Role of Gadd45a in the Density-Dependent G1 Arrest Induced by P27kip1

Role of Gadd45a in the density-dependent G1 arrest induced by p27Kip1

Xiaohong Zhang1,2,LeMa1,2, Steven A Enkemann1,2 and WJ Pledger*,1,2

1Molecular Oncology Program, H. Lee Mofﬁtt Cancer Center and Research Institute, 12902 Magnolia Drive, Tampa, FL 33612, USA; 2Department of Interdisciplinary Oncology, University of South Florida College of Medicine, Tampa, FL 33612, USA

p27Kip1, an inhibitor of cyclin-dependent kinases, is an multiorgan hyperplasia and are larger than their control important regulator of cell cycle progression. We have littermates (Fero et al., 1996; Kiyokawa et al., 1996; previously shown that p27Kip1 inhibits the G0 to S Nakayama et al., 1996). Conversely, p27Kip1 decreases in transition when ectopically expressed in p27-47 mouse abundance when quiescent cells are exposed to mitogens fibroblasts arrested at high but not low densities. In the or replated at lower densities (Firpo et al., 1994; Nourse study described here, we identify Gadd45a, a member of et al., 1994; Agrawal et al., 1996; Winston et al., 1996). the growth arrest- and DNA damage-inducible family of Conditions that prevent p27Kip1 downregulation prevent proteins, as a potential mediator of the density-dependent cell cycle re-entry; such conditions include ectopic effects of p27Kip1 on cell proliferation. Gadd45a mRNA expression of p27Kip1 and treatment of T cells and and protein were more abundant in p27-47 cells arrested macrophages with rapamycin and cyclic AMP agonists, at high densities than at low densities. Amounts of both respectively (Kato et al., 1994; Nourse et al., 1994; decreased and remained low when cells arrested at high Toyoshima and Hunter, 1994; Zhang et al., 2000). densities were exposed to mitogens in the absence, but not Depending on the system, p27Kip1 expression is modu- in the presence, of ectopically expressed p27Kip1. Impor- lated at the transcriptional, translational, and/or post- tantly, enforced expression of Gadd45a prevented density- translational level (Pagano et al., 1995; Agrawal et al., arrested mouse fibroblasts from initiating DNA synthesis 1996; Hengst and Reed, 1996; Millard et al., 1997; in response to mitogens. We suggest that amounts of Nguyen et al., 1999; Dijkers et al., 2000; Yang et al., Gadd45a above a certain threshold are growth inhibitory 2001). and that such amounts are achieved in cells arrested at The major targets of p27Kip1 are members of the CDK high but not low densities. For cultures arrested at high family of nuclear serine/threonine kinases. CDKs densities, the resumption of cell cycle traverse requires a associate with cyclins, which are expressed periodically, sustained reduction in Gadd45a abundance, a process that to form catalytically active complexes that orchestrate is induced by mitogens and inhibited by p27Kip1. the passage of cells through the cell cycle (Pines, 1995). Oncogene (2003) 22, 4166–4174. doi:10.1038/sj.onc.1206599 p27Kip1 interacts with and inhibits the activity of complexes containing the D cyclins and cdk4, cyclin E Keywords: Gadd45a; p27Kip1; culture density and cdk2, and cyclin A and cdk2. When active, these complexes mediate G0/G1 progression and S phase entry. Although best known as a CDK inhibitor, p27Kip1 Kip1 Introduction may have additional functions. We found that p27 antisense oligonucleotides blocked the suspension-in- p27Kip1 is a ubiquitously expressed member of the Cip/ duced differentiation of primary keratinocytes without Kip family of cyclin-dependent kinase (CDK)inhibitors resulting in CDK activation (Hauser et al., 1997). (Polyak et al., 1994; Toyoshima and Hunter, 1994). Its Inhibition of intestinal cell differentiation (but not Kip1 primary function is to induce and maintain the quiescent restoration of CDK activities)by p27 antisense state in response to antiproliferative signals such as cDNA has also been reported (Deschenes et al., 2001). Kip1 mitogen deprivation and cell–cell contact (Sherr and Moreover, in this system, ectopic expression of p27 Roberts, 1995). p27Kip1 accumulates in cells approaching induced differentiation, whereas ectopic expression of Cip1 quiescence (Coats et al., 1996; Dong et al., 1998a), and the related CDK inhibitor, p21 , did not. Miskimins Kip1 cells lacking p27Kip1 do not readily growth arrest. For et al. (2002)showed that p27 increased the activity of example, cells receiving p27Kip1 antisense oligonucleo- the promoter for myelin basic protein when expressed in Kip1 tides or expressing p27Kip1 antisense cDNA continue to oligodendrocytes. This action of p27 was not cycle when placed in mitogen-depleted medium (Coats duplicated by the cdk2 inhibitor, roscovitine, and thus et al., 1996; Rivard et al., 1996). Moreover, because of apparently was unrelated to CDK inactivation. Other Kip1 inappropriate cell proliferation, p27Kip1-null mice exhibit studies demonstrated an association between p27 and Jab1, a coactivator of the c-Jun and JunD transcription factors (Tomoda et al., 1999). Conditions in which cdk2 *Correspondence: WJ Pledger, H. Lee Moffitt Cancer Center and Kip1 Research Institute, 12902 Magnolia Drive, Tampa, FL 33612, USA; was not active favored p27 –Jab1 interaction, and E-mail: pledgerw@moffitt.usf.edu formation of these complexes resulted in their transloca- Received 25 July 2002; revised 12 March 2003; accepted 21 March 2003 tion from the nucleus to the cytoplasm. In anergic T Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4167 cells, the presence of p27Kip1–Jab1 complexes correlated (PDGF)and serum in the presence or absence of IPTG with reduced Jun-dependent promoter activity and for 8 h (microarray analysis)or 18 h (cell cycle analysis). reduced interleukin-2 expression (Boussiotis et al., The cell cycle analysis verified the effectiveness of the 2000). Thus, in addition to regulating CDK activity, IPTG treatment, and the results of two separate p27Kip1 may also play a role in gene expression, and experiments are presented in Figure 1. In both experi- changes in gene expression elicited by p27Kip1, either ments, over 60% of the cells in the control populations directly or indirectly, may contribute to such processes were in S phase 18 h after addition of PDGF and serum. as differentiation and proliferation. In contrast, populations stimulated in the presence of In a previous study, we prepared a Balb/c-3T3 mouse IPTG contained less than 15% S phase cells, as did fibroblast cell line (termed p27-47)that ectopically quiescent populations. Thus, in these experiments, expresses p27Kip1 in response to isopropyl-b-d-thioga- IPTG treatment and consequent p27Kip1 overexpression lactoside (IPTG)(Zhang et al., 2000). Surprisingly, efficiently prevented mitogen-treated cells from initiat- subconfluent p27-47 cells grew normally in IPTG- ing DNA synthesis. containing medium, despite marked reductions in cdk4 We chose the 8 h (i.e., a mid-G0/G1)time point for and cdk2 activities. Differences in growth rate became the microarray analysis to allow time for p27Kip1 apparent as cells approached confluency, with the induction to effect changes in gene expression and to IPTG-treated cultures doubling more slowly and reach- exclude the numerous cell cycle-dependent genes ex- ing lower saturation densities than the control cultures. pressed in late G1 from the analysis (DeGregori et al., Thus, the capacity of cells to grow in the presence of 1995). Western blots confirmed that p27Kip1 was more supraphysiological amounts of p27Kip1 decreases as abundant in IPTG-treated cultures than in control culture density increases. Moreover, the degree of cultures at 8 h (data not shown). The microarray confluency, rather than confluency itself, determines analysis was done twice, using the Mu11 K and whether cells are susceptible or refractory to the MG74A GeneChip arrays in experiments 1 and 2, antiproliferative actions of p27Kip1. When exposed to respectively, which correspond to the cell cycle analyses mitogens and IPTG, cells arrested at lower saturation shown in Figure 1. The Mu11 K arrays contain targets densities efficiently entered the S phase, whereas cells arrested at higher saturation densities remained in G0/G1. G0/G1 S G2 + M The data described above suggest that cells at high density require a mitogenic signal(s)that is not required 100 _ experiment 1 by cells at low density and that this signal does not occur when amounts of p27Kip1 exceed a certain threshold. This 80 _ signal may involve the expression of a gene product that stimulates proliferation or the loss of a gene product 60 _ that inhibits proliferation. The studies presented here show that the stress-inducible protein Gadd45a accu- 40 _ mulates in cells arrested at high densities and that amounts of Gadd45a mRNA and protein decrease and 20 _ remain low in cells stimulated in the absence, but not in the presence, of ectopic p27Kip1. We also find that ectopic experiment 2 expression of Gadd45a prevents density-arrested cells 100 _ from initiating DNA synthesis. We suggest that p27Kip1 inhibits the G0 to S phase progression of cells arrested at 80 _ high densities by maintaining Gadd45a at levels that are PERCENT CELLS nonpermissive for proliferation. 60 _

