PDK1 Regulates Cell Proliferation and Cell Cycle Progression through Control of Cyclin D1 and p27Kip1 Expression*

PDK1 (3-phosphoinositide-dependent protein kinase 1) is a key mediator of signaling by phosphoinositide 3-kinase. To gain insight into the physiological importance of PDK1 in cell proliferation and cell cycle control, we established immortalized mouse embryonic fibroblasts (MEFs) from mice homozygous for a “floxed” allele of Pdk1 and from wild-type mice. Introduction of Cre recombinase by retrovirus-mediated gene transfer resulted in the depletion of PDK1 in Pdk1lox/lox MEFs but not in Pdk1+/+ MEFs. The insulin-like growth factor-1-induced phosphorylation of various downstream effectors of PDK1, including Akt, glycogen synthase kinase 3, ribosomal protein S6, and p70 S6 kinase, was markedly inhibited in the PDK1-depleted (Pdk1-KO) MEFs. The rate of serum-induced cell proliferation was reduced; progression of the cell cycle from the G0-G1 phase to the S phase was delayed, and cell cycle progression at G2-M phase was impaired in Pdk1-KO MEFs. These cells also manifested an increased level of p27Kip1 expression and a reduced level of cyclin D1 expression during cell cycle progression. The defect in cell cycle progression from the G0-G1 to the S phase in Pdk1-KO MEFs was rescued by forced expression of cyclin D1, whereas rescue of the defect in G2-M progression in these cells required both overexpression of cyclin D1 and depletion of p27Kip1 by RNA interference. These data indicate that PDK1 plays an important role in cell proliferation and cell cycle progression by controlling the expression of both cyclin D1 and p27Kip1.

The physiological functions of PDK1 in living organisms have been investigated by engineering of the corresponding gene. Deletion of the gene in Saccharomyces cerevisiae (9,10), Schizosaccharomyces pombe (11), Caenorhabditis elegans (12), and Drosophila (13,14) revealed that PDK1 is required for the normal development and viability of these organisms, highlighting its functional importance. Although the Pdk1 knockout mouse is not viable (15), mice that harbor a hypomorphic Pdk1 allele, in which the abundance of PDK1 is reduced to 10% of that in wild-type mice, are viable but fail to produce mature T cells as a result of a defect in the differentiation and proliferation of this lineage (16). Furthermore, 7-hydroxystaurosporine (UCN-01), an inhibitor of PDK1, was shown to abrogate tumor cell growth and to promote apoptosis (17). Depletion of PDK1 with the use of either an antisense oligonucleotide or RNA interference also inhibited proliferation of a glioblastoma cell line (18). These various observations thus suggested that PDK1 regulates cell proliferation. However, mouse embryonic stem cells in which both copies of Pdk1 were disrupted were viable and proliferated at a rate similar to that of wild-type cells (3). The relevance of PDK1 to cell proliferation and the underlying mechanism of any regulatory role have thus remained unclear.
To examine the role of PDK1-dependent signaling in cell proliferation and control of the cell cycle, we have now investigated the effects of PDK1 deficiency in mouse embryonic fibroblasts (MEFs). We here show that lack of PDK1 resulted in inhibition of cell proliferation but did not promote apoptosis in MEFs, and that this inhibition of proliferation was attributable to a delayed transition from the G 0 -G 1 to the S phase and to slowed progression at the G 2 -M phase of the cell cycle. We also show that control of cell cycle progression by PDK1 requires the induction of cyclin D1 expression at the transition from the G 0 -G 1 to the S phase and the coordinated regulation of cyclin D1 and p27 Kip1 expression at G 2 -M. Our findings thus reveal an important regulatory role for PDK1 in cell proliferation and cell cycle progression.

EXPERIMENTAL PROCEDURES
Antibodies-Rabbit polyclonal antibodies to PDK1 were kindly provided by F. Liu (University of Texas, Health Science Center, San Antonio) (19). Polyclonal antibodies to Akt, to Akt phosphorylated on Thr 308 or on Ser 473 , to phosphorylated glycogen synthase kinase 3 (GSK3) ␣ or ␤ (Ser 21 and Ser 9 , respectively), to ribosomal protein S6, to phosphorylated ribosomal protein S6 (Ser 235 and Ser 236 ), to p70 S6 kinase, to extracellular signal-regulated kinase (ERK) 1 or 2, to phosphorylated ERK1 or ERK2 (Thr 202 and Tyr 204 ), to phosphorylated retinoblastoma (Rb) protein (Ser 780 ), and to phosphorylated FOXO1 (Ser 256 ), as well as monoclonal antibodies to phosphorylated p70 S6 kinase (Thr 389 ), were obtained from Cell Signaling Technologies (Beverly, MA). Polyclonal antibodies to cyclindependent kinase (CDK) 2, to CDK4, to cyclin A, to cyclin E, and to FOXO1 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), and those to Skp2 were from Zymed Laboratories Inc.. Monoclonal antibodies to cyclin D1 and to p27 Kip1 were obtained from Pharmingen and BD Transduction Laboratories, respectively.
