinguish the eects of p27 and p15. Alternatively, we cannot rule out the possibility that cki's that have not been tested here or are yet to be discovered may play a role in inhibition of cdk4 and cdk6. The ink4 family inhibitor, p18, has been described as being more speci®c for cdk6 (Guan et al., 1994), but we have not detected changes in the total cellular levels of p18 protein in IFNg-treated MECs (unpublished observations). IFNg-induced growth arrest of MEC is irreversible and its kinetics are closely associated with the appearance of early markers for squamous differentiation (unpublished observations). This process is probably complex and involves multiple steps. Although we acknowledge that these studies do not clearly reveal the mechanism for the initial immediate block in cell cycle progression, we propose that p27 has an important role in mediating the actions of IFNg. Cells arrested in G0/early G1 have been reported to
p27Kip1 and MEC growth arrest BL Harvat et al
contain elevated amounts of p27 in cdk2 complexes (Soos et al., 1996). Our ®nding that p27 levels increase in cdk2 complexes within 2 h of treatment of MECs and cdk2 activity is undetectable within 4 h is consistent with the initiation of a rapid accumulation of cells in early G1 and the role of p27 in inhibiting cdk activation. In addition to Rb phosphorylation, cdk2 is thought to have additional substrates regulating cell cycle progression. E2F-1 over-expression has been shown to overcome growth arrest induced by the over-expression of p16 but not p21 or p27 (Mann and Jones, 1996), implying that the transcription activation function of E2F released by hyperphosphorylation of Rb can compensate for lack of cdk4/6 activity, but cannot compensate for lack of cdk2 activity. p27 expression has also been shown to cause growth arrest in Saos-2 cells which lack functional Rb (Polyak et al., 1994b; Toyoshima and Hunter, 1994). In addition, we have found that human epithelial keratinocytes in which Rb has been inactivated by the HPV type 18 virus remain sensitive to IFNg and demonstrate increases in p27 levels with IFNg treatment (unpublished results). Thus, the role that p27 has in IFNginduced growth arrest may be important for targeting addition function(s) of cdk2 besides Rb phosphorylation and may provide multiple blocks in progression, some of which would be independent of Rb status or cdk4/6 activation. The persistent elevation in p27 protein levels initiated by IFNg could maintain the irreversible nature of the growth arrest and the continued inhibition of cdk2 activity, similar to the ®ndings that have been reported for neuronal dierentiation (Kranenburg et al., 1995; Lee et al., 1996). We reported previously that mammary carcinoma cell lines are resistant to growth inhibition by IFNg and that hyperphosphorylation of Rb was not blocked by IFNg treatment. Here, we demonstrate that all the carcinoma cell lines tested contain high levels of cdk2 kinase activity and constitutively express higher levels of p27 protein in proliferating cells than that found in MECs. It has been proposed that basal levels of p27 act as a threshold regulator of G1 cdk activation (Sherr and Roberts, 1995; Massague and Polyak, 1995). The high levels of kinase activity in the presence of p27 protein in untreated carcinoma cells suggest that at least some portion of the cdk2 molecules have escaped regulation by p27, which may be part of the mechanism by which carcinoma cells become resistant to IFNg. Consistent with this idea, IFNg treatment was unable to inhibit cdk2 kinase activity, indicating the dysfunction renders the carcinoma cells insensitive to external growth arrest signals. Changes in levels of p27 have recently been shown to regulate the length of the G1 phase of the cell cycle (Coats et al., 1996). Cells refractory to the inhibitory eects of p27 might be expected to have a shortened G1, suggesting that the lack of p27 regulation demonstrated here may directly relate to the growth rate of tumors. The lack of cdk2 inhibition or increase in p27 levels in mammary carcinoma cells in response to IFNg may be explained by several possibilities. (a) In some cell lines, the activation of p27 expression may already be stimulated to a maximum output. Recent evidence suggests that over-expression of cyclin D1 or cyclin E causes an up-regulation of p27 protein as
