SHORT COMMUNICATION

The kinase-inhibitory domain of p21-activated kinase 1 (PAK1) inhibits cell

cycle progression independent of PAK1 kinase activity

M Thullberg1, A Gad1,3, A Beeser2, J Chernoff2 and S Stromblad1

1Department of Biosciences and Nutrition, Karolinska Institutet, Huddinge, Sweden; 2Tumor Cell Biology Program, Fox ChaseCancer Center, Philadelphia, PA, USA and 3Department of Laboratory Medicine, Karolinska Institutet, Huddinge, Sweden

p21-activated kinase 1 (PAK1) is a mediator of downstreamsignaling from the small GTPases Rac and Cdc42. In itsinactive state, PAK1 forms a homodimer where two kinasesinhibit each other in trans. The kinase inhibitory domain(KID) of one molecule of PAK1 binds to the kinase domainof its counterpart and keeps it inactive. Therefore, theisolated KID of PAK1 has been widely used to specificallyinhibit and study PAK function. Here, we show that theisolated KID induced a cell cycle arrest with accumulationof cells in the G1phase of the cell cycle with an inhibition ofcyclin D1 and D2 expression. This cell cycle arrest requiredthe intact KID and was also induced by a mutated KIDunable to block PAK1 kinase activity. Furthermore, theKID-induced cell cycle arrest could not be rescued by theexpression of a constitutively active PAK1-T423E mutant,concluding that this arrest occurs independently of PAK1kinase activity. Our results suggest that PAK1 through itsKID inhibits cyclin D expression and thereby enforces a cellcycle arrest. Our results also call for serious precaution inthe use of KID to study PAK function.Oncogene (2007) 26, 1820–1828. doi:10.1038/sj.onc.1209983;published online 25 September 2006

Keywords: p21-activated kinase 1 (PAK1); cell cycle;cyclin D1; Rac; GTPase

The p21-activated kinase (PAK) family members areserine/threonine protein kinases named by the fact thatthey can be activated by the small p21-GTPases Racand Cdc42 (Jaffer and Chernoff, 2002). Some of thePAKs have been implicated to be involved in cellu-lar transformation. For example, the PAK1 gene isamplified in ovarian and breast cancers (Bekri et al.,1997; Schraml et al., 2003). Furthermore, invasive breastcancer cells and tumors have increased PAK1 activity.In addition, the kinase-dead dominant-negative PAK1-K299R mutant (dn PAK1) inhibited transformationinduced by oncogenic ras whereas a kinase-active T423EPAK1 (ca PAK1) induced anchorage-independent

growth in breast epithelial cells (Tang et al., 1997;Vadlamudi et al., 2000).

As mediators of Cdc42 and Rac signaling, PAKs areinvolved in the control of cytoskeletal rearrangementssuch as disassembly of stress fibers and focal adhesioncomplexes (Manser et al., 1997, 1998; Zhao et al., 1998).These functions might mediate transformation-promot-ing effects of activated PAK1 leading to invasive motilecells and tumor progression and metastasis. In addition,PAKs have been implicated in the control of cell cycleprogression. In particular, the yeast PAK homologuesSte20 and Cla4 in Saccharomyces cerevisiae are primaryregulators of cell and actin polarization throughout thecell cycle and are essential for cytokinesis and mitosis(Cvrckova et al., 1995; Holly and Blumer, 1999; Weisset al., 2000). Recent studies show that mammalianPAK1 is also involved in the regulation of mitosis (Liet al., 2002; Zhao et al., 2005). Furthermore, by usinginhibitors of PAK1-3, Nheu et al. (2004) showed thatPAK activity is essential for renin–angiotensin system-induced upregulation of cyclin D1 and by preventingPAK1 expression by siRNA, Balasenthil et al. (2004)could reduce cyclin D1 mRNA expression. Thesefindings suggest a role for PAK1 in G1 progression.

Consequently, PAKs may play a role in oncogenesisboth by affecting cell motility and cell proliferation.However, the role of PAKs during mammalian cell cycleprogression is not fully clarified.

We wanted to elucidate whether PAK1 alone isinvolved in cell cycle regulation. To this end, dnPAK1, wt PAK1 and ca PAK1 were induced byremoval of tetracycline in mouse immortalized fibro-blasts S2-6-derived clones CL8, 1A1 and 6A9, respec-tively (Shockett et al., 1995; Sells et al., 1999), and cellcycle progression after synchronization by contactinhibition was analysed (Figure 1a). Surprisingly, wefound that both the dn PAK1 and the wt PAK1inhibited cell cycle progression and that these cells couldnot progress through the G1 phase. Induction of caPAK1 in the same cell line did not affect cell cycleprogression. The three clones expressed reasonablelevels of the respective PAK1 protein with approxi-mately 2 times higher expression compared to theendogenous (Figure 1b). We also transiently transfectedNIH 3T3 cells and the S2-6 subline 25A3 cultured inthe presence of tetracycline with myc-tagged constructsof wt PAK1, dn PAK1 and ca PAK1 (Figure 1c)

