Takeshi Sekiguchi 1,2, Eishi Noguchi 2, Toshiro Hayashida 2, Torahiko Nakashima 2, Hideo Toyoshima 1, Takeharu Nishimoto 2 and Tony Hunter 1,*

Transcription

1 D-type cyclin expression is decreased and p21 and p27 CDK inhibitor expression is increased when tsbn462 CCG1/TAF II 250 mutant cells arrest in G1 at the restrictive temperature Takeshi Sekiguchi 1,2, Eishi Noguchi 2, Toshiro Hayashida 2, Torahiko Nakashima 2, Hideo Toyoshima 1, Takeharu Nishimoto 2 and Tony Hunter 1,* 1 Molecular Biology and Virology Laboratory, North Torrey Pines Road, The Salk Institute, La Jolla, CA 92037, USA; 2 Department of Molecular Biology, Graduate School of Medical Science, Kyushu University, Higashi-ku, Maidashi, Fukuoka 812, Japan Communicated by: Fumio Hanaoka Abstract Background: The tsbn462 temperature-sensitive mutant hamster cell line exhibits cell cycle arrest and apoptosis at the restrictive temperature of C, due to a point mutation in the CCG1/ TAF II 250 gene, which encodes a component of the general transcription factor TFIID. Results: We now report that CCG1/TAF II 250 persisted as a complex with TBP and associated proteins (TAFs) in tsbn462 cells at the restrictive temperature. FACScan analysis revealed that the tsbn462 mutation resulted in a failure to progress out of G0 into G1. Using two-dimensional gel electrophoresis we observed a decrease in the synthesis of several proteins, starting in the middle of the G1 phase, becoming very pronounced during late G1. The expression of the immediate early genes c-fos, c-jun and c-myc was normally induced by serum treatment of quiescent cells at the restrictive temperature, whereas expression of cyclins A, D1 and D3 was reduced. Expression of the cyclindependent kinase (CDK) inhibitor proteins p21 and p27 was enhanced. Consistent with the decreased cyclin D and increased p21/p27 expression, we found that phosphorylation of Rb was decreased at C. Cyclin A-, E- and Cdk2-associated histone H1 kinase activity was reduced concomitantly with the increase in p21 protein. Conclusion: Decreased cyclin/cdk kinase activity and decreased Rb phosphorylation are possible causes of G1 cell cycle arrest in tsbn462 cells at the restrictive temperature. Introduction The cell cycle is composed of four consecutive phases, namely G1, S, G2 and M. In vertebrate cells, there are at least two types of genes, which can be distinguished based on how their expression is regulated. One type of gene is constitutively expressed, whereas the other type is expressed in a cell cycle-dependent manner. Among the latter group, immediate early genes, such as c-fos and c-jun, are induced just after serum stimulation, followed * Corresponding author: Fax: by induction of c-myc. In late G1, genes such as cyclin A and cdc2 are expressed. The decision whether or not to replicate is taken during the G1 phase. The G1/S transition is controlled by positive cell cycle regulators, the cyclin D/Cdk4(6), cyclin E/Cdk2, and cyclin A/ Cdk2 cyclin/cdk complexes, which act in that order to phosphorylate target proteins including Rb reviewed in Sherr (1993). Tumour suppressor gene products, such as Rb, act as negative regulators of G1 progression. Rb is phosphorylated and inactivated by cyclin/cdk complexes at the G1/S boundary (Buchkovich et al. 1989; Chen et al. 1989; DeCaprio et al. 1989; Ewen et al. # Blackwell Science Limited Genes to Cells (1996) 1,

2 T Sekiguchi et al. 1993) and this phosphorylation releases cells arrested in G1 and allows them to enter into S phase (Goodrich et al. 1991). G1 progression may also be regulated by cyclin/cdk inhibitors, which fall into two distinct families. The p21 family, which includes p21, p27 and p57, inhibits cyclin/cdk2 and cyclin/cdk4 (or Cdk6) complexes, and the p16 family, which includes p15, p16, p18 and p19, inhibits cyclin Cdk4(6) complexes (reviewed in Sherr & Roberts 1995). Conditional expression of the wild-type p53 tumour suppressor protein induces growth arrest in late G1 phase (Lin et al. 1992), and recent work indicates that p53 induces the expression of p21 and that this is important for arrest in G1 in response to DNA damage. Although the biological significance of many of these genes has been elucidated (reviewed by Hunter & Pines 1994), how their expression is regulated is largely unknown. The tsbn462 cell line is a temperature-sensitive mutant hamster cell line that cannot grow at the nonpermissive temperature of C and which belongs to the same complementation group as the ts13 cell line (Talavera & Basilico 1977). At C these cells do not progress through G1 into S phase due to a defect in CCG1 gene function (Sekiguchi et al. 1987), and undergo apoptosis (Sekiguchi et al. 1995). In previous studies we isolated the CCG1 gene, which complements the tsbn462 mutation (Sekiguchi et al. 1988, 1991). The CCG1 gene maps to human chromosome X q11 q13 (Brown et al. 1989), and in both tsbn462 and ts13 cells there is the same point mutation in CCG1, resulting in the change of Gly 690 to Asp 690 (Hayashida et al. 1994). The TFIID complex is one of the basal transcription factors needed for RNA polymerase II-directed transcription. TFIID consists of the TATA-binding protein, TBP, and at least six TBP-associated factors or TAFs in Drosophila (Dynlacht et al. 1991), and consists of TBP and at least eight TAFs in mammals (Zhou et al. 1992; Zhou et al. 1993). Recently, TAF II 250/p250, which is a component of TFIID that binds directly to TBP, is shown to be identical to CCG1 (Ruppert et al. 1993; Hisatake et al. 1993). TAF II 250/p250 is proposed to be a coactivator for mrna synthesis, which is required for sequence-specific transcription factoractivated transcription but not basal transcription (reviewed by Gill & Tjian 1992). CCG1/TAF II 250 is believed to be a central component of TFIID, and has been shown to associate independently with at least six TAFs and TBP (Yokomori et al. 1993). CCG1/ TAFII250 can also function as a protein kinase (Dikstein et al. 1996). ytaf II 145 is a 145 kda yeast protein that is similar to CCG1/TAF II 250 but lacks the bromodomains and acidic region (Reese et al. 1994). Interestingly, the minimal region of CCG1/ TAF II 250 needed to complement the tsbn462 mutation is a 140 kda N-terminal fragment of CCG1 that lacks the bromodomains and acidic region, but can still form a complex with TBP (Noguchi et al. 1994). To determine whether CCG1/TAF II 250 plays a role in the expression of specific genes during the cell cycle, especially during the G1 phase, we have examined the state of CCG1/TAF II 250-containing complexes and the expression of a number of genes and cyclin/ CDK kinase activities in tsbn462 cells at C and C. We found that CCG1/TAF II 250 persisted in a complex with TBP in tsbn462 cells at C. We also observed the differential expression of certain cell cycle genes in tsbn462 cells. After the shift to C, expression of D-type cyclins was reduced, whereas expression of the cyclin-dependent kinase (CDK) inhibitors p21 and p27 was induced. These inhibitors were found to be associated with cyclin D1 and Cdk2 complexes. As a result of decreased cyclin D expression and increased expression of p21 and p27, phosphorylation of Rb and cyclin A, E/Cdk2 activities were reduced at C, and these events may underlie the arrest of tsbn462 cells in G1 at the restrictive temperature. Results Persistence of the CCG1/TAF II 250-containing complex in tsbn462 cells at the restrictive temperature Many temperature sensitive mutant proteins are unstable at the restrictive temperature. Indeed, this is true for RCC1 and DAD1, the mutant proteins in tsbn2 and tsbn7 cells, respectively, which disappear after less than 10 h when cells are shifted up to C (Nishitani et al. 1991; Nakashima et al. 1993). In contrast, CCG1/ TAF II 250 was detected by immunoblotting for at least 26 h after shifting tsbn462 cells to C (Fig. 1a), although its level declined somewhat. Changes in components of the CCG1/TAF II 250 complex at C could result in the loss of expression of essential genes and/or the induction of cell cycle arrest genes and underlie the temperature-sensitive phenotype of tsbn462 cells. Therefore, we investigated whether or not CCG1/TAF II 250 was complexed with TBP at the restrictive temperature. A CCG1/TAF II 250/ TBP complex was detected at equivalent levels 688 Genes to Cells (1996) 1, # Blackwell Science Limited

