Brief Communication 1439

Inhibition of PDK-1 activity causes a reduction in cellproliferation and survivalPeter Flynn*†, Michelle Wong*†, Maryam Zavar*, Nicholas M. Dean‡

and David Stokoe*

3-Phosphoinositide-dependent kinase-1 (PDK-1) wasidentified by its ability to phosphorylate and activateprotein kinase B (PKB) in vitro [1,2] and canphosphorylate and activate additional protein kinasesin the AGC family in vitro [3–6]. Its role in vivo has,however, only begun to be addressed. We usedantisense oligonucleotides directed against PDK-1expression to explore the role of PDK-1 in humanglioblastoma cells (U87-MG), which express a mutantPTEN allele. Reduction in PDK-1 levels resulted ininhibition of PKB activity, and a reduction inphosphorylation on Thr308 and Ser473 of PKB. p70 S6kinase (p70S6K) activity was also reduced. Cellproliferation was dramatically inhibited followingtreatment with PDK-1 antisense oligonucleotides, dueto a combination of decreased cell doubling and anincrease in apoptosis. This is in contrast to directinhibition of phosphoinositide 3-OH kinase (PI3-kinase), which results in G1 arrest with no effect onapoptosis. This study confirms both PKB and p70S6K asin vivo substrates for PDK-1. The effect of acute PDK-1loss on cell proliferation and survival suggests theinvolvement of PI 3-kinase dependent and independentsignaling events, and implicates PDK-1 as a potentialtherapeutic target for human neoplasms.

Addresses: *Cancer Research Institute, University of California, SanFrancisco, California 94115, USA. ‡Department of Pharmacology, ISISPharmaceuticals, Carlsbad, California 92008, USA.

Correspondence: David StokoeE-mail: stokoe@cc.ucsf.edu

†These authors contributed equally to this work.

Received: 26 May 2000Revised: 14 August 2000Accepted: 20 September 2000

Published: 3 November 2000

Current Biology 2000, 10:1439–1442

0960-9822/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.

Results and discussionTo evaluate the importance of the role of PDK-1 in PI3-kinase and PKB signaling in vivo we designed and syn-thesized a panel of 15 chimeric phosphorothioate anti-sense oligonucleotides targeted against the nucleotidesequence of human PDK-1 [7,8]. All 15 oligonucleotides(ISIS 20466–20480, here termed 66–80) exhibited activity

to various degrees when assayed for their ability to reducePDK-1 mRNA levels without affecting β-actin mRNAlevels. The results for oligonucleotides 71–75 are shown inFigure 1a. We chose two oligonucleotides that were activein several independent experiments (66 5′-AGCTGG-CTGGTGGTCCTGGC-3′ and 71 5′-TGAATGATGCC-CTTGCCGTG-3′, which correspond to nucleotides 4–23(amino acids 2–8) and 640–659 (amino acids 197–203),respectively), and synthesized control oligonucleotidescontaining 2, 4, 6 and 8 base-pair (bp) mismatches. Intro-duction of just two mismatches almost completely abol-ished the ability to inhibit PDK-1 RNA levels (Figure 1b).Oligonucleotide 71 consistently inhibited PDK-1 RNAexpression by > 90%, so this, together with its 8 bp mis-match 98, was chosen for further experiments. All experi-ments were also repeated and confirmed with antisenseoligonucleotide 66 and its 8 bp mismatch 94.

To determine the effects of PDK-1 antisense oligonu-cleotides on levels of PDK-1 protein and on the activityof PKB, we used U87-MG cells, a glioblastoma cell linethat expresses a truncated and inactive form of the tumorsuppressor phospholipid phosphatase, PTEN, resultingin increased PKB activity [9,10]. Over a 96 hour post-transfection time course, oligonucleotide 71 dramaticallyreduced PDK-1 protein levels. Cell numbers increased(data not shown), hence PKB and PDK-1 protein(Figure 1) increased during this period in cells treatedwith oligonucleotide 98. Reduction in PDK-1 levelscaused a reduction in PKB activity, similar to that causedby the PI 3-kinase inhibitor LY294002 (Figure 1d). Thissuggests that PDK-1 is absolutely required for PKB acti-vation in U87-MG cells. The polyclonal antisera we usedto immunoprecipitate PKB recognizes all isoforms (α, βand γ), and the activity of all three was reduced followingtreatment with PDK-1 antisense oligonucleotides (datanot shown).