Retroviral Vectors-To produce retroviral vectors for Cre recombinase or PDK1, we subcloned cDNAs for Cre or mouse PDK1 into pWZL containing the blasticidin resistance gene (kindly provided by G. P. Nolan, Stanford University, Stanford, CA) or into pMXs containing the puromycin resistance gene (kindly provided by T. Kitamura, University of Tokyo, Japan), respectively. Plat-E retrovirus-packaging cells were kindly provided by T. Kitamura.
Generation of PDK1 Knock-out MEFs and Rescue of PDK1 Expression-Mice homozygous for a "floxed" allele of Pdk1 in which exons 3 and 4 are flanked by LoxP sequences (Pdk1 lox/lox mice) were described previously (20). Embryonic fibroblasts were isolated from Pdk1 lox/lox and the corresponding wild-type (Pdk1 ϩ/ϩ ) mice at embryonic day 13.5 and were immortalized by the 3T3 protocol (21). To disrupt Pdk1 in the immortalized MEFs harboring the floxed allele, we introduced a retroviral vector for Cre recombinase into the cells as described previously (22). Cells resistant to blasticidin S (10 g/ml) (Invitrogen) were selected, and the resulting PDK1 knock-out (Pdk1-KO) MEFs were maintained at 37°C under 5% CO 2 in Dulbecco's modified Eagle's medium containing glucose at 4.5 g/liter and supplemented with 10% heat-inactivated fetal bovine serum, penicillin (50 units/ml), and streptomycin sulfate (50 g/ml), referred to hereafter as culture medium. Pdk1-KO MEFs stably expressing PDK1 (Pdk1-KO/PDK1 MEFs) were similarly generated by infection with a retroviral vector for PDK1 and were selected on the basis of resistance to puromycin (2 g/ml) (Sigma). Cells were stimulated with recombinant human insulin-like growth factor-1 (IGF-1) (Sigma) after incubation for 16 h in serum-free medium.
Adenoviral Vectors and Infection-Adenoviral vectors for mouse cyclin D1 tagged with a nuclear localization signal (AdD1NLS) and for a small interfering RNA (siRNA) specific for rat and mouse p27 Kip1 mRNA (Adp27siRNA) were described previously (23). An adenoviral vector encoding ␤-galactosidase (AxLacZ) was kindly provided by I. Saito (University of Tokyo, Japan). MEFs were infected with adenoviral vectors essentially as described previously (24).
Northern Blot Analysis, Immunoblot Analysis, and Assay of Akt and PI 3-Kinase Activities-Total RNA (10 g) extracted from cells was subjected to Northern blot analysis with fulllength mouse cyclin D1 or p27 Kip1 cDNAs as probes. Cell lysates (ϳ50 g of protein) were subjected to SDS-PAGE and immunoblot analysis as described (25). The activity of Akt in immunoprecipitates prepared with antibodies to Akt was assayed with the synthetic Crosstide peptide as substrate (26). The activity of PI 3-kinase in immunoprecipitates prepared with antibodies to phosphotyrosine was assayed as described (27).
Oligonucleotide Microarray Analysis-Total RNA (10 g) isolated from Pdk1 ϩ/ϩ MEFs expressing Cre recombinase (Pdk1 ϩ/ϩ /Cre MEFs) or from Pdk1-KO MEFs was used as a template for the synthesis of biotin-labeled cRNA, which was then used to probe Genechip Mouse MU74 microarrays (Affymetrix, Santa Clara, CA). The arrays were scanned and analyzed as described previously (22).
Assay of DNA Fragmentation-Cells that had been treated with staurosporine (Sigma) were suspended in a lysis buffer (10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 10 mM EDTA, 0.5% SDS) and incubated for 2 h at 50°C with proteinase K (100 g/ml) (Wako, Osaka, Japan). The reaction mixture was then subjected to phenol/chloroform extraction and ethanol precipitation, and DNA was transferred to TE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA), incubated with RNase A (10 g/ml) (Sigma) for 45 min at 37°C and subjected to electrophoresis on a 2% agarose gel. The gel was stained with ethidium bromide for visualization of DNA fragments.