3 Cell cycle arrest in the mutant of CCG1/TAF II 250 Figure 1 Expression of the CCG1/TAF II 250 protein in tsbn462 cells. (a) Persistence of the mutant CCG1/TAF II 250 protein in tsbn462 cells at C. Exponentially growing tsbn462 cells were shifted up to C. At the indicated times, protein samples were prepared and immunoblotted with anti-ccg1 antibodies. Arrowheads indicate the position of the CCG1/TAF II 250 protein. (b) Complex formation between CCG1/TAF II 250 and TBP in tsbn462 cells. Protein samples were extracted from exponentially growing tsbn462 cells at C (lanes 1, 3, 5) and at C for 13 h (lanes 2, 4, 6). CCG1/TAF II 250 and TBP were immunoprecipitated with protein A-Sepharose beads (lanes 1, 2), MBL anti-ccg1 (lanes 3, 4) and anti-tbp (lanes 5, 6) antibodies, respectively, and were blotted with anti-ccg1 antibodies. An arrowhead shows the position of the CCG1/TAF II 250 protein. (c) Complex formation between CCG1/TAF II 250 and other proteins. tsbn462 cells were synchronized in G0 by growth in 0.25% calf serum for 3 days at C, and then released by the addition of 10% calf serum. Cells were labelled with 35 S-methionine for 3 h starting either at 13 h (lanes 1, 2, 3, 4) or at 16 h after serum addition (lanes 5, 6, 7, 8). tsbn462 cells were grown at C (lanes 1, 5, 6) or C (lanes 2, 7, 8); BHK21 cells treated similarly were grown at C (lane 3) or C (lane 4). Cell extracts were prepared and immunoprecipitated initially with preimmune serum. The supernatants were then immunoprecipitated with anti-ccg1 (S-4) antibodies as described in Experimental procedures. For competition experiments, 50 g/ml of the synthetic peptide were mixed with anti-ccg1 (S-4) antibodies (lanes 6, 8). Protein samples were analysed by SDS gel electrophoresis on a 10% polyacrylamide gel (lanes 1 8). Small arrowheads show the position of CCG1/TAF II 250-associated proteins. The large arrowhead between lanes 2 and 3 and lanes 6 and 7 shows the position of the 90 kda protein. The positions of 205, 116.5, 80 and 49.5 kda molecular weight markers (Bio-Rad) are shown on the right of each panel. # Blackwell Science Limited Genes to Cells (1996) 1,

4 T Sekiguchi et al. after 13 h at C and C (Fig. 1b). The persistence of CCG1/TAF II 250 and the fact that it remained complexed with TBP at C suggests that it retains some activity in tsbn462 cells at the restrictive temperature. We also examined the other components of the complex at both temperatures by immunoprecipitation of 35 S-methionine-labelled cells with an anti-ccg1 (S- 4) antiserum (Fig. 1c). CCG1/TAF II 250 and seven other associated proteins of 220, 210, 150, 140, 120, 100 and 70 kda were detected at C (indicated by small arrowheads in Fig. 1c, lane 5). That these proteins were specifically associated with CCG1/TAF II 250 was confirmed by competition experiments with the synthetic immunizing peptide (Fig. 1c, lanes 6 and 8). CCG1/ TAF II 250 was found to be associated with the same proteins in tsbn462 cells grown at C, but, in addition, with a 90 kda protein (Fig. 1c, lanes 2 and 7) that was absent from tsbn462 cells grown at C (Fig. 1c, lanes 1 and 5) and BHK21 cells grown at C (Fig. 1c, lane 4). Inhibition of G1 phase progression by the tsbn462 CCG1/TAF II 250 mutation To determine precisely where tsbn462 cell cycle progression is blocked at C, cells were arrested at the G1/S boundary by hydroxyurea treatment, and then released from the block at C. As determined by FACScan analysis, the cells were able to progress as far as the next G1 phase, and a fraction of the cells even appeared to enter the next S phase (9 h) when compared to the cells grown continuously at C (13 h) (Fig. 2a). However, when tsbn462 cells were synchronized in G0, and then released by serum stimulation, the cells were incapable of entering S phase (Fig. 2b), indicating that a major function of CCG1/TAF II 250 is to regulate the expression of genes essential for progression out of G1 into S. To determine exactly when the function of CCG1/TAF II 250 is required during G1 progression, a shift-up experiment was carried out. The execution point was found to be around 10 h after serum stimulation (Fig. 2c), where the Figure 2 tsbn462 cells arrest in G1 at C. (a) tsbn462 cells were arrested by hydroxyurea treatment at the G1/S boundary at C. One series of dishes was then shifted up to C, while the other series was kept at C. At the indicated times, cells were trypsinized and processed for FACScan analysis. (b) tsbn462 cells were arrested in G0 by growth in 0.25% calf serum for 3 days at C. One series of dishes was then shifted up to C, while the other series was kept at C. (c) Determination of the execution point of tsbn462 mutation. Cells were synchronized in G0 by growth in 0.25% calf serum and released by the addition of 10% calf serum. 3 H-thymidine was added at 0 h, and then cells were shifted up to C at the indicated times C: cells were kept at C after serum addition C: cells were shifted up to C after serum addition. 4, 6, 8, 10, 12, 15 indicate that cells were kept for 4, 6, 8, 10, 12, and 15 h after serum addition, and were then shifted up to C, respectively. 690 Genes to Cells (1996) 1, # Blackwell Science Limited

