Differential targeting of MAP kinases to the ETS-domain transcription

The EMBO Journal Vol.17 No.6 pp.1740–1749, 1998

Differential targeting of MAP kinases to the

ETS-domain transcription factor Elk-1

Shen-Hsi Yang, Alan J.Whitmarsh

Roger J.Davis

and Andrew D.Sharrocks

Department of Biochemistry and Genetics, The Medical School,

University of Newcastle upon Tyne, Newcastle upon Tyne NE2 4HH,

UK and

Program in Molecular Medicine, Department of Biochemistry

and Molecular Biology, Howard Hughes Medical Institute, University

of Massachusetts Medical School, Worcester, MA 01605, USA

Corresponding author

e-mail: [email protected]

The activation of MAP kinase (MAPK) signal transduc-

tion pathways results in the phosphorylation of tran-

scription factors by the terminal kinases in these

cascades. Different pathways areactivated by mitogenic

and stress stimuli, which lead to the activation of

distinct groups of target proteins. The ETS-domain

transcription factor Elk-1 is a substrate for three

distinct classes of MAPKs. Elk-1 contains a targeting

domain, the D-domain, which is distinct from the

phosphoacceptor motifs and is required for efﬁcient

phosphorylation and activation by the ERK MAPKs.

In this study, we demonstrate that members of the

JNK subfamily of MAPKs are also targeted to Elk-1

by this domain. Targeting via this domain is essential

for the efﬁcient and rapid phosphorylation and activa-

tion of Elk-1 both in vitro and in vivo. The ERK and

JNK MAPKs use overlapping yet distinct determinants

in the D-domain for targeting to Elk-1. In contrast,

members of the p38 subfamily of MAPKs are not

targeted to Elk-1 via this domain. Our data therefore

demonstrate that different classes of MAPKs exhibit

differential requirements for targeting to Elk-1.

Keywords: Elk-1/ETS-domain proteins/MAP kinase/

TCFs/transcription factor

Introduction

The MAP kinase (MAPK) pathways play major roles

in converting mitogenic and stress stimuli into nuclear

responses (reviewed in Treisman, 1996; Whitmarsh and

Davis, 1996). In humans, at least three parallel pathways

exist which can be classiﬁed according to the sequence

conservation in the terminal MAPKs. The ERK pathway

primarily transmits mitogenic and differentiation stimuli,

whereas the JNK and p38 pathways primarily transduce

stress and cytokine stimuli to the nucleus. These pathways

are conserved in a diverse range of organisms including

Saccharomyces cerevisiae, Drosophila melanogaster and

Caenorhabditis elegans (reviewed in Treisman, 1996).

Several distinct MAPKs have been identiﬁed in each class

of pathway. For example, the ERK subclass contains

ERK1 and ERK2, the JNK subclass contains JNK1, JNK2

and JNK3, and the p38 subclass contains p38α, p38β,

1740

p38γ and p38δ (Gupta et al., 1996; Jiang et al., 1996;

Lechner et al., 1996; Stein et al., 1997; Cuenda et al.,

1997; Goedert et al., 1997; Wang et al., 1997; Enslen

et al., 1998; reviewed in Whitmarsh and Davis, 1996;

Robinson and Cobb, 1997). In addition, multiple MAPK

isoforms are generated by alternative splicing which is

best illustrated by the JNK MAPKs (Gupta et al., 1996).

The large number of MAPKs potentially allows differential

phosphorylation of nuclear substrates and thereby speciﬁc

responses to upstream signals. However, the substrate

speciﬁcity determinants for these related kinases are just

beginning to be understood.

A combination of in vitro and in vivo approaches has

led to the identiﬁcation of several nuclear targets for

MAPK pathways. For example, c-Myc (Gupta and Davis,

1994) and Spi-B (Mao et al., 1996) are ERK substrates,

c-Jun (Derijard et al., 1994; Kyriakis et al., 1994) and

NFAT4 (Chow et al., 1997) represent JNK targets, whereas

CHOP (Wang and Ron, 1996) and MEF2C (Han et al.,

1997) are targets for the p38 pathways. ATF-2 (Gupta

et al., 1995; Livingstone et al., 1995; Van Dam et al.,

1995) and ATFa (Gupta et al., 1995; Bocco et al., 1996)

are targets for both the JNK and p38 stress-activated

pathways. Members of the ternary complex factor (TCF)

sub-family of ETS-domain transcription factors are also

targets of MAPK pathways (reviewed in Treisman, 1996;

Whitmarsh and Davis, 1996). The TCF Elk-1 is a target

for all three pathways (Price et al., 1996; Janknecht and

Hunter, 1997a; Whitmarsh et al., 1997; reviewed in

Treisman, 1996). However, another family member,

SAP-1, appears to only be phosphorylated and activated

efﬁciently by the ERK and p38 pathways (Price et al.,

1996; Whitmarsh et al., 1995, 1997; Strahl et al., 1996),

although it can also act as a JNK substrate (Janknecht and

Hunter, 1997b). Phosphorylation of the TCFs takes place

at multiple residues in the conserved C-terminal transcrip-

tional activation domain (C-domain) which leads to

enhanced DNA binding and TCF-mediated transcriptional

activation (reviewed in Treisman, 1994, 1996; Price

et al., 1996).

It is becoming apparent that the substrate speciﬁcity of

MAPKs is determined by two components: the sequence

and local context of the phosphoacceptor motifs, and

binding of the kinase to docking sites on the substrate.

For example, in the case of phosphorylation of c-Jun by

the JNK MAPKs, the local context of the phosphoacceptor

motifs plays a major role in substrate speciﬁcity determina-

tion in combination with targeting via a kinase docking

domain on the transcription factor (Derijard et al., 1994;

Kallunki et al., 1994, 1996; Sluss et al., 1994; Dai et al.,

1995; Gupta et al., 1996). A short region of c-Jun, the

δ-domain, appears to be sufﬁcient for this interaction

(Derijard et al., 1994; Kallunki et al., 1994, 1996; Sluss

et al., 1994; Dai et al., 1995; Gupta et al., 1996). Similar

MAPK targeting of Elk-1

interactions occur between JNK MAPKs and ATF-2 via

a short motif which is distinct from the phosphoacceptor

sites (Gupta et al., 1995, 1996; Livingstone et al., 1995).

It has recently been demonstrated that the ERK MAPKs

are also targeted to a nuclear substrate, Elk-1, by a docking

motif known as the D-domain which is conserved amongst

the TCFs (Yang et al., 1998). The D-domain is located

N-terminally from the transcriptional activation domain

and is required for efﬁcient phosphorylation of Elk-1

within this adjacent domain and hence enhancement of its

transcriptional activation potential (Yang et al., 1998).

In this study, we have investigated the targeting of a

panel of mitogen- and stress-activated MAPKs to Elk-1.

Elk-1 has been implicated as a substrate for members of

all three classes of MAPKs and hence represents an

excellent candidate to investigate targeting of MAPKs.