Cell Cycle Synchronization and Cell Cycle Analysis by Flow Cytometry-Cells were cultured overnight in serum-free medium and then subjected to cell cycle arrest by incubation for 16 or 24 h in culture medium containing aphidicolin (1 g/ml) (Sigma) or nocodazole (200 ng/ml) (Sigma), respectively. They were then released from arrest by incubation in drug-free culture medium. Cells were collected by exposure to trypsin and centrifugation at 200 ϫ g for 5 min, washed with phosphate-buffered saline (PBS), and fixed in 75% ethanol at Ϫ20°C. The fixed cells were washed with PBS, incubated for 30 min at 37°C with RNase A (100 g/ml), stained with propidium iodide (50 g/ml), and analyzed by flow cytometry (FACSCalibur; BD Biosciences).
BrdUrd Staining-Cells were exposed to 10 M bromodeoxyuridine (BrdUrd) for 30 min, harvested, washed with PBS, and subjected to cell cycle analysis with the use of a BrdUrd flow kit (Pharmingen). Labeled cells were detected by flow cytometry (FACSCalibur).
Immunocytofluorescence Analysis-Cells were seeded on glass coverslips before infection with adenoviral vectors. They were subsequently fixed with 3% paraformaldehyde, washed, and incubated consecutively with monoclonal antibodies to p27 Kip1 and with fluorescein isothiocyanate-conjugated goat antibodies to mouse immunoglobulin G (Jackson Immuno-Research, West Grove, PA) together with propidium iodide. Images were acquired with a confocal laser-scanning microscope (LSM5 PASCAL, Carl Zeiss).

Lack of PDK1 Expression Prevents IGF-1 Signaling-To
determine the biological function of PDK1 in cell proliferation, we prepared MEFs from Pdk1 ϩ/ϩ and Pdk1 lox/lox mouse embryos and disrupted Pdk1 in the latter cells by infecting them with a retroviral vector encoding Cre recombinase. Although the abundance of PDK1 in Pdk1 ϩ/ϩ MEFs was not affected by infection with the retroviral vector for Cre, PDK1 was not detected in the Cre-expressing Pdk1 lox/lox MEFs (Fig. 1A).
To confirm that disruption of Pdk1 specifically prevented signaling downstream of PDK1, we first measured the activity of Akt in the Cre-expressing Pdk1 lox/lox MEFs (Pdk1-KO MEFs). Whereas IGF-1 induced a marked increase in Akt activity in Pdk1 lox/lox MEFs, no such effect was apparent in Pdk1-KO MEFs (Fig. 1B). We next examined the effects of IGF-1 on the activation of downstream effectors of PDK1 signaling by immunoblot analysis with antibodies specific for phosphorylated forms of these proteins. The IGF-1-induced phosphorylation of Akt on Thr 308 , which is a target site of PDK1, was completely inhibited in Pdk1-KO MEFs, whereas the basal level of Akt phosphorylation on Ser 473 was increased in these cells (Fig. 1C). The IGF-1-induced phosphorylation of GSK3␣ (on Ser 21 ) and GSK3␤ (on Ser 9 ), two well characterized substrates of Akt, was inhibited in Pdk1-KO MEFs, as was the IGF-1-induced phosphorylation of both p70 S6 kinase and ribosomal protein S6 as well as the basal level of S6 phosphorylation. The basal level of phosphorylation of ERK1 and ERK2 was also markedly increased in Pdk1-KO MEFs. The IGF-1-induced phosphorylation of these various signaling molecules was not affected by expression of Cre recombinase in Pdk1 ϩ/ϩ MEFs. These results indicated that IGF-1 signaling mediated by PDK1 was specifically inhibited in Pdk1-KO MEFs.