5 Cell cycle arrest in the mutant of CCG1/TAF II 250 execution point is defined as the time after which there is no effect of shifting to the nonpermissive temperature on cell cycle progression (Pringle 1978). In this experiment, DNA synthesis began at around 13 h at C. Thus, the CCG1/TAF II 250-induced genes required for cells to enter S phase appear to be induced 10 h after serum stimulation. To look at the global effects of the CCG1/TAF II 250 mutation on gene expression, we examined newly synthesized proteins in tsbn462 cells at C and C by two-dimensional gel electrophoresis of 35 S- labelled cell proteins (Fig. 3). In wild-type BHK21 cells, protein synthesis at C was essentially the same as that at C, as judged by two-dimensional gel electrophoresis (data not shown). In exponentially growing tsbn462 cells, the number and intensity of protein spots were almost same between 0 and 2 h and h after the shift to C (Fig. 3a), suggesting that only a small number of proteins were influenced by the CCG1/TAF II 250 mutation. In the case of G0- arrested cells, the number and intensity of protein spots at C were almost same between 6 h and 8 h and h after serum stimulation. However, the number and intensity of protein spots decreased progressively at C, and there was a significant difference by h after serum stimulation (Fig. 3b). These results suggest that the number of gene products affected by the CCG1/TAF II 250 mutation is not that great. Early G1 genes are expressed in tsbn462 cells at C following serum stimulation To learn what kind of genes are affected by the CCG1/ TAF II 250 mutation and therefore might be involved in G1 arrest, we compared the expression of a number of cell cycle genes in tsbn462 cells grown at C and C, either at the RNA level by Northern blotting and promoter-reporter gene assay, or at the protein level by immunoblotting. For this purpose, tsbn462 cells were arrested in G0 at C and were then released by serum stimulation to progress into G1 phase at C or C. Northern analysis showed that several typical immediate early genes, c-fos, fosb, c-jun and junb were induced as effectively at C as at C (Fig. 4a). c-myc RNA levels were high, even in G0 phase tsbn462 cells, and exhibited a relative increase after 6 h at C (Fig. 4a, panel 6), with parallel amounts of c- Myc protein also being detected (Sekiguchi et al. 1995). Figure 3 New protein synthesis visualized by two-dimensional gel electrophoresis. (a) Exponentially growing tsbn462 cells were shifted to C and exposed to 35 S- methionine for 2 h at the indicated times. Labelled proteins were analysed by twodimensional gel electrophoresis as described in Experimental procedures. Left, cells labelled between 0 and 2 h at C. Right, cells labelled at h at C. (b) tsbn462 cells synchronized in G0 were stimulated with serum and exposed to 35 S-methionine for 2 h at the indicated times, and labelled proteins were analysed by two-dimensional gel electrophoresis. Upper left, cells labelled at 6 8 h after serum addition at C. Upper right, cells labelled at h after serum addition at C. Lower left, cells labelled at 6 8 h after serum addition at C. Lower right, cells labelled at h after serum addition at C. # Blackwell Science Limited Genes to Cells (1996) 1,

6 T Sekiguchi et al. To assess the ability of various promoters to be utilized in tsbn462 cells we employed promoter reporter gene assays using transient transfection of the promoter reporter gene constructs into growing tsbn462 cells. After incubation for 17 h at C, cells were then grown at either 33.5 or C for a further 24 h before assay. Northern analysis indicated that the c-fos promoter was active at both temperatures, and the promoter assay confirmed that the c-fos promoter was used as efficiently at C as at C, as observed previously by Wang & Tjian (1994). On the other hand, the transcription of the c-myc promoter was significantly greater at C than at C in tsbn462 cells, but not in a ts + transformant of tsbn462 cells expressing wild-type human CCG1/TAF II 250 or in BHK21 cells (Fig. 4b), suggesting that the c-myc promoter was superactivated at C as a result of the CCG1/TAF II 250 mutation. Reduced D-type cyclin and increased p21 CDK inhibitor expression in tsbn462 cells at C following serum stimulation Expression of D-type cyclins is known to be essential for progression through G1 (Quelle et al. 1993) in normal fibroblasts. Therefore, we studied the expression of cyclins D1 and D3 (we could not detect cyclin D2 in BHK21 cells). In BHK cells, cyclin D1 RNA was induced at 10 h after serum stimulation to approximately the same extent at both C and C (Fig. 5a, panel 1b). In contrast, in tsbn462 cells the basal level of cyclin D1 RNA was higher than in BHK cells, and showed only modest serum-stimulated induction at C (Fig. 5a, panels 1a). At C the cyclin D1 RNA level dropped rapidly and showed only a very weak induction at later times (Fig. 5a, panels 1a). Similar results were obtained for cyclin D3 RNA (Fig. 5a, panels 2a and b). The reduced level of cyclin D3 RNA observed at C in tsbn462 cells was restored to a normal level by the expression of wild-type CCG1/ TAF II 250 in ts + tsbn462 cells (Fig. 5a, panel 2c). Since D-type cyclin mrnas have a short half-life (Matsushime et al. 1991), this suggests that the transcription of D-type cyclins was reduced at the restrictive temperature by the CCG1/TAF II 250 mutation. To test Figure 4 The expression of early G1 phase genes in tsbn462. (a) tsbn462 cells were arrested in G0 phase by growth in 0.25% calf serum for 3 days at C, and then stimulated by addition of serum for the indicated times. The cells were preincubated at C for 2 h before addition of serum. Northern blotting analysis of RNA extracted from these cells was performed with c- fos (1), fosb (2), c-jun (3), junb (4), -actin (5) and c-myc (6) probes. The filters were exposed to screens and analysed by a Fuji image analyser L, 30 0 L, 1 L, 2 L, 3 L, 6 L, 9 L, 13 L and 16 L stand for 15 min, 30 min, 1, 2, 3, 6, 9, 13 and 16 h after serum addition at C, respectively H, 30 0 H, 1H, 2H, 3H, 6H, 9H, 13H and 16H stand for 15 min, 30 min, 1, 2, 3, 6, 9, 13 and 16 h after serum addition at C, respectively. (b) c-myc and c-fos promoter reporter gene assays. c-myc and c-fos promoter- CATreporter gene plasmids were transfected into tsbn462, BHK21 and tsbn462(ccg1) (tsbn462 cells expressing wild-type hamster CCG1) cells and promoter activities at C and 39.58C were measured as described in Experimental procedures. Briefly, six dishes of growing cells were transfected with a DNA Ca-phosphate coprecipitate at C for 17 h, and three dishes were kept at C while the other three dishes were shifted up to C. After 24 h, cells were collected and CAT activity were measured using a Phosphorimager. The ratio of the average activity value at C divided by that at C is plotted. Black bars represent c- fos and white bars represent c-myc. 692 Genes to Cells (1996) 1, # Blackwell Science Limited