part of a self-regulatory feedback (Han et al., 1996; Sgambato et al., 1996). Our results show that cyclin E and cdk2 are expressed at considerably higher levels in carcinoma cell lines than in MECs (Figure 7a), and others have shown ampli®cation or overexpression of either cyclin D or cyclin E in a signi®cant percentage of primary mammary carcinomas (Wang et al., 1994; Dickson et al., 1995; Peters et al., 1995). These ®ndings suggest that the overexpression of positive growth regulators may overcome the ability of the feedback to stimulate sucient p27 expression for adequate control of cdk2 kinase activity. (b) Mutations in cell cycle proteins may have made the cdk complexes refractory to the inhibiting eects of p27. Eorts to detect mutated forms of p27 in various primary solid tumors and carcinoma cell lines have been negative to date (Kawamata et al., 1995; Ponce-Castaneda et al., 1995; Pietenpol et al., 1995; Ferrando et al., 1996). However, aberrant truncated forms of cyclin E have been reported to be elevated in both mammary carcinoma cell lines, including all of the lines used here (Figure 6b; Keyomarsi et al., 1995; Said and Medina, 1995), as well as in primary tumor tissue samples, where their occurrence correlated with the clinical stage of the disease. (c) A component of the IFNg signal transduction pathway may be faulty. We have previously demonstrated variable abilities in these cell lines to induce the IFNg-responsive gene, GBP, although all respond to IFNg by up-regulation of the two transcription factors in the IFNg signaling pathway, Stat 1 (Harvat and Jetten, 1996) and IRF-1 (unpublished results). (d) Finally, there could be dysfunction in the speci®c post-transcriptional mechanism involved in increasing p27 protein levels. Our results suggest that, in addition to the previously reported possibility of defects in the IFNg signal transduction pathway, resistance of carcinoma cells to IFNg may also reside in an inability to further induce p27 protein levels. We have shown that this inability of carcinoma cells to upregulate p27 protein suciently to inhibit cdk2 activity may be an inherent result of overexpression of cdk2 and cyclin E in these cells. Introduction of increased levels of p27 both in vivo and in vitro had inhibitory eects respectively on growth and cdk2 activity and indicate that cdk2/cyclin E complexes in mammary carcinoma cells are not intrinsically resistant to inhibition by p27. However, the data also imply that certain carcinoma cells require higher levels of p27 (MDA-MB-549) or higher p27:cdk2 ratios (MCF-7) for kinase inhibition than MECs and may therefore have genetic defects which impact on the relative anity or activity of p27 for the cdk2/cyclin E complex. Such defects could also cause carcinoma cells to require a higher threshold of p27 for kinase inhibition than normal cells. These possibilities require further investigation. In support of the hypothesis that p27 has a role in irreversible growth arrest of MECs, we have shown that normal MECs can be growth arrested and their cdk2 activity inhibited in a dose dependent manner solely by increasing the levels of p27 protein. The results obtained in this study with mammary carcinoma cell lines, in which existing p27 levels in proliferating cells appear ineective in inhibiting cdk2 activity and are not regulated by IFNg and additional
2119
p27Kip1 and MEC growth arrest BL Harvat et al
2120
exogenous p27 protein can inhibit kinase activity, also supports such a hypothesis. p27 appears to have a role in the rapid, IFNg-induced inhibition of cdk2 activity, as p27 levels in cdk2 complexes rapidly increase upon exposure of MECs to IFNg. In addition, we propose that the persistent elevation of p27 protein levels seen with IFNg treatment is an important contributing factor in sustaining the irreversible nature of the growth arrest until the squamous dierentiation program is well in place and other positive cell cycle components, such as cdk4, are down-regulated. Whether p27 also promotes the dierentiation program itself is an intriguing question. Similar roles for p27 are being investigated for neuronal dierentiation, which has been shown to be induced by p27 overexpression in neuroblastoma cells (Kranenburg et al., 1995), for terminal dierentiation of myoblasts (Halevy et al., 1995) and for dierentiation of promyelocytic leukemia HL60 cells into monocytes (Hengst and Reed, 1996). Studies pursuing this question are currently in progress.
