Cell surface receptors activate p21-activated kinase 1 via multiple Ras and PI3-kinase-dependent pathways

www.elsevier.com/locate/cellsig

Cellular Signalling 15 (2003) 1099–1109

Cell surface receptors activate p21-activated kinase 1 via multiple Ras and

PI3-kinase-dependent pathways

Raymond E. Menarda, Raymond R. Mattinglya,b,*

aDepartment of Pharmacology, Wayne State University, 540 E. Canfield, Room 6326, Detroit, MI 48201, USAbBarbara Ann Karmanos Cancer Institute, 540 E. Canfield, Detroit, MI, USA

Received 10 February 2003; received in revised form 15 April 2003; accepted 12 May 2003

Abstract

p21-activated kinases (PAKs) were the first identified mammalian members of a growing family of Ste20-like serine– threonine protein

kinases. In this study, we show that PAK1 can be stimulated by carbachol, lysophosphatidic acid (LPA), epidermal growth factor (EGF), and

phorbol 12-myristate 13-acetate (PMA) by multiple independent and overlapping pathways. Dominant-negative Ras, Rac, and Cdc42

inhibited PAK1 activation by all of these agonists, while active Rac1 and Cdc42 were sufficient to maximally activate PAK1 in the absence of

any treatment. Active Ras induced only a weak activation of PAK1 that could be potentiated by muscarinic receptor stimulation. Studies

using inhibitors of the EGF receptor tyrosine kinase, phosphatidylinositol 3-kinase (PI3-kinase) and protein kinase C (PKC) revealed that all

of the cell surface agonists could activate PAK1 through pathways independent of PKC, that EGF stimulated a PI3-kinase dependent pathway

to stimulate PAK1, and that muscarinic receptor stimulation of PAK1 was predominantly mediated through this EGF-R-dependent

mechanism. Activation of PAK1 by LPA was independent of PI3-kinase and the EGF receptor, but was inhibited by dominant-negative

RhoA. These results identify multiple Ras-dependent pathways to activation of PAK1.

D 2003 Elsevier Inc. All rights reserved.

Keywords: p21-activated kinase; Protein kinase C; Muscarinic receptor; Epidermal growth factor receptor

1. Introduction

Cell transformation requires modulation of signals from

oncogenes to stimulate proliferation, actin cytoskeleton

rearrangement, and cell survival. Mitogen-activated protein

kinase (MAPK) cascades control the expression of genes

that are important for the regulation of these cell functions.

In mammalian cells, several parallel MAPK pathways have

been extensively studied in recent years [1]. The ERK/

MAPK cascade is regulated by Ras GTPase in response to

agonists of tyrosine kinase and G protein-coupled receptors

(GPCRs). The MAPK p38 and Jun kinase (JNK) cascades

0898-6568/$ - see front matter D 2003 Elsevier Inc. All rights reserved.

doi:10.1016/S0898-6568(03)00087-1

Abbreviations: PAK, p21-activated kinase; MBP, myelin basic protein;

LPA, lysophosphatidic acid; EGF, epidermal growth factor; PMA, phorbol

12-myristate 13-acetate; GPCR, G protein coupled receptor; MAPK,

mitogen-activated protein kinase; HA, haemagglutinin; PAGE, polyacryla-

mide gel electrophoresis; PI3-kinase, phosphoinositide 3-kinase; GAP,

GTPase-activating protein.

* Corresponding author. Department of Pharmacology, Wayne State

University, 540 E. Canfield, Room 6326, Detroit, MI 48201, USA. Tel.: +1-

313-577-6022; fax: +1-313-577-6739.

E-mail address: [email protected] (R.R. Mattingly).

are involved in the stress response of mammalian cells. The

JNK pathway has been shown to be activated by the Rho

family GTPases Cdc42 and Rac1 [2].

The members of the Rho subfamily of small GTPases are

thought to be primarily involved in the organization of the

actin cytoskeleton; Rac regulates lamellipodia and then

ruffling behavior, Rho controls stress fiber formation, and

Cdc42 has been shown to control the formation of filopodia

[3]. Rac- and Rho-controlled pathways involved in the

organization of the actin cytoskeleton are distinct from

those involved in cell transformation [3,4]. In Swiss 3T3

fibroblasts, Cdc42, Rac, and Rho have been placed in a

hierarchical cascade, where Cdc42 activates Rac, which in

turn activates Rho; additionally, Ras has been found to

activate Rac [4–7]. The Rho family GTPases thus link

plasma membrane receptors to the assembly and organiza-

tion of the actin cytoskeleton. In a variety of cell types,

including fibroblasts, extracellular stimuli have been shown

to activate the Rho GTPase cascade at different points [8].

Addition of LPA to quiescent fibroblasts induces the for-

mation of actin stress fibers, while growth factors such as

PDGF and insulin stimulate polymerization of actin at the

R.E. Menard, R.R. Mattingly / Cellular Signalling 15 (2003) 1099–11091100

plasma membrane of many cell types to induce lamellipodia

formation and surface ruffling in a PI3-kinase-dependent

manner [6]. Active Rac and Cdc42 regulate distinct down-

stream signaling pathways by interacting with specific

effector proteins, including a family of serine–threonine

protein kinases termed PAKs (p21-activated kinases).