7 Cell cycle arrest in the mutant of CCG1/TAF II 250 Figure 5 Expression of D-type cyclins and the p21 and p27 cyclin-dependent kinase inhibitors. (a) tsbn462, ts + tsbn462 and BHK21 cells were arrested in G0 by growth in 0.25% CS for three days at C, and then stimulated by addition of serum at C or C for the indicated times. Northern blotting analysis of RNA samples from tsbn462 (a) and BHK21 (b) cells was carried out with probes for cyclins D1 (1) and D3 (2) and -actin (3), which was used as a control. ts + tsbn462 (CCG1) cells, which stably express CCG1 cdna, were analysed by Northern blotting (2c). 1 L, 2 L, 3 L, 6 L, 9 L, 13 L, 15 L, 16 L and 19 L indicate h after serum addition at C. 1H, 2H, 3H, 6H, 9H, 13H, 15H, 16H and 19H indicate h after serum addition at C. (b) Cyclin D1 and p21 promoter reporter gene assay. Cyclin D1 and p21 promoter-luciferase reporter gene plasmids were transfected into tsbn462 and BHK21 cells, and their promoter activities at C and C were determined as described in the legend to Fig. 4b using a luminometer to measure luciferase activities. The ratio of the average activity value at C divided by that at C is plotted. The different promoter reporter gene constructs are defined on the figure. (c) Protein extracts from tsbn462 and BHK21 cells were prepared as in (a), separated by SDS gel electrophoresis and immunoblotted with anti-cyclin D1, anti-p21 and anti-p27 antibodies. (d) tsbn462 cells were labelled with 35 S-methionine at C for 1 h, washed with TD, and chased with DME medium containing 10% serum at C (open circles) for the indicated times. The intensities of the bands were quantified by Phosphorimager. Cell lysates were prepared and immunoprecipitated with anti-cyclin D1 antibodies as described in Experimental procedures. Separate cultures of tsbn462 cells were shifted to C for 4 h, and then labelled with 35 S-methionine at C for 1 h, washed with TD and chased at C (closed circle). whether cyclin D1 transcription was decreased at C, we transiently transfected a cyclin D1 promoter reporter construct into tsbn462 and BHK21 cells. As measured by this assay, cyclin D1 transcription was decreased at C to 7% of that at C in tsbn462 cells (Fig. 5b), suggesting that the reduction in D-type cyclin RNAs at the restrictive temperature in tsbn462 cells is due to decreased transcription. Next, we studied the level of cyclin D1 protein in tsbn462 and BHK21 cells by immunoblotting (we were unable to measure cyclin D3 protein levels due to lack of suitable antibodies). The level of cyclin D1 in quiescent tsbn462 cells was significantly higher than that in BHK21 cells. Serum stimulation induced an increase in cyclin D1 protein in both cell lines at C and in BHK21 cells at C (Fig. 5c). The level of cyclin D1 protein was not decreased in tsbn462 cells at C, which contrasts with the decline in cyclin D1 RNA at C. To investigate this discrepancy, we studied the stability of cyclin D1 protein in tsbn462 # Blackwell Science Limited Genes to Cells (1996) 1,

8 T Sekiguchi et al. Figure 6 State of Rb phosphorylation in G0-synchronized tsbn462 cells. (a) Immunoprecipitates were prepared from lysates of tsbn462 and BHK21 cells with an anti-rb antibody as described in Experimental procedures, separated by SDS gel electrophoresis and immunoblotted with the same anti-rb antibody. (b) Rb promoter reporter gene assay. Rb promoter luciferase reporter gene plasmids were transfected into tsbn462 and BHK21 cells and assayed for promoter activity at C and C as described in Fig. 4b. The ratio of the average activity value at C divided by that at C is plotted. cells at both temperatures by pulse-chase analysis (Fig. 5d). The half-life of cyclin D1 at C was less than 30 min. However, when we incubated tsbn462 cells at C for 4 h and labelled with 35 S-methionine thereafter, the half-life of cyclin D1 protein was greater than 2 h. Therefore, the relatively constant level of cyclin D1 protein in tsbn462 cells at C maybe due to the stabilization of cyclin D1 protein at this temperature. CDK activity can be controlled, not only by the level of its partner cyclins, but also by CDK inhibitors. For this reason, we determined the levels of the p21 and p27 CDK inhibitors in tsbn462 cells (we could not detect p16 in BHK21 cells (data not shown)). Using a promoter reporter gene assay, we found that p21 transcription was increased about twofold at C in tsbn462 cells, but not in BHK21 cells (Fig. 5b). Consistent with the promoter assay, the level of p21 protein was increased in serum-simulated tsbn462 cells at C, whereas there was no increase in BHK21 cells at C or in either cell type at C (Fig. 5c). The level of the p27 CDK inhibitor was high in quiescent cells, and did not change upon serum stimulation at either temperature in tsbn462 or BHK21 cells. This suggests that an increased level of p21 protein, perhaps in combination with p27, could play a role in inactivating G1 cyclin/cdk kinases in tsbn462 cells at C. The Rb protein is a major substrate of cyclin D/CDK complexes in G1 (Kato et al. 1993). As one means of assessing whether cyclin D/CDK kinases were active at C in tsbn462 cells, the state of Rb phosphorylation was studied. In serum-stimulated tsbn462 cells, Rb phosphorylation was induced at C but not at C (Fig. 6a), whereas Rb phosphorylation was induced at both temperatures in BHK21 cells. There was some Rb phosphorylation at C in serumstarved tsbn462 cells, which may result from the presence of a significant level of cyclin D1 in them (Fig. 5c). It was also evident that the level of Rb protein did not increase at C in tsbn462 cells. To test if this was due to decreased transcription of Rb, an Rb promoter reporter gene construct was transiently transfected into tsbn462 cells and BHK21 cells (Fig. 6b). Rb transcription in tsbn462 cells was decreased about threefold at C, whereas it was increased about twofold in BHK21 cells. Thus, the lack of accumulation of Rb at C in tsbn462 cells can be accounted for at least in part by decreased Rb transcription. Essential late G1 genes are not expressed in tsbn462 cells at C following serum stimulation The expression of various cell cycle-regulator genes at the G1/S boundary was examined by Northern analysis and immunoblotting following the serum stimulation of quiescent tsbn462 cells at C and C. The expression of Cdk2 RNA was induced around 13 h after the addition of serum at both temperatures (Fig. 7). In contrast, Cdk2 protein levels were already high in G0 phase cells and remained relatively constant throughout G1 progression. However, the state of Cdk2 694 Genes to Cells (1996) 1, # Blackwell Science Limited