The ERK and JNK MAPKs are both targeted to Elk-1

via the D-domain. Targeting is essential for efﬁcient

phosphorylation of Elk-1 in vitro and in vivo. However,

different residues in the D-domain are important for ERK

and JNK binding. In contrast, the p38 MAPKs are not

targeted to Elk-1 via the D-domain. Our results therefore

demonstrate that Elk-1 contains a docking site that contains

speciﬁcity determinants which allow recognition by two

different classes of MAPKs but exclude binding of a third

class. Such speciﬁcity determinants within transcription

factor substrates are likely to play a pivotal role in

producing unique nuclear responses to the activation of

diverse MAPK signal transduction pathways.

Results

Differential requirement for the Elk-1 D-domain for

phosphorylation by MAPKs in vitro

Elk-1 has been shown to be an in vitro substrate for all

three classes of MAPKs and to be regulated by these

kinase cascades in vivo (Price et al., 1996; Janknecht and

Hunter, 1997a; Whitmarsh et al., 1997; reviewed in

Treisman, 1996; Whitmarsh and Davis, 1996). It has

recently been demonstrated that Elk-1 contains an ERK

targeting motif that maps to the D-domain which is located

N-terminally from the transcriptional activation domain

(Figure 1A). The integrity of this domain is required to

allow efﬁcient Elk-1 phosphorylation in vitro (Yang et al.,

1998). In order to investigate whether the efﬁciency of

Elk-1 phosphorylation by members of the stress-activated

MAPKs is enhanced by the presence of the D-domain,

we investigated the ability of a representative member of

each subclass, JNK-1 and p38γ, to phosphorylate a series

of truncated Elk-1 derivatives (Figure 1).

N-terminal truncations of Elk-1 up to amino acid 310

(Elk-310) did not alter the efﬁciency of Elk-1 phosphoryl-

ation by ERK2. However, truncations up to and beyond

amino acid 330 (Elk-330 and Elk-349), which delete the

D-domain, resulted in a signiﬁcant reduction in Elk-1

phosphorylation (Figure 1A; Figure 1B, lanes 1–4; Yang

et al., 1998). Similarly, the efﬁciency of phosphorylation

of the proteins Elk-330 and Elk-349 by JNK-1 was

dramatically reduced (Figure 1A; Figure 1B, lanes 5–8).

In contrast, the efﬁciency of phosphorylation of all the

Elk-1 deletion proteins by p38γ was virtually identical

(Figure 1A; Figure 1B, lanes 9–12). Taken together, these

results indicate that the integrity of the D-domain is

1741

Fig. 1. Differential requirement for the D-domain for efﬁcient Elk-1

phosphorylation by MAPKs. (A) A diagram illustrating a series of

truncated Elk-1 proteins fused to GST. The black box represents the

D-domain of Elk-1 (amino acids 312–334), and the numbers of the

N- and C-terminal amino acids in the Elk-1 moiety are indicated. The

degree of phosphorylation of each protein in the immune complex

protein kinase assay (relative to Elk-310) is indicated. (B) The

phosphorylation of GST–Elk fusion proteins by MAPKs was examined

using either activated ERK2 or epitope-tagged JNK-1 and p38γ which

had been immunopuriﬁed from UV-activated COS-1 cells. The activity

of each protein kinase towards the substrate GST–Elk205 was

standardized with respect to activated ERK2 (1U, NEB). Kinase

assays were performed for 15 min at 30°C with equal molar quantities

(5 pmol) of GST–Elk-1 fusion proteins as substrates.

required for efﬁcient phosphorylation of Elk-1 by the

ERK and JNK MAPKs but not by p38γ.

In order to determine whether the differential require-

ment for the D-domain is conserved within each class of

MAPK, we tested the ability of several members of

each subfamily to phosphorylate full-length Elk-1 in the

presence or absence of the D-domain. The activity of each

MAPK towards Elk-1 was initially standardized (Figure

2B, lanes 1–4; Figure 2C). The kinetics of phosphorylation

of wild-type Elk-1 were virtually indistinguishable (Figure

2B, lanes 1–4), although graphical analysis indicated that

phosphorylation by ERK-2, JNK-1 and, to a lesser extent,

JNK-2 was non-linear and rapidly reached maximal levels

(Figure 2C). In contrast, differential phosphorylation of

Elk-1∆D (which lacks the D-domain) by individual

MAPKs was observed (Figure 2B, lanes 5–8; Figure 2C).

The kinetics and overall efﬁciency of phosphorylation

of Elk-1∆D by p38α, p38β

and p38γ were virtually

indistinguishable from wild-type Elk-1 (compare Figure

2B, lanes 1–4 and 5–8; Figure 2C). In contrast, the kinetics

of Elk-1∆D phosphorylation by ERK2, JNK-1 and JNK-2

were delayed and the overall efﬁciency greatly reduced.

It appears, therefore, that the differential requirement for

the D-domain for efﬁcient phosphorylation of Elk-1 is a

common property of all members of a particular subclass

of MAPK.

Identiﬁcation of important residues in the

D-domain

The D-domain plays an important role in enhancing Elk-1

phosphorylation by the ERK and JNK MAPKs. In order

S.-H.Yang et al.

Fig. 2. Kinetic analysis of Elk-1 phosphorylation by MAPKs in the

presence and absence of the D-domain. (A) Diagrammatic illustration

of full-length Elk-1 and Elk-1 with a nine amino acid deletion in the

D-domain (Elk-1∆D). Bacterially expressed FLAG epitope-tagged full-

length Elk-1 and Elk-1∆D proteins were puriﬁed and equimolar

concentrations were used in all assays. (B) Full-length Elk-1 (lanes 1–

4) and Elk-1∆D (lanes 5–8) (3.75 pmol of each) were phosphorylated

by MAPKs (as indicated on the right panel of the ﬁgure) for the times

indicated above each lane. The activity of each kinase was

standardized to ERK-2 as described in Figure 1. (C) Graphical

representation of the data from (B). Open and closed squares denote

wild-type Elk-1 and Elk-1∆D respectively used as MAPK substrates.

Data are presented relative to phosphorylation of wild-type Elk-1 after

120 min (taken as 100).

to investigate whether identical residues within this domain

are involved in this function, pairs of amino acids were

mutated to alanine residues (Figure 3A). Such mutations

should preserve any structural motifs which are present but

remove side chains which are available for intermolecular

interactions. The mutant proteins were tested as substrates

for a panel of MAPKs (Figure 3B). Mutations in the N-

terminal half of the D-domain resulted in signiﬁcant

reductions in the efﬁciency of phosphorylation by ERK2

(Figure 3B, mutants M1 and M2; Yang et al., 1998).