Effects of PDK1 Depletion on Cell Proliferation and Cell Cycle Progression-We next examined whether PDK1 deficiency affected the proliferation of MEFs. Pdk1 ϩ/ϩ MEFs, Pdk1 ϩ/ϩ MEFs expressing Cre recombinase (Pdk1 ϩ/ϩ /Cre MEFs), Pdk1 lox/lox MEFs, and Pdk1-KO MEFs were seeded at equal densities, and their numbers were determined at various times up to 8 days. The rate of increase in the number of Pdk1-KO MEFs was substantially reduced compared with that for the other three types of MEFs ( Fig. 2A). To investigate whether this slower increase in the number of Pdk1-KO MEFs was because of the induction of apoptosis, we examined the extent of DNA fragmentation in Pdk1 lox/lox and Pdk1-KO MEFs cultured in the presence of 50 or 500 nM FIGURE 1. Effects of depletion of PDK1 on IGF-1-induced activation of various signaling proteins. A, MEFs derived from Pdk1 lox/lox or Pdk1 ϩ/ϩ mice were infected with a retroviral vector encoding Cre recombinase (ϩ) or with the corresponding empty vector (Ϫ). Cell lysates were subjected to immunoblot analysis with antibodies to PDK1 or to Akt (control). B, Pdk1 lox/lox MEFs infected as in A were cultured in serum-free medium for 16 h and then incubated for 10 min in the absence (Ϫ) or presence (ϩ) of 100 nM IGF-1. Cell lysates were then subjected to immunoprecipitation with antibodies to Akt, and the resulting precipitates were assayed for Akt activity. Data are means Ϯ S.E. from three independent experiments. C, Pdk1 ϩ/ϩ and Pdk1 lox/lox MEFs infected as in A and deprived of serum for 16 h were incubated for 10 min in the absence or presence of 100 nM IGF-1, lysed, and subjected to immunoblot analysis with antibodies to phosphorylated (p-) or total forms of the indicated proteins. Data in A and C are representative of at least three independent experiments. staurosporine for 1, 4, or 24 h. The level of staurosporineinduced apoptosis did not differ between the two cell types (Fig. 2B), suggesting that the reduced rate of increase in the number of Pdk1-KO MEFs was because of inhibition of proliferation rather than to the induction of apoptosis.
We next examined cell cycle progression in Pdk1 lox/lox and Pdk1-KO MEFs by flow cytometric analysis. Cells were synchronized in G 0 -G 1 by serum deprivation for 24 h and were stimulated to re-enter the cell cycle by re-exposure to 10% serum. Whereas ϳ60% of Pdk1 lox/lox MEFs had entered S phase 14 h after serum stimulation, only ϳ20% of Pdk1-KO MEFs had done so (Fig. 3, A and B). At 18 h, ϳ40% of Pdk1-KO MEFs had entered S phase, whereas the Pdk1 lox/lox cells had progressed to G 2 -M. The reduced number of Pdk1-KO MEFs in S phase was confirmed by analysis of BrdUrd incorporation. The number of BrdUrd-labeled cells at 14 h after the onset of serum stimulation was thus markedly smaller for Pdk1-KO MEFs than for Pdk1 lox/lox MEFs (Fig. 3C). These results indicated that the transition from G 0 -G 1 to S phase is delayed in Pdk1-KO MEFs, suggesting that PDK1 may regulate cell proliferation by controlling molecular events at this transition.
PDK1 Regulates the Expression of Cyclin D1 and p27 Kip1 during Cell Cycle Progression-We examined the expression of various proteins that contribute to regulation of the transition from the G 0 -G 1 to the S phase of the cell cycle. Whereas serum stimulation induced a marked increase in the amount of cyclin D1 in Pdk1 lox/lox cells, the increase in cyclin D1 expression was inhibited, and the amount of this protein was subsequently decreased in Pdk1-KO MEFs (Fig.  3D). These data suggested that PDK1 might regulate the induction of cyclin D1 associated with serum-stimulated cell cycle progression. Degradation of the CDK inhibitor p27 Kip1 is required for progression through G 0 -G 1 to S phase (28,29). The amount of p27 Kip1 was markedly higher in Pdk1-KO MEFs than in Pdk1 lox/lox MEFs before serum stimulation. The addition of serum induced a pronounced down-regulation of p27 Kip1 in both cell types, but the abundance of this protein remained higher in Pdk1-KO MEFs than in Pdk1 lox/lox MEFs during progression through the S and G 2 -M phases (Fig.  3D). The degradation of p27 Kip1 during the S and G 2 -M phases is regulated by the ubiquitin ligase SCF Skp2 (30). The seruminduced up-regulation of Skp2 was delayed in Pdk1-KO MEFs compared with that apparent in Pdk1 lox/lox MEFs (Fig. 3D), suggesting that PDK1 might contribute to Skp2 induction and that the delayed induction of Skp2 in Pdk1-KO MEFs is responsible for the increased level of p27 Kip1 in these cells during progression from S to G 2 -M. The level and time course of CDK2, CDK4, and ERK expression were not affected by the lack of PDK1 (Fig. 3D). The abundance of cyclin E was increased in serum-deprived Pdk1-KO MEFs, and the induction of cyclin A expression by serum stimulation was delayed in these cells in parallel with the delayed entry into the S phase (Fig. 3D). Cyclin D1 binds to and activates CDK4 and CDK6, resulting in the phosphorylation of Rb protein, which is required for G 1 -S transition (31,32). The phosphorylation of Rb protein on Ser 780 , which is targeted by the cyclin D1-CDK4 complex (33), was detected 6 h after serum stimulation in Pdk1 lox/lox MEFs but remained undetected in Pdk1-KO MEFs even 22 h after serum stimulation (Fig. 3D). These results suggested that the impairment in the induction of cyclin D1 in the PDK1-deficient cells results in failure of CDK4-mediated Rb phosphorylation.