Received 13 January 2006; revised 19 July 2006; accepted 10 August2006; published online 25 September 2006

Correspondence: Dr M Thullberg, Department of Biosciences andNutrition, Karolinska Institutet, Huddinge 141 57, Sweden.E-mail: minna.thullberg@ki.se

Oncogene (2007) 26, 1820–1828& 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00

www.nature.com/onc

Figure 1 Overexpression of wt PAK1 and dn PAK1 inhibits cell cycle progression. (a) Expression of HA-tagged wt PAK1, dn PAK1and ca PAK1 was induced by tetracycline removal in the 1A1, CL8 and 6A9 cell lines of modified S26 mouse fibroblasts (Shockettet al., 1995; Sells et al., 1997) that were synchronized by contact inhibition for at least 48 h (Thullberg et al., 2000). Target proteins wereinduced 12 h before the cells were released to progress through the cell cycle. The cells were fixed and DNA stained with propidiumiodide (PI) and the cell cycle phase distribution was analysed by flowcytometry (FACS) as described by Bakhiet et al. (2001). Thepercentage of cells in the G1phase is given above the respective DNA profile. Displayed results are representative figures out of threeexperiments. (b) The levels of induced proteins compared to endogenous PAK1 were determined by immunoblot using the 12CA5anti-HA mAb, N-20 anti PAK1 pAb and b-actin as an indicator of equal loading. (c) Myc-tagged PAK1 and derivatives weretransiently expressed in NIH 3T3 mouse fibroblasts and the 25A3 cell line of S26 cells cultured with tetracycline. Forty-eight hoursafter transfection, BrdU was added to the culture media and after 1 h cells, were fixed and stained with 9E10 mAb against the myc tagand with G3G4 mAb against BrdU. BrdU incorporation in transfected cells was analysed by fluorescence microscopy with at least 300cells counted per condition. Expression levels of the different PAK1 constructs were similar. Shown is the average of three experiments(NIH3T3). The experiment was also repeated in 25A3 cells (right panel). Error bars indicate s.d. values and statistical analyses wereperformed using heteroscedastic one-tailed t-test.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1821

Oncogene

(Sells et al., 1999). Forty-eight hours after transfection,we pulsed the cells with bromodeoxyuridine (BrdU) for60 min and the fraction of transfected cells in S phasewas revealed by double immunofluorescence microscopyas described in Figure 2c legend. Expression levels of thedifferent proteins were similar (data not shown). Wefound that the fractions of cells in S phase were smallerin cells transfected with either dn PAK1 or wt PAK1compared to in cells transfected with empty vector,whereas ca PAK1 had no effect on BrdU incorporation(Figure 1c). Our results suggest that expression ofinactive PAK1 inhibits cell cycle progression in contrastto active PAK1, which did not. The fact that alsoexponentially growing cells were affected by dn andwt PAK1 suggests that the G1-phase progression isinhibited and not the G0 to G1 transition.

Given that PAK1 also binds the small GTPases Racand Cdc42, which were previously shown to have effectson cell cycle progression (Welsh et al., 2001), the effectsobserved with the PAK1 constructs may, not only bedue to a block in PAK1 signaling, but also to GTPase-inhibition and a block of their downstream signalingpathways. Consistent with this reasoning, dn PAK1inhibited cell cycle progression more efficiently than wtPAK1 in both NIH 3T3 and in 25A3 cells (Figure 1c).To block PAK1 activity in a different way independentof small GTPases, we used a 67-residue polypeptidederived from the PAK1 N-terminus amino acids 83–149corresponding to the PAK1 N-terminal kinase inhibi-tory domain (KID) (Zhao et al., 1998). In its inactivestate, PAK1 forms homodimers where the KID on onePAK1 molecule binds to and inhibits the kinase of itscounterpart in trans (Lei et al., 2000). Expression of theglutathione-S-transferase (GST)-tagged KID inhibitsPAK1 kinase activity without blocking the guaninetriphosphate loading of Rac or Cdc42 both in vivo andin vitro (Zhao et al., 1998) and has been widely used as aspecific inhibitor of PAK family members (Zenke et al.,1999; Lee et al., 2001; Li et al., 2001; Dadke et al., 2003;Balasenthil et al., 2004; Chi et al., 2004; DerMardi-rossian et al., 2004).