PAKs were the first identified mammalian members of a

growing family of Ste20-like serine– threonine protein

kinases. PAKs were found in screens for binding targets of

Rac and Cdc42 GTPases in rat brain [9]. This provided some

of the first evidence that Cdc42 and Rac act like other

traditional GTP-binding proteins to stimulate a target/effec-

tor protein. Additional studies revealed three proteins in

neutrophil cytosol of molecular size 62, 65 and 68 kDa,

which interacted in a GTP-dependent manner with Rac and

Cdc42, but not Rho [10]. Peptide sequencing of these

proteins showed identity to rat brain PAK65. It is generally

accepted that in mammalian tissues there are at least three

main PAK isoforms; PAK1, a 68-kDa protein that is highly

expressed in brain, muscle, and spleen; PAK2, a 62-kDa

protein with ubiquitous tissue distribution; and PAK3, a 65-

kDa protein that is expressed in the brain, but with a cell type

distribution that is different from that of PAK1 [11]. Three

additional PAKs have since been identified. These PAKs,

termed 4–6, have been classified as Group II PAKs [12–15].

All PAK proteins identified to date share a similar N-

terminal motif stretching over 18 amino acids that mediates

the interaction with Rac or Cdc42 and is referred to as CRIB

(Cdc42/Rac interactive binding domain) [16,17]. There are

many proteins from a wide range of species that contain a

CRIB homology domain, including WASP. WASP, the

Wiskott–Aldrich Syndrome Protein, may link Cdc42 to

the actin cytoskeleton. WASP binds to Cdc42 in a GTP-

dependent manner and has been shown to associate with

Nck, an SH3 domain containing adaptor protein [18]. The

adaptor protein Nck has also been shown to recruit PAK1 to

cell membranes and this membrane localization is sufficient

for PAK1 activation [19]. Autophosphorylation of PAK

blocks binding of Nck and PIX to PAK1, which provides

a way of regulating PAK interactions [20,21].

The PAK proteins also contain a highly homologous C-

terminal kinase domain [22]. PAKs 1 and 2 play similar roles

in cytoskeletal rearrangements and cellular signaling [23].

PAK2 is also involved in apoptotic cell death, following its

proteolytic cleavage by DEVD-sensitive caspases [24,25].

The role of PAKs in Raf1 activation has come to the forefront

in recent years. PAK3 is capable of phosphorylating Raf-1 on

serine 338 in vitro and in vivo [26], in a process that requires

the interaction of PI3-kinase and Ras [27]. PAK1 has also

been shown to be involved in Raf1 regulation via PI3-kinase

[28] and the interaction between active PAK1 and Raf1 is

necessary for Raf1 phosphorylation [29]. Microtubule integ-

rity also has been shown to regulate PAK and Raf1 activation

in a Ras-independent manner [30].

In this study, we delineated multiple overlapping and

independent pathways through which PAK1 could be acti-

vated. We stimulated COS7 cells, SKNSH neuroblastoma

cells and NIH-3T3 fibroblasts with agonists for G protein

coupled receptor pathways, with epidermal growth factor

(EGF), and with a cell permeable activator of protein kinase

C, phorbol 12-myristate 13-acetate (PMA). Our results

support a model where multiple signaling pathways con-

verge on the activation of PAK1 through G protein and

tyrosine kinase-dependent mechanisms.

2. Materials and methods

2.1. Plasmids and antibodies

Antibodies to the N and C terminus (N20 and C19) of

PAK1 were purchased from Santa Cruz Biotechnology. The

PhosphoPAK1 (Thr 423)/PAK2 (Thr 402) antibody was

purchased from Cell Signalling Technology. Plasmids used

in this study include, pCDhM2 [31], pRK7mycPAK1 pre-

pared from pCMV6mycPAK1 plasmid (that was a kind gift

from Gary Bokoch) by subcloning a BamH1/EcoR1 frag-

ment into pRK7, and pKH3 with inserts for RacN17,

Cdc42N17, Ras N17, RhoAN19, RacV12, Cdc42L61,

RasV12, and RhoAV14. The pKH3 plasmid expresses

inserts with a triple HA epitope tag at the N-terminus as

described previously [8,32].

2.2. Cell culture and transfection assays

COS7 cells, NIH-3T3 fibroblasts, and SKNSH neuro-

blastoma cells were cultured as described previously [32–

34]. COS7 cells were transfected by calcium phosphate

precipitation with 1 Ag pRK7myc-PAK1 plasmid, 1 AgpCDhM2, and 3 Ag test or vector control plasmids. Cells

were grown for 48 h after transfection and for cells treated

with LPA, serum starved for 24 h prior to treatment.

2.3. Agonist treatment and immunoprecipitation for COS7

transfectants

The COS7 cells were treated with 100 AM carbachol, 4

AM LPA, 200 nM PMA or 100 nM EGF for time points

ranging from 0 to 60 min. The maximal time to activation

was 5 min for EGF, 15 min for LPA and PMA and 30 min

for carbachol. For inhibition studies, pretreatment with 200

nM Wortmannin, 200 nM AG1478, or 5 AM bisindolylma-

leimide was done for 30 min prior to agonist treatment. The

medium was then aspirated, the cells washed twice with

cold PBS, and lysed on ice as described [35], except that 0.5

mM DTT, 20 mM h-glycerophosphate, 20 Ag/ml aprotinin,

20 Ag/ml leupeptin, 20 Ag/ml pepstatin, 1 mM PMSF, and 1

AM okadaic acid were added to the lysis buffer just prior to

use. Material was scraped into microfuge tubes and centri-

fuged at maximum speed for 10 min. The supernatant was

pre-cleared with 20 Al of a 50% suspension of Protein A

Sepharose beads and then re-centrifuged.

R.E. Menard, R.R. Mattingly / Cellular Signalling 15 (2003) 1099–1109 1101

The cleared extracts were then split into two aliquots.