To determine whether the effects of PDK1 deficiency on cyclin D1 and p27 Kip1 expression during cell cycle progression are mediated at the mRNA level, we performed Northern blot analysis. The amount of cyclin D1 mRNA was decreased in Pdk1-KO MEFs before serum stimulation, and the serum-induced increase in the abundance of this mRNA apparent in Pdk1 lox/lox MEFs was markedly suppressed in the Pdk1-KO cells (Fig. 3E). In contrast, the amount of p27 Kip1 mRNA in Pdk1-KO MEFs was similar to that in Pdk1 lox/lox MEFs at all times during cell cycle progression (Fig. 3E). These findings were confirmed by microarray-based analysis of gene expression in Pdk1 ϩ/ϩ /Cre and Pdk1-KO MEFs (supplemental Table 1). With the exception of cyclin D1 mRNA, the microarray data revealed no differences between the two cell types in the abundance of mRNAs for G 1 cyclins (such as cyclins D2, D3, E1, and E2) or for members of the Ink4 family of CDK inhibitors (such as p15, p18, and p19). These results suggested that PDK1 regulates the expression of cyclin D1 at the level of transcription or mRNA stability and that of p27 Kip1 at the level of translation or protein stability.
Expression of PDK1 Rescues the Cell Proliferation Defect and the Delayed Transition from G 0 -G 1 to S Phase in PDK1-deficient MEFs-To confirm that the defect in cell proliferation and the delayed transition from the G 0 -G 1 to S phase in Pdk1-KO MEFs was because of depletion of PDK1, we infected these cells with a retroviral vector encoding PDK1. The abundance of PDK1 in the resulting cells (Pdk1-KO/PDK1 MEFs) was only ϳ10% of that in Pdk1 lox/lox MEFs (Fig. 4A). Ectopic expression of PDK1 in Pdk1-KO MEFs resulted in partial rescue of the IGF-1-induced phosphorylation of Akt on Thr 308 , of GSK3␣ (Ser 21 ) and GSK3␤ (Ser 9 ), of p70 S6 kinase on Thr 389 , and of S6 on Ser 235 and Ser 236 (Fig. 4B).
Although these IGF-1-induced phosphorylation events were restored only partially, the rate of increase in the number of Pdk1-KO MEFs was fully restored by ectopic expression of PDK1 (Fig. 4C). Furthermore, the delay in the transition from the G 0 -G 1 to S phase apparent in Pdk1-KO MEFs was virtually eliminated in Pdk1-KO/PDK1 MEFs (Fig. 4D). The various defects in the expression of cyclin D1, cyclin A, cyclin E, Skp2, and p27 Kip1 as well as in the phosphorylation of Rb protein on Ser 780 during cell cycle progression of Pdk1-KO MEFs were also largely corrected by forced expression of PDK1 (Fig. 4E).

Role of Altered Regulation of Cyclin D1 and p27 Kip1 Expression in the Proliferation Defect of PDK1-deficient MEFs-We
further investigated whether the changes in the cell cycledependent expression of cyclin D1 or p27 Kip1 induced by loss of PDK1 were responsible for the proliferation defect and delayed transition from G 0 -G 1 to S phase in Pdk1-KO MEFs. Infection of Pdk1-KO MEFs with an adenoviral vector encoding cyclin D1 tagged with a nuclear localization signal (AdD1NLS) resulted in a marked increase in the total abundance of cyclin D1, whereas that with an adenoviral vector encoding an siRNA specific for p27 mRNA (Adp27siRNA) resulted in a pronounced decrease in the amount of p27 Kip1 (Fig. 5A). Infection with both AdD1NLS and Adp27siRNA increased the proliferation rate of Pdk1-KO MEFs, whereas infection with either vector alone had no such effect (Fig.  5B). Furthermore, infection with AdD1NLS alone or with both AdD1NLS and Adp27siRNA promoted cell cycle progression from the G 0 -G 1 to S phase in Pdk1-KO MEFs, whereas this transition was not affected by infection with Adp27siRNA alone (Fig. 5C).