We used 25A3 cells which are mouse immortalizedfibroblasts S2-6 modified to express GST-KID in aninducible manner. 25A3 cells were cultured with orwithout the expression of GST-KID and cell cycledistribution was analysed by flow cytometry (Figure 2aand b). When we induced GST-KID in cells synchro-nized in G0/G1phase by contact inhibition, the cellswere unable to progress through the G1 phase within20 h after release (Figure 2a). However, the cells notexpressing KID had progressed through G1 phase andentered S phase (Figure 2a). At the time of replating (0 hafter release), both the cells with or without KIDexpression were synchronized in G0 phase to an equalextent (Figure 2a). To further analyse the effect of KIDon the cell cycle, we induced KID in exponentiallygrowing cells and subsequently analysed cell cycledistribution 48 h after induction. We found that KIDinduced an almost complete cell cycle arrest with anaccumulation of cells in the G1 phase of the cell cycle.Forty-eight hours after the start of GST-KID induction,

89% of the cells were in the G1 phase, as compared to65% of the asynchronously growing cells withoutinduction of KID (Figure 2b). In addition, we expressedGST and GST-KID in NIH 3T3 mouse fibroblasts andin the COS-7 and ECV304 tumor cell lines by transienttransfections and then analysed the fraction of trans-fected cells in S phase (Figure 2c and d). We found thatKID induced a cell cycle arrest in all cells tested,indicating that KID blocks proliferation of transformedtumor cells. The fact that cells expressing KID onlyaccumulated in the G1 phase indicated that KIDblocked critical events in the G1 phase but had no effectduring the S, G2 or Mphases. That KID induced a cellcycle arrest when induced in exponentially growing cellsindicates that the G1-phase progression is disturbed andnot the G0/G1 transition. These results also indicatethat the KID domain may be responsible for the cellcycle arrest enforced by wt PAK1 overexpression andthat this arrest occurs independently of Cdc42 and Rac.

In order to further study the role of KID during cellcycle progression and to map the part of KIDresponsible for the cell cycle arrest, we constructedmammalian expression vectors expressing GST alone orGST fusion proteins with full-length KID correspond-ing to the 83–149 amino-acid residues in PAK1, theN-terminal half of KID (aa 83–118), the C-terminal halfKID (aa 118–149), and two fragments covering themiddle of KID (aa 113–133 and aa 83–133) (Figure 3a).We expressed these constructs in 25A3 mouse fibroblastscultured with tetracycline and analysed how many of thetransfected cells were in S phase. This cell line made itpossible for us to induce GST-KID by the removalof tetracycline and use this as an internal control ofS-phase inhibition (data not shown). The expressionlevels of the different fragments and the full-length KIDwere similar but expression of only GST is generallymuch higher (data not shown). We found that only thefull-length intact KID significantly blocked cell cycleprogression (Figure 3a). In addition, we used a pointmutant of GST-KID to investigate whether inhibition ofPAK1 was essential for the cell cycle arrest (Figure 3band c). The point mutant KID-L107F, where theLeucine at position 107 is changed to Phenylalanine,was suggested not to inhibit PAK1 as this point mutantin the full-length PAK1 kinase creates a constitutivelyactive kinase. The Leucine resides at the PAK1 interfaceand is involved in KID binding and inhibiting theC-terminal kinase domain when PAK1 is in its inactivestate as a homodimer (Brown et al., 1996; Lei et al.,2000). The KID-L107F has been frequently used as aninactive KID control in studies of PAK1 functions(Dadke et al., 2003; Balasenthil et al., 2004; Chi et al.,2004; DerMardirossian et al., 2004). The GST-KID,GST-KID-L107F and GST alone were used in transienttransfections in NIH 3T3 and in 25A3 cultured withtetracycline (Figure 3b and c). Surprisingly, we foundthat expression of the GST-KID-L107F induced a cellcycle arrest to a similar degree as the GST-KID itself(Figure 3b and c). However, in spite of higher levels ofGST-KID-L107F, this construct could not efficientlyinhibit the activity of PAK1 induced by Rac1

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1822

Oncogene

Figure 2 Expression of the KID of PAK1 induces a cell cycle arrest. (a, b) 25A3 mouse fibroblasts modified to express GST-KID in atet-off system (Shockett et al., 1995), were grown with or without induction of the GST-KID. (a) 25A3 cells were synchronized in G0/G1 by contact inhibition and released with or without GST-KID and the cell cycle phase distribution was determined by PI-FACS 20hafter release (left panel) with percentages of cells in G1phase given above the profiles. The expression of GST-KID was analysed byimmunoblot (boxed panel) with anti-GST mAb, with expression of PAK1 as a control of equal loading. Note that the different lanesare from the same gel but are assembled for presentation purpose. (b) GST-KID was induced for 48 h in exponentially growing 25A3cells. Cell cycle distributions were analysed by PI-FACS with indicated percentages of cells in G1phase. (c, d) NIH 3T3 mousefibroblasts, COS-7 and ECV304 bladder carcinoma cells were transiently transfected with GST-KID or with GST as a control. Cellswere exposed to 30 mM BrdU for 1 h and fixed 48 h after transfection. (c) Representative microscopic images of NIH 3T3 cellstransfected with GST (upper panel) or GST-KID (lower panel). Transfected cells expressing BrdU were detected by doubleimmunoflourescence combining anti-GST mAb (left) and anti-BrdU mAb (middle). Nuclear staining was detected by Hoechst (right).Arrowheads indicate GST-expressing cells with BrdU incorporation whereas arrows display cells expressing GST-KID, which did notincorporate BrdU. (d) Bar graph shows quantification of the fraction of transfected cells in S phase that was quantified by counting 300cells per condition. All experiments were repeated at least twice and the graph shows representative experiments for each cell type.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1823