40 _ Results 20 _ Regulation of Gadd45a expression by p27Kip1 Microarray analysis was carried out to identify genes 0 hr 0 hr + 18 hr 18 hr + whose expression is modulated in a p27Kip1-dependent IPTG IPTG manner in conditions in which p27Kip1 is growth Figure 1 Inhibition of S phase entry by ectopic p27Kip1. p27-47 inhibitory. p27-47 cells, which ectopically express cells were grown to conﬂuency in medium containing 10% serum p27Kip1 in the presence but not in the absence of IPTG, and treated with or without 1 mm IPTG for 18 h. Cells were then were grown to conﬂuency in medium containing 10% harvested (G0, G0 þ IPTG)or incubated for 18 h in medium serum. As a result, cells became quiescent at high containing 20 ng/ml PDGF and 10% serum (18 h)or 20 ng/ml PDGF, 10% serum, and 1 mm IPTG (18 h þ IPTG). The percen- densities, which rendered them responsive to the tages of cells in G1, S, and G2/M were determined by FACS Kip1 antiproliferative effects of p27 . Density-arrested cells analysis. Experiments 1 and 2 correspond to experiments 1 and 2 in were stimulated with platelet-derived growth factor Table 1

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4168 Table 1 Identiﬁcation of p27Kip1 gene targets by microarray analysis a GENE Effect of ectopic expression of p27Kip1 IPTG: hr: 0 2 8 18 0 28 18 growing Mu11 K MG-74A (experiment 1) (experiment 2) Gadd45α mRNA

Rrm2 3.2 2.7 À À 28S PTX3 4.8 3.2 18S cdc2 À5.5 À5.3 gadd45a 3.2 6.4 1234568 7 910

p27-47 cells were grown to conﬂuency in medium containing 10% b 8 hr + serum and treated with or without 1 mm IPTG for 18 h. Cultures were 0 hr 8 hr IPTG then refed with medium containing 20 ng/ml PDGF and 10% serum, and IPTG was readded to IPTG-pretreated cultures. Cells were cyclin D1 mRNA harvested 8 h after stimulation, and total mRNA was isolated. The experiment was done twice, and two separate microarray analyses were performed using the Mu11 K and MG-74A GeneChip arrays. The table GAPDH mRNA lists the genes whose expression was increased or decreased more than twofold by p27Kip1 expression in both analyses. Experiments 1 and 2 correspond to experiments 1 and 2 in Figure 1 c IPTG IPTG