One 200-Al portion was used for TCA precipitation of total

protein. To the other 200-Al portion, 0.48 Ag 9e10 anti-

body was added for 1 h at 4 jC, followed by 0.24 Agrabbit anti-mouse antibody for 1 h and then 40 Al of a 50%suspension of Protein A Sepharose beads for an additional

hour. The beads were washed once in lysis buffer, two

times in RIPA buffer, pH 8.0, containing 300 mM NaCl,

1% NP-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris, 5

mM orthovanadate, 50 mM Na h-glycerophosphate, 50

mM NaF, 20 mM Na pyrophosphate, and 1 mM PMSF

and finally once in a kinase buffer containing 12.5 mM

HEPES, 12.5 mM Na h-glycerophosphate, 7.5 mM Mg

Cl2, 0.5 mM EGTA, 0.5 mM NaF, and 0.5 mM Na

Vanadate, pH 7.0. One half of the immunoprecipitated

sample was used for the kinase assay and one half was

used for Western blot analysis. The blots were probed with

9e10 to detect myc-tagged proteins and 12Ca5 to detect

HA-tagged proteins.

2.4. In vitro kinase assay

The in vitro kinase assay is a modification of a procedure

by Knaus et al. [35]. Forty microliters of kinase buffer

containing 4 ACi g-32P-ATP, 500 ng GTP, and 1 Ag RacV12,per sample was added to the immunoprecipitates. Where

indicated, 4 Ag Myelin Basic Protein (MBP) was used as a

substrate to detect PAK1 activity. After 20 min at 30 jC, thekinase reaction was stopped by the addition of sample

loading buffer and boiling of samples for 5 min. The results

were visualized using SDS-polyacrylamide electrophoresis

(PAGE), autoradiography, and phosphoimaging. Data

reported are meanF S.E.M. from three independent experi-

ments.

2.4.1. Agonist treatment and immunoprecipitation of

endogenous PAK1

The procedure is the same as described above expect that

NIH-3T3/hM1 fibroblasts or SKNSH neuroblastoma cells

were lysed with a buffer, pH 7.4 containing 25 mM HEPES,

0.3 M NaCl, 1.5 mM MgCl2, 1 mM Na Vanadate, 0.1%

Triton X-100, 0.5 mM DTT, 20 mM h-glycerophosphate, 20Ag/ml aprotinin, 20 Ag/ml leupeptin, 1 mM PMSF, and 1

AM okadaic acid, and then PAK1 was immunoprecipitated

using 0.6 Ag N-20 anti-PAK1 antibody.

Fig. 1. Time course of PAK1 activation in COS7 cells treated with EGF,

carbachol, LPA, or PMA. COS7 cells were transfected with myc-tagged

PAK1, treated as shown and the myc-PAK immunoprecipitated and assayed

for activity against an MBP substrate. (A) Kinase assay of PAK1

phosphorylation following treatment with agonists. Maximal activation of

mycPAK1 activity was determined using phosphoimaging. The times to

maximal activation were 5 min for EGF, 15 min for LPA and PMA, and 30

min for carbachol. (B) Representative Western blot showing quantitative

recovery of mycPAK1 in the kinase assay protocol is independent of

treatment protocol. –Ve shows cells transfected with an empty vector

control. (C) Graph representing time course activation of mycPAK1.

MeanF S.E.M. from n= 5.

3. Results

3.1. Time course of PAK1 activation in COS7 cells

To investigate the pathways through which PAK1 can be

activated, COS7 cells were transiently transfected with a

myc-tagged PAK1 construct and subtype 2 human musca-

rinic receptors, and then stimulated with EGF (to activate

growth factor tyrosine kinase signaling), carbachol or LPA

(to stimulate G protein-coupled receptors), and PMA (to

activate PKC). Time course analysis showed that maximal

PAK1 activation with EGF was at 5 min, while LPA and

PMA took 15 min to reach peak activation, and carbachol

was the slowest to activate PAK1 taking 30 min (Fig. 1).

The peak increase in activity compared to untreated cells

was 4.8 fold for EGF, 4.5 fold for carbachol, and 3.1 fold for

both LPA and PMA. These results may indicate a hierarchy

to PAK1 activation in these cells. Once GPCRs are activat-

ed, a transactivation with the EGF receptor may take place

[36,37], and the Ghg subunits that are released may also

activate down stream effectors such as PI3-kinase [38–40]

to achieve maximal activation of PAK1, but with slower

kinetics.

3.2. Activation of PAK1 by EGF requires Ras, Rac and

Cdc42 activity

Mutations that occur outside of the GTPase effector

domain can determine the activity of Ras, Rac, Cdc42,

Fig. 2. Activation of PAK1 by EGF requires Ras, Rac and Cdc42 activity.

COS7 cells were co-transfected with mycPAK1 and active GTPases,

HARacV12, HACdc42L61, HARasV12, or dominant-negative GTPases,

HARacN17, HARasN17, or HACdc42N17. The cells were treated with

EGF for 5 min. (A) EGF-activated myc PAK1-transfected cells, and the

dominant-negative Ras, Cdc42, and Rac inhibited this activation. (B)

PAK1 was activated in cells expressing active Cdc42 and Rac, and

slightly activated when RasV12 was expressed. MBP was used as a

substrate to detect PAK1 phosphorylation in an in vitro kinase assay.

Western blots were probed with 9e10 antibody to detect myc-tagged

PAK1, and probed with 12Ca5 to detect HA-tagged proteins. (C) Graph

depicting activation of PAK1 in untreated and EGF treated COS7 cells.


Fig. 3. Activation of PAK1 by carbachol requires Ras, Rac and Cdc42

activity. Cells co-transfected with mycPAK1 and dominant-negative (A) or

active (B) GTPases were treated with carbachol for 30 min. MBP was used

as a substrate to detect PAK1 phosphorylation in an in vitro kinase assay.