Similarly, reductions in the efﬁciency of phosphorylation

of the M1 and M2 mutants by JNK-1 and JNK-2 were

observed (Figure 3B, lanes 2 and 3). However, in the case

of JNK-1 and JNK-2, additional important residues were

identiﬁed by the M3 mutant which was also phosphorylated

to a reduced level (Figure 3B, lane 4). In contrast, none

of the mutations resulted in largedecreases in the efﬁciency

1742

Fig. 3. Identiﬁcation of residues within the D-domain required for

efﬁcient phosphorylation by MAPKs. (A) Amino acid sequence of the

wild-type (WT) and D-domain mutants R314A/K315A (M1), R317A/

L319A (M2), L323A/S324A (M3), L327A/L328A (M4) and L319A/

L321A (M5) are indicated. Numbers above the sequence represent the

N- and C-terminal residues in the D-domain. (B) Kinase assays of

wild-type and mutant GST–Elk205 fusion proteins by MAPKs were

carried out as described in Figure 1. Amounts of phosphorylation of

the mutant M1, M2, M3, M4 and M5 Elk-1 proteins relative to WT

Elk-1 (taken as 1) are 0.47 6 0.06, 0.40 6 0.04, 1.10 6 0.22, 1.12 6

0.28 and 0.23 6 0.11 (ERK2; Yang et al., 1998); 0.45 6 0.06, 0.39 6

0.11, 0.63 6 0.10, 1.19 6 0.20 and 0.21 6 0.06 (JNK-1); 0.39 6

0.04, 0.28 6 0.06, 0.46 6 0.09, 1.01 6 0.01 and 0.14 6 0.05

(JNK-2); 0.79 6 0.07, 0.72 6 0.04, 1.0 6 0.09, 1.17 6 0.24 and

0.75 6 0.35 (p38α); 0.71 6 0.15, 0.66 6 0.04, 0.97 6 0.01, 1.05 6

0.07 and 0.77 6 0.09 (p38β

); and 0.97 6 0.08, 0.9 6 0.18, 0.98 6

0.17, 1.08 6 0.18 and 1.22 6 0.31 (p38γ).

of Elk-1 phosphorylation by p38γ and, in comparison with

the effect on ERK and JNK MAPKs, only small differences

were seen for phosphorylation of the M1 and M2 mutants

by p38α and p38β

(Figure 3B). Finally, an additional

mutant was created (M5) which has mutations in both of

the leucine residues which make up the conserved ‘LXL’

central core of several MAPK targeting motifs (Yang

et al., 1998). The efﬁciency of phosphorylation of the M5

mutant by the ERK and JNK MAPKs is severely reduced

whereas little effect is seen on phosphorylation by the

p38 MAPKs (Figure 3B, lane 7).

These results demonstrate that residues in the D-domain

play different roles in directing phosphorylation of Elk-1

by MAPKs. Residues in the N-terminal end of this domain

are important for both the ERK and JNK MAPKs, with

further C-terminal residues also being important for the

JNKs. In contrast, mutations within the D-domain have

minimal effects on phosphorylation by the p38 MAPKs,

consistent with the lack of effect of deleting this domain.

Differential binding of MAPKs to the Elk-1

D-domain

It previously has been demonstrated that the ability of

ERK2 to phosphorylate Elk-1 correlates with its ability

to bind to its substrate via the D-domain (Yang et al.,

1998). Moreover, JNKs have also been demonstrated to

MAPK targeting of Elk-1

interact physically with their substrate c-Jun via the

δ-domain (Derijard et al., 1994; Kallunki et al., 1994;

Sluss et al., 1994; Dai et al., 1995) and with Elk-1 (Gille

et al., 1996), although in the latter case the binding site

was not determined. In comparison with ERK2, binding

of JNKs to Elk-1 was barely detectable using the GST

pull-down assay (Gupta et al., 1996; data not shown),

suggesting that the kinetics of interaction differ between

these two different classes of kinases. In order to investi-

gate binding of the kinases to Elk-1 under the same

conditions, we therefore adopted a different approach

using a peptide competition assay. Peptides were synthe-

sized which correspond to Elk-1 amino acids 311–328

and encompass the D-domain (Figure 4A). Increasing

amounts of these peptides were included in kinase assays

to compete for binding of the MAPKs to Elk-1 via the

D-domain. Mutant D-domain peptides and a peptide from

JIP-1, a speciﬁc JNK inhibitor (Dickens et al., 1997),

were used as controls (Figure 4A).

The wild-type and M3 mutant D-domain peptides both

acted as inhibitors of ERK2 in a concentration-dependent

manner (Figure 4B, lanes 2–4 and 8–10). In contrast, the

M5 and JIP peptides did not act as competitors (Figure

4B, lanes 5–7 and 12–13). These results are consistent

with the observation that the M3 mutation affects neither

the efﬁciency of phosphorylation nor binding by ERK2,

whereas the M5 mutation inhibits both these functions

(Figure 3; Yang et al., 1998; data not shown). Similarly,

the wild-type peptide inhibits phosphorylation by the

JNK kinases whereas the M5 peptide is an ineffective

competitor. In comparison with ERK2, the wild-type

peptide inhibits phosphorylation by JNK-1 and JNK-2 at

a 10-fold lower concentration (Figure 4B, compare lanes

3 and 4). Signiﬁcantly, inhibition of JNK-1 and JNK-2 by

the M3 peptide is reduced in comparison with the wild-

type peptide (Figure 4B, compare lanes 3 and 9), which

correlates with the reduction in phosphorylation observed

in the M3 mutant protein (Figure 3). The control JIP

peptide acts as an efﬁcient JNK inhibitor in this assay

(Figure 4B, lanes 12 and 13). In contrast, none of the

peptides used act as efﬁcient inhibitors of the p38 MAPKs

(Figure 4B).

In order to conﬁrm that the peptides are blocking

substrate binding rather than impairing the catalytic

activity of the MAPKs, they were used as competitors in

kinase reactions containing ERK2 and JNK-2 and the

substrates myelin basic protein (MBP) and c-Jun respect-

ively. The binding of MAPKs to a docking domain on

MBP is not thought to occur and, consistent with this

concept, phosphorylation of MBP by ERK2 was not

inhibited by the presence of either the wild-type, M3

or M5 peptides (Figure 4C). The JIP peptide inhibited

phosphorylation of c-Jun by JNK-2 (Figure 4D, lanes 8

and 9). This inhibition previously has been attributed to

competition for binding of the kinase to the c-Jun δ-domain

(Dickens et al., 1997). Similarly, the wild-type D-domain

peptide inhibited phosphorylation of c-Jun by these kin-

ases, albeit to a lesser extent (Figure 4D, lanes 2 and 3).