Oncogene

Figure 3 The complete KID of PAK1 induces cell cycle arrest independent of PAK1 inhibition. 25A3 cells cultured in the presence oftetracycline (a, c) and NIH 3T3 cells (b) were transfected with the full-length GST-KID corresponding to the amino acids 83–149 inPAK1, fragments of KID (a), or with a GST-KID L107F point mutant (b, c) as indicated. GST alone served as a control. The fractionsof transfected cells in the S phase were quantified as described above. (a) Bar graph illustrates the average of three experiments with s.d.indicated by error bars. The fraction of cells transfected with the full-length GST-KID was statistically different from the GST control:P-value given in the figure according to heteroscedastic one tailed t-test, whereas the other KID fragments did not enforce anystatistically significant effect. (b, c) The point mutant GST-KID-L107F incapable of inhibiting PAK1 (Zhao et al., 1998; Lei et al.,2000) was as efficient as the wild type GST-KID in inhibiting cell cycle progression in both NIH 3T3 (b) and 25A3 cells (c). Shown isone representative experiment (b) or averages of three experiments with s.d. shown as error bars and the P-values from statisticalanalyses by t-test indicated above the bars (c). (d) The efficiency of GST, GST-KID and GST-KID-L107F mutant to inhibit Rac-induced PAK1 activity was analysed by an in vitro kinase assay. NIH 3T3 cells were transfected with the indicated DNA and PAK1kinase activity was detected as 32P-incorporation to myelin basic protein as a substrate essentially as described (Sells et al., 1999), upperpanel. Quantification was made by phosphorimager (cyclon server, Packard) with the quantiquest software and given as DLU/mm2

arbitrary units, below. Expression levels of transfected proteins were evaluated by immunoblot from the same lysates, middle. Theinput plasmids indicated, below. (e) GST-KID was induced in 25A3 cells, which were transiently transfected with an empty vector (v),ca PAK1, wt PAK1 or dn PAK1. Vector-tranfected cells grown with tetracycline (vþ tet) served as control. The fraction of cells inS phase was determined by immostaining of BrdU incorporation in transfected cells as described above, upper panel. Middle panelshows an in vitro kinase assay with quantification as above, where ca PAK1 has high kinase activity even though GST-KID isexpressed. The immunoblot illustrates equal expression of tranfected proteins.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1824

Oncogene

(Figure 3d), verifying the inability of this mutant toaffect PAK1 activity (Zenke et al., 1999; Dadke et al.,2003; DerMardirossian et al., 2004).

To clarify whether the KID and PAK1 may act withinthe same molecular pathway in block of cell prolifera-tion, we investigated if ca PAK1 could rescue the cellcycle arrest enforced by KID. We hypothesized that ifKID induces a cell cycle block by binding and inhibitingPAK1 and/or hits the same molecular pathway as PAK1kinase activity, then the ca PAK1 would be able torescue this block. We transfected 25A3 cells with dnPAK1, wt PAK1 and ca PAK1 together with theexpression of GST-KID and analysed the fraction oftransfected cells that were in S phase (Figure 3e).Unexpectedly, but consistent with our results using theKID-mutant, expression of ca PAK1 did not rescue thecell cycle arrest induced by GST-KID. However, it wasnot obvious that the ca PAK1 mutant should render aPAK1 kinase that was resistant to the inhibition byKID. To test this, we transfected a myc-tagged ca PAK1into 25A3 cells with or without the induction of GST-KID and measured the kinase activity (Figure 3e). Thein vitro kinase assay showed that ca PAK1 is equallyactive with or without the expression of KID andconsequently demonstrated that KID did not inhibit caPAK1, unless KID dissociated from ca PAK1 in vitro.Dn PAK1 was used as a negative control and comparedwith wt PAK1 (Sells et al., 1997). Consistent withprevious reports, we found that expression of the KIDinhibited the kinase activity of wt PAK1 (Figure 3d–e),although the inhibition was not very dramatic using wtPAK1 alone that displays low activity without thecoexpression of Rac or Cdc42 (compare Figure 3d ande). Taken together, our results indicate firstly that KIDhas a target separate from PAK1 and secondly, that thistarget does not reside upstream of PAK1 kinase activityin PAK1 signaling pathways controlling cell prolifera-tion.