In vitro, PDK-1 catalyzes phosphorylation of PKB exclu-sively on Thr-308 [1,2,7,8]. In vivo, however, PKBbecomes phosphorylated on both Thr308 and Ser473 fol-lowing growth factor stimulation [11]. Both phosphoryla-tion sites are dramatically reduced following treatment ofU87-MG cells with antisense oligonucleotide 71 but not98 (Figure 2a). This suggests that Ser473 is not phospho-rylated by a distinctly regulated protein kinase, but doesnot distinguish between direct, PDK-1 mediated, phos-phorylation or PKB autophosphorylation. In addition toPKB, PDK-1 can also phosphorylate other members of

the AGC protein kinase family, including p70S6K, p90ribosomal S6 kinase (p90rsk) and protein kinase C relatedkinases 1 and 2 (PRKs 1 and 2), at the equivalent phos-phorylation site in the activation loop [3–6]. p70S6K activ-ity is inhibited following treatment with PDK-1 antisenseoligonucleotides (Figure 2b). In contrast to p70S6K, p90rsk

and PRKs 1 and 2 were not inhibited by PDK-1 antisenseoligonucleotides (data not shown). These AGC kinase

family members cannot be ruled out as in vivo PDK-1 sub-strates, however, as kinase inhibition via our short-termtreatment depends on both phosphorylation and dephos-phorylation of the PDK-1 site. Indeed, for both p90rsk andthe PRKs this site is highly resistant to protein phos-phatases ([12], and M. Frodin, personal communication).

We next looked at the biological consequences of adecrease in PDK-1 protein. PDK-1 antisense caused a dra-matic inhibition of cell proliferation over a 7 day periodfollowing transfection (Figure 3a). The degree of inhibi-tion was similar to that achieved by LY294002, whichcauses arrest of U87-MG cells in the G1 phase of the cellcycle [13–15]. We therefore investigated whether theeffects of PDK-1 antisense on cell proliferation were alsomediated through this mechanism. PDK-1 antisense hadno effect on the cell-cycle profile of proliferating U87-MGcells (Figure 3b). They also had only a very modest abilityto prevent G2 arrest induced by nocodazole (Figure 3b),unlike LY294002 (Figure 3b), and PTEN ([14], and datanot shown). Therefore, a G1 arrest is unlikely to be thesole reason for the inhibition of cell proliferation. Ourresults suggest a pathway emanating from PI 3-kinase thatcontrols a G1 checkpoint distinct from that involvingPDK-1 and PKB. In support of this, we have been unableto show G1 arrest following expression of a catalyticallycompromised form of PKB which inhibits endogenousPKB activity, nor have we been able to prevent the G1arrest induced by LY294002 or PTEN by expressing anactivated form of PKB (data not shown).

Despite previous experiments implicating PI 3-kinaseand PKB in cell survival signaling, no observableincreases in apoptosis are induced by either LY294002

1440 Current Biology Vol 10 No 22

Figure 1

Current Biology

Con

trol

71 72 73 74 75 Con

trol

7195969798

PDK1mRNA

β-actin

98 71

Con

trol

LY (h)

PK

B A

ctiv

ity

0

Contro

l

LY294002

98 24 h

98 48h

98 72h

98 96h

71 24h

71 48h

71 72h

71 96h

25,000

50,000

75,000

100,000(a) (b) (c) (d)

PDK1mRNA

β-actin

PDK1protein

PKBprotein

24 48

72 96

24 48

72

96

02468Number ofmismatches

Identification of active PDK-1 antisense oligonucleotides. (a) HeLacells were transfected with oligonucleotides ISIS 20471–20475(designated 71–75) or with lipofectin alone (Control), and mRNAwas extracted 24 h later. Northern blots were probed with humanPDK-1 nucleotide sequence, then stripped and reprobed for β-actin.(b) HeLa cells were transfected with either lipofectin alone oroligonucleotides ISIS 20471, 30795–30798 (denoted as 71 and95–98), which contain 0, 2, 4, 6 and 8 bp of mismatch. (c) U87-MG

cells were transfected with either lipofectin alone or oligonucleotides71 and 98 (300 nM). Cells were lysed at the times indicated (48 h inthe case of the control and LY294002 (LY)-treated cells (20 µM LYfor 20 min)). Equal volumes of lysate were separated by SDS–PAGEand western blotted using PDK-1- and PKB-specific antibodies.(d) The lysates from (c) were used to immunoprecipitate PKB whichwas then assayed for in vitro kinase activity as described inSupplementary material.