9 Cell cycle arrest in the mutant of CCG1/TAF II 250 Figure 7 The expression of G1 genes in tsbn462 cells at C and C. (a) tsbn462 cells were arrested in G0 phase by growth in 0.25% calf serum for 3 days at C, and then stimulated by addition of serum for the indicated times. Northern blotting analysis was carried out on RNA extracted from these cells using probes for the Cdk2, cyclin A, Cdc2, cyclin B1 and GAPDH genes. 1 L, 2 L, 6 L, 9 L, 13 L and 16 L stand for 1, 2, 6, 9, 13 and 16 h after serum addition at C, respectively. 1H, 2H, 6H, 9H, 13H and 16H stand for 1, 2, 6, 9, 13 and 16 h after serum addition at C, respectively. (b) Protein samples from tsbn462 cells treated as in (a) were prepared, separated by SDS gel electrophoresis and immunoblotted with antibodies against Cdk2, cyclin A, PSTAIR (detects Cdc2 and related proteins) and Cdc25C. 6 L, 9 L, 13 L, 16 L and 22 L stand for 6, 9, 13, 16 and 22 h after serum addition at C, respectively. 6H, 9H, 13H, 16H and 22H stand for 6, 9, 13, 16 and 22 h after serum addition at C, respectively. (c) BHK21 cells were arrested in G0 phase, and stimulated with serum as described in panels (a) and (b). Protein extracts were prepared, separated by SDS gel electrophoresis and immunoblotted with anti-pstair, anti-cyclin A and anti-cdc25c antibodies. 6 L, 9 L, 13 L, 16 L and 22 L stand for 6, 9, 13, 16 and 22 h after serum addition at C, respectively. 6H, 9H, 13H, 16H and 22H stand for 6, 9, 13, 16 and 22 h after serum addition at C, respectively. (d) Immunoprecipitates from lysates of tsbn462 cells stimulated with serum for different times as described in panels were prepared with anti-cdk2 and anti-cyclin E antibodies as described in Experimental procedures and were used to measure Cdk2- and cyclin E-associated kinase activities using histone H1 as a substrate. The graphs to the right of the figure show the amounts of 32 P in histone H1, as measured using a Phosphorimager. # Blackwell Science Limited Genes to Cells (1996) 1,

10 T Sekiguchi et al. Figure 8 Expression of cyclin A in tsbn462 cells. (a) tsbn462 cells were arrested in G0 phase by growth in 0.25% calf serum at C, and then stimulated by addition of serum for the indicated times (top line) and shifted to C at the indicated times (right hand side). Protein samples were prepared, separated by SDS gel electrophoresis and immunoblotted with anti-cyclin A antibodies. (b) Cyclin A promoter reporter gene assay. Cyclin A promoter luciferase reporter gene plasmids were transfected into tsbn462, tsbn462(ccg1), SV40 T antigen transformed tsbn462, and BHK21 cells and assayed for promoter activity at C and C as described in Fig. 4b. The ratio of the average activity value at C divided by that at C is plotted. phosphorylation changed at C, but not at C, with the Thr160 phosphorylated form of Cdk2, which is the lower band in the 33 kda doublet (Gu et al. 1992), being detected between 13 and 16 h at C. Cdc2, cyclin A and cyclin B1 RNAs were present at low but detectable levels up to 9 h after the addition of serum, and were then rapidly induced at C. In contrast, no induction of any of these RNAs was observed at C (Fig. 7a). Anti-PSTAIR antibodies, which recognize Cdc2 and Cdc2-related proteins that have the canonical PSTAIR sequence, detected a 34 kda protein ( p34) in tsbn462 and BHK cells. Since p34 Cdc2 is the most abundant 34 kda protein detected by the anti-pstair antibodies, p34 is presumably mainly Cdc2. p34 levels were high in G0 cells, but increased further at the G1/S boundary in tsbn462 cells grown at C, but not in those grown at C. Cyclin A and Cdc25C proteins accumulated at C, but were not detected at C in tsbn462 cells (Fig. 7b). As a control, protein samples from BHK21 cells were prepared after growth under similar conditions, and were immunoblotted with anti-pstair, cyclin A and Cdc25C antibodies. This analysis showed that cyclin A, Cdc25C and p34 proteins were induced at both temperatures (Fig. 7c). To test whether cyclin/cdk kinase activities were influenced by the CCG1/TAF II 250 mutation, we measured Cdk2- and cyclin E-associated histone H1 kinase activities (Fig. 7d). Cdk2- and cyclin E-associated kinase activities were not induced at C following serum stimulation of quiescent tsbn462 cells, but were induced at C. Since cyclin E is a G1 cyclin and is essential for cell cycle progression (Ohtsubo et al. 1995), the lack of induction of cyclin E- and Cdk2-associated kinase activity at C could be a cause of G1 arrest in tsbn462 cells. Moreover, cyclin A-associated kinase activity was not induced (see below), because cyclin A itself was not induced at C. Expression of a cyclin A promoter reporter gene was significantly reduced at C (Fig. 8b), indicating that the decrease in cyclin A RNA expression is due to reduced transcription, a conclusion consistent with the reported defect in cyclin A promoter activity in ts13 cell extracts at the restrictive temperature (Wang & Tjian 1994). However, cyclin A protein was synthesized normally starting at 10 h when the cells were incubated for 5 h at C followed by a shift up to C (Fig. 8a), yet this protocol did not allow tsbn462 cells to enter S phase (Fig. 2c). This suggests that cyclin A expression is not sufficient for the G1-to-S transition in these cells and that gene products in addition to cyclin A are needed for entry into S phase, although it is possible that the increase in p21 at C could block these cells in G1 despite the accumulation of cyclin A (Fig. 5c). Since cyclin A synthesis was induced normally in cells that had been at C for 5 h, this experiment also suggests that CCG1/TAF II 250 does not regulate cyclin A transcription directly, but rather regulates factors needed for its induction that are expressed during the first 5 h after serum stimulation. 696 Genes to Cells (1996) 1, # Blackwell Science Limited