Ectopic expression of p27 Kip1 inhibits CDK2 activity and thereby induces cell cycle arrest in the G 1 phase (34 -37).
Expression of D1NLS promoted cell cycle progression through the G 1 -S transition in Pdk1-KO MEFs despite the high level of p27 Kip1 in these cells (Fig. 5, A and C). Most p27 Kip1 is restricted to the nucleus of quiescent cells (38), but p27 Kip1 shuttles between the nucleus and cytoplasm in a manner dependent on specific phosphorylation events (39 -43). Given that nuclear localization of p27 Kip1 is required for its inhibition of CDK activity, the nuclear export of this protein promotes G 1 -S progression. We examined the localization of p27 Kip1 in Pdk1-KO MEFs expressing D1NLS by immunofluorescence analysis. Most p27 Kip1 was present in the nucleus of serum-deprived Pdk1 lox/lox MEFs infected with a control adenoviral vector for ␤galactosidase (AdLacZ) as well as in that of serum-deprived Pdk1-KO cells infected with AdLacZ or AdD1NLS (Fig. 5D). 9 h after re-exposure to serum, AdLacZ-infected Pdk1 lox/lox MEFs and AdD1NLS-infected Pdk1-KO MEFs showed a marked decrease in the amount of p27 Kip1 in the nucleus, whereas the localization of p27 Kip1 remained predominantly nuclear in AdLacZ-infected Pdk1-KO MEFs (Fig. 5D). These results suggested that cyclin D1 is required for the cytoplasmic translocation of p27 Kip1 and that this function of exogenous cyclin D1 negated the inhibitory effect of endogenous p27 Kip1 on G 1 -S progression in Pdk1-KO MEFs. Kip1 -Although overexpression of cyclin D1 promoted cell cycle progression from the G 0 -G 1 to S phase in Pdk1-KO MEFs (Fig. 5C), this effect was not accompanied by an increase in the rate of cell proliferation (Fig. 5B). To determine whether cell cycle progression was impaired at an additional stage in Pdk1-KO MEFs, we arrested Pdk1-KO and Pdk1-KO/PDK1 MEFs in early S phase by culturing them first in serum-free medium and then in complete medium containing aphidicolin. Aphidicolin binds directly to DNA polymerase and thereby blocks DNA synthesis, but it does not inhibit RNA or protein synthesis or prevent progression through the G 1 phase (44). The cells were harvested at various times after release from the  JUNE 20, 2008 • VOLUME 283 • NUMBER 25

JOURNAL OF BIOLOGICAL CHEMISTRY 17707
aphidicolin block and analyzed for cell cycle progression by flow cytometry (Fig. 6A). At 6 h after release from the aphidicolin block, most Pdk1-KO/PDK1 MEFs were in G 2 -M phase.
The proportion of these cells in G 2 -M phase had begun to decrease at 8 h after release, and most of the cells had entered G 1 phase by 10 h. However, the proportion of Pdk1-KO MEFs in G 2 -M phase was maximal at 8 h, remained largely unchanged at 10 h, and was greater than the corresponding value for Pdk1-KO/PDK1 MEFs at 12 h. These data indicated that the lack of PDK1 resulted in slowed progression through G 2 -M phase or in delayed mitosis.
To characterize further the progression from G 2 -M to G 1 phase in Pdk1-KO MEFs, we first cultured the cells in serum-free medium and then released them into medium containing the microtubule-destabilizing agent nocodazole to induce arrest at the G 2 -M phase (45). After exposure to nocodazole for 24 h, 72.6 and 84.3% of Pdk1-KO and Pdk1-KO/PDK1 MEFs, respectively, were arrested at G 2 -M (Fig.  6B). Removal of nocodazole allowed both Pdk1-KO and Pdk1-KO/PDK1 MEFs to re-enter the G 1 from the G 2 -M phase in a similar manner (Fig. 6B). The loss of PDK1 thus did not appear to affect progression of the cell cycle from G 2 -M to G 1 phase. These results suggested that PDK1 regulates the G 2 -M transition and that the defect in this transition in Pdk1-KO MEFs contributes to their reduced rate of proliferation.
The forced expression of cyclin D1 combined with the depletion of p27 Kip1 rescued the proliferation defect of Pdk1-KO MEFs (Fig. 5B).