The D-type cyclins are normally induced in the earlyG1 phase and this accumulation is a prerequisite forG1 phase progression (Bartek et al., 1999). Therefore,we investigated whether the expression of KID mightaffect cyclin D protein levels. We used cells synchro-nized in the G0/G1 phase and released them with orwithout KID and analysed the levels of cyclin D1 andcyclin D2 by immunoblotting within 24 h after release(Figure 4a). The cells without KID progressed normallythrough the cell cycle and induced cyclin D1 and D2during G1-phase progression with detectable levelsalready 4 h after restart but with increased levelsthroughout the time course of observation, with thehighest cyclin D1 and D2 levels 24 h after release(Figure 4a). In contrast, the cells expressing KID didnot induce any detectable D-type cyclins within 24 hafter release (Figure 4a). Interestingly, the expression ofKID did not reduce the levels of activated extracellularsignal-regulated kinase (ERK) 1/2 nor activated AKT(S-473-PO4), suggesting that the downregulation ofcyclin D1 was not mediated through these signallingmolecules (Figure 4a). To control that KID wasexpressed after induction in these cells, we analysed

the protein levels of GST-KID during the experiment.Indeed, GST-KID could not be detected in the cellsgrown with tetracycline, but was heavily expressed in thecells cultured without tetracycline (Figure 4a). Similarly,in 1A1 cells expressing wt PAK1, there was a significantdecrease in cyclin D1 levels with no change in the levelsof activated ERK 1/2 (Figure 4b). However, theexpression of ca PAK1 had no reducing effect on cyclinD1 levels (data not shown), consistent with its lack ofcell cycle inhibitory activity (Figure 1a). We alsoanalysed cyclin D1 levels in cells expressing KIDcombined with ca PAK1 and found no detectableincrease of cyclin D1 by immunoblot (data not shown).

To investigate whether the KIDL107F mutant had acomparable impact on cyclin D1 levels, we transientlyexpressed wt KID and the mutant KIDL107F in 25A3cells and analysed cyclin D1 levels by immunofluore-scence (Figure 4c). We found fewer cells with cyclin D1in cells expressing any of the KID constructs comparedto GST alone (Figure 4c). Differences in detection levelsmay explain why 8–10% of cells have detectable cyclinD1 by immunofluorescence, while we cannot detectcyclin D1 in KID-expressing cells by immunoblot(Figure 4a and c). Our results suggest that the KIDmutant induced cell cycle arrest with low cyclin D1levels similar to the KID wt. To test the function of theKID regulation of cyclin D in the accompanying cellcycle arrest, GST-KID expression by removal oftetracycline was combined with transient transfectionof cyclin D1 in 25A3 cells (Figure 4d). In fact, cyclin D1overexpression rescued cell proliferation in the presenceof GST-KID (Figure 4d). With cyclin D1 expressed,we could detect 11% of the cells in S phase comparedto 1.5% in vector-transfected cells. Given that thetransient cyclin D1 transfection rate was 35% andthe fraction of cell in S phase in the control notexpressing KID was 32%, implicates that the maximumfraction of rescued cells could be 35% out of 32, whichequals 11%. Our data shows that expression of cyclinD1 could impose a complete rescue of the cell cyclearrest enforced by KID. Together, this shows that theKID-enforced block of cyclin D provides a functionalmechanism for the observed cell cycle inhibition.Because cyclin D1 alone cannot push cell cycle progres-sion from G0 (Resnitzky et al., 1994), this result alsodemonstrate that expression of GST-KID induces aG1-phase arrest. Our results further suggest that theKID region in wt PAK1 could act by a similarmechanism to suppress cell cycle G1 phase with down-regulation of cyclin D1.

This study shows that the KID of PAK1 inhibits cellcycle progression and arrests cells in the G1 phase.Furthermore, we show that the KID-mediated cell cyclearrest occurs independently of inhibition of PAK1kinase activity. Importantly, the GST-tagged KIDpeptide has been used in many functional studiesinvestigating the role of PAK kinases (Zhao et al.,1998; Zenke et al., 1999; Adam et al., 2000; Lee et al.,2001; Li et al., 2001; Dadke et al., 2003; Balasenthilet al., 2004; Chi et al., 2004; DerMardirossian et al.,2004). However, our finding that KID may have other

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1825

Oncogene

targets than PAK1–3 illuminates the fact that precau-tions must be taken when interpreting the results of suchstudies. The observation that KID arrested cells in theG1 phase demands further precautions interpretingresults from studies using this inhibitor, as signal

transduction might be very different in a G1 phase-arrested cell compared to a cycling cell.