Mutant D-domain peptides which are defective in JNK

binding are also defective in inhibiting phosphorylation

of c-Jun by JNKs (Figure 4D, lanes 4–7). Thus, the

D-domain peptide is acting as expected to disrupt inter-

1743

Fig. 4. The Elk-1 D-domain acts as a binding site for the ERK and

JNK MAPKs. (A) The sequences of peptide competitors and a

diagrammatic illustration of the substrate (GST–Elk310, 5 pmol) used

in the kinase reactions. Identical or highly conserved (Arg/Lys)

residues conserved in the Elk-1- and JIP-derived peptides are indicated

by dashes. (B) Phosphorylation of Elk-1 by MAPKs in the presence of

competitor peptides. The peptide competition assay was based on the

kinase assays described in Figure 1 except that the MAPKs were pre-

incubated in the absence (lanes 1 and 11) or presence of competitor

peptides (10- to 1000-fold excess over Elk-1 substrate), 50 pmol (lanes

2, 5, 8 and 12), 500 pmol (lanes 3, 6, 9 and 13) and 5 nmol (lanes 4,

7 and 10), respectively. (C) and (D) Competition assays were carried

out as in (B) by using 5 pmol of MBP and GST–c-Jun as substrates,

respectively. The assays were performed in the absence (C and D, lane

1) or presence of peptide competitors, 50 pmol (C, lanes 2, 5 and 8;

D, lanes 2, 4, 6 and 8), 500 pmol (C, lanes 3, 6 and 9; D, lanes 3, 6, 9

and 12) and 5 nmol (C, lanes 4, 7 and 10), respectively. Increases in

the concentration of added competitor peptides are indicated

schematically above each set of lanes.

actions between the JNKs and c-Jun but is an ineffective

competitor of phosphorylation of MBP by ERK2.

Collectively, these data demonstrate that the D-domain

acts as a binding site for both the ERK and JNK MAPKs,

but phosphorylation of Elk-1 by the p38 MAPKs does not

take place via this mechanism.

S.-H.Yang et al.

The JNK targeting domains of Elk-1 and c-Jun act

as discrete interchangeable modules

The Elk-1 D-domain is sufﬁcient to bind to ERK2 in a

heterologous context when fused to another protein (Yang

et al., 1998). In order to investigate whether the D-domain

can act in a similar manner to allow JNK targeting,

reciprocal chimeric proteins were created in which the

MAPK targeting domains of c-Jun and Elk-1 were

exchanged (Figure 5A). Deletion of the Elk-1 D-domain

(Elk-330; Figure 1; Figure 5B, lanes 1 and 2) or the

δ-domain from c-Jun (cJun43–223; Figure 5B, lanes 7

and 8) resulted in a loss in the efﬁciency of phosphorylation

by both JNK-1 and JNK-2. However, addition of the

reciprocal targeting domains to these truncated proteins

converted them back to efﬁcient JNK substrates (cJunδ–

Elk-330 and ElkD–cJun; Figure 5B, lanes 4 and 5).

Interestingly, of the two chimeric proteins, only cJunδ–

Elk-330 exhibited enhanced phosphorylation by ERK2

1744

despite the D-domain in the ElkD–cJun chimera represent-

ing an ERK-binding motif (Figure 5B, top panel, lanes 3

and 4). This is in agreement with the observation that the

local context of the phosphoacceptor motifs also plays

a major role in determining the speciﬁcity of c-Jun

phosphorylation (Gupta et al., 1996; Kallunki et al., 1996).

Finally, in order to rule out that the D-domain itself was

phosphorylated by the MAPKs, two control constructs

were made and tested. Neither ElkD (which contains

the D-domain alone) nor ElkD–cJun(S63A/S73A) (which

lacks the primary JNK phosphoacceptor motifs) represent

good ERK or JNK substrates (Figure 5B, lanes 3 and 6).

In order to investigate binding of the kinases and the

relative strengths of the different binding motifs, peptide

competition assays were carried out with the chimeric

Elk-1–c-Jun proteins (Figure 5C). The wild-type D-domain

and JIP peptides acted as efﬁcient competitors of ElkD–

cJun phosphorylation by JNKs (Figure 5C, lanes 2 and

5). Competition was also exhibited by the M3 peptide,

albeit to a lesser extent with JNK-2, whereas the M5

peptide was an ineffectual competitor (Figure 5C, lanes 3

and 4). These results are essentially the same as observed

with wild-type Elk-1 (Figure 4), thereby demonstrating that

the D-domain acts in a similar manner in a heterologous

context. Competition assays were also carried out with

cJunδ–Elk-330, and again the JIP peptide acted as an

efﬁcient competitor of phosphorylation by JNKs (Figure

5C, lane 10). The wild-type and M3 D-domain peptides

also acted as competitors, but to a lesser extent than

observed with ElkD–cJun (Figure 5C, compare lanes 7

and 9 with lanes 2 and 4), indicating that the δ-domain

acts as a stronger JNK-binding motif. Furthermore, com-

petition for JNK-1 was more pronounced than for JNK-2

(Figure 5C, lanes 7 and 9; compare JNK-1 and JNK-2

panels). This is consistent with the observation that the

c-Jun δ-domain preferentially binds JNK-2 (Kallunki et al.,

1994; Sluss et al., 1994; Gupta et al., 1996) whereas the

Elk-1 D-domain preferentially binds JNK-1 (Figure 4;

data not shown). Residual ERK phosphorylation of cJunδ–

Elk-330 was inhibited by the wild-type and M3 peptides

Fig. 5. The JNK-binding domain of Elk-1 (D-domain) and c-Jun

(δ-domain) can be functionally interchanged. (A) A diagram

illustrating fusions of GST to Elk-310 (amino acids 310–428), Elk-330

(amino acids 330–428), ElkD (amino acids 310–329), cJunδ–Elk-330

chimera (amino acids 1–58 of c-Jun, Elk-1 amino acids 330–428),

ElkD–cJun chimera (amino acids 321–348 of Elk-1, c-Jun amino acids

55–223), ElkD–cJun (S63A/S73A), cJun43–223 (c-Jun amino acids

43–223) and c-Jun (amino acids 1–223). The numbers of the N- and

C-terminal c-Jun (normal type) and Elk-1 (italics) amino acids are

indicated above and below each construct respectively. The locations

of the major JNK phosphorylation sites in c-Jun (Ser63 and Ser73) are

indicated as the most functionally important phosphoacceptor sites in

Elk-1 (Ser383 and Ser389). The sequences of the Elk-1 D-domain and

c-Jun δ-domain are shown and are indicated by solid and open boxes,

respectively. The Elk-1 transcriptional activation domain (C-) is shown

by a grey box. Identical or highly conserved (Arg/Lys) residues

conserved between the two domains are indicated by dashes. The

central ‘LXL’ motif is bracketed. (B) Kinase assays of GST fusion

proteins as substrates for the indicated MAPKs were carried out using

5 pmol of each protein for 30 min reactions as described in Figure 1.

the presence of competitor peptides. Peptide competition assays were

carried out as described in Figure 4. Five pmol of each chimeric

protein were used in the reactions with 500 pmol of the indicated

peptides. The exposure of the ERK2 panel was selected in order to

show equivalent ERK and JNK phosphorylation of cJunδ–Elk-330.

MAPK targeting of Elk-1

Fig. 6. Role of the D-domain in Elk-1 phosphorylation in vivo in IL-1-stimulated CHO and NIH-3T3 cells. CHO (A–D) or NIH-3T3 (E–G) cells

were transfected with 2 µg of CMV-driven expression vectors encoding either wild-type Elk-1 or Elk-1∆D. Total cell extracts were taken at the

indicated times after IL-1 stimulation and analysed by Western blot using the anti-Phospho Elk-1(383) antibody to detect phosphorylation of Ser383

(A and E) or anti-FLAG antibody to examine the total levels of Elk-1 and Elk-1∆D in each sample (B and F). Samples were also analysed by gel

retardation analysis in the presence of core

SRF

and the c-fos SRE (SRE*) (C, D, G and H). Anti-phospho Elk-1(383) antibody was also included.