Gadd45 0 4 812 4 8 12 hr from the Unigene data base for B11 000 mouse genes, and the MG-74A arrays are an updated and overlapping Gadd45α version of the Mu11 K arrays. Table 1 lists the four

genes whose expression was increased (gadd45a, PTX3) p27Kip1 or decreased (cdc2, Rrm2)more than twofold in IPTG- treated versus control cultures in both analyses. The 1234568 7 Rrm2 gene encodes a subunit of ribonucleotide reduc- Figure 2 Effects of p27Kip1 induction on Gadd45a expression in tase, an enzyme involved in DNA replication (Thelander growing and arrested p27-47 cells. (a)Lanes 1–8: p27-47 cells were and Berg, 1986), and the protein product of the PTX3 grown to confluency in medium containing 10% serum and pretreated with or without 1 mm IPTG for 18 h. Cells were refed gene plays a role in inflammation (Breviario et al., 1992). with medium containing 20 ng/ml PDGF and 10% serum, and The remaining genes encode the cell cycle-regulatory IPTG was readded to the IPTG-pretreated cultures. Cells were proteins, cdc2 and Gadd45a. cdc2 is a CDK that harvested at the times indicated. Lanes 9 and 10: exponentially combines with cyclin B to signal entry into mitosis growing p27-47 cells received 1 mm IPTG or no addition for 18 h. All lanes: amounts of Gadd45a mRNA were determined by (Nurse, 1990), whereas Gadd45a arrests cells in G1 or Northern blotting. To ensure that all the lanes contained equal G2 (Fornace et al., 1988; Fan et al., 1999; Wang et al., amounts of RNA, the agarose gel was stained with ethidium 1999; Zhang et al., 2001). As we were interested in bromide for detection of 28S and 18S rRNA. (b)Density-arrested potential targets of p27Kip1 during G0/G1, we selected p27-47 cells were harvested (0 h)or stimulated for 8 h in medium Gadd45a for further analysis. containing 20 ng/ml PDGF and 10% serum (8 h)or 20 ng/ml PDGF, 10% serum, and 1 mm IPTG (8 h þ IPTG). Amounts of Northern blots were performed to corroborate the cyclin D1 mRNA and GAPDH mRNA (loading control)were microarray data and to examine the kinetics of Gadd45a quantitated by RNase protection assay. (c)Density-arrested p27-47 mRNA expression in control and IPTG-treated p27-47 cells were stimulated with PDGF and serum in the presence or cells. Density-arrested cultures pretreated with or with- absence of IPTG for the indicated times. Amounts of Gadd45a and p27Kip1 were determined by Western blotting. Lane 1: as a marker out IPTG expressed readily detectable amounts of for Gadd45a, Balb/c-3T3 cells were transfected with an expression Gadd45a mRNA (Figure 2a, lanes 1 and 5). Amounts plasmid encoding Gadd45a of Gadd45a mRNA declined within 2 h of addition of PDGF and serum to quiescent cells and were substantially reduced at 8 h (lanes 2 and 3). By 18 h, Gadd45a may reflect the capacity of p27Kip1 to induce events that mRNA levels had partially returned to starting levels increase the synthesis or stability of the Gadd45a (compare lanes 1 and 4). In cells receiving PDGF, transcript. Alternatively, p27Kip1 may inhibit the mito- serum, and IPTG, Gadd45a mRNA levels decreased gen-activated pathways that downregulate Gadd45a initially, but were fully restored to quiescent levels by 8 h expression. We favor the latter hypothesis because (lanes 6 and 7). These findings are consistent with the p27Kip1 did not increase Gadd45a mRNA levels when microarray data and suggest that enforced expression of overexpressed in cycling, subconfluent p27-47 cells p27Kip1 counteracts the decrease in Gadd45a mRNA (Figure 2a, lanes 9 and 10). Amounts of Gadd45a abundance induced by PDGF and serum. The pattern of mRNA were low in cycling cells treated with or without Gadd45a mRNA loss and recovery seen in p27-47 cells IPTG for 18 h and were comparable to those of density- stimulated in the absence of IPTG is similar to that seen arrested cells receiving PDGF and serum for 8 h in quiescent smooth muscle cells exposed to PDGF or (compare lane 3 with lanes 9 and 10). serum (Kettenhofen et al., 2001). To confirm that ectopically expressed p27Kip1 does not The greater abundance of Gadd45a mRNA in cells repress gene expression in general, we examined its stimulated in the presence than in the absence of IPTG effects on cyclin D1 mRNA expression, which is

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4169 upregulated by PDGF in ﬁbroblasts (Winston and a Pledger, 1993). As demonstrated by RNase protection 0 0.5 1 2 4 hr after act D addition assay, amounts of cyclin D1 mRNA were greater in mitogen-stimulated p27-47 cells than in quiescent cells Gadd45α mRNA and were equivalent in cells stimulated with IPTG and cells stimulated without IPTG (Figure 2b). The microarray analyses also showed similar amounts of cyclin D1 28S mRNA in extracts of control and IPTG-treated cells. The effect of p27Kip1 overexpression on the abundance 18S of the Gadd45a protein was determined by Western b blotting. Gadd45a was present in quiescent p27-47 cells act D added at 2 hr (Figure 2c, lane 2), but was almost undetectable in cells stimulated with PDGF and serum for 4, 8, or 12 h (lanes 02 4 4 hr after stimulation 6, 7, 8). Amounts of Gadd45a also declined in cells receiving mitogens and IPTG for 4 h (lane 3). This Gadd45α mRNA decrease, however, was not sustained, and by 12 h, Gadd45a was equally abundant in IPTG-treated cultures and quiescent cultures (compare lanes 2 and 5). 28S Thus, the pattern of Gadd45a protein expression parallels that of Gadd45a mRNA expression in control 18S and IPTG-treated cultures. Western blots conﬁrmed c that p27Kip1 was overexpressed in cells stimulated in the 8 hr + presence of IPTG. 0 hr 8 hr CHX