Materials and methods Cell culture MECs and mammary carcinoma cell lines were grown and maintained as described (Harvat and Jetten, 1996). Except where noted, IFNg was used at a concentration of 200 U/ ml. Western blotting Cells were lysed and analysed by immunoblot as described previously (Harvat and Jetten, 1996). Antibodies used were cdk4 (1:2500) and p27 (1:1000, gifts from Dr Yue Xiong, University of North Carolina, Chapel Hill); p15/16 (JC6, (1:5) a gift from Dr Jim Koh, Mass. General Hosp., Boston); cdk2 (1:5000), cyclin D1 (1:5000) and cyclin E (1:2500) antibodies from Upstate Biotechnology, Inc. (Lake Placid, NY); p27Kip1 (C-19, 1:500), cyclin D1 (HD11, 1:500), cyclin D2 (C-17, 1:500), cyclin E (HE12, 1:1000), cdk2 (M2, 1:1000) and cdk6 (C-21, 1:500) antibodies from Santa Cruz Technologies, Inc. (Santa Cruz, CA); p27Kip1 (1:1250) antibody from Transduction Laboratories (Lexington, KY); and Waf1 (Ab-1, 1:500) and cyclin D3 (1:500) antibodies from Oncogene Science (Cambridge, MA). Northern analysis Total cellular RNA was isolated using the TriReagent protocol (Molecular Research Center, Inc., Cincinnati, OH). Poly-(A)+-RNA was isolated using the PolyATract System 1000 kit (Promega, Madison, WI). Hybridization and wash steps used the protocol for Quikhyb (Stratagene, La Jolla, CA). Probes were made using gel puri®ed restriction fragments as follows: for p16, an EcoRI/XhoI 800 bp cDNA insert (a gift from Dr David Beach, Howard Hughes Medical Institute, Cold Spring Harbor, NY) and for p27, a 600 bp NheI/XhoI cDNA insert (Luo et al., 1995, a gift from Dr Joan Massague , Memorial SloanKettering Cancer Center, NY). Immunoprecipitations Proliferating MECs or cells treated for 24 h with IFNg were lysed in the culture plate with NP40 lysis buer
(10 mM NaHPO4, pH 7.2, 150 mM NaCl, 1% NP-40, 1 mM EDTA, 5 mM EDTA, 2 mM DTT) or Tween 20 lysis buer (50 mM HEPES, pH 7.5, 150 mM NaCl, 0.1% Tween 20, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol; Matsushime et al., 1994). All lysis buers contained protease and phosphatase inhibitors as follows: 1 mM PMSF, 10 mg/ml leupeptin, 5 mg/ml aprotinin, 10 mg/ml soybean trypsin inhibitor, 150 mg/ml benzamidine, 1 mM sodium ¯uoride, 10 mM betaglycerolphosphate and 0.1 mM sodium vanadate. NP-40 lysates, either frozen immediately at 7708C until they could be used or processed immediately, were rocked at 48C for 10 ± 30 min and then centrifuged for 10 min at 10 000 g at 48C. Tween 20 lysates were frozen immediately in liquid nitrogen and stored at 7708C until the assay was done. After thawing, Tween 20 lysates were put on ice for 2 ± 3 h and then centrifuged for 10 min at 10 000 g. From this point, extracts prepared by NP40 or Tween 20 lysis were treated the same and all procedures were carried out at 48C. Each lysate was preabsorbed with protein A agarose for 30 min, spun for 2 min at 2000 g and analyzed for protein concentration by the BioRad Protein Assay reagent (Hercules, CA). Lysates representing 250 mg of protein were then incubated with 2 mg (cyclin E, cyclin D1, cyclin D2, cdk2, cdk6) or 1 ml (cdk4) of the appropriate antibody for 1.5 ± 2 h. Where indicated, antibodies were ®rst neutralized by 2 h of incubation at room temperature with a 10-fold excess (w/w) of cognate peptide. Antibody-associated complexes were absorbed onto 15 ml of a 50% slurry of prewashed protein A agarose (Upstate Biotechnology, Inc., Lake Placid, NY) with rocking for 1 h and precipitated by centrifugation at 2000 g for 2 min. Pellets were washed three times in the appropriate lysis buer using the above gentle centrifuge conditions and, for kinase assays, two times in kinase buer (50 mM HEPES, pH 7.5, 10 mM MgCl2, 1 mM DTT) with transfer to a clean tube during the last wash. After removal of as much liquid as possible, pellets were resuspended either in SDS ± PAGE sample buer for Western analysis as above or in 15 ml of kinase buer for kinase assays. All quantitation was done using a Personal Densitometer SI and ImageQuaNT software (Molecular Dynamics, Sunnyvale, CA). Kinase assays Lysis buers were NP-40 or Tween 20, as above, for histone H1 or GST-Rb kinase assays, respectively. Using the immunoprecipitated complexes in kinase buer, the kinase reaction was started by adding 12 ml to each reaction of a master mix containing kinase buer with 50 mM ATP and per reaction: 10 mCi [g32P]ATP and 1 mg Histone H1 (#382150, Calbiochem, LaJolla, CA) or 1 ± 2 mg GST-RB-C-term (Guan et al., 1994). Histone H1 assays were incubated 15 min at room temperature and Rb assays were incubated for 30 min at 308C. Reactions were stopped with addition of 30 ml of 26 SDS protein sample buer containing 100 mM DTT and 0.01% bromophenol blue dye. Assays were boiled for 5 min and loaded onto SDS ± PAGE Supra-minigels (Enprotech, Natick, MA). Following electrophoresis, gels were ®xed in 15% methanol/10% acetic acid, dried without heat and exposed to Hyper®lm (Amersham, Arlington Heights, IL) for 1 ± 8 h. The GST-Rb substrate was isolated essentially using the method of Matsushime et al. (1994) except the NETN lysis buer contained protease inhibitors as described above and the eluting kinase buer was pH 8.3. Puri®ed p27 protein was isolated by means of a histidine tag from BL-21 bacteria carrying a p27 cDNA expression vector (Luo et al., 1995, a gift from Dr Joan MassagueÂ, Memorial Sloan-Kettering Cancer Center, NY) and incubated with immunoprecipitated, washed complexes for 30 min on ice prior to adding the kinase reaction starter mix.