Figure 2

Inhibition of PDK-1 levels reduces PKB phosphorylation on both Thr308and Ser473, and inhibits p70S6K activity. U87-MG cells were transfectedwith either lipofectin alone (Control), or with PDK-1 oligonucleotides 71and 98 and harvested 48 h later. LY294002 (20 µM) was added to thecells 20 min before harvest. (a) PKB activity was assessed as describedfor Figure 1. Equal portions of the lysate were separated by SDS–PAGEand western blotted using antibodies specific for the Thr308 andSer473 phosphorylated forms of PKB, and with antibody that recognizestotal PKB. (b) p70S6K was immunoprecipitated from equal amounts oftotal protein and assayed for activity as described in the Supplementarymaterial. A portion of the lysate was separated by SDS–PAGE andwestern blotted using antibodies specific for p70S6K.

Current Biology

(a) (b)

PK

B a

ctiv

ity

p70S

6K a

ctiv

ity

Total PKB

Ser473

Thr308

p70S6K

010,00020,00030,00040,00050,00060,000

0

50,000

100,000

150,000

Contro

l 9

8 71 LY

Contro

l 9

8 71 LY

or PTEN in attached U87-MG cells [14,16]. TerminaldUTP nick end labeling (TUNEL) experiments showed,however, that PDK-1 antisense oligonucleotides causedsubstantial apoptosis 48 and 72 hours after oligo-nucleotide administration, effects similar in magnitude tothose of tumor necrosis factor (TNF) in the presence ofcycloheximide, although somewhat delayed (Figure 4a,b).The difference between PI 3-kinase inhibition andPDK-1 antisense treatment in their relative abilities to

cause apoptosis suggests at least two possibilities. Thefirst is that PDK-1 has substrates in addition to PKB orp70S6K that could mediate a survival signal independentof PI 3-kinase signaling. The second possibility is thatPKB (or another protein in the PI 3-kinase pathway) isthe relevant survival substrate, but that inhibition of PI3-kinase provides additional signals that prevent apopto-sis. One possibility in this regard is that cells arrestedin G1 are protected from apoptosis. This situation has been

Brief Communication 1441

Figure 3

Current Biology

G1 62%S 7%G2/M 25%

G1 11%S 5%G2/M 78%

– Nocodazole + Nocodazole – Nocodazole + Nocodazole

G1 77%S 1%G2/M 20%

G1 78%S 1%G2/M 19%

G1 59%S 6%G2/M 26%

G1 31%S 5%G2/M 58%

G1 59%S 7%G2/M 27%

G1 18%S 5%G2/M 71%

Cel

l num

ber

3 4 5 6

Days post-transfection

LY294002

66

94

Control

0

100,000

200,000

300,000

400,000

500,000

600,000

(b)(a)

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Cou

nts

PI

Lipofectin

LY294002

71

98

PDK-1 antisense oligonucleotides abolish cell proliferation butcause little G1 arrest. (a) Subconfluent U87-MG cells weretransfected in 15 cm dishes with the indicated oligonucleotides, ortreated with 20 µM LY294002. At the indicated times aftertransfection cells were trypsinized and viable cells counted by trypan

blue exclusion. (b) U87-MG cells were transfected with the indicatedoligonucleotides or lipofectin alone, or treated with 20 µMLY294002, and harvested for FACS analysis 72 h later. Whereindicated, nocodazole (70 ng/ml) was added as described inSupplementary material.