Gadd45a mRNA is unstable in quiescent and stimulated Gadd45α mRNA cultures When added to quiescent cells, mitogens may repress the 28S transcription of the gadd45a gene or reduce the stability of the Gadd45a message. To assess stability, we 18S monitored Gadd45a mRNA abundance in p27-47 cells Figure 3 Rapid turnover of Gadd45a mRNA in quiescent and exposed to actinomycin D, an inhibitor of mRNA stimulated cultures exposed to actinomycin D. (a–c)p27-47 cells synthesis. When actinomycin D was added to quiescent were grown to confluency in medium containing 10% serum, and cells, amounts of Gadd45a mRNA decreased within Gadd45a mRNA levels were determined by Northern blotting. 30 min, were substantially reduced at 2 h, and were Equal loading was ascertained by detection of 18S and 28S rRNA. (a)Quiescent cultures received 20 mg/ml actinomycin D (act D)for below detection limits at 4 h (Figure 3a). Based on these the indicated times. (b)Quiescent cells were refed with medium kinetics, the Gadd45a transcript has a half-life of 50 min containing 20 ng/ml PDGF and 10% serum. Cells received 20 mg/ and thus is relatively unstable, even in the absence of ml actinomycin D 2 h after stimulation and were harvested 2 h after mitogens. The half-life of Gadd45a mRNA in quiescent addition of actinomycin D. (c)Quiescent cells (0 h)were incubated for 9 h in medium containing PDGF and serum with or without p27-47 cells is similar to the half-life of Gadd45a mRNA 10 mg/ml cycloheximide (CHX) in quiescent rat fibroblasts (20–80 min)and breast carcinoma cells (45 min)(Marhin et al., 1997; Abcouwer et al., 1999). Since Gadd45a mRNA degrades rapidly, it appears that a high rate of gadd45a gene transcription PDGF and serum for 8 h, and this drop was largely accounts for the accumulation of Gadd45a mRNA in prevented by cycloheximide (Figure 3c). Cycloheximide density-arrested p27-47 cells. Within 2 h of addition of had no effect on Gadd45a mRNA abundance in the actinomycin D to mitogen-stimulated cells, amounts of absence of mitogens (data not shown). Thus, mitogens Gadd45a were barely detectable (Figure 3b). Thus, the reduce the expression of the Gadd45a transcript by a Gadd45a transcript decays rapidly in both stimulated process that is dependent on the synthesis of a particular and quiescent cells. These findings suggest that mitogens protein or proteins. downregulate Gadd45a by repressing the transcription of the gadd45a gene. Effects of p27Kip1 on the stability of Density-dependent expression of Gadd45a the Gadd45a transcript could not be determined because actinomycin D reduced the abundance of p27Kip1 in As described above, amounts of Gadd45a mRNA were IPTG-treated p27-47 cells. similar in exponentially growing p27-47 cells treated To determine whether the decrease in Gadd45a with and without IPTG (Figure 2a). To determine mRNA abundance in mitogen-stimulated cells requires whether ectopic expression of p27Kip1 affects Gadd45a protein synthesis, density-arrested p27-47 cells were expression at any time during the growth cycle, p27-47 exposed to PDGF and serum in the presence or absence cells were seeded at low density and grown to confluence of the protein synthesis inhibitor cycloheximide. Gad- in medium containing or lacking IPTG. Cells were d45a mRNA levels fell substantially in cells receiving harvested daily for 5 days, and amounts of p27Kip1 were

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4170 Kip1 a p27 expression levels. To test this premise, we grew Day: 12345 p27-47 cells for several days in medium containing 1, 5, IPTG: or 10% serum. Cultures were harvested on the day at which they became conﬂuent, and cell numbers were Gadd45α mRNA determined. Cells grown in medium supplemented with 5% serum and in medium supplemented with 10% 6 6 GAPDH mRNA serum achieved similar densities, 12 Â 10 and 15 Â 10 cells/plate, respectively. Cells cultured in the presence of

p27Kip1 1% serum were larger than those cultured in the presence of 5 or 10% serum and reached a density of only 3.7 Â 106 cells/plate. All cultures were quiescent and b expressed similar amounts of endogenous p27Kip1

3 control (Figure 5 and data not shown). Amounts of Gadd45a mRNA and protein were similar in cells receiving 5% 2 serum and 10% serum and were greater in cells exposed to 5 or 10% as compared with 1% serum. These data 1 IPTG DOUBLINGS demonstrate a direct correlation between Gadd45a expression and culture density, and suggest that 12345 Kip1 DAYS IN CULTURE supraphysiological amounts of p27 reduce Gadd45a expression by lowering the density at which the cells Figure 4 Gadd45a expression during the growth cycle of control become confluent. and IPTG-treated p27-47 cells. (a, b)p27-47 cells were seeded at a density of 0.8 Â 106 cells/100 mm plate into medium containing 10% serum. Some cultures received IPTG (final concentration, 1mm)at the time of seeding. Cells were harvested daily for 5 days. Effect of enforced expression of Gadd45a on the G0 to S Cell numbers on day 5 were 5.7 Â 106 and 3.5 Â 106 for control and phase transition IPTG-treated cultures, respectively. (a)Amounts of Gadd45 a mRNA and GAPDH mRNA (loading control)were determined by To address the biological consequences of Gadd45a Northern blotting, and amounts of Gadd45a protein were expression, we prepared Balb/c-3T3 cell lines that determined by Western blotting. (b)Cell numbers were determined ectopically express Gadd45a in response to IPTG. with a hemacytometer and are expressed as cumulative doublings Growth curves and Gadd45a expression levels for one of these cell lines, termed Gadd45-24, are shown in Figure 6. Gadd45a protein was much more abundant in greater in IPTG-treated cultures than in control cultures IPTG-treated cultures than in control cultures at all time at all time points (Figure 4a). During the first 3 days of growth, control and IPTG-treated cultures grew at similar rates, and both populations expressed minimal amounts of Gadd45a mRNA (Figure 4a, b). In the 1% 5% 10% serum absence of IPTG, amounts of Gadd45a mRNA increased on day 4, coincident with the cessation of Gadd45α mRNA proliferation, and continued to increase in density- arrested cultures (day 5). IPTG-treated cultures became quiescent on day 3, at a density approximately 65% of Gadd45α that of control cultures. In these cultures, Gadd45a upregulation occurred subsequent to growth arrest (days 4 and 5)and to a much lesser extent than in control cultures. These findings show that ectopic p27Kip1 expression of p27Kip1 suppresses the increase in Gad- d45a mRNA abundance that occurs when cells become 28S quiescent and suggest that Gadd45a does not signal the exit of cells from the cell cycle. 18S Given the capacity of p27Kip1 to prevent maximal Gadd45a downregulation in cells exiting G0, its capacity to prevent maximal Gadd45a upregulation in cells actin entering G0 seems paradoxical. Since IPTG-treated p27-47 cells arrest at lower densities than do their Figure 5 Modulation of Gadd45a expression by culture density. p27-47 cells were grown in 100 mm tissue culture plates in medium untreated counterparts (Figure 4a)(Zhang et al., 2000), containing 1, 5, or 10% serum and were harvested on the day at it is possible that culture density dictates the extent of which cultures became confluent. Cell numbers at confluency were Gadd45a expression in cells entering quiescence. In this 3.7 Â 106,12Â 106, and 15 Â 106 for cells receiving 1, 5, and 10% situation, the effects of p27Kip1 on Gadd45a expression serum, respectively. Amounts of Gadd45a mRNA were determined Kip1 would be limited to its effects on culture density, and by Northern blotting, and amounts of Gadd45a and p27 were determined by Western blotting. As normalization controls for conditions that increase culture density would be Northern and Western blotting, amounts of 18S and 28S rRNA predicted to increase Gadd45a expression regardless of and actin, respectively, are shown