p27Kip1 and MEC growth arrest BL Harvat et al
Adenovirus infection Recombinant adenovirus containing human p27 cDNA sequences (Adp27) or empty of cDNA (Adnull) was added to RPMI without serum or MEGM minus EGF and pituitary bovine extract for carcinoma cells and MECs respectively. Adp27 was constructed as described (Katayose et al., 1995) using the p27 cDNA from Dr Joan Massague mentioned above. Dishes of cells approximately 40 ± 50% con¯uent were incubated with a minimum of adenovirus-containing media for 1 h with periodic rocking and then brought to normal levels with regular growth media. Plates were then incubated for 48 h, at which time
cells were pulsed for [3H]thymidine incorporation or harvested for immunoblot analysis as described (Harvat and Jetten, 1996).
Acknowledgements We thank David Beach for the p16Ink4 cDNA, Joan Massague for the p27Kip1 cDNA expression vector, Jim Koh for the JC6 antibody and Yue Xiong for the GST-RbC terminal construct and antibodies to cdk4 and p27. Thanks also to David Alcorta and Richard Paules for critical reading of this manuscript.
References Adachi Y and Laemmli UK. (1994). Embo J., 13, 4153 ± 4164. Adams PD and Kaelin WG Jr. (1995). Sem. Cancer Biol., 6, 99 ± 108. Akiyama T, Ohuchi T, Sumida S, Matsumoto K and Toyoshima K. (1992). Proc. Natl. Acad. Sci., 89, 7900 ± 7904. Alcorta DA, Xiong Y, Phelps D, Hannon G, Beach D and Barrett CJ. (1996). Proc. Natl. Acad. Sci., 93, 13742 ± 13747. Alexandrow MG and Moses HL. (1995). Cancer Res., 55, 1452 ± 1457. Aprelikova O, Xiong Y and Liu ET. (1995). J. Biol. Chem., 270, 18195 ± 18197. Bates S, Bonetta L, MacAllan D, Parry D, Holder A, Dickson C and Peters G. (1994). Oncogene, 9, 71 ± 79. Coats S, Flanagan WM, Nourse J and Roberts JM. (1996). Science, 272, 877 ± 880. Dickson C, Fantl V, Gillett C, Brookes S, Bartek J, Smith R, Fisher C, Barnes D and Peters G. (1995). Cancer Letters, 90, 43 ± 50. Dowdy SF, Hinds PW, Louie K, Reed SI, Arnold A and Weinberg RA. (1993). Cell, 73, 499 ± 511. Duronio RJ and O'Farrell PH. (1995). Genes Dev., 9, 1456 ± 1468. Eblen ST, Fautsch MP, Anders RA and Leof EB. (1995). Cell Growth Dier., 6, 915 ± 925. El-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Trent JM, Lin D, Mercer WE, Kinzler KW and Vogelstein B. (1993). Cell, 75, 817 ± 825. Elledge SJ and Harper JW. (1994). Curr. Opin. Cell Biol., 6, 847 ± 852. Ewen ME, Sluss HK, Sherr CJ, Matsusime H, Kato J and Livingston DM. (1993a). Cell, 73, 487 ± 497. Ewen ME, Sluss HK, Whitehouse LL and Livingston DM. (1993b). Cell, 74, 1009 ± 1020. Ferrando AA, Balbin M, Pendas AM, Vizoso F, Velasco G and Lopez-Otin C. (1996). Hum. Genet., 97, 91 ± 94. Firpo EJ, Ko A, Solomon MJ and Roberts JM. (1994). Mol. Cell. Biol., 14, 4889 ± 4901. Guan K-L, Jenkins CW, Li Y, Nichols MA, Wu X, O'Keefe CL, Matera AG and Xiong Y. (1994). Genes Dev., 8, 2939 ± 2952. Halevy O, Novitch BG, Spicer DB, Skapek SX, Rhee J, Hannon GJ, Beach D and Lassar AB. (1995). Science, 267, 1018 ± 1021. Hall M. Bates S and Peters G. (1995). Oncogene, 11, 1581 ± 1588. Han EK, Begemann M, Sgambato A, Soh J, Doki Y, Xing W, Liu W and Weinstein IB. (1996). Cell Growth Dier., 7, 699 ± 710. Hannon GJ and Beach D. (1994). Nature, 371, 257 ± 261. Harper JW, Elledge SJ, Keyomarsi K, Dynlacht B, Tsai L-H, Zhang P, Dobrowski S, Bai C, Connell-Crowley L, Swindell E, Fox MP and Wei N. (1995). Mol. Biol. Cell, 6, 387 ± 400.