To determine the effects of these two manipulations on the G 2 -M transition, we infected Pdk1-KO MEFs with AdD1NLS, Adp27siRNA, or both vectors and then synchronized the cells first at G 0 -G 1 phase by serum deprivation and then in early S phase by their release into complete medium containing aphidicolin. The combination of D1NLS and p27 siRNA promoted progression from G 2 -M to G 1 phase at 10 h after removal of aphidicolin, compared with cells infected with AdLacZ (Fig. 7). In contrast, neither AdD1NLS nor Adp27siRNA alone manifested such an effect. These results, together with those shown in Fig. 5B, dem- onstrate that the combination of both overexpression of cyclin D1 and depletion of p27 Kip1 is required for correction of the proliferation defect of Pdk1-KO MEFs.

DISCUSSION
We have shown that PDK1 is required for normal cell proliferation, with analysis of synchronized cells revealing that PDK1dependent signaling contributes to regulation not only of the transition from G 0 -G 1 to S phase but also of the G 2 -M transition during cell cycle progression in MEFs. Pdk1-KO MEFs thus exhibited a reduced rate of cell proliferation and a delay in cell cycle progression from G 0 -G 1 phase to S phase. Mice that lack PDK1 specifically in pancreatic ␤ cells were recently shown to have a reduced number of ␤ cells as a result of an increase in the prevalence of apoptosis (46). PDK1-deficient Drosophila embryos also showed extensive apoptosis throughout the entire body (13), suggesting that PDK1 modulates apoptosis during Drosophila development. However, we did not detect an increase in the level either of spontaneous apoptosis or of staurosporine-induced apoptosis in Pdk1-KO MEFs. Glioblastoma cells depleted of PDK1 either with an antisense oligonucleotide or by RNA interference did not show an increased frequency of apoptosis (18). The inhibition of proliferation observed in the glioblastoma cells expressing PDK1 siRNA was attributed predominantly to induction of G 1 arrest. The inhibition of cell cycle progression rather than the induction of apoptosis thus appears to be responsible for the reduced rate of increase in the number of Pdk1-KO MEFs observed in culture.
Proliferation of mammalian cells is regulated by complexes of cyclins and CDKs (47)(48)(49). Cyclin D1 is a key regulator of progression of the cell cycle from the G 1 to S phase (50,51). The amounts of cyclin D1 mRNA and protein remained low in Pdk1-KO MEFs after re-exposure of quiescent cells to serum. The FOXO subfamily of forkhead transcription factors regulates the transcription of several genes important for cell proliferation or survival (52)(53)(54), with FOXO1 having been shown to inhibit transcription of the genes for cyclins D1 and D2 (53). We found that the phosphorylation of FOXO1 on Ser 256 , which is catalyzed by Akt and inhibits FOXO1 function, was not induced by IGF-1 in Pdk1-KO MEFs (supplemental Fig.  1). Moreover, ectopic expression of PDK1 rescued this defect in IGF-1-induced phosphorylation of FOXO1 in Pdk1-KO MEFs (supplemental Fig. 1). These results suggest that the amount of cyclin D1 mRNA is regulated by a PDK1-Akt-FOXO1 signaling pathway and that the defect in cyclin D1 induction in Pdk1-KO MEFs may be attributable to the defect in FOXO1 phosphorylation.
Cyclin D1 controls cell cycle progression through the G 1 phase and the transition from the G 1 to S phase. In fibroblast cell lines or MEFs, overexpression of cyclin D1 accelerates transit from the G 0 -G 1 to S phase and promotes cell proliferation (55)(56)(57). Although forced expression of cyclin D1 promoted transition from the G 0 -G 1 to S phase in Pdk1-KO MEFs, it did not rescue the proliferation defect or the delayed G 2 -M progression also apparent in these cells. However, depletion of p27 Kip1 by RNA interference together with cyclin D1 overexpression corrected the proliferation defect and the delay in G 2 -M progression in Pdk1-KO MEFs, indicating that both upregulation of cyclin D1 and down-regulation of p27 Kip1 at the G 2 -M transition are essential for proliferation in MEFs.