Western blots were probed with 9e10 antibody to detect myc-tagged PAK1,

and probed with 12Ca5 to detect HA-tagged proteins. (C) Graph depicting

activation of mycPAK1 in untreated and carbachol treated COS7 cells.



and Rho. For Ras, the mutation of glycine to valine in

position 12 (G12 to V) results in a decrease in intrinsic

GTPase and GAP-stimulated activity that causes the persis-

tence of the GTP-bound state. This mutation maintains Ras

in its active form for a much longer duration and thus it

produces a ‘‘dominant-positive’’ or ‘‘constitutively active’’

form of Ras [41]. Conversely, the S17 to N mutation results

in a conformational change in Ras that decreases affinity for

nucleotides, particularly GTP, thus rendering Ras inactive.

This Ras N17 is considered to be a dominant-negative form

of Ras since it can preferentially bind to upstream exchange

factors and so block activation of the wild type Ras

proteins as well [42]. The same is also true for the Rho

GTPases where RacV12, Cdc42L61, and RhoAV14

mutants are constitutively active and RacN17, Cdc42N17,

Fig. 4. Activation of PAK1 by LPA requires Ras, Rac and Cdc42 activity:

Cells co-transfected with mycPAK1 and dominant-negative (A) or active

(B) GTPases were treated with LPA for 15 min. MBP was used as a




activation of mycPAK1 in untreated and LPA-treated COS7 cells.


Fig. 5. Activation of PAK1 by PMA requires Ras, Rac and Cdc42 activity.

Cells co-transfected with mycPAK1 and dominant-negative (A) or active

(B) GTPases were treated with PMA for 15 min. MBP was used as a




activation of mycPAK1 in untreated and PMA-treated COS7 cells.


R.E. Menard, R.R. Mattingly / Cellular

and RhoAN19 are dominant-negative forms of these

GTPases [42].

To investigate the pathway through which EGF acti-

vates PAK1, COS7 cells were co-transfected with myc-

PAK1 and active GTPases, RacV12, Cdc42L61, RasV12,

or dominant-negative GTPases, RacN17, RasN17, or

Cdc42N17, and cells were treated with EGF for 5

min. In the treated and untreated cells, mycPAK1 was

activated when active Rac and Cdc42 were expressed

(Fig. 2). Interestingly, active Ras, although only a weak

activator of PAK1, prevented further stimulation of

PAK1 by EGF. The activation of PAK1 by EGF was

inhibited when the dominant-negative GTPases were

expressed (Fig. 2).

3.3. Activation of PAK1 by carbachol requires Ras, Rac and

Cdc42 activity

To investigate the pathway through which muscarinic

receptors activate PAK1, COS7 cells were co-transfected

with mycPAK1 plus hM2 receptors, together with active or

dominant-negative GTPases, and the cells were treated with

carbachol for 30 min. In the treated and untreated cells,

mycPAK1 was activated when active Rac and Cdc42 were

expressed (Fig. 3). Treatment with carbachol was able to

enhance PAK1 activation beyond that induced by expression

of RasV12. The activation of PAK1 by carbachol was

inhibited when any of the dominant-negative GTPases were

expressed (Fig. 3).

Signalling 15 (2003) 1099–1109 1103

llular Signalling 15 (2003) 1099–1109

3.4. Activation of PAK1 by LPA requires Ras, Rac and

Cdc42 activity

To investigate the pathway through which LPA activates

PAK1, COS7 cells were co-transfected with mycPAK1 and

active or dominant-negative GTPases, and the cells were

treated with LPA for 15 min. In the treated and untreated

cells, mycPAK1 was activated when active Rac and Cdc42

were expressed (Fig. 4). As with the other agonists, the

R.E. Menard, R.R. Mattingly / Ce1104

Fig. 6. Role of the EGF receptor (EGFR), PI3-kinase and protein kinase C in the ac

mycPAK1 transfectants were pretreated with the EGFR inhibitor AG1478 (200 n

agonist treatment. (B) COS7 cells were pretreated with the protein kinase C in

depicting activation of mycPAK1 in untreated and EGF, carbachol, LPA and PM

from n= 3.

activation of PAK1 by LPA was blocked by dominant-

negative GTPases.

3.5. Activation of PAK1 by PMA requires Ras, Rac and

Cdc42 activity

To investigate a pathway for PAK1 activation that may

bypass cell surface receptors, COS7 cells were co-trans-

fected with mycPAK1 and active or dominant-negative

tivation of PAK1 by EGF, carbachol, LPA and PMA stimulation. (A) COS7

M), or the PI3-kinase inhibitor wortmannin (200 nM) for 30 min prior to

hibitor bisindolylmaleimide (5 AM) prior to agonist treatment. (C) Graph

A treated COS7 cells, pretreated with inhibitors as shown. MeanF S.E.M.

Fig. 7. Detection of active PAK1 using an antibody that recognizes

phosphorylated threonine 423. COS7 cells were transfected with myc-

tagged PAK1 and treated as shown. Western blots were probed first with

9e10 to detect myc-tagged PAK1, then stripped and re-probed with a

PhosphoPAK1 (Thr 423)/PAK2 (Thr 402) antibody (Cell Signalling

Technology).

Fig. 8. LPA stimulation of COS7 cells may regulate PAK1 activation

through RhoA. COS7 cells were transfected with mycPAK1 alone or co-

transfected with HARhoAN19 or HARhoV14 and then treated with LPA.

(A) Data from a representative experiment. (B) Graph depicting activation

of PAK1 by LPA in COS7 cells. MeanF S.E.M. from n= 3.