Figure 4

Apo

ptot

ic In

dex

(%)

20 30 40 50 60 70 80Hours post-transfection

Current Biology

Lipofectin 71 98(b)(a) TNF+

cycloheximide

0

5

10

15

20

25

Loss of PDK-1 activity results in apoptosis. (a) TUNEL staining ofU87-MG cells 48 h after transfection with PDK-1 antisense andmismatch oligonucleotides, or treatment with 10 ng/ml TNF and10 µg/ml cycloheximide. Hoechst DNA staining is shown in the toppanels to indicate total cells present, and cells positive forincorporation of rhodamine dUTP are stained red in the bottom panels.

(b) Quantitation of the percentage of cells staining positive for theTUNEL assay. Four fields of approximately 400 cells were picked atrandom and the number of rhodamine-staining cells over the number oftotal cells is indicated. Filled squares, control; filled diamonds,oligonucleotide 98; filled triangles, oligonucleotide 71; filled circles,LY294002; stars, TNF + cycloheximide.

described in primary keratinocytes, where induction ofgrowth arrest in G1 protected cells from apoptosis in-duced by ultraviolet radiation [17].

We used time-lapse videomicroscopy to analyze the effectsof PDK-1 antisense oligonucleotides on U87-MG prolifera-tion and survival over 48 hours, beginning 24 hours post-transfection. The PDK-1-antisense-treated cells weremorphologically distinct from controls, with enlarged, flat-tened cell bodies and reduced motility. Increased apopto-sis was also seen towards the end of the time period,confirming the TUNEL experiments (Figure 4). Surpris-ingly, and in contrast to the FACS results, there was also adramatic effect on the cell cycle, as seen by a decrease inthe number of mitoses in cells treated with PDK-1 anti-sense relative to controls when normalized for cell number(see Supplementary material).

While this paper was in preparation, Williams et al. [18]published an analysis of embryonic stem (ES) cells inwhich a lack of PDK-1 protein expression had been gen-erated by homologous recombination. They showed thatPKB activity was abolished, in agreement with our study.However, in contrast to our results , which show simulta-neous inhibition of Thr308 and Ser473 phosphorylation,they showed that although Thr308 phosphorylationwas bolished, Ser473 phosphorylation was almost unaf-fected. The reason for this discrepancy is unclear, andshows that the mechanism and role of PKB Ser473 phos-phorylation in different cell types under different condi-tions remains to be clarified. The PDK-1–/– ES cells werealso defective in their ability to cause p70S6K activation inresponse to IGF-1; this is consistent with our datashowing inhibition of p70S6K by PDK-1 antisenseoligonucleotides. Also, whereas loss of PDK-1 expressionin U87-MG cell causes a dramatic decrease in cell prolif-eration, the PDK-1–/– ES cells were viable and grew nor-mally. All these discrepancies suggest that the role ofPDK-1 may be distinct in different cell types and thatthe acute removal of PDK-1 protein is not functionallythe same as genetic deletion.

The effect of PDK-1 antisense oligonucleotides on prolif-eration and apoptosis in U87-MG cells suggests thatPDK-1 inhibition might be a useful therapeutic strategyunder conditions of chronic PI 3-kinase signaling. Investi-gation into which PDK-1 substrates mediate these effectsis currently underway.

Supplementary materialSupplementary material including detailed methodologies, a figure showingmitotic index and movies of U87-MG cell proliferation after antisense treat-ments is available at http://current-biology.com/supmat/supmatin.htm.

AcknowledgementsWe thank Vanessa Taylor, Daphne Haas-Kogan and Frank McCormick forvaluable discussions, Clodagh O’Shea for help with FACS analysis and

David Hancock for help with time-lapse videomicroscopy. This work wassupported by NIH grant RO1CA79548 to D.S. and grant IRG-97-150-01from the American Cancer Society to D.S.

References1. Stokoe D, Stephens LR, Copeland T, Gaffney PRJ, Reese CB,

Painter GF, et al.: Dual role of phosphatidylinositol-3,4,5-trisphosphate in the activation of protein kinase B. Science 1997,277:567-570.

2. Alessi DR, James SR, Downes CP, Holmes AB, Gaffney PRJ,Reese CB, et al.: Characterisation of a 3-phosphoinositide-dependent protein kinase which phosphorylates and activatesprotein kinase Bαα. Curr Biol 1997, 7:261-269.

3. Alessi DR, Kozlowski MT, Weng QP, Morrice N, Avruch J:3-Phosphoinositide-dependent protein kinase 1 (PDK1)phosphorylates and activates the p70 S6 kinase in vivo andin vitro. Curr Biol 1998, 8:69-81.