Supershifted bands representing complexes containing phosphorylated Elk-1 derivatives are shown in (C) and (G). Ternary complexes with core

SRF

and a 134 bp fragment of c-fos promoter containing the SRE (SRE*) are shown in (D) and (H). The locations of unphosphorylated ternary complex

(3°I) and multiple phosphorylated forms of ternary complex (3°II) are indicated. (C and D) and (G and H) are from separate experiments using the

same extracts.

but not by the M5 and JIP peptides (Figure 5C, top

panel, lanes 6–10) as predicted from previous experiments

(Figure 4).

Collectively, these data demonstrate that the Elk-1

D-domain can act as a JNK-binding motif in a

heterologous context. However, differences in the relative

strength of kinase binding are apparent, with the c-Jun

δ-domain acting as a stronger JNK-2-binding site than

the Elk-1 D-domain and the Elk-1 D-domain preferably

binding JNK-1 rather than JNK-2.

Differential targeting of the JNK and p38 MAPKs

to Elk-1 in vivo

Evidence has been gathered to suggest that Elk-1 is a

substrate for both the JNK (Cavigelli et al., 1995; Gille

et al., 1995; Whitmarsh et al., 1995, 1997) and p38 (Price

et al., 1996; Whitmarsh et al., 1997) MAPK subclasses

in vivo. However, the response of Elk-1 to these different

stress-activated MAPK pathways differs depending on the

cell type and stimulus. For example, UV activates Elk-1

via the p38 pathway in HeLa and NIH-3T3 cells (Price

et al., 1996). In contrast, interleukin-1 (IL-1) activates

Elk-1 via the JNK pathway in CHO cells, but by a

combination of the JNK and p38 pathways in NIH-3T3

cells (Whitmarsh et al., 1997). We took advantage of this

cell type-speciﬁc response to IL-1 to investigate the role

of the D-domain in targeting stress-activated MAPKs to

Elk-1 in vivo.

CHO and NIH-3T3 cells were transfected with either

Elk-1 or Elk-1∆D, stimulated with IL-1, and cell extracts

subsequently were prepared over a 15 min time period.

1745

The phosphorylation status of the Elk-1 derivatives was

monitored indirectly by gel retardation analysis (Figure

6D and H) and directly by supershift analysis of the

Elk-1-containing ternary complexes using an anti-phospho

Elk-1(Ser383) antibody (Figure 6C and G) which was

also used in Western blotting experiments (Figure 6A and

E). In CHO cells, the phosphorylation of Ser383 in Elk-1

occurred rapidly within 5 min of stimulation (Figure 6A

and C, lane 3), and was accompanied by a shift in the

mobility of the ternary DNA-bound complex (Figure

6D, lane 3). In comparison, the speed and degree of

phosphorylation of Ser383 and induction of a lower

mobility ternary complex containing Elk-1∆D was

reduced. Maximal phosphorylation of Ser383 was not

reached for 15 min (Figure 6A and C), and the stimulation

of a lower mobility DNA-bound complex was reduced in

comparison with wild-type Elk-1 (Figure 6D, compare

lanes 1–5 and 6–10). Taken together with the observation

that JNKs represent the major IL-1-activated protein

kinases in CHO cells that activate Elk-1 (Whitmarsh et al.,

1997), these data provide strong evidence for a role of

the D-domain in targeting these cascades to Elk-1 in vivo.

In contrast, in NIH-3T3 cells, phosphorylation of Ser383

in response to IL-1 stimulation took place in Elk-1 and

Elk-1∆D with virtually identical kinetics (Figure 6E and

G). However, in comparison with CHO cells, the magni-

tude of Ser383 phosphorylation was reduced, which is

consistent with the observation that minimal effects on

the mobility of ternary complexes containing either Elk-1

or Elk-1∆D were observed (Figure 6H). Together with the

observation that p38 is a major effector of IL-1 signalling

S.-H.Yang et al.

Fig. 7. The role of the Elk-1 D-domain in IL-1-inducible

transcriptional activation in vivo.(A) A diagrammatic representation of

wild-type and mutant Elk-1 proteins fused to the DNA-binding domain

of GAL4. CHO (B and C) or NIH-3T3 cells (D and E) were

transfected with CMV promoter-driven constructs encoding GAL4

fusions to either wild-type or mutant Elk-1 derivatives and a GAL4-

driven luciferase reporter plasmid. Cells were either unstimulated or

stimulated with IL-1. Transfection efﬁciency was monitored by using

the β-galactosidase expression vector pCH110. The luciferase activities

relative to the unstimulated cells of each wild-type or mutant (means

6 standard deviations; n 5 3) are presented. Expression levels of the

GAL4 fusion proteins in CHO (C) and NIH-3T3 (E) cell lines were

examined by Western blotting using total cell extracts with an anti-

GAL4 antibody.

to Elk-1 in NIH-3T3 cells (Whitmarsh et al., 1997), these

results are consistent with the notion that the Elk-1 D-

domain is not required for targeting by p38 in vivo.

The response of GAL4 fusion proteins containing the

Elk-1 transcriptional activation domain to IL-1 stimulation

in these cell types was also investigated. Wild-type Elk-1

and proteins containing mutations in the D-domain were

constructed which either did not affect (M4) or reduced

(M5) the efﬁciency of phosphorylation by the JNKs

in vitro (Figure 7A). These proteins were all expressed at

equivalent levels in CHO and NIH-3T3 cells (Figure 7C

and E). In CHO cells, the response of the M5 mutant to

IL-1 stimulation was severely reduced (65% reduction)

whereas, in comparison, the M4 mutant was virtually

unaffected (15% reduction; Figure 7B). However, in NIH-

3T3 cells, the response of the M5 mutant was only slightly

reduced (20% reduction) and again the M4 mutant was

virtually unaffected (Figure 7D). These results therefore

add further weight to the notion that the D-domain is

required for targeting of JNK MAPKs but not the p38

MAPKs to Elk-1 in vivo.

Discussion

MAPK signalling cascades play a pivotal role in converting

extracellular signals into speciﬁc nuclear responses

1746

(reviewed in Treisman, 1996; Whitmarsh and Davis,

1996). In order to achieve such speciﬁc responses, MAPKs

must recognize their substrates with high speciﬁcity. A

general mechanism is emerging to generate such speciﬁcity

in which initial binding of the kinases to a docking domain

is followed by recognition of the local context of the

phosphorylation motifs (Gupta et al., 1996; Kallunki et al.,

1996; Yang et al., 1998). In the case of c-Jun, the δ-

domain directs binding by the JNK subclass of MAPKs

(Derijard et al., 1994; Kallunki et al., 1994; Sluss et al.,

1994; Dai et al., 1995). Similarly, the Elk-1 D-domain

directs binding of the ERK subclass of MAPKs (Yang

et al., 1998). However, in this study, we demonstrate that

the Elk-1 D-domain possesses dual speciﬁcity and is also

a target for the JNK MAPKs. This domain does not,

however, allow promiscuous MAPK binding but discrim-

inates against binding by the p38 MAPK subclass. This

is the ﬁrst reported example of such a motif which allows

diverse MAPK pathways to be integrated via a single

transcription factor.