GTPases, and the cells were then treated with PMA for 15

min (Fig. 5). All of the active GTPases caused stimulation

of PAK1 that was insensitive to further activation by PMA

and all of the dominant-negative GTPases blocked PMA

stimulation of PAK1.

3.6. Role of the EGF receptor (EGFR), PI3-kinase and

protein kinase C in the activation of PAK1 by EGF,

carbachol, LPA and PMA stimulation

To further investigate the signaling pathways through

which PAK1 is activated, COS7mycPAK1 transfectants were

pretreated with the selective EGFR inhibitor AG1478 (200

nM), or the PI3-kinase inhibitor wortmannin (200 nM), or the

protein kinase C inhibitor bisindolylmaleimide (5 AM) for 30

min prior to agonist treatment. Pretreatment of cells with

AG1478 decreased PAK1 activation by EGF from 7.4 to 1.2

fold and by carbachol from 5.2 to 2.4 fold. Pretreatment with

wortmannin also inhibited the stimulatory effects of EGF,

from 7.4- to 1.8-fold increase, and carbachol from 5.2- to 2.3-

fold increase. These inhibitors did not significantly affect the

stimulation of PAK1 by LPA or PMA (Fig. 6). Pretreatment

with bisindolylmaleimide did not affect PAK1 activation by

EGF, carbachol, or LPA, however activation by PMA was

inhibited, as there was a decrease in fold activation from 3.6

to 1.2 (Fig. 6). These results indicate that PAK1 stimulation

by EGF and carbachol relies on EGFR and PI3-kinase

activation, and that LPA and PMA stimulate independent

pathways to PAK1 activation.

3.7. Detection of active PAK1 using an antibody which

recognizes phosphorylated threonine 423

In order to confirm the results of the PAK1 kinase

activity assays performed by MBP phosphorylation, West-

ern blots from experiments in which COS7 cells were

pretreated with the inhibitors for 30 min prior to agonist

treatment were stripped and re-probed with a PhosphoPAK1

(Thr 423)/PAK2 (Thr 402) antibody [43]. The results are

shown in Fig. 7. Activation of PAK1 by carbachol, EGF,

LPA, and PMA can be detected by this antibody. Further-

more, the results of the inhibition study depicted in Fig. 6

are confirmed by this antibody. This antibody may thus

prove to be a more time efficient method of PAK1 activation

than radioactive kinase assays.

3.8. LPA stimulation of COS7 cells may regulate PAK1

activation through RhoA

As none of the pharmacological inhibitors blocked LPA

stimulation of PAK1 activity, we concluded that another

R.E. Menard, R.R. Mattingly / Cellular1106

signal transduction pathway was involved. We therefore

investigated the possible role of RhoA as a potential effector

of LPA activation [7,8,44]. COS7 cells were transfected

with mycPAK1 alone or co-transfected with RhoAN19 or

RhoV14 and then treated with LPA. Cotransfection with

RhoAN19 inhibits PAK1 stimulation from 2.8- to 1.1-fold

activation in LPA-treated cells (Fig. 8). The mycPAK1 co-

transfectants containing RhoAV14, revealed increased

PAK1 activity of 2.3-fold activation in the presence or

absence of LPA (Fig. 8). These results indicate a critical

role for RhoA in PAK1 activation.

Fig. 9. Activation of endogenous PAK1 in SKNSH neuroblastoma, NIH-

3T3/hM1 fibroblasts, and COS7 cells. (A) Autophosphorylation and

phosphorylation of MBP used as an exogenous substrate were investigated

using an in vitro kinase assay of PAK1 immunoprecipitated from SKNSH,

NIH3T3/hM1, and COS7 cells. PAK1 autophosphorylation was detected in

all three-cell lines. MBP was also a substrate but BSA was not. (B) NIH-

3T3 fibroblasts were treated for 5 min with EGF, carbachol, LPA, and

PMA. Endogenous PAK1 was immunoprecipitated and kinase assays were

performed by autophosphorylation. (C) Graph depicting PAK1 activation in

NIH-3T3/hM1 cells. The 67- and 63-kDa bands refer to PAK1 and PAK 2,

respectively. MeanF S.E.M. from n= 5.

3.9. Activation of endogenous PAK1 in SKNSH neuro-

blastoma, NIH-3T3/hM1 and COS7 cells

Our results have identified four agonist pathways for

activation of PAK1 in COS7 cells transfected to express a

mycPAK1 construct. To confirm that endogenous PAK1

could be activated by those same pathways, the autophos-

phorylation of PAK1 and phosphorylation of MBP as a

PAK1 substrate were investigated using an in vitro kinase

assay of endogenous PAK1 immunoprecipitated from COS7

cells, SKNSH neuroblastoma cells, and NIH3T3/hM1fibro-

blasts [32]. PAK1 autophosphorylation was detected in all

three cell lines. The addition of MBP as a substrate reduced

the level of autophosphorylation detected, but showed that

MBP is also a good substrate for endogenous PAK1 (Fig.

9A). Treatment of the NIH3T3/hM1 fibroblasts with EGF,

carbachol, LPA, and PMA showed that endogenous PAK1 is

activated by these agonists in these cells (Fig. 9).