We hypothesize that KID may bind to and inhibitother target(s) important for G1-phase progression andthereby either inhibits the transcription of D-type

Figure 4 Role of cyclin D in the regulation of the KID-imposed G1-arrest. (a) 25A3 cells were released from contact inhibition withor without the expression of GST-KID and were harvested at the indicated time points and protein levels of indicated proteins wereanalysed with PAK1 as loading control are shown. (b) 1A1 cells were released from contact inhibition with or without the expression ofwt PAK1. Protein levels of active ERK 42/44 and cyclin D1 were analysed by immunoblot at the indicated time points. (c) GST, GST-KID and the GST-KIDL107F mutant were expressed in 25A3 cells for 48 h. The cells were immunostained against GST and cyclin D1and the fraction of transfected cells expressing cyclin D1 was determined by immunofluorescence. The results from two experiments areshown and at least 200 cells were analysed for each treatments. (d) Cyclin D1 or vector was transfected into 25A3 cells with or withoutthe induction of GST-KID and the fraction of cells in S phase was determined by BrdU/PI FACS (Bakhiet et al., 2001). Shown isaverage of 3 experiments with s.d. error bars. The difference between cyclin D1-transfected and vector-transfected cells when KID wasexpressed was statistically significant as determined by student’s t-test. Analysis of cotransfected GFP showed that the averagetransfection efficiency was 35% for the three experiments. The rescue observed of 11% corresponds to what would be expected in thecase of a complete rescue, when comparing to the 32% of cells in S phase in the controls without KID and considering the transfectionefficiency. The immunoblot shows one representative experiment with expression of cyclin D1 and GST-KID and beta-actin as controlof equal loading.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1826

Oncogene

cyclins or enhances the degradation of those cyclins.One such candidate target might be CIB 1, a Ca2þ andkinase binding protein that interacts with PAK1 in theKID (Leisner et al., 2005). Importantly, loss of G1-phase control has been shown to be fundamental duringcancer development and expression of several tumorsuppressors enforces a G1-phase arrest. This studyshows that KID is able to block cell proliferation alsoof cancer cells. The 67 amino-acid long peptide of KIDmight therefore be useful to inhibit tumor growth iftransduced or expressed in tumor cells.

An interesting question is whether the KID region inits normal environment as a part of PAK1 equallyinhibits the cell cycle. In fact, our findings support thishypothesis. KID as a peptide imposed a solid G1-phasearrest and also dn PAK1 and wt PAK1 inhibited cellcycle progression with accompanying inhibition ofcyclin D expression (Figure 4b and data not shown).However, ca PAK1 did not affect the cell cycle norcyclin D1 levels (Figure1a and data not shown). Thedifferential effects by different PAK1 constructs mightbe due to the different PAK1 conformations formedwithin cells where only certain conformations expose theKID in PAK1 to the putative cell cycle regulator. Theconformation of PAK1 can be regulated by autophos-phorylation, lipid binding as well as binding to smallGTPases (Lei et al., 2000; Chong et al., 2001). AlsoPAK1, in most inhibited states, forms homodimerswhere the KID might not be accessible to other proteinsthan PAK1 (Lei et al., 2000; Parrini et al., 2002). Thismight explain the fact that the isolated KID had a muchmore pronounced effect on the cell cycle as compared towt PAK1 and dn PAK1. Thus, in the physiologicalsituation, it is likely that only a fraction of the PAK1 hasan exposed KID that could bind to other proteins andthereby inhibit the cell cycle. In addition, phosphory-lated PAK1 may not be accessible for the putative cellcycle regulator, as ca PAK1 did not inhibit cell cycleprogression. Although we found that the isolated KIDdomain of PAK1 causes cell cycle arrest, we cannot rule

out additional mechanisms for the cell cycle inhibitionby wt PAK1 and dn PAK1. PAK1 contains severalprotein–protein interaction domains, and interacts withproteins like PIX and Nck in addition to its Cdc42 andRac binding, and also has proven functions as scaffold(Manser et al., 1994, 1998; Bokoch et al., 1996; Galisteoet al., 1996; Thiel et al., 2002; Li et al., 2003; Sundberg-Smith et al., 2005). In fact, two recent studies show thatPAK1 is necessary for cell cycle progression (Balasenthilet al., 2004; Nheu et al., 2004). In addition, three studiesshow that ca PAK1 induces transcription of cyclin D1promotor (Dadke et al., 2003; Balasenthil et al., 2004;Chi et al., 2004).Considering also our findings, PAK1may have both cell cycle promoting and inhibitingproperties and the levels of phosphorylated active versusinactive PAK1 might finetune this function. It is alsopossible that PAK1 may regulate cyclin D1 both attranscriptional and post-transcriptional levels.

In summary, we found that KID inhibits cell cycleprogression independent of inhibition of PAK1 kinaseactivity, which demands serious precautions when usingthis peptide for investigating PAK1 functions. Wesuggest that KID and the KID domain in PAK1 mayinteract with a critical cell cycle regulator important forG1-phase progression through promotion of D-typecyclin expression.

Acknowledgements

We thank Dr Edward Manser for kindly providing thepXJGST-KID and pXJGSTKID-L107F constructs and theDevelopmental Studies Hybridoma Bank, University of Iowafor providing anti-BrdU mab G3G4 and anti-myc mab 9E10.This work was supported by grants to SS from the SwedishCancer Society and the Swedish Research Council and to MTfrom The Swedish Society of Medicine. SS holds a seniorscientist position from the Swedish Research Council and MTwas supported from The Swedish Cancer Society and TheSwedish Society of Medicine. AB and JC were supported bygrants from the National Institutes of Health.