MAP kinase targeting domains

Short motifs which bind to the JNK MAPKs have been

identiﬁed in the JNK inhibitor protein JIP-1 (Dickens

et al., 1997) and the transcription factor c-Jun (Derijard

et al., 1994; Kallunki et al., 1994; Sluss et al., 1994; Dai

et al., 1995; Gupta et al., 1996). In this study, we have

identiﬁed a further motif in Elk-1 which targets JNKs to

transcription factors. Similarities between the binding

motifs from c-Jun and Elk-1 (Figure 5A) and JIP-1 and

Elk-1 (Figure 4A) are apparent, with the consensus motif

XXXXL

L representing a core binding site. The

JNK-binding site on ATF-2 also contains a motif which

loosely conforms to this consensus (KHEMTLKF) (Gupta

et al., 1995; Livingstone et al., 1995). However, examina-

tion of the binding motifs of several JNK substrates reveals

little similarity outside this central core motif although

more extensive similarities between pairs of proteins can

be observed (e.g. c-Jun/JIP-1 and Elk-1/NFAT4; Figure

8). Moreover, c-Jun preferentially binds to JNK-2

(Kallunki et al., 1994; Sluss et al., 1994; Gupta et al., 1996)

whereas Elk-1 preferentially binds to JNK-1 (Figures 4

and 5). Similarly, ATF-2 exhibits preferential binding of

certain JNK isoforms (Gupta et al., 1996). The non-

conserved residues surrounding this central core motif

must dictate these subtly different binding preferences.

Further speciﬁcity determinants must be built into the Elk-

1 D-domain as this is also bound efﬁciently by the ERK

MAPKs (Yang et al., 1998). Similarly, further speciﬁcity

determinants must be built into the ATF-2 docking site as

this serves as both a JNK- and p38-binding motif (N.Jones,

personal communication). Binding of the ERK and JNK

MAPKs to Elk-1 probably occurs with different kinetics

as, although binding can be detected by a competition

assay (Figure 4), only ERK binding can be detected by a

GST pull-down assay (Yang et al., 1998; data not shown).

Moreover, as wild-type D-domain peptides displace JNKs

more readily than ERK-2 from Elk-1, this suggests that

the JNKs may bind with high afﬁnity but rapidly associate

and dissociate from the substrate in the presence of ATP

as observed with p38 binding to MEF2C (Han et al.,

1997). Stable complexes between Elk-1 and JNKs have

been demonstrated previously (Gille et al., 1996), but we

MAPK targeting of Elk-1

Fig. 8. Similarity amongst JNK-binding motifs. The sequences of the

JNK-binding domains found in Elk-1, c-Jun, JunB, JIP-1, NFAT4,

ATF-2 and ATFa are shown. The sequence of the D-domain of SAP-1

is also shown. The asterisk denotes that SAP-1 is predominantly an

ERK substrate and is poorly phosphorylated by JNKs. The MAPK-

binding motifs in Elk-1 and ATF-2 exhibit dual speciﬁcity with both

JNK and ERK MAPKs targeted to Elk-1 (this study) and both JNK

and p38 MAPKS targeted to ATF-2 via a single motif (N.Jones,

personal communication). Amino acid numbers of the N- and

C-terminal residues are given based on their location in either human

c-Jun, JunB, Elk-1, NFAT4, ATF-2, ATFa, SAP-1 or mouse JIP-1

proteins. Identical or highly conserved (Arg/Lys) amino acids

comprising the R/KXXXXLXL motif are highlighted. Brackets

indicate the conserved central MAPK-binding motif.

are unable to detect such complexes under conditions in

which we detect ERK2 binding. However, although we

can detect stable complexes of ERK2 and Elk-1 both

in vitro (Yang et al., 1998) and in vivo when both partners

are overexpressed (data not shown), the physiologically

relevant binding event is likely to involve a transient

interaction of the kinase and substrate after cellular stimu-

lation. An emerging theme amongst MAPK targeting

motifs is that variations on a simple core consensus can

dictate binding of speciﬁc subsets of transcription factors.

Speciﬁc rules are difﬁcult to discern from the currently

characterized motifs, although all are short sequences (18

amino acids in the case of the D-domain) and the central

‘LXL’ motif clearly plays a pivotal role in kinase targeting

to transcription factors (e.g. M5 mutant, Figures 3, 4 and

7). However, MAPKs may also be targeted to substrates

by different motifs, as binding of the ERK and p38

MAPKs to Mnk1 and Mnk2 kinases does not appear to

be mediated by sequences related to the Elk-1 D-domain

(Waskiewiecz et al., 1997).

In contrast to the targeting of ERK and JNK MAPKs

to Elk-1 via the D-domain, this domain does not play a

role in the binding of p38 MAPKs (Figure 4). The

D-domain, therefore, does not represent a promiscuous

MAPK targeting motif but allows discrimination between

different classes of MAPKs.

Stress-activated MAPKs and Elk-1 activation

Elk-1 can be phosphorylated and activated by both the

JNK (Cavigelli et al., 1995; Gille et al., 1995; Whitmarsh

et al., 1995, 1997) and p38 (Price et al., 1996; Whitmarsh

et al., 1997) MAPK subclasses in vitro and in vivo.