Signalling 15 (2003) 1099–1109

4. Discussion

Mammalian PAKs have been implicated in the regulation

of a number of cellular activities. These include regulation

of MAPK signaling pathways (including JNK, p38, and

ERK), apoptosis, and the cell cycle. The involvement of

PAKs in many important biological signaling pathways and

processes suggests that abnormalities of PAK function are

likely to contribute to various disease states. Possibilities

include inflammatory diseases involving enhanced leuko-

cyte influx and activation, cancers involving high levels of

metastasis, defects in neuronal formation or degeneration,

and autoimmune disorders, developmental abnormalities,

and cancers in which defective apoptotic cell death plays

an etiologic role [22]. Two important endpoints of PAK

signaling have emerged: nuclear events that influence gene

expression, and cytoskeletal events that impact upon cellular

dynamics. While these biological endpoints provide inter-

esting possibilities regarding cellular signaling by Cdc42

and Rac, they also raise questions regarding the nature of

PAK regulation and the identity of downstream targets [23].

In recent years, there has been a great interest in upstream

components that lead to PAK1 activation. In the present

study, we show PAK1 activation in COS7 transfectants and

NIH3T3/hM1 fibroblasts by EGF, carbachol, LPA and

PMA.

Functional studies of the Ras and Rho GTPases have

relied on the use of dominant-negative and constitutively

active forms of these proteins. Dominant-negative Cdc42N17

was observed to inhibit nerve growth factor (NGF) induced

neurite outgrowth in PC12 cells, and the active Cdc42L61

caused morphological changes in these same cells in the

absence of growth factors [45]. It has been shown that

microinjection of dominant-negative RacN17 inhibits plate-

let-derived growth factor (PDGF) stimulated ruffles in Swiss

3T3 cells, and that in these same cells, microinjection of


active RacV12 is sufficient to induce membrane ruffles in the

absence of growth factors [5,6]. The use of dominant-nega-

tive Cdc42 and Rac has been utilized in studying anchorage

independent growth in Ras transformation [46], and also Rac

regulation of actin polymerization [47]. Active GTPases can

also effect cell transformation. Active RacV12 can cause

malignant transformation on its own [48], and active RasV12

can induce focus formation in NIH3T3 fibroblasts that can be

inhibited by co-expression of dominant-negative RhoA19

[49]. It has also been reported that dominant-negative forms

of Rac, Cdc42, and RhoA are able to inhibit Ras activation of

PAK in Rat-1 fibroblasts [50].

The activation of PAK1 by active Cdc42 and Rac1 (Figs.

2–5) has been reported by many groups [9,10,22,35,51,52],

and our results coincide with what others have found. In

contrast, the effects of active Ras on PAK1 are more

surprising. The low level of PAK1 activation by RasV12

in these cells, except when cells are treated with carbachol,

is intriguing. It has been reported that RasV12 activates

PAK1 in certain cell lines, such as RAT1and Swiss-3T3, but

not NIH3T3 fibroblasts [50,53]. It is possible that other

factors are required for the activation of PAK by RasV12

and that muscarinic receptor stimulation can supply these

signals in COS7 cells. Alternatively, there may be down-

regulation of endogenous receptors and the PKC pathway

by RasV12, but the transfected muscarinic receptors may

remain functional under these conditions. All agonist stim-

ulation of PAK1 could be inhibited by dominant-negative

forms of Ras, Rac, and Cdc42, indicating that EGF, carba-

Fig. 10. A potential scheme whereby multiple signaling pathways converge on

mechanisms. The results show that PAK1 is activated by four separate pathways

activate PAK1. Carbachol can transactivate EGFR and go through PI3-kinase to a

EGFR-independent pathways (as indicated by the dashed arrows). LPA activation

activation of PAK1. The activation by EGF goes through PI3-kinase and PMA act

chol, LPA, and PMA act through mechanisms in which

these small GTPases are required. Thus, there is a definite

involvement of Ras in PAK1 activation, but the exact

pathway through Ras to PAK1 remains unclear.

In recent years, evidence of cross-communication be-

tween EGFR and GPCRs has emerged. It has been observed

that GPCR-induced EGFR tyrosine phosphorylation is rap-

id, transient, and comparable with receptor activation by

low amounts of EGF [36,54–56]. The signal capacity of the

transactivated EGFR is not restricted to Ras–MAP kinase

pathway but rather functions as an important mediator for

Gq, Gi, and G12/13-coupled receptors to induce a diversity

of cellular responses such as activation of PI3-kinase,

modulation of ion channels, and cytoskeletal reorganization

[36,57]. The time course study we performed showed EGF

and carbachol activate PAK1 following 5 and 30 min of

treatment, respectively (Fig. 1). In addition, the carbachol-

dependent activation of PAK1 was blocked by the selective

EGFR inhibitor AG1478 (Fig. 6). The results of our study

clearly indicate that muscarinic receptors require EGFR

transactivation for the major part of their ability to activate

PAK1. It is likely that muscarinic receptors can also couple

to additional pathways for PAK1 activation, however, as

carbachol still stimulates PAK1 in the presence of active Ras

when EGF cannot, and the inhibition of carbachol-depen-

dent PAK1 activation by AG1478 is incomplete. It is

noteworthy that the pathway from endogenous G protein-

coupled receptors for LPA does not require the EGF

receptor [36,54].

the activation of PAK1 through G protein and tyrosine kinase-dependent

. The two G protein coupled receptor agonists act differently in how they

ctivate PAK1 via Rac/Cdc42. It can also apparently couple weakly through

of PAK1 goes through RhoA, which can possibly lead to Ras-dependent

ivates PAK1 via PKC. PKC may activate an upstream effector such as Ras.