References

Adam L, Vadlamudi R, Mandal M, Chernoff J,Kumar R. (2000). Regulation of microfilament reorganiza-tion and invasiveness of breast cancer cells by kinasedead p21-activated kinase-1. J Biol Chem 275: 12041–12050.

Bakhiet M, Tjernlund A, Mousa A, Gad A, Stromblad S,Kuziel WA et al. (2001). RANTES promotes growth andsurvival of human first-trimester forebrain astrocytes. NatCell Biol 3: 150–157.

Balasenthil S, Sahin AA, Barnes CJ, Wang RA, Pestell RG,Vadlamudi RK et al. (2004). p21-activated kinase-1 signal-ing mediates cyclin D1 expression in mammary epithelialand cancer cells. J Biol Chem 279: 1422–1428.

Bartek J, Lukas J, Bartkova J. (1999). Perspective:defects in cell cycle control and cancer. J Pathol 187:95–99.

Bekri S, Adelaide J, Merscher S, Grosgeorge J, Caroli-Bosc F,Perucca-Lostanlen D et al. (1997). Detailed map of a regioncommonly amplified at 11q13 – >q14 in human breastcarcinoma. Cytogenet Cell Genet 79: 125–131.

Bokoch GM, Wang Y, Bohl BP, Sells MA, Quilliam LA,Knaus UG. (1996). Interaction of the Nck adapterprotein with p21-activated kinase (PAK1). J Biol Chem271: 25746–25749.

Brown JL, Stowers L, Baer M, Trejo J, Coughlin S,Chant J. (1996). Human Ste20 homologue hPAK1 linksGTPases to the JNK MAP kinase pathway. Curr Biol 6:598–605.

Chi S, Chang S, Park D. (2004). Pak regulates calpain-dependent degradation of E3b1. Biochem Biophys ResCommun 319: 683–689.

Chong C, Tan L, Lim L, Manser E. (2001). The mechanism ofPAK activation. Autophosphorylation events in bothregulatory and kinase domains control activity. J Biol Chem276: 17347–17353.

Cvrckova F, De Virgilio C, Manser E, Pringle JR, Nasmyth K.(1995). Ste20-like protein kinases are required for normallocalization of cell growth and for cytokinesis in buddingyeast. Genes Dev 9: 1817–1830.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1827

Oncogene

Dadke D, Fryer BH, Golemis EA, Field J. (2003). Activationof p21-activated kinase 1-nuclear factor kappaB signaling byKaposi’s sarcoma-associated herpes virus G protein-coupledreceptor during cellular transformation. Cancer Res 63:8837–8847.

DerMardirossian C, Schnelzer A, Bokoch GM. (2004).Phosphorylation of RhoGDI by Pak1 mediates dissociationof Rac. GTPase Mol Cell 15: 117–127.

Galisteo ML, Chernoff J, Su YC, Skolnik EY, Schlessinger J.(1996). The adaptor protein Nck links receptor tyrosinekinases with the serine-threonine kinase. Pakl J Biol Chem271: 20997–21000.

Holly SP, Blumer KJ. (1999). PAK-family kinases regulatecell and actin polarization throughout the cell cycle ofSaccharomyces cerevisiae. J Cell Biol 147: 845–856.

Jaffer ZM, Chernoff J. (2002). p21-activated kinases: threemore join the Pak. Int J Biochem Cell Biol 34: 713–717.

Lee SH, Eom M, Lee SJ, Kim S, Park HJ, Park D. (2001).BetaPix-enhanced p38 activation by Cdc42/Rac/PAK/MKK3/6-mediated pathway. Implication in the regulationof membrane ruffling. J Biol Chem 276: 25066–25072.

Lei M, Lu W, Meng W, Parrini MC, Eck MJ, Mayer BJ et al.(2000). Structure of PAK1 in an autoinhibited conformationreveals a multistage activation switch. Cell 102: 387–397.

Leisner TM, Liu M, Jaffer ZM, Chernoff J, Parise LV. (2005).Essential role of CIB1 in regulating PAK1 activation andcell migration. J Cell Biol 170: 465–476.

Li F, Adam L, Vadlamudi RK, Zhou H, Sen S, Chernoff Jet al. (2002). p21-activated kinase 1 interacts with andphosphorylates histone H3 in breast cancer cells. EMBORep 3: 767–773.

Li W, Chong H, Guan KL. (2001). Function of the Rho familyGTPases in Ras-stimulated Raf activation. J Biol Chem 276:34728–34737.

Li Z, Hannigan M, Mo Z, Liu B, Lu W, Wu Y et al. (2003).Directional sensing requires G beta gamma-mediated PAK1and PIX alpha-dependent activation of Cdc42. Cell 114:215–227.