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4171 a or absence of IPTG for 18 h, and cell cycle position was determined by FACS analysis. When exposed to Day: 1345 mitogens, 35% of the cells in the control population IPTG: entered S phase, as compared with less than 6% of the cells in the IPTG-treated population (Figure 6c). These Gadd45α ﬁndings show that ectopic expression of Gadd45a activates a G1 checkpoint and suggest that p27Kip1 inhibits the G0 to S phase transition by promoting the b accumulation of Gadd45a. 3

control Discussion 2 Gadd45a was originally identiﬁed as a gene product that was enriched in hamster ﬁbroblasts exposed to DNA-

DOUBLINGS 1 damaging agents (Fornace et al., 1988). Subsequent IPTG studies showed that Gadd45a was also highly expressed in serum-deprived and contact-inhibited cells of mouse, human, and hamster origin (Fornace et al., 1989). 12345 Gadd45a is a highly charged, acidic protein of B21 kDa DAYS IN CULTURE that is localized primarily in the nucleus (Carrier et al., 1994; Zhan et al., 1994). Biological activities of Gadd45a c include the repair of damaged DNA and the modulation G0/G1 100 of chromatin structure and accessibility (Carrier et al., S 1999; Smith et al., 2000). Gadd45a also functions as a 80 growth inhibitor. Cells that overexpress Gadd45a form G2/M 60 fewer colonies and grow at slower rates that do their vector-transfected counterparts (Zhan et al., 1994; 40 Zhang et al., 2001). Conversely, Gadd45a-null cells 20 grow faster than do Gadd45a-wild-type cells (Hollander PERCENT CELLS et al., 1999). Mice lacking Gadd45a develop normally Kip1 0 hr 18 hr 18 hr but exhibit genomic instability and, like p27 -null + IPTG mice, are larger than their wild-type counterparts (Fero et al., 1996; Kiyokawa et al., 1996; Nakayama et al., Figure 6 Overexpression of Gadd45a activates a G1 checkpoint. 1996; Hollander et al., 1999). (a, b)Gadd45-24 cells were seeded at a density of 0.6 Â 106 cells/ 100 mm plate into medium containing 10% serum. Some cultures The studies described here provide new information received 1 mm IPTG at the time of seeding, and cells were harvested about Gadd45a. We show that Gadd45a expression is daily for 5 days. (a)Amounts of Gadd45 a were determined by regulated by p27Kip1 and culture density, and that Western blotting. (b)Cell numbers were determined with a enforced expression of Gadd45a activates a G1 check- hemacytometer and are expressed as cumulative doublings. (c) point in mouse fibroblasts. Our studies were carried out Density-arrested Gad45a-24 cells were harvested (0 h)or stimulated for 18 h in medium containing 25 ng/ml PDGF and 10% serum in on cell lines derived from Balb/c-3T3 cells, and the presence (18 h þ IPTG)or absence (18 h)of IPTG. The specifically, we show the following: (1)cells arrested at percentages of cells in G0/G1, S, and G2/M were determined by high densities express more Gadd45a mRNA and FACS analysis of cells stained with propidium iodide protein than do cells arrested at low densities; (2) mitogenic stimulation of density-arrested cells reduces Gadd45a expression, and this decrease is inhibited by points (Figure 6a). However, Gadd45-24 cells grew in ectopic expression of p27Kip1; and (3)overexpression of medium containing IPTG, albeit at a slower rate than Gadd45a prevents density-arrested cells from entering S did cells in medium lacking IPTG (Figure 6b). phase in response to mitogens. In a previous study, we When density-arrested p27-47 cells are exposed to demonstrated a direct correlation between culture PDGF and serum, Gadd45a mRNA and protein levels density and the susceptibility of cells to the antiproli- decrease and remain low for at least 12 h, and S phase ferative actions of p27Kip1 (Zhang et al., 2000). On the entry begins 12–15 h after stimulation (Figure 2a and basis of our current findings, we propose a model of cell data not shown). In cells stimulated in the presence of cycle regulation in which culture density modulates IPTG, Gadd45a mRNA and protein levels decline Gadd45a expression and the extent to which Gadd45a is briefly but return to quiescent levels at or before 12 h, expressed determines how cells respond to p27Kip1.We and IPTG-treated cells do not enter S phase. Thus, when suggest that amounts of Gadd45a above a certain present in cells at sufficient quantities, Gadd45a may threshold are growth inhibitory and that such amounts prevent quiescent cells from initiating DNA synthesis in are achieved in cells arrested at high but not low response to mitogens. To test this premise, Gadd45-24 densities. When cells arrested at high densities are cells were treated with PDGF and serum in the presence exposed to mitogens, Gadd45a levels fall below the