The role of PAKs in cytoskeletal re-organization led us to

investigate what other proteins may be involved in this

process. Phorbol esters, such as PMA, can induce actin

assembly in neutrophils and this cytoskeletal alteration is

due to PKC activation [58]. Various isotypes of PKC may

play different roles in cells. Both PKCa and PKCu can be

activated by diacylglycerol but PKCa can potentially activate

PKC~ leading to InB stimulation, and PKCu may activate

Ras leading to Erk2 activation [59,60]. The PMA activation

of PAK1 is inhibited by dominant-negative forms of Rac and

Cdc42, therefore it is likely that PKC interacts with a protein

upstream of PAK1. It is likely that PMA induces Ras

activation, as dominant-negative Ras can inhibit PMA stim-

ulation of PAK1. The inhibition of PMA-dependent PAK1

activation by bisindolylmaleimide shows that there is a role

for protein kinase C in PAK1 stimulation, but the lack of

effect of PKC inhibition on any of the cell surface receptor

agonists suggests that PKC is not a necessary component of

the activation of PAK1 by any of these pathways.

The lack of inhibition of LPA-dependent PAK1 activity

by inhibitors of PKC and PI3-kinase led us to investigate

other proteins that may be involved in PAK1 activation.

LPA receptors have been shown to act through G12,13 G-

proteins to stimulate RhoA and PAK1 [6,61,62]. The exact

mechanism by which RhoA induces PAK1 activation is

unclear. In a study of PAK1 and Rho GTPases in rat brain,

there was an association between Rac and PAK1, Cdc42 and

PAK1, but not between RhoA and PAK1 [63]. In our

experiments, RhoAN19 was able to block the activation of

PAK1 by LPA, and in addition RhoAV14 was able to

activate PAK1 (Fig. 8). How then could RhoA lead to

PAK activation? It has been reported that co-transfection

of active RhoA and active Raf can activate PAK in Rat1

fibroblasts [50]. In the COS7 cells in this study, active RhoA

was sufficient to induce PAK1 activation. The role of RhoA

in cell adhesion has been studied and it has been determined

that FAK and paxillin are phosphorylated in response to

RhoA and this may lead to Ras/Raf activation [64,65].

There is also evidence that cAMP-dependent protein kinase

(PKA) phosphorylates PAK1, paxillin and Nck [66], and it

is known PAK1 interacts with paxillin and Nck [19,21,67–

70]. It is thus quite possible that activation of PAK1 by

RhoA is indirect through other upstream effectors (Fig. 10).

Further investigation of the role of components upstream of

PAK1, such as Ghg and Akt, may yield a more complete

insight as to the exact mechanism of PAK1 activation.

Acknowledgements

We thank Dr. G. Bokoch for the gift of the plasmid

encoding PAK1, Dr. C. Neudauer for the subcloning of

PAK1 and A. Jovanovski for technical assistance.

This work was supported in part by RO1-CA81150

(to R.R.M.) and T32-CA09531 (to R.E.M.).

References

[1] Lopez-Ilasaca M. Biochem Pharmacol 1998;56(3):269–77.

[2] Brown JL, Stowers L, Baer M, Trejo J, Coughlin S, Chant J. Curr

Biol 1996;6(5):598–605.

[3] Symons M. Curr Opin Biotechnol 1995;6(6):668–74.

[4] Downward J. Nature 1992;359(6393):273–4.

[5] Ridley AJ, Hall A. Cell 1992;70(3):389–99.

[6] Nobes CD, Hawkins P, Stephens L, Hall A. J Cell Sci 1995;

108(Pt. 1):225–33.

[7] Mackay DJ, Hall A. J Biol Chem 1998;273(33):20685–8.

[8] Beqaj S, Jakkaraju S, Mattingly RR, Pan D, Schuger L. J Cell Biol

2002;156(5):893–903.

[9] Manser E, Leung T, Salihuddin H, Zhao ZS, Lim L. Nature

1994;367(6458):40–6.

[10] Martin GA, Bollag G, McCormick F, Abo A. EMBO J 1995;14(9):

1970–8.

[11] Sells MA, Knaus UG, Bagrodia S, Ambrose DM, Bokoch GM,

Chernoff J. Curr Biol 1997;7(3):202–10.

[12] Abo A, Qu J, Cammarano MS, Dan C, Fritsch A, Baud V, et al.

EMBO J 1998;17(22):6527–40.

[13] Yang F, Li X, Sharma M, Zarnegar M, Lim B, Sun Z. J Biol Chem

2001;276(18):15345–53.

[14] Pandey A, Dan I, Kristiansen TZ, Watanabe NM, Voldby J, Kajikawa

E, et al. Oncogene 2002;21(24):3939–48.

[15] Jaffer ZM, Chernoff J. Int J Biochem Cell Biol 2002;34(7):713–7.

[16] Hoffman GR, Cerione RA. Cell 2000;102(4):403–6.

[17] Knaus UG, Wang Y, Reilly AM, Warnock D, Jackson JH. J Biol

Chem 1998;273(34):21512–8.

[18] Van Aelst L, D’Souza-Schorey C. Genes Dev 1997;11(18):2295–322.

[19] Lu W, Katz S, Gupta R, Mayer BJ. Curr Biol 1997;7(2):85–94.

[20] Manser E, Loo TH, Koh CG, Zhao ZS, Chen XQ, Tan L, et al. Mol

Cell 1998;1(2):183–92.

[21] Zhao ZS, Manser E, Lim L. Mol Cell Biol 2000;20(11):3906–17.

[22] Knaus UG, Bokoch GM. Int J Biochem Cell Biol 1998;30(8):

857–62.

[23] Bagrodia S, Cerione RA. Trends Cell Biol 1999;9(9):350–5.

[24] Rudel T, Bokoch GM. Science 1997;276(5318):1571–4.

[25] Rudel T, Zenke FT, Chuang TH, Bokoch GM. J Immunol 1998;

160(1):7–11.

[26] King AJ, Sun H, Diaz B, Barnard D, Miao W, Bagrodia S, et al.