However, the response of Elk-1 to these different stress-

activated MAPK pathways differs depending on the cell

type and stimulus. IL-1 activates Elk-1 via the JNK

pathway in CHO cells but by a combination of the JNK

and p38 pathways in NIH-3T3 cells (Whitmarsh et al.,

1747

1997), whereas UV activates Elk-1 via a combination of

the p38 and ERK pathways in HeLa and NIH-3T3 cells

(Price et al., 1996). In the present study, several lines of

evidence indicate that phosphorylation of Elk-1 requires

targeting of the kinase by binding to the D-domain for

the JNK but not the p38 MAPKs. First, deletions of the

D-domain reduce the efﬁciency of Elk-1 phosphorylation

by members of the JNK but not the p38 MAPK subclasses

(Figures 1 and 2). Secondly, point mutations in the

D-domain cause minimal reductions in the efﬁciency of

Elk-1 phosphorylation by the p38 subclass in comparison

with the JNK MAPKs (Figure 3). Thirdly, peptide competi-

tion assays demonstrate a role for the D-domain in binding

the JNK but not the p38 MAPKs (Figure 4). Interestingly,

although not as marked as the effect on the ERK and JNK

MAPKs, subtle changes in the proﬁciency of Elk-1 as a

substrate for p38α and, to a lesser extent, for p38β2 are

uncovered in the M1 and M2 mutants (Figure 3). As

deletion of the D-domain does not affect phosphorylation

of Elk-1 by the p38 MAPKs (Figures 1 and 2), this result

may reﬂect that these mutations cause subtle conforma-

tional changes in Elk-1 which may restrict kinase access

to the phosphoacceptor motifs. However, as the D-domain

peptide acts as a weak competitor of Elk-1 phosphorylation

by p38β

(Figure 4), this might reﬂect that the Elk-1-

binding site on p38 MAPKs retains some similarity to the

other MAPKs and hence binds weakly to Elk-1 via

the D-domain. Finally, in vivo, the efﬁciency of Elk-1

phosphorylation and activation by p38 MAPK is not

dependent upon the integrity of the D-domain (Figures 6

and 7). The latter result is also consistent with the

observation that stimulation of Elk-1 phosphorylation and

activity in vivo by the p38 MAPK pathways is not as

pronounced as activation by either the JNK (compare

Figure 6D and H and Figure 7B and D; Whitmarsh et al.,

1997) or ERK pathways (Yang et al., 1998). Furthermore,

in response to UV stimulation in HeLa and NIH-3T3

cells, the p38 pathway appears insufﬁcient for full Elk-1

activation and requires cooperation with the ERK pathway

(Price et al., 1996). Finally, in 293 cells, Elk-1 does not

appear to be activated by p38 MAPKs (Janknecht and

Hunter, 1997a). The reduced activation by and lack of

targeting of p38 MAPKs to Elk-1 in vitro and in vivo in

comparison with the strong activation elicited by the ERK

and JNK MAPKs suggests that the residual activation by

p38 may be a consequence of overexpressing pathway

components and overriding normal speciﬁcity determin-

ants. The p38 MAPKs may, however, phosphorylate Elk-1

in a more constitutive manner which does not require rapid

and efﬁcient kinase targetingto the substrate. Alternatively,

sustained p38 activation may be required to activate Elk-

1. Further studies are required to differentiate between

these possibilities. Finally, p38 may bind to a docking site

located elsewhere on Elk-1. However, none of the deleted

proteins we have tested exhibit markedly reduced phos-

phorylation by p38 isoforms (Figure 1; data not shown).

Moreover, no binding of p38α, β

and γ to Elk-1 could

be detected in GST pull-down assays carried out in the

absence of ATP (data not shown). Together these results

strongly suggest that p38 does not require or bind to a

docking motif on Elk-1.

In conclusion, our data contribute to the understanding

of how MAPKs recognize and phosphorylate their nuclear

S.-H.Yang et al.

targets. Elk-1 contains a MAPK-binding motif which

allows efﬁcient targeting and subsequent transcription

factor activation by both the ERK and JNK subclasses of

MAPKs. Similarly, both JNK and p38 subclasses of

MAPKs are targeted via a single binding motif to ATF-2

(N.Jones, personal communication). Future studies are

likely to uncover further motifs which direct binding of

the p38 subclass of MAPKs to transcription factors (e.g.

MEF2C; Han et al., 1997). Investigation of these targeting

domains will provide signiﬁcant insights into how the

speciﬁcity of signalling via MAPK cascades is achieved.

Materials and methods

Plasmid constructs

The following plasmids were used for expressing GST fusion proteins in

Escherichia coli. pAS407 (encoding GST–Elk205; Elk-1 amino acids

205–428), pAS545 (encoding GST–Elk310; Elk-1 amino acids 310–428),

pAS406 (encoding GST–Elk330; Elk-1 amino acids 330–428), pAS405

(encodingGST–Elk349;Elk-1 aminoacids 349–428)and pAS547(encod-

ing GST–ElkD–cJun; Elk-1 amino acids 310–348 fused to c-Jun amino

acids55–223)havebeendescribedpreviously(Yanget al.,1998).Plasmids

encoding GST–cJun1–223 and GST–cJun43–223 have been described

previously (Hibi et al., 1993). pAS767, encoding GST fused to Elk-1

aminoacids307–329(GST–ElkD), wasconstructedbyinsertingaBamHI–

EcoRI-cleaved PCR-derived fragment into the same sites of pGEX-3X.

pAS768, encoding c-Jun amino acids 1–58 fused to Elk-1 amino acids

330–428 (cJunδ–Elk-330), was constructed by ligating a BamHI–BglII-

cleavedPCRfragment(encodingc-Junaminoacids1–58)intotheBglIIsite

of pAS406. pAS548, pAS549, pAS550, pAS564 and pAS769 (encoding

GST–Elk205 mutants) are derivatives of pAS407 with the site-directed

mutations R314A/K315A (GST–Elk205M1), R317A/L319A (GST–

Elk205M2), L323A/S324A (GST–Elk205M3), L327A/L328A (GST–

Elk205M4) and L319A/L321A (GST–Elk205M5), respectively. pAS565

(GST–Elk307M2), pAS566 (GST–Elk307M3), pAS567 (GST–

Elk310

S383A/S389A

) and pAS568 (GST–Elk310

M2/S383A/S389A

) were

described previously (Yang et al., 1998). Mutations were introduced by a

two-step PCR protocolusinga mutagenic primer and two ﬂanking primers

as described previously (Shore et al., 1996).

pAS278 and pAS380 [encoding full-length His/FLAG-tagged Elk-1

and the same protein with an internal deletion of amino acids 312–321

(Elk-1∆D)] were used for expression of Elk-1 derivatives in E.coli (Yang

et al., 1998).

The following plasmids were constructed for use in mammalian cell

transfections. pG5E1b contains ﬁve GAL4 DNA-binding sites cloned

upstream of a minimal promoter element and the ﬁreﬂy luciferase gene

(Seth et al., 1992). pSG424 encodes the GAL4 DNA-binding domain

(Sadowski and Ptashne, 1989). The vectors pCDNA3-F-JNK1 (Derijard

et al., 1994), pCDNA3-F-JNK2 (Sluss et al., 1994), pCMV5-F-p38α

(Raingeaud et al., 1995), pCDNA3-F-p38β

(Enslen et al., 1998)

and pCDNA3-F-p38γ (J.Raingeaud and R.J.Davis, unpublished data)

encoding ﬂag-tagged MAPKs have been described previously. pAS572

(pCMV-GAL4-Elk205) and pAS577 (pCMV-GAL4-Elk205M4) were

described previously (Yang et al., 1998). pAS770 (GAL4-Elk205M5)

was constructed by ligating BamHI–XbaI fragments from pAS769 into

the same sites of pSG424. pAS771 (pCMV-GAL4-Elk205M5) was

constructed by ligating the HindIII–XbaI fragments from pAS770 into

the same sites of pCMV5. The cytomegalovirus (CMV) promoter-driven

expression vectors, pAS383 and pAS387, encoding full-length Elk-1

and Elk-1∆D, respectively, with C-terminal FLAG tags were described

previously (Yang et al., 1998). All plasmid constructs encoding Elk-1-

derived proteins made by PCR were veriﬁed by automated dideoxy

sequencing.