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4172 threshold and remain low for several hours, thus Dong et al., 1998b). In contrast, Gadd45a mRNA levels allowing cells to move through G0/G1 and into S are restored to quiescent levels within 8 h of addition of phase. However, in cells stimulated in conditions in PDGF, serum, and IPTG to quiescent p27-47 cells. which p27Kip1 is overexpressed, Gadd45a levels do not These kinetics suggest that p27Kip1 promotes the remain below the threshold, and G0/G1 progression accumulation of Gadd45a mRNA by a process that does not occur. In cells arrested at low densities and in does not involve CDK activation. exponentially cycling cells, Gadd45a is expressed at Gadd45a-null mouse embryo fibroblasts enter G0 subthreshold levels and, consequently, would not when deprived of serum, thus indicating that Gadd45a is impede proliferation, even in the presence of large not required for G0 arrest (Hollander et al., 1999). In amounts of p27Kip1. Thus, the capacity of p27Kip1 to p27-47 cells grown to confluence in the presence of inhibit G0 to S phase progression is dependent on the IPTG (shown here)and in smooth muscle cells deprived amount of Gadd45a in cells. When overexpressed, of serum (Kettenhofen et al., 2001), increases in amounts of Gadd45a obviously exceed the threshold Gadd45a expression occurred after cells had ceased and inhibit cell proliferation regardless of culture proliferation. Thus, Gadd45a upregulation apparently is density. How culture density regulates Gadd45a expres- a consequence rather than a cause of cell cycle arrest. sion and how p27Kip1 blocks mitogen-induced reductions Rather than inducing quiescence, Gadd45a may provide in Gadd45a expression remain to be determined. a growth-suppressive signal that contributes to the Maximal Gadd45a expression occurs in conditions in maintenance of quiescence. We suggest that conditions which growth inhibition is combined with stress (Zhan that are highly restrictive for growth (e.g., high culture et al., 1994). For example, density-dependent growth density)induce the expression of large amounts of arrest coupled with the depletion of essential nutrients Gadd45a, which in turn decrease the responsiveness of from the culture medium substantially increases the cells to mitogenic signals. amount of Gadd45a mRNA in hamster fibroblasts (Zhan et al., 1994). In these experiments, medium depletion was accomplished by incubating confluent Materials and methods cultures for 2–4 days without refeeding. The growth of cells to high densities also induces stress by causing Preparation of cell lines and cell culture medium depletion and by imposing spatial constraints Balb/c-3T3 cell lines expressing p27Kip1 or Gadd45a under that restrict cell movement and spreading. Thus, our IPTG regulation were prepared by a two-step procedure data showing that Gadd45a expression is a function of originally devised by Labow et al. (1990)and Pestov and Lau culture density are consistent with the concept of stress- (1994), and recently described by Zhang et al. (2000). Cells induced Gadd45a expression. were first transfected with a plasmid encoding the inducer- Human fibroblasts arrest in G2 when microinjected binding domain of the lac repressor inserted upstream of the with an expression vector encoding Gadd45a (Wang VP16 transcriptional activation domain. Neomycin-resistant et al., 1999; Yang et al., 2000). Conversely, loss of colonies were screened for repressor activity using a plasmid Gadd45a abrogates the G2 checkpoint induced by UV containing a luciferase reporter gene positioned downstream of a multimerized lac operator sequence. A cell line (designated irradiation and methyl methanesulfonate (Hollander LAP-31)that exhibited high levels of luciferase activity in the et al., 1999; Wang et al., 1999). The mechanism by which absence of IPTG and basal levels in the presence of IPTG was Gadd45a elicits G2 arrest involves the interaction of used for step two of the procedure. LAP-31 cells were Gadd45a with cyclin B–cdc2 complexes and the transfected with a plasmid (pX12)encoding the entire mouse consequent disruption of these complexes (Wang et al., p21Kip1 or the entire human Gadd45a coding sequence 1999; Jin et al., 2000; Yang et al., 2000). In some (provided by Dr A Fornace). The full-length Gadd45a instances, however, Gadd45a blocks cell cycle progres- fragment was prepared by polymerase chain reaction (PCR) sion in G1; such instances include the ectopic expression using pCMV-Gadd45a as template and a 50 NheI primer (50GCCTAGCTAGCATGACTTTGGAGGAATTCTCG 30) of Gadd45a in H1299 lung carcinoma cells and U2OS 0 0 osteosarcoma cells (Smith et al., 1994; Fan et al., 1999; and a 3 EcoRV primer (5 CTAGTGATATCTCACCGTTCA- GGGAGATTAATCAC30). The PCR product was sequenced Zhang et al., 2001). Gadd45a has also been implicated in and subcloned into pX12 at the NheI and EcoRV sites. the G1 arrest that occurs in A431 cells exposed to After transfection of LAP-31 cells, hygromycin-resistant epidermal growth factor (Fong et al., 2001). Our data colonies were screened for IPTG-inducible expression of show that supraphysiological amounts of Gadd45a p27Kip1 or Gadd45a by Western blotting. Two cell lines were block the G0 to S transition in quiescent Gadd45-24 used in the studies described here: p27-47 and Gadd45-24, cells exposed to mitogens and thus provide further which express supraphysiological amounts of p27Kip1 and evidence in support of Gadd45a-mediated G1 arrest. Gadd45a, respectively, in response to 1 mm IPTG (see below) Our studies do not exclude the existence of additional (Zhang et al., 2000). Both cell lines were cultured in Dulbecco’s checkpoints in IPTG-treated p27-47 cells, and the modified Eagle’s medium supplemented with 10% fetal calf mechanistic basis of the Gadd45a-activated G1 check- serum, 50 mg/ml streptomycin, 50 U/ml penicillin, 200 mg/ml G418, and 50 mg/ml hygromycin at 371C in a humidified point remains to be determined. The addition of atmosphere containing 5% CO2. mitogens to density-arrested Balb/c-3T3 cells results in To induce cell cycle progression, confluent quiescent cultures the activation of cdk4-containing complexes beginning were refed with fresh medium containing 20 ng/ml PDGF (the 9 h poststimulation and the activation of cdk2-contain- BB isoform)and 10% serum. For ectopic expression of p27 Kip1 ing complexes at the G1/S border (Agrawal et al., 1996; or Gadd45a, IPTG was added to confluent p27-47 or