Nature 1998;396(6707):180–3.

[27] Sun H, King AJ, Diaz HB, Marshall MS. Curr Biol 2000;10(5):

281–4.

[28] Chaudhary A, King WG, Mattaliano MD, Frost JA, Diaz B, Morrison

DK, et al. Curr Biol 2000;10(9):551–4.

[29] Zang M, Hayne C, Luo Z. J Biol Chem 2002;277(6):4395–405.

[30] Zang M, Waelde CA, Xiang X, Rana A, Wen R, Luo Z. J Biol Chem

2001;276(27):25157–65.

[31] Mattingly RR, Macara IG. Nature 1996;382(6588):268–72.

[32] Mattingly RR, Sorisky A, Brann MR, Macara IG. Mol Cell Biol

1994;14(12):7943–52.

[33] Mattingly RR, Milstein ML, Mirkin BL. Cell Signal 2001;13(7):

499–505.

[34] Mattingly RR, Gibbs RA, Menard RE, Reiners Jr JJ. J Pharmacol Exp

Ther 2002;303(1):74–81.

[35] Knaus UG, Morris S, Dong HJ, Chernoff J, Bokoch GM. Science

1995;269(5221):221–3.

[36] Leserer M, Gschwind A, Ullrich A. IUBMB Life 2000;49(5):405–9.

[37] Prenzel N, Zwick E, Daub H, Leserer M, Abraham R, Wallasch C,

et al. Nature 1999;402(6764):884–8.

[38] Nakano T, Kontani K, Kurosu H, Katada T, Hoshi M, Chiba K. Dev

Biol 1999;209(1):200–9.

[39] Maier U, Babich A, Nurnberg B. J Biol Chem 1999;274(41):29311–7.

[40] Kurosu H, Maehama T, Okada T, Yamamoto T, Hoshino S, Fukui Y,

et al. J Biol Chem 1997;272(39):24252–6.


[41] McCormick F. Curr Opin Genet Dev 1994;4(1):71–6.

[42] Lim L, Manser E, Leung T, Hall C. Eur J Biochem 1996;242(2):

171–85.

[43] Vadlamudi RK, Adam L, Wang RA, Mandal M, Nguyen D, Sahin A,

et al. J Biol Chem 2000;275(46):36238–44.

[44] Ridley AJ. Curr Biol 1996;6(10):1256–64.

[45] Daniels RH, Hall PS, Bokoch GM. EMBO J 1998;17(3):754–64.

[46] Qiu RG, Abo A, McCormick F, Symons M. Mol Cell Biol 1997;17(6):

3449–58.

[47] Joneson T, McDonough M, Bar-Sagi D, Van Aelst L. Science

1996;274(5291):1374–6.

[48] Qiu RG, Chen J, Kirn D, McCormick F, Symons M. Nature

1995;374(6521):457–9.

[49] Qiu RG, Chen J, McCormick F, Symons M. Proc Natl Acad Sci

U S A 1995;92(25):11781–5.

[50] Tang Y, Yu J, Field J. Mol Cell Biol 1999;19(3):1881–91.

[51] Manser E, Leung T, Lim L. Methods Mol Biol 1998;84:295–305.

[52] Manser E, Lim L. Prog Mol Subcell Biol 1999;22:115–33.

[53] Reeder MK, Serebriiskii IG, Golemis EA, Chernoff J. J Biol Chem

2001;276(44):40606–13.

[54] Daub H, Wallasch C, Lankenau A, Herrlich A, Ullrich A. EMBO J

1997;16(23):7032–44.

[55] Eguchi S, Numaguchi K, Iwasaki H, Matsumoto T, Yamakawa T,

Utsunomiya H, et al. J Biol Chem 1998;273(15):8890–6.

[56] Keely SJ, Uribe JM, Barrett KE. J Biol Chem 1998;273(42):

27111–7.

[57] Gohla A, Offermanns S, Wilkie TM, Schultz G. J Biol Chem

1999;274(25):17901–7.

[58] Downey GP, Chan CK, Lea P, Takai A, Grinstein S. J Cell Biol

1992;116(3):695–706.

[59] Goekjian PG, Jirousek MR. Curr Med Chem 1999;6(9):877–903.

[60] Mellor H, Parker PJ. Biochem J 1998;332(Pt. 2):281–92.

[61] Nobes C, Hall A. Curr Opin Genet Dev 1994;4(1):77–81.

[62] Schmitz U, Thommes K, Beier I, Vetter H. Biochem Biophys Res

Commun 2002;291(3):687–91.

[63] Terashima T, Yasuda H, Terada M, Kogawa S, Maeda K, Haneda M,

et al. J Neurochem 2001;77(4):986–93.

[64] Nobes CD, Hall A. J Cell Biol 1999;144(6):1235–44.

[65] Clark EA, King WG, Brugge JS, Symons M, Hynes RO. J Cell Biol

1998;142(2):573–86.

[66] Howe AK, Juliano RL. Nat Cell Biol 2000;2(9):593–600.

[67] Brown MC, West KA, Turner CE. Mol Biol Cell 2002;13(5):

1550–65.

[68] Hashimoto S, Tsubouchi A, Mazaki Y, Sabe H. J Biol Chem

2001;276(8):6037–45.

[69] Bokoch GM, Wang Y, Bohl BP, Sells MA, Quilliam LA, Knaus UG.

J Biol Chem 1996;271(42):25746–9.

[70] Wunderlich L, Goher A, Farago A, Downward J, Buday L. Cell

Signal 1999;11(4):253–62.

Cell surface receptors activate p21-activated kinase 1 via multiple Ras and PI3-kinase-dependent pathways

Documents