11 Cell cycle arrest in the mutant of CCG1/TAF II 250 Figure 9 Cyclin/CDK kinase activity in exponentially growing tsbn462 cells. (a) Northern blotting analysis cells from tsbn462 and BHK21 cells were seeded on to six dishes. The next day cells were shifted up to C at the indicated time points and collected. RNAs were isolated and probed with: (1) cyclin D1; (2) cyclin D3; (3) GAPDH. One-tenth volume of the samples (about 10 5 cells) were used. Times at C in h are shown at the top of the panel. (b) Immunoblotting analysis. Cells as prepared as in (a) were collected and boiled with sample buffer. One tenth volume of sample (about 10 5 cells) were electrophoresed and transferred onto a nitrocellulose membrane. The filters were blotted with the indicated antibodies. Times at C in h are shown at the top of the panel. (c) Cdk2-, cyclin A-, E- and Cdc2-associated histone HI kinase activity. Lysates of tsbn462 and BHK21 cells prepared from the indicated time points at C were immunoprecipitated with anti-cyclin A, anti-cyclin E, anti-cdk2 or anti-cdc2 antibodies. Immunoprecipitates were assayed for histone HI kinase activity as described in Experimental procedures. (d) Protein samples prepared as in (a) were processed as described in Fig. 6 to determine the state of Rb phosphorylation. (e) Transcriptional activity of the cyclin D1 promoter was evaluated by measuring the luciferase activity of a mixed population of stable transformants of the cyclin D1 promoter-luciferase vector in growing tsbn462 cells (closed circle) and BHK21 cells (open circle) after temperature shift to C for the indicated times. From these experiments, it appeared likely that the expression during the late stage of G1 phase of at least one gene essential for entry into S phase is dependent on the activity of CCG1/TAF II 250, although this gene appears not to be cyclin A itself. Indeed, the lack of cyclin A expression at C in tsbn462 cells resulting from a G1 arrest could be caused by a defect in expression earlier in G1 of a gene essential for cyclin A induction. To test this idea, we expressed SV40 large T antigen in tsbn462 cells, which is known to obviate the need for expression of some early cell cycle regulators such as cyclin D. In an SV40 T antigen-expressing tsbn462 cell line cyclin A promoter activity was completely restored at C, as it was in tsbn462 # Blackwell Science Limited Genes to Cells (1996) 1,

12 T Sekiguchi et al. cells expressing wild-type CCG1, ts + tsbn462 (CCG1) cells (Fig. 8b). Therefore, reduced cyclin A expression at C appears to be a consequence of a G1 arrest caused by a failure to express another G1 gene, such as cyclin D1, and by an increase in the level of the cyclin/ CDK inhibitors p21 and p27. Association of cyclin/cdk kinase with the cyclin-dependent kinase inhibitors at C Logarithmically growing tsbn462 cells arrest in G1 phase when shifted to C. To investigate the cause of this arrest we determined the effect of shifting growing tsbn462 cells to C on the levels of cyclins A, D1 and D3, and the CDK inhibitors p21 and p27. After the shift to C, cyclin D1 and D3 RNA levels gradually decreased in tsbn462 cells, whereas they increased in BHK21 cells (Fig. 9a, panels 1 and 2). To confirm that cyclin D1 promoter activity was decreased at C, we established pools of tsbn462 and BHK21 cells harbouring a cyclin D1 promoter luciferase reporter gene (Fig. 9e). Incubation of the tsbn462 cyclin D1-luciferase cells at C for only 5 h resulted in a sharp 90% decrease in luciferase activity, which then stayed at a low level. In the BHK21 cyclin D1-luciferase cells, cyclin D1 promoter activity initially decreased to 40%, but recovered thereafter. Cyclin D1 protein levels initially increased in tsbn462 cells shifted to C but then fell, whereas they continued to rise in BHK21 cells (Fig. 9b, panel 1). The fact that levels of cyclin D1 protein did not parallel those of cyclin D1 RNA could be explained by the stabilization of cyclin D1 protein at C (Fig. 5d). p21 protein levels increased about sevenfold, 4 h after the shift to C in tsbn462 but not in BHK21 cells (Fig. 9b, panel 4; Fig. 10b, right panel), which is similar to the timing of the increase observed after serum stimulation at C (Fig. 5c). These results coincide with results of the p21 promoter assay, which showed that there was increased transcription at C in tsbn462 cells (Fig. 5b). p27 protein levels also increased at 15 h after the shift of growing cells to C in tsbn462 cells (Fig. 9b, panel 5). Since the p21 CDK inhibitor binds A-, E- and D- type cyclin/cdk complexes and inhibits their activity, the induction of p21 would be expected to result in the loss of cyclin A/Cdk2, cyclin E/Cdk2 and cyclin D1/ CDK kinase activity. Consistent with this idea, we found that Rb phosphorylation was rapidly inhibited in tsbn462 cells after the shift to C (Fig. 9d), which is an indication that cyclin D1-associated kinase activity was decreased at this restrictive temperature in tsbn462 cells. Moreover, cyclin A-, E- and Cdk2- associated kinase activities, as measured by histone H1 phosphorylation, were decreased about fivefold after 8 h and more than 20-fold after 24 h at C, in tsbn462 but not in BHK21 cells (Fig. 9c, panels 1, 2, 3). Since dephosphorylation of Thr-160 of Cdk2 was not detected in kinase assay samples at C in tsbn462 cells (data not shown), p21 might play a role in inactivating Cdk2-associated kinase activity. p27, which was induced after 15 h at C in tsbn462 cells, when the cells showed G1 cell cycle arrest (Sekiguchi et al. 1995), may collaborate with p21 to inhibit G1 cyclin/cdk kinase activities in tsbn462 cells. Consistent with this model, the activity associated with cyclin/cdc2 complexes, which function in G2 and are not inhibited by p21 or p27, was not decreased until 24 h at C (Fig. 9c, panel 4). Finally, we analysed the association of p21 and p27 with cyclin/cdk complexes in growing tsbn462 cells at C and C (Fig. 10). Cyclin D1 was found to be associated with both p27 and p21 (Fig. 10a and b). The level of p21 bound to cyclin D1 was increased about fourfold after 8 h at C, compared to that at C (Fig. 10a and b). In contrast, there was no increase in the level of p27 bound to cyclin D1 at C. As observed for cyclin D1, Cdk2 immunoprecipitates from 35 S-labelled cells contained increased p21 at C (Fig. 10c, left panel). This was confirmed by releasing 35 S-labelled proteins from Cdk2 immunoprecipitates and then reprecipitating with anti-cdk2 and p21 antibodies (Fig. 10c, middle panel). Consistent with the increase in cyclin/cdk-associated p21 at C, an increase in total p21 was detected at C, both by direct immunoblotting of cell lysates (Fig. 9b, panel 4) or immunoblotting of anti-p21 immunoprecipitates (Fig. 10c, right panel). Moreover, when boiled lysates were added to a lysate of cells grown at C, followed by immunoprecipitation of Cdk2 and an H1 kinase assay, the C cell lysate was found to contain a higher level of heat stable Cdk2 inhibitors (i.e. p21 plus p27) than the C cell lysate (Fig. 10d). Discussion The tsbn462 BHK21 mutant cell line arrests specifically in G1 when shifted to C, and yet the causative mutation is in the CCG1/TAF II 250/p250 protein, which is a core component of TFIID, a general transcription factor for RNA polymerase IIdependent transcription. CCG1/TAF II 250 might be expected to be necessary for the expression of almost all RNA polymerase II transcribed genes, 698 Genes to Cells (1996) 1, # Blackwell Science Limited