Harvat BL and Jetten AM. (1996). Cell Growth Dier., 7, 289 ± 300. Hatakeyama M, Brill JA, Fink GR and Weinberg RA. (1994). Genes Dev., 8, 1759 ± 1771. Heichman KA and Roberts JM. (1994). Cell, 79, 557 ± 562. Hengst L and Reed SI. (1996). Science, 271, 1861 ± 1864. Hengst L, Dulic V, Slingerland JM, Lees E and Reed SI. (1994). Proc. Natl. Acad. Sci., 91, 5291 ± 5295. Herzinger T, Wolf DA, Eick D and Kind P. (1995). Oncogene, 10, 2079 ± 2084. Hinds PW, Mittnacht S, Dulic V, Arnold A, Reed SI and Weinberg RA. (1992). Cell, 70, 993 ± 1006. Hirai H, Roussel MF, Kato J, Ashmun RA and Sherr CJ. (1995). Mol. Cell. Biol., 15, 2672 ± 2681. Hunter T and Pines J. (1994). Cell, 79, 573 ± 582. Jackson PK, Chevalier S, Philippe M and Kirschner MW. (1995). J. Cell Biol., 130, 755 ± 769. Johnson DG. (1995). Oncogene, 11, 1685 ± 1692. Katayose D, Wersto R, Cowan KH and Seth P. (1995). Cell Growth Dier., 6, 1207 ± 1212. Kawamata N, Morosetti R, Miller CW, Park D, Spirin KS, Nakamaki T, Takeuchi S, Hatta T, Simpson J, Wilczynski S, Lee YY, Bartram CR and Koeer HP. (1995). Cancer Res., 55, 2266 ± 2269. Kato J, Matsuoka M, Polyak K, Massague J and Sherr CJ. (1994). Cell, 79, 487 ± 496. Keyomarsi K, Conte D Jr, Toyofuku W and Fox MP. (1995). Oncogene, 11, 941 ± 950. Ko A, Ohtsuki M, Polyak K, Roberts JM and Massague J. (1993). Science, 260, 536 ± 539. Kranenburg O, Scharnhorst V, Van der Eb AJ and Zantema A. (1995). J. Cell Biol., 131, 227 ± 234. Lee M-H, Nikolic M, Baptista CA, Lai E, Tsai L-H and Massague J. (1996). Proc. Natl. Acad. Sci., 93, 3259 ± 3263. Li Y, Jenkins CW, Nichols MA and Xiong Y. (1994). Oncogene, 9, 2261 ± 2268. Lukas J, Bartkova J, Rohde M, Strauss M and Bartek J. (1995). Mol. Cell. Biol., 15, 2600 ± 2611. Luo Y, Hurwitz J and Massague J. (1995). Nature, 375, 159 ± 161. Mann DJ and Jones NC. (1996). Curr. Biol., 6, 474 ± 483. Massague J and Polyak K. (1995). Curr. Opinion Genetics Dev., 5, 91 ± 96. Matsushime H, Quelle DE, Shurtle SA, Shibuya M, Sherr CJ and Kato J. (1994). Mol. Cell. Biol., 14, 2066 ± 2076. MuÈller R. (1995). TIG, 11, 173 ± 178. Noda A, Ning Y, Venable SF, Pereira-Smith OM and Smith JR. (1994). Exp. Cell Res., 211, 90 ± 98. Nourse J, Firpo E, Flanagan WM, Coats S, Polyak K, Lee M-H, Massague J, Crabtree GR and Roberts JM. (1994). Nature, 372, 570 ± 573. Ohtsubo M, Theodoras AM, Schumacher J, Roberts JM and Pagano M. (1995). Mol. Cell. Biol., 15, 2612 ± 2624.