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4173 Gadd45a-24 cultures either at the same time as or 18 h before Biobond nylon membranes (Sigma)and crosslinked to the PDGF and serum. IPTG does not appreciably increase the membrane in a Stratalinker. Sequence-verified DNA fragments ectopic expression of p27Kip1 or Gadd45a when added to from expressed sequence-tagged clones (Image clones, Re- quiescent cultures; that is, both IPTG and mitogens are search Genetics)were gel-purified and 32P-labeled using a required for efficient expression (data not shown). Thus, the DECAprime II kit (Ambion, Inc.). Hybridizations using the kinetics of p27Kip1 induction are similar in cells exposed to radiolabeled probe were carried out overnight in PerfectHyb IPTG before the addition of mitogens and cells simultaneously Plus (Sigma)at 65 1C. The blots were washed twice at room exposed to mitogens and IPTG. temperature with 2 Â SSC/0.1% SDS, and radiolabeled bands were visualized by autoradiography. Flow cytometry For RNase protection assays, RNA (7 mg)was hybridized overnight at 561C with 32P-labeled probes corresponding to the Cells were removed from the plates with 0.125%. trypsin and mCYC-1 probe set (PharMingen). Samples were then digested m 0.5 m EDTA in phosphate-buffered saline (PBS); an equal with RNase T1 and RNase A for 45 min at 301C and volume of medium containing 10% serum was added to proteinase K for 15 min at 371C. Samples were extracted with neutralize the trypsin. Cells were pelleted and resuspended in phenol/chloroform, collected by ammonium acetate/ethanol PBS (1 ml), and 100% ethanol (4 ml) was added drop by drop precipitation, denatured at 901C for 3 min, and electrophor- while vortexing. After incubation at 41C for a minimum of 2 h, esed on a 5% polyacrylamide gel. Gels were dried and exposed cells were pelleted and resuspended in PBS containing 0.1% to X-ray film. Tween 20, 0.05% bovine serum albumin, 10 mg/ml RNase A, and 50 mg/ml propidium iodide. After a further incubation at room temperature for at least 30 min, cell cycle distributions Western blots were determined with a Becton Dickinson FACScan. For determination of Gadd45a protein levels, cells were scraped from the plates, resuspended in PBS containing 1% Microarray analysis Nonidet-P40, 10% glycerol, and Complete protease inhibitors (Roche), and sonicated. Proteins (B150 mg)were resolved by The microarray analysis was done on an Affymetrix GeneChip SDS–PAGE and transferred to nitrocellulose membranes. Instrument System following protocols established by Affyme- Blots were blocked in PBS containing 0.05% Tween and 5% trix Inc. Total cellular RNA was prepared with Trizol Reagent instant milk and incubated with Gadd45a antibody (H-165; (Invitrogen)and further purified using an RNeasy kit Santa Cruz)in PBS containing 0.05% Tween. This antibody (Qiagen). RNA was converted to double-stranded cDNA recognizes both mouse and human Gadd45a. Proteins using an oligo dT24 primer that contains an RNA polymerase recognized by the antibody were detected by enhanced T7 recognition sequence, and the resulting double-stranded chemiluminescence using a horseradish peroxidase-coupled cDNA was transcribed into biotin-labeled cRNA using T7 secondary antibody as specified by the manufacturer (Pierce). RNA polymerase. Biotinylated cRNA was hybridized to p27Kip1 Western blots were carried out as described previously GeneChip probe arrays, and hybridized RNA was visualized (Zhang et al., 2000). by staining the biotinylated cRNA with streptavidin-conju- gated phycoerythrin. Scanned chip images were analysed using GeneChip algorithms. Two GeneChip arrays were used in our Acknowledgements experiments: the Mu11k set (A and B chips), which contains This work was supported by the Cortner-Couch Endowed B11 000 functionally characterized mouse genes and EST Chair for Cancer Research and NIH Grant CA67360 (WJP). clusters, and the Murine Genome U74A (MG-74A)set, which We thank Dr Albert Fornace for providing the Gadd45a contains B11 000 known genes and ESTs (Affymetrix Inc.). coding sequence, Dr Nancy Olashaw for the manuscript preparation, and Dr Shrinkant Mane of the Microarray Core Facility at the Moffitt Cancer Center for his assistance in the Northern blotting and RNase protection assays performance and interpretation of the microarray analyses. For Northern blotting, total cellular RNA was prepared using We also acknowledge the helpful service of the Flow Trizol Reagent, and aliquots (15 mg/lane)were separated on Cytometry and Molecular Imaging Core Laboratories at the 1% agarose–formaldehyde gels. RNA was transferred to Moffitt Cancer Center.

References

Abcouwer SF, Schwarz C and Meguid RA. (1999). J. Biol. Carrier F, Smith ML, Bae I, Kilpatrick KE, Lansing TJ, Chem., 274, 28645–28651. Chen CY, Engelstein M, Friend SH, Henner WD Agrawal D, Hauser P, McPherson F, Dong F, Garcia A and and Gilmer TM. (1994). J. Biol. Chem., 269, 32672– Pledger WJ. (1996). Mol. Cell. Biol., 16, 4327–4336. 32677. Boussiotis VA, Freeman GJ, Taylor PA, Berezovskaya A, Coats S, Flanagan WM, Nourse J and Roberts JM. (1996). Grass I, Blazar BR and Nadler LM. (2000). Nat. Med., 6, Science, 212, 877–880. 290–297. DeGregori J, Kowalik T and Nevins JR. (1995). Mol. Cell. Breviario F, d’Aniello EM, Golay J, Peri G, Bottazzi B, Biol., 15, 4215–4224. Bairoch A, Saccone S, Marzella R, Predazzi V and Rocchi Deschenes C, Vezina A, Beaulieu JF and Rivard N. (2001). M et al . (1992). J. Biol. Chem., 267, 22190–22197. Gastroenterology, 120, 423–438. Carrier F, Georgel PT, Pourquier P, Blake M, Kontny HU, Dijkers PF, Medema RH, Pals C, Banerji L, Thomas NS, Antinore MJ, Gariboldi M, Myers TG, Weinstein JN, Lam EW, Burgering BM, Raaijmakers JA, Lammers JW, Pommier Y and Fornace Jr AJ. (1999). Mol. Cell. Biol., 19, Koenderman L and Coffer PJ. (2000). Mol. Cell. Biol., 20, 1673–1685. 9138–9148.