13 Cell cycle arrest in the mutant of CCG1/TAF II 250 Figure 10 Association of cyclin D1 and Cdk2 with CDK inhibitors, p21 and p27. (a) Protein samples from tsbn462 cells growing exponentially at C and from growing cells shifted to C for 8 h were prepared as described in Experimental procedures and were subjected to immunoprecipitation with anti-cyclin D1 antibodies. The immunoprecipitates were separated by SDS gel electrophoresis and immunoblotted with anti-cyclin D1 (upper panel), anti-p27 (middle panel) and anti-p21 (lower panel) antibodies. (b) BHK21 and tsbn462 cells were labelled with 35 S-methionine for 4 h either at C or at C directly after shifting from C. Protein samples from these cells were prepared as described in Experimental procedures and were immunoprecipitated with anti-cyclin D1 antibodies (left and middle panels) and with anti p21 and p27 antibodies (right panel). (c) tsbn462 cells were labelled with 35 S- methionine for 8 h either at C or at C directly after shifting from C. Protein samples were immunoprecipitated with anti-cdk2 antibodies (left panel). In a separate experiment similar 35 S-labelled immunoprecipitates were boiled with sample buffer to release proteins from the protein A beads, which were then reprecipitated with anti-cdk2 or anti-p21 antibodies as described in Experimental procedures (middle panel). Lysates of tsbn462 cells prepared as in panel (a) were immunoprecipitated with anti-p21 antibodies, separated by SDS gel electrophoresis, and immunoblotted with anti-p21 antibodies (right panel). (d) Cdk2-associated histone H1 kinase activities were assayed on anti-cdk2 immunoprecipitates made from lysates of cells grown at C and cells grown at C and shifted to C for 8 h. Quantification was done using a Phosphorimager. Cells were lysed in H1 kinase buffer as described in Experimental procedures. Boiled extracts were prepared by boiling cell extracts from C and C for 5 min, followed by centrifugation at g for 10 min. One volume of lysate of cells grown at C was mixed with an equal volume of boiled extract from cells grown at C (33.5(B)) or C (39.5(B)) and incubated at room temperature for 30 min. Subsequently Cdk2 was immunoprecipitated, and the immunoprecipitates were assayed for histone H1 kinase activity. Right panel shows the autoradiogram. # Blackwell Science Limited Genes to Cells (1996) 1,

14 T Sekiguchi et al. and therefore inactivation of CCG1/TAF II 250 should lead to random arrest in all phases of the cell cycle. However, we show here that the mutation of Gly 690 to Asp 690 in CCG1/TAF II 250 resulted in differential effects on gene expression in G1; some genes like the D-type cyclins and cyclin A were not induced by serum treatment of quiescent tsbn462 cells at C, whereas early G1 phase genes like c-fos, c-jun and c- myc were induced normally. The CCG1/TAF II 250 Gly 690 -to-arg 690 mutation in hamster tsbn462 and ts13 cells occurs at a conserved Gly in human and Drosophila CCG1/TAF II 250 s (Hayashida et al. 1994) and is located in a well-conserved region in the yeast, Drosophila and human proteins (Noguchi et al. 1994). Although this region does not have any recognizable motifs, it could be a site for protein protein interaction. In the TFIID complex, CCG1/TAF II 250 has been shown to interact directly with TBP (Hisatake et al. 1993; Ruppert et al. 1993), potentially via the N- and C-terminal CCG1/TAF II 250 regions (Kokubo et al. 1993), TAF II 55 (Chiang & Roeder 1995), TAF II 30a, TAF II 30b, TAF II 60, TAF II 110, TAF II 150 (Chen et al. 1994). There are two possible explanations for the specific G1 arrest in tsbn462 cells. One is that tsbn462 CCG1/ TAF II 250 is only partially inactivated at C and thus only some genes are affected. A less likely possibility is that CCG1/TAF II 250 is inactivated completely, but is only required for the transcription of some genes, such as the D-type cyclins, and/or for the repression of a subset of genes such as c-myc. Since CCG1/TAF II 250 is a large multidomain protein, it would not be surprising if the point mutation affected only one of the many interactions that CCG1/TAF II 250 can make. For instance, the CCG1/TAF II 250 mutation might prevent the incorporation of a TAF into the TFIID complex, or affect the interaction of CCG1/TAF II 250 with a transcription factor and/or coactivator, either of which might be responsible for the altered pattern of gene expression. The CCG1/TAF II 250 protein persisted at C in tsbn462 cells, and, with one exception, was complexed with the same proteins as CCG1/TAF II 250 at the permissive temperature. Therefore, although we do not have reagents to determine if all the hamster TAFs are present, it seems likely that the TFIID complex is properly assembled at C and it is plausible that the mutant CCG1/TAF II 250 retains significant activity. However, the 90 kda protein associated with CCG1/ TAF II 250 in tsbn462 cells at C might affect some of the activities of the TFIID complex by competing for access to sequence-specific transcription factors. Based on its size and ability to associate with other ts mutant proteins, the 90 kda protein could be a member of the hsp90 heat shock protein family that recognizes and binds denatured proteins. The key to understanding the defect in the mutant CCG1/TAF II 250 in tsbn462 cells is to determine which G1 genes are directly affected by the mutation in CCG1/TAF II 250 in tsbn462 cells, and which genes are affected indirectly as a result of cell cycle arrest caused by the tsbn462 mutation. Thus, it seems likely that the reduced cyclin A expression at C is an indirect result of the G1 arrest, since cyclin A expression occurred at C when SV40 T antigen was expressed in tsbn462 cells, which will override the need for cyclin D/Cdk function by inactivating Rb. The earliest defect in G1 cell cycle gene expression that we have detected in tsbn462 cells at the restrictive temperature is a decrease in cyclin D1 mrna accumulation. This occurs at the transcriptional level, as indicated by decreased cyclin D1 promoter reporter gene activity in tsbn462 cells at C. We do not know, however, whether the reduction in cyclin D1 promoter activity is a direct consequence of the CCG1/TAF II 250 mutation. Preliminary analysis of cyclin D1 promoter DNA binding proteins indicates that even cyclin D1 may not be a direct target, since we observed the loss of specific gel shift bands using a cyclin D1 promoter fragment with nuclear extracts of tsbn462 cells grown at C and a decrease in cyclin D1 expression by overexpression of p21 at C in BHK21 cells (T.S., unpublished observation). Thus, the mutation in CCG1/TAF II 250 could affect the expression of transcription factors such as Rb (see below) that are induced in early G1 and are needed for cyclin D1 promoter function. The cyclin D1 promoter lacks a conventional TATA element, but contains an initiator element (Inr) similar to that in the adenovirus major late promoter, and TFIID may be required for Inr function on this promoter (Roy et al. 1993). The cyclin D1 promoter also contains two Sp1 binding sites, a CRE, a TRE, an octamer binding site and an E2F site (Herber et al. 1994; Muller et al. 1994; Philipp et al. 1994). Sp1, AP-1 and CREB were present at normal levels at C in tsbn462 cells, but Oct-1 DNA-binding activity was decreased (T.S., unpublished observation), suggesting that the expression of Oct-1 requires CCG1/TAFII250. The fact that Sp1 levels were normal at C, yet Sp1-mediated transcription is reportedly decreased in vitro in ts13 cell extracts (Wang & Tjian 1994) might suggest that an interaction between Sp1 and CCG1/ TAFII250 is affected by the mutation. This defect, 700 Genes to Cells (1996) 1, # Blackwell Science Limited