2121
p27Kip1 and MEC growth arrest BL Harvat et al
2122
Pagano N, Tam SW, Theodoras AM, Beer-Romero P, Del Sal G, Chau V, Yew PR, Draetta GF and Rolfe M. (1995). Science, 269, 682 ± 685. Parry D, Bates S, Mann DJ and Peters G. (1995). EMBO J., 14, 503 ± 511. Peter M and Herskowitz I. (1994). Cell, 79, 181 ± 184. Peters G, Fantl V, Smith R, Brookes S and Dickson C. (1995). Breast Cancer Res. Treatment, 33, 125 ± 135. Pietenpol JA, Bohlander SK, Sato Y, Papadopoulos N, Liu B, Friedman C, Trask BJ, Roberts JM, Kinzler KW, Rowley JD and Vogelstein B. (1995). Cancer Res., 55, 1206 ± 1210. Polyak K, Kato J, Solomon MJ, Sherr CJ, Massague J, Roberts JM and Ko A. (1994a). Genes Dev., 8, 9 ± 22. Polyak K, Lee M-H, Erdjument-Bromage H, Ko A, Roberts JM, Tempst P and Massauge J. (1994b). Cell, 78, 59 ± 66. Ponce-Castaneda MV, Lee MH, Latres E, Polyak K, Lacombe L, Montgomery K, Mathew S, Krauter K, Sheinfeld J, Massague J and Cordon-Cardo C. (1995). Cancer Res., 55, 1211 ± 1214. Resnitzky D and Reed SI. (1995). Mol. Cell. Biol., 15, 3463 ± 3469. Resnitzky D, Hengst L and Reed SI. (1995). Mol. Cell. Biol., 15, 4347 ± 4352. Reynisdottir I, Polyak K, Iavarone A and Massague J. (1995). Genes Dev., 9, 1831 ± 1845. Said TK and Medina D. (1995). Carcinogenesis, 16, 823 ± 830. Sardet C, Vidal M, Cobrinik D, Geng Y, Onufryk C, Chen A and Weinberg RA. (1995). Proc. Natl. Acad. Sci., 92, 2403 ± 2407.
Saunders NA and Jetten AM. (1994). J. Biol. Chem., 269, 2016 ± 2022. Serrano M, Hannon GJ and Beach D. (1993). Nature, 366, 704 ± 707. Sgambato A, Han EK, Zhou P, Schieren I and Weinstein IB. (1996). Cancer Res., 56, 1389 ± 1399. Sheikh MS, Li X-S, Chen J-C, Shao Z-M, Ordonez JV and Fontana JA. (1994). Oncogene, 9, 3407 ± 3415. Sherr CJ. (1995). TIBS, 20, 187 ± 190. Sherr CJ and Roberts JM. (1995). Genes Dev., 9, 1149 ± 1163. Slingerland JM, Hengst L, Pan C-H, Alexander D, Stampfer MR and Reed SI. (1994). Mol. Cell. Biol., 14, 3683 ± 3694. Soos TJ, Kiyokawa H, Yan JS, Rubin MS, Giordano A, DeBlasio A, Bottega S, Wong B, Mendelsohn J and Ko A. (1996). Cell Growth Dier., 7, 135 ± 146. Strausfeld UP, Howell M, Rempel R, Maller JL, Hunt T and Blow JJ. (1994). Cur. Biol., 4, 876 ± 883. Toyoshima H and Hunter T. (1994). Cell, 78, 67 ± 74. Vairo G, Livingston DM and Ginsberg D. (1995). Genes Dev., 9, 869 ± 881. Wang TC, Cardi RD, Zukerberg L, Lees E, Arnold A and Schmidt EV. (1994). Nature, 369, 669 ± 671. Xiong Y, Zhang H and Beach D. (1992). Cell, 71, 505 ± 514. Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R and Beach D. (1993a). Nature, 366, 701 ± 704. Xiong Y, Zhang H and Beach D. (1993b). Genes Dev., 7, 1572 ± 1583. Zhang H, Hannon GJ and Beach D. (1994). Genes Dev., 8, 1750 ± 1758.