The CDK inhibitor p27 Kip1 regulates progression of the cell cycle from G 1 to S phase. Overexpression of p27 Kip1 results in the accumulation of cells in G 1 phase and inhibits cell proliferation (58 -61). Most p27 Kip1 is located in the nucleus in the G 0 and early G 1 phase, but the protein appears transiently in the cytoplasm at the G 1 -S transition (62). Although Pdk1-KO MEFs exhibited a high level of p27 Kip1 expression, overexpression of cyclin D1 in the nucleus induced the export of p27 Kip1 from the   Fig. 5C were synchronized in early S phase by serum deprivation and aphidicolin treatment. They were harvested at the indicated times after release from aphidicolin block for analysis of DNA content by flow cytometry. The percentage of cells in G 0 -G 1 , S, and G 2 -M phases was determined. Data are means Ϯ S.E. from three independent experiments. nucleus to the cytoplasm and promoted the transition from G 0 -G 1 to S phase in these cells. Overexpression of cyclin D2 at G 0 phase was recently shown to promote the translocation of p27 Kip1 from the nucleus to the cytoplasm, whereas cyclins D1 and D3 did not show the same effect (63). We found that overexpression of cyclin D1 in the nucleus promoted the seruminduced nuclear export of p27 Kip1 at the transition from the G 0 -G 1 to S phase but not at G 0 phase in Pdk1-KO MEFs. It is thus possible that the serum-independent translocation of p27 Kip1 from the nucleus to the cytoplasm at the G 0 phase is regulated by cyclin D2 and that serum-dependent p27 Kip1 translocation at the transition from G 0 -G 1 to S phase is promoted by cyclin D1.
The abundance of p27 Kip1 is regulated by proteolysis and decreases markedly during progression from the G 0 to S phase (64,65). The degradation of p27 Kip1 in the cytoplasm by Kip1 ubiquitination-promoting complex is associated with the progression of cells from G 0 to S phase, whereas the nuclear degradation of p27 Kip1 by SCF Skp2 occurs during the S and G 2 phases (30). The serum-induced expression of Skp2 in Pdk1-KO MEFs was delayed compared with that in Pdk1 lox/lox MEFs, and this delay paralleled that of p27 Kip1 down-regulation during progression from the S to G 2 -M phase. Although the mechanisms responsible for regulation of Skp2 expression during cell cycle progression are not fully understood, treatment of human glioblastoma cell lines with the PI 3-kinase inhibitor LY294002 was shown to result in the selective depletion of Skp2 and accumulation of p27 Kip1 (66). PTEN (phosphatase and tensin homolog deleted on chromosome 10) functions as a negative regulator of PI 3-kinase signaling as a result of its activity as a lipid phosphatase for phosphatidylinositol 3,4,5-trisphosphate (67). PTEN deficiency in mouse embryonic stem cells resulted in an increase in the amount of Skp2 and a concomitant decrease in that of p27 Kip1 (68). Furthermore, the expression of Skp2 induced by PI 3-kinase signaling is mediated through regulation of binding of the transcription factor E2F1 to the proximal promoter of the SKP2 gene in pancreatic ductal adenocarcinoma cells (69). Depletion of Akt1 and Akt2 by RNA interference also resulted in down-regulation of Skp2 in the PTEN-negative prostate cancer cell lines PC3 and DU145 (70). These previous observations, together with our present results, indicate that PDK1 may control cell cycle progression from S to G 2 -M phase by inducing the expression of Skp2.
Although the role of PI 3-kinase signaling in progression through G 1 and into S phase has been well characterized, the demonstration that activation of PI 3-kinase and Akt is also required for progression through G 2 -M is relatively recent (71). This previous study also showed that the activity of Akt in HeLa cells increases as cells enter G 2 phase and progress to mitosis. Another study revealed that inhibitors of Akt or PI 3-kinase block the progression of PC12 cells released from nocodazoleinduced G 2 -M arrest (72). However, we found that Pdk1-KO MEFs released from nocodazole-induced G 2 -M arrest progressed normally to G 1 phase. ERK has been implicated in regulation of progression through M phase and into G 1 as well as through G 1 and G 2 -M (73). Given that ERK was found to be activated in serum-deprived Pdk1-KO MEFs, ERK signaling may act to coordinate passage through the M-G 1 transition in these cells. In addition to the increased basal activity of ERK in serum-deprived Pdk1-KO MEFs, we also detected an increased amount of PI 3-kinase activity associated with anti-phosphotyrosine immunoprecipitates prepared from these cells after their incubation with or without IGF-1 (supplemental Fig. 2). The catalytic p110 subunit of class IA PI 3-kinase is a downstream target and binding partner of Ras (74). Although the mechanism responsible for this phenomenon remains to be elucidated, the increased PI 3-kinase activity in serum-deprived Pdk1-KO MEFs may contribute to the increased ERK activity.
In conclusion, we have identified an important role for PDK1 in cell proliferation and cell cycle progression. Activation of PDK1 in response to mitogenic stimulation is thus necessary not only for the transcriptional up-regulation of cyclin D1 during cell cycle progression from G 0 -G 1 to S phase but also for the post-transcriptional down-regulation of p27 Kip1 and the induction of cyclin D1 at the G 2 -M transition. Our results thus indicate that cyclin D1 and p27 Kip1 are key components of the signaling pathway by which PDK1 regulates cell proliferation.