15 Cell cycle arrest in the mutant of CCG1/TAF II 250 combined with the loss of Oct-1, could contribute to the decreased cyclin D1 transcription at C. Rb may also play a role, since Rb activates the cyclin D1 promoter (Muller et al. 1994). Reduced cyclin D1 expression at C could be due to the observed lack of Rb induction (Fig. 6a and 9d) or to a failure of Rb to activate cyclin D1 transcription. A direct interaction of Rb and CCG1/TAF II 250 has been reported, which may explain the partial abrogation of Rb-stimulated Sp1-mediated transcription at the restrictive temperature in ts13 cells (Shao et al. 1995). Another cyclin D1 regulator is c-myc, which represses cyclin D1 transcription when expressed constitutively (Philipp et al. 1994), possibly through competing with USF (upstream stimulatory factor) binding to the Inr. Since c-myc is induced at C, this could actively repress cyclin D1 transcription. Thus, several effects could combine to prevent the induction of cyclin D1 at C. We noted that some genes, such as p21 sdi1/cip1/waf1 and c-myc, are rapidly induced after a temperature shift in tsbn462 cells. p21 can be induced in a p53-dependent or p53-independent manner (El-Deiry et al. 1993; Michieli et al. 1994; Sheikh et al. 1994). We know that p21 can be induced in a p53-independent manner in tsbn462 cells, because it was induced in tsbn462 cells expressing SV40 T antigen, which binds to and inactivates p53 (Sekiguchi et al. 1995) with the same kinetics and to the same level as in parental tsbn462 cells. However, this does not rule out the possibility that p53-dependent induction of p21 occurs in the parental tsbn462 cells. c-myc was also induced upon temperature-shift in tsbn462 cells, and it is possible that there is a common mechanism of induction. We do not know whether these inductions reflect a decrease in the expression of repressor proteins that negatively regulate the p21 and c-myc promoters or whether this is a gainof-function property of the mutant CCG1/TAF II 250. What is the mechanism underlying G1 arrest in tsbn462 cells? Cyclin D1 is essential for the progression through G1 in fibroblasts, and the overexpression of cyclin D1 ( Jiang et al. 1993; Quelle et al. 1993; Resnitzky et al. 1994) and cyclin E (Ohtsubo & Roberts 1993; Resnitzky et al. 1994; Wimmel et al. 1994) accelerates G1 phase in fibroblasts. Cyclin D1 may be limiting in tsbn462 cells at C due to its reduced expression, because tsbn462 cells are Rb-positive cells and require cyclin D1 expression for the G1-to-S phase transition. Moreover, the induction of the p21 and p27 CDK inhibitors at C, in combination with the reduced expression of cyclin D1, could explain the observed decrease in cyclin D1 kinase activity and Rb phosphorylation in tsbn462 cells at C. Although Fig. 6a shows that Rb synthesis was also decreased at C, the unphosphorylated form of Rb was present. There is a possibility that this low level of unphosphorylated Rb inhibits cell cycle progression. The ability of p21 and p27 to inhibit cyclin A/Cdk2 and cyclin E/Cdk2 kinase activity could account for the fact that even when tsbn462 cells were able to express cyclin A in a delayed shift-up experiment, they were still unable to enter S phase (Fig. 2c), since cyclin A kinase activity which is required for the initiation of S phase (Girard et al. 1991; Pagano et al. 1992) might be inhibited by the elevated levels of p21 and p27 in these cells. Thus, tsbn462 cells may arrest in G1 at C because of reduced D-type cyclin expression and increased p21 and p27 expression, which together result in a reduction of the G1 cyclin/cdk activities and Rb phosphorylation that are essential for the progression through G1 into S phase. The effects of temperature shift on gene expression in tsbn462 cells were not identical in serum-stimulated G0- arrested cells and growing cells. In G0-arrested cells, the induction of cell cycle regulators such as cyclin D1 and cyclin A was decreased at C, and in growing cells the transcription of these genes was reduced after temperature shift. Moreover, p21 protein levels were low at C in both growing cells and G0-arrested cells and increased after serum stimulation at C and temperature-shift, respectively. In contrast, while p27 protein levels were high in G0-arrested cells at C, as noted in other cell types (Nourse et al. 1994), and did not increase upon temperature shift, p27 levels were low at C in growing cells and were increased upon temperature-shift, possibly as a result of cell cycle arrest. In conclusion, the tsbn462 G1 arrest is probably not due to a single factor, but is a combination of events which include a decrease in cyclin D1/Cdk kinase activity and in Rb phosphorylation due to decreased cyclin D1 and increased p21/p27 expression, which in turn leads to reduced cyclin E/Cdk2 kinase activity and reduced cyclin A expression and cyclin A/Cdk2 kinase activity as judged both by H1 kinase assay and by Thr160 phosphorylation. The fact that ectopic expression of cyclin D1 did not rescue S phase entry (data not shown) also supports the conclusion that a combination of events rather than a single factor is involved. Experimental procedures Cell lines and cell culture conditions The tsbn462 cell line is a ts mutant of the BHK21/13 cell line (Nishimoto et al. 1982). tsbn462, SV40 T antigen-expressing # Blackwell Science Limited Genes to Cells (1996) 1,