4. Pullen N, Dennis PB, Andjelkovic M, Dufner A, Kozma SC,Hemmings BA, et al.: Phosphorylation and activation of p70s6k byPDK1. Science 1998, 279:707-710.

5. Jensen CJ, Buch MB, Krag TO, Hemmings BA, Gammeltoft S,Frodin M: 90 kDa ribosomal S6 kinase is phosphorylated andactivated by 3-phosphoinositide-dependent protein kinase-1.J Biol Chem 1999, 274:27168-27176.

6. Flynn P, Mellor H, Casamassima A, Parker PJ: Rho GTPase control ofprotein kinase C-related protein kinase activation by 3-phosphoinositide-dependent protein kinase. J Biol Chem 2000,275:11064-11070.

7. Alessi DR, Deak M, Casamayor A, Caudwell FB, Morrice N,Norman DG, et al.: 3-Phosphoinositide-dependent proteinkinase-1 (PDK1): structural and functional homology with theDrosophila DSTPK61 kinase. Curr Biol 1997, 7:776-789.

8. Stephens L, Anderson K, Stokoe D, Erdjument-Bromage H,Painter GF, Holmes AB, et al.: Protein kinase B kinases thatmediate phosphatidylinositol 3,4,5-trisphosphate dependentactivation of protein kinase B. Science 1998, 279:710-714.

9. Haas-Kogan D, Shalev N, Wong M, Mills G, Yount G, Stokoe D:PKB/akt activity is elevated in glioblastoma cells due to mutationof PTEN/MMAC1. Curr Biol 1998, 8:1195-1198.

10. Myers MP, Pass I, Batty IH, Van der Kaay J, Stolarov JP,Hemmings BA, et al.: The lipid phosphatase activity of PTEN iscritical for its tumor suppressor function. Proc Natl Acad Sci USA1998, 95:13513-13518.

11. Alessi DR, Andjelkovic M, Caudwell B, Cron P, Morrice N, Cohen P,et al.: Mechanism of activation of protein kinase B by insulin andIGF-1. EMBO J 1996, 15:6541-6551.

12. Balendran A, Biondi RM, Cheung PC, Casamayor A, Deak M,Alessi DR: A 3-Phosphoinositide-dependent protein kinase-1(PDK1) docking site is required for the phosphorylation of proteinkinase cζζ and PKC-related kinase 2 by PDK1. J Biol Chem 2000,275:20806-20813.

13. Li DM, Sun H: PTEN/MMAC1/TEP1 suppresses the tumorigenicityand induces G1 cell cycle arrest in human glioblastoma cells.Proc Natl Acad Sci USA 1998, 95:15406-15411.

14. Furnari FB, Huang HJ, Cavenee WK: The phosphoinositolphosphatase activity of PTEN mediates a serum-sensitive G1growth arrest in glioma cells. Cancer Res 1998, 58:5002-5008.

15. Cheney IW, Neuteboom ST, Vaillancourt MT, Ramachandra M,Bookstein R: Adenovirus-mediated gene transfer of MMAC1/PTENto glioblastoma cells inhibits S phase entry by the recruitment ofp27Kip1 into cyclin E/CDK2 complexes. Cancer Res 1999,59:2318-2323.

16. Tamura M, Gu J, Danen EH, Takino T, Miyamoto S, Yamada KM:PTEN interactions with focal adhesion kinase and suppression ofthe extracellular matrix-dependent phosphatidylinositol 3-kinase/Akt cell survival pathway. J Biol Chem 1999, 274:20693-20703.

17. Chaturvedi V, Qin JZ, Denning MF, Choubey D, Diaz MO,Nickoloff BJ: Apoptosis in proliferating, senescent, andimmortalized keratinocytes. J Biol Chem 1999, 274:23358-23367.

18. Williams MR, Arthur JSC, Balendran A, van der Kaay J, Poli V,Cohen P, et al.: The role of 3-phosphoinositide-dependent proteinkinase 1 in activating AGC kinases defined in embryonic stemcells. Curr Biol 2000, 10:439-448.

1442 Current Biology Vol 10 No 22