Manser E, Huang HY, Loo TH, Chen XQ, Dong JM, Leung Tet al. (1997). Expression of constitutively active alpha-PAKreveals effects of the kinase on actin and focal complexes.Mol Cell Biol 17: 1129–1143.

Manser E, Leung T, Salihuddin H, Zhao ZS, Lim L. (1994). Abrain serine/threonine protein kinase activated by Cdc42and Rac1. Nature 367: 40–46.

Manser E, Loo TH, Koh CG, Zhao ZS, Chen XQ, Tan L et al.(1998). PAK kinases are directly coupled to the PIX familyof nucleotide exchange factors. Mol Cell 1: 183–192.

Nheu T, He H, Hirokawa Y, Walker F, Wood J, Maruta H.(2004). PAK is essential for RAS-induced upregulation ofcyclin D1 during the G1 to S transition. Cell Cycle 3: 71–74.

Parrini MC, Lei M, Harrison SC, Mayer BJ. (2002).Pak1 kinase homodimers are autoinhibited in trans anddissociated upon activation by Cdc42 and Rac1. Mol Cell 9:73–83.

Resnitzky D, Gossen M, Bujard H, Reed SI. (1994).Acceleration of the G1/S phase transition by expression of

cyclins D1 and E with an inducible system. Mol Cell Biol 14:1669–1679.

Schraml P, Schwerdtfeger G, Burkhalter F, Raggi A,Schmidt D, Ruffalo T et al (2003). Combined array com-parative genomic hybridization and tissue microarrayanalysis suggest PAK1 at 11q13.5-q14 as a critical oncogenetarget in ovarian carcinoma. Am J Pathol 163: 985–992.

Sells MA, Boyd JT, Chernoff J. (1999). p21-activated kinase 1(Pak1) regulates cell motility in mammalian fibroblasts. JCell Biol 145: 837–849.

Sells MA, Knaus UG, Bagrodia S, Ambrose DM,Bokoch GM, Chernoff J. (1997). Human p21-activatedkinase (Pak1) regulates actin organization in mammaliancells. Curr Biol 7: 202–210.

Shockett P, Difilippantonio M, Hellman N, Schatz DG.(1995). A modified tetracycline-regulated system providesautoregulatory, inducible gene expression in culturedcells and transgenic mice. Proc Natl Acad Sci USA 92:6522–6526.

Sundberg-Smith LJ, Doherty JT, Mack CP, Taylor JM.(2005). Adhesion stimulates direct PAK1/ERK2 associationand leads to ERK-dependent PAK1 Thr212 phosphoryla-tion. J Biol Chem 280: 2055–2064.

Tang Y, Chen Z, Ambrose D, Liu J, Gibbs JB, Chernoff Jet al. (1997). Kinase-deficient Pak1 mutants inhibitRas transformation of Rat-1 fibroblasts. Mol Cell Biol 17:4454–4464.

Thiel DA, Reeder MK, Pfaff A, Coleman TR, Sells MA,Chernoff J. (2002). Cell cycle-regulated phosphorylation ofp21-activated kinase 1. Curr Biol 12: 1227–1232.

Thullberg M, Bartek J, Lukas J. (2000). Ubiquitin/protea-some-mediated degradation of p19INK4d determinesits periodic expression during the cell cycle. Oncogene 19:2870–2876.

Vadlamudi RK, Adam L, Wang RA, Mandal M, Nguyen D,Sahin A et al. (2000). Regulatable expression of p21-activated kinase-1 promotes anchorage-independent growthand abnormal organization of mitotic spindles in humanepithelial breast cancer cells. J Biol Chem 275: 36238–36244.

Weiss EL, Bishop AC, Shokat KM, Drubin DG. (2000).Chemical genetic analysis of the budding-yeast p21-activated kinase Cla4p. Nat Cell Biol 2: 677–685.

Welsh CF, Roovers K, Villanueva J, Liu Y, Schwartz MA,Assoian RK. (2001). Timing of cyclin D1 expression withinG1 phase is controlled by Rho. Nat Cell Biol 3: 950–957.

Zenke FT, King CC, Bohl BP, Bokoch GM. (1999).Identification of a central phosphorylation site in p21-activated kinase regulating autoinhibition and kinaseactivity. J Biol Chem 274: 32565–32573.

Zhao ZS, Lim JP, Ng YW, Lim L, Manser E. (2005). TheGIT-Associated Kinase PAK Targets to the Centrosomeand Regulates Aurora-A. Mol Cell 20: 237–249.

Zhao ZS, Manser E, Chen XQ, Chong C, Leung T, Lim L.(1998). A conserved negative regulatory region in alpha-PAK: inhibition of PAK kinases reveals their morphologicalroles downstream of Cdc42 and Rac1. Mol Cell Biol 18:2153–2163.

PAK1 kinase-inhibitory domain induces cell cycle arrestM Thullberg et al

1828

Oncogene