Oncogene Gadd45a mediates growth inhibition by p27Kip1 X Zhang et al 4174 Dong F, Agrawal D, Bagui T and Pledger WJ. (1998a). Mol. Nourse J, Firpo E, Flanagan WM, Coats S, Polyak K, Lee Biol. Cell, 9, 2081–2092. MH, Massague J, Crabtree GR and Roberts JM. (1994). Dong F, Cress Jr WD, Agrawal D and Pledger WJ. (1998b). Nature, 372, 570–573. J. Biol. Chem., 273, 6190–6195. Nurse P. (1990). Nature, 344, 503–508. Fan W, Richter G, Cereseto A, Beadling C and Smith KA. Pagano M, Tarn SW, Theodoras AM, Beer-Romero P, Del Sal (1999). Oncogene, 18, 6573–6582. G, Chau V, Yew PR, Draetta GF and Rolfe M. (1995). Fero ML, Rivkin M, Tasch M, Porter P, Carow CE, Firpo E, Science, 269, 682–685. Polyak K, Tsai LH, Broudy V, Perlmutter RM, Kaushansky Pestov DG and Lau LF. (1994). Proc. Natl. Acad. Sci. USA, K and Roberts JM. (1996). Cell, 85, 733–744. 91, 12549–12553. Firpo EJ, Koff A, Solomon MJ and Roberts JM. (1994). Mol. Pines J. (1995). Biochem. J., 308, 697–711. Cell. Biol., 14, 4889–4901. Polyak K, Lee MH, Erdjument-Bromage H, Koff A, Roberts Fong WF, Leung CH, Lam W, Wong NS and Cheng SH. JM, Tempst P and Massague J. (1994). Cell, 78, 59–66. (2001). Biochim. Biophys. Acta, 1517, 250–256. Rivard N, L’Allemain G, Bartek J and Pouyssegur J. (1996). Fornace Jr AJ, Alamo Jr I and Hollander MC. (1988). Proc. J. Biol. Chem., 271, 18337–18341. Natl. Acad. Sci. USA, 85, 8800–8884. Sherr CJ and Roberts JM. (1995). Genes Dev., 9, 1149–1163. Fornace Jr AJ, Nebert DW, Hollander MC, Luethy JD, Smith ML, Chen IT, Zhan Q, Bae I, Chen CH, Gilmer TM, Papathanasiou M, Fargnoli J and Holbrook NJ. (1989). Kastan MB, O’Connor PM and Fornace Jr AJ. (1994). Mol. Cell. Biol., 9, 4196–4203. Science, 266, 1376–1380. Hauser PJ, Agrawal D, Flanagan M and Pledger WJ. (1997). Smith ML, Ford JM, Hollander MC, Bortnick RA, Amund- Cell Growth Differ., 8, 203–211. son SA, Seo YR, Deng CX, Hanawalt PC and Fornace Jr Hengst L and Reed SI. (1996). Science, 271, 1861–1864. AJ. (2000). Mol. Cell. Biol., 20, 3705–3714. Hollander MC, Sheikh MS, Bulavin DV, Lundgren K, Augeri- Thelander L and Berg P. (1986). Mol. Cell. Biol., 6, 3433–3442. Henmueller L, Shehee R, Molinaro TA, Kim KE, Tolosa E, Tomoda K, Kubota Y and Kato J. (1999). Nature, 398, Ashwell JD, Rosenberg MP, Zhan Q, Fernandez-Salguero 160–165. PM, Morgan WF, Deng CX and Fornace Jr AJ. (1999). Nat. Toyoshima H and Hunter T. (1994). Cell, 78, 67–74. Genet., 23, 176–184. Wang XW, Zhan Q, Coursen JD, Khan MA, Kontny HU, Yu Jin S, Zhao H, Fan F, Blanck P, Fan W, Colchagie AB, L, Hollander MC, PM OC, Fornace Jr AJ and Harris CC. Fornace Jr AJ and Zhan Q. (2000). Oncogene, 19, (1999). Proc. Natl. Acad. Sci. USA, 96, 3706–3711. 4050–4057. Winston J, Dong F and Pledger WJ. (1996). J. Biol. Chem., Kato JY, Matsuoka M, Polyak K, Massague J and Sherr CJ. 271, 11253–11260. (1994). Cell, 79, 487–496. Winston JT and Pledger WJ. (1993). Mol. Biol. Cell, 4, 1133– Kettenhofen R, Hoppe J, Eberhard G, Seul C, Ko Y and 1144. Sachinidis A. (2001). Cell Signal., 13, 787–799. Yang Q, Manicone A, Coursen JD, Linke SP, Nagashima M, Kiyokawa H, Kineman RD, Manova-Todorova KO, Scares Forgues M and Wang XW. (2000). J. Biol. Chem., 275, VC, Hoffman ES, Ono M, Khanam D, Hayday AC, 36892–36898. Frohman LA and Koff A. (1996). Cell, 85, 721–732. Yang W, Shen J, Wu M, Arsura M, FitzGerald M, Suldan Z, Labow MA, Bairn SB, Shenk T and Levine AJ. (1990). Mol. Kim DW, Hofmann CS, Pianetti S, Romieu-Mourez R, Cell. Biol., 10, 3343–3356. Freedman LP and Sonenshein GE. (2001). Oncogene, 20, Marhin WW, Chen S, Facchini LM, Fornace Jr AJ and Penn 1688–1702. LZ. (1997). Oncogene, 14, 2825–2834. Zhan Q, Lord KA, Alamo Jr I, Hollander MC, Carrier F, Ron Millard SS, Van JS, Nguyen H, Pagano M, Kiyokawa H and D, Kohn KW, Hoffman B, Liebermann DA and Fornace Jr Koff A. (1997). J. Biol. Chem., 272, 7093–7098. AJ. (1994). Mol. Cell. Biol., 14, 2361–2371. Miskimins R, Srinivasan R, Marin-Husstege M, Miskimins Zhang W, Hoffman B and Liebermann DA. (2001). Int. J. WK and Casaccia-Bonneﬁl P. (2002). J. Neurosci. Res., 67, Oncol., 18, 749–757. 100–1005. Zhang X, Wharton W, Donovan M, Coppola D, Croxton R, Nakayama K, Ishida N, Shirane M, Inomata A, Inoue T, Cress WD and Pledger WJ. (2000). Mol. Biol. Cell, 11, Shishido N, Horii I and Loh DY. (1996). Cell, 85, 707–720. 2117–2130. Nguyen H, Gitig DM and Koff A. (1999). Mol. Cell. Biol., 19, 1190–1201.

Oncogene