Protein expression and puriﬁcation

GST fusion proteins were expressed in the E.coli JM101 strain and

puriﬁed asdescribedpreviously (Shore et al.,1995). Full-length hexahisti-

dine-tagged polypeptides were expressed in E.coli BL21(DE3)pLysS

with the pET vector system and quantiﬁed as described previously (Yang

et al., 1998).

Tissue culture, cell transfection and reporter gene assays

COS-1 cells were maintained in Dulbecco’s modiﬁed Eagle’s medium

(DMEM) supplemented with 10% fetal bovine serum (FBS) (Gibco-

1748

BRL). CHO cells were maintained in F12 medium supplemented with

5% FBS. NIH-3T3 cells were maintained in DMEM supplemented with

10% FBS. Transfection experiments were carried out using Superfect

transfection reagent (Qiagen) as described previously (Yang et al., 1998).

For reporter gene assays, GAL4-driven promoters were co-transfected

with various vectors encoding GAL4–Elk-1 fusion proteins. The activities

of the GAL4 DNA-binding domain (amino acids 1–147) and GAL4–

Elk-1 deletion fusion proteins (50 ng of plasmid DNA) were measured

in co-transfection assays in all cell lines using 1 µg of reporter plasmid

pG5E1bLuc. Transfection efﬁciencies were normalized by measuring

the activity from a co-transfected plasmid (1 µg) which expresses

β-galactosidase (pCH110, Pharmacia KB Biotechnology Inc.). Cell

extracts were prepared, and luciferase and β-galactosidase assays were

carried out as described previously (Yang et al., 1998).

Protein kinase assays

In order to prepare recombinant JNK and p38 MAPKs, COS-1 cells

were transfected with constructs encoding FLAG epitope-tagged MAPKs.

Kinases were activated by stimulation with UV light for 30 min. Puriﬁed

kinases were then eluted from beads by competing with 0.1 mg/ml of

FLAG peptide. Recombinant active ERK2 was obtained from New

England Biolabs (NEB) and MBP was obtained from Sigma. The kinase

assays were carried out in 20 µl reaction volumes as described previously

(Yang et al., 1998). The phosphorylation of substrate proteins was

examined following SDS–PAGE by autoradiography, and quantiﬁed by

phosphorimaging (Fuji BAS1500; TINA 2.08e software). Data were

quantiﬁed by phosphorimaging and the data presented graphically after

curve ﬁtting with the appropriate equation using BIOSOFT Fig.P or

Microsoft Excel software. Peptide competition experiments were carried

out essentially as described above, except that pre-incubation with

50–5000 pmol of the peptide competitors with MAPKs was carried out

before the kinase reactions. Final peptide concentrations were

2.5–25 µM (10- to 100-fold excess over Elk-1 substrate).

Western blot analysis

FLAG-tagged Elk-1 and Elk-1∆D in extracts from CHO or NIH-3T3

cells were detected by immunoblot analysis using a mouse monoclonal

anti-M2 FLAG antibody (Kodak) or anti-phosphoplus™ Elk-1 (S383)

antibody (NEB). GAL4 fusion proteins were detected using the anti-

GAL4 antibody directed against the amino-terminal DNA-binding

domain (Santa Cruz). Immune complexes were detected by using

horseradish peroxidase-conjugated secondary antibody followed by ECL

(Amersham).

Gel retardation assays

Gel retardation assays were performed with a

P-labelled 134 bp c-fos

promoter fragment containing the SRE (SRE*) as described previously

(Whitmarsh et al., 1995). Total cell extracts from transfected cells

containing ~0.02 pmol of Elk-1 or Elk-1∆D were used in DNA-binding

reactions. Binding reactions on SRE-containing sites also contained

puriﬁed bacterially expressed core

SRF

(Shore and Sharrocks, 1994).

Antibody supershift experiments were described previously (Yang et al.,

1998). Protein–DNA complexes were analysed on non-denaturing 5%

polyacrylamide gels cast in 0.53 Tris–borate–EDTA and visualized by

autoradiography and phosphorimaging.

Figure generation and data quantiﬁcation

All ﬁgures were generated electronically from scanned images of

autoradiographic images by using Picture Publisher (Micrograﬁx) and

Powerpoint version 7.0 (Microsoft) software. Final images are represent-

ative of the original autoradiographic images. Phosphorimager data were

quantiﬁed using Tina software (version 2.08e).

Acknowledgements

We thank Margaret Bell and Catherine Pyle for excellent technical and

secretarial assistance, Bob Liddell for DNA sequencing and oligonucleo-

tide synthesis and Arthur Moir for peptide synthesis. We are grateful to

Steve Yeaman, Janet Quinn and members of our laboratories for

comments on the manuscript and for stimulating discussions, and to Nic

Jones for communicating data prior to publication. This work was

supported by the North of England Cancer Research Campaign, the

Wellcome Trust and the National Cancer Institute (USA). R.J.D. is an

investigator of the Howard Hughes Medical Institute. A.D.S. is supported

by the Lister Institute of Preventative Medicine.

MAPK targeting of Elk-1

References

Bocco,J.L., Bahr,A., Goetz,J., Hauss,C., Kallunki,T., Kedinger,C. and

Chatton,B. (1996) In vivo association of ATFa with JNK/SAP kinases

activities. Oncogene, 12, 1971–1980.

Cavigelli,M., Dolﬁ,F., Claret,F.X. and Karin,M. (1995) Induction of c-fos

expression through JNK-mediatedTCF/Elk-1 phosphorylation. EMBO.

J., 14, 5957–5964.

Chow,C.-W., Rincon,M., Cavanagh,J., Dickens,M. and Davis,R.J. (1997)

Nuclear accumulation of NFAT4 opposed by the JNK signal

transduction pathway. Science, 278, 1638–1641.

Cuenda,A., Cohen,P., Buee-Scherrer,V. and Goedert,M. (1997) Activation

of stress-activated protein kinase-3 (SAPK3) by cytokines and cellular

stresses is mediated via SAPKK3 (MKK6); comparison of the

speciﬁcities of SAPK3 and SAPK2 (RK/p38). EMBO J., 16, 295–305.

Dai,T., Rubie,E., Franklin,C.C., Kraft,A., Gillespie,D.A.F., Avruch,J.,

Kyriakis,J.M. and Woodgett,J.R. (1995) Stress-activated protein

kinases bind directly to the δ domain of c-Jun in resting cells:

implications for repression of c-jun function. Oncogene, 10, 849–855.

rijard,B., Hibi,M., Wu,I-H., Barrett,T., Su,B., Deng,T., Karin,M. and

Davis,R.J. (1994) JNK1: a protein kinase stimulated by UV light and

Ha-Ras that binds and phosphorylates the c-jun activation domain.

Cell, 76, 1025–1037.

Dickens,M., Rogers,J., Cavanagh,J., Raitano,A., Xia,Z., Halpern,J.R.,

Greenberg,M.E., Sawyers,C. and Davis,R.J. (1997) A cytoplasmic

inhibitor of the JNK signal transduction pathway. Science, 277,

693–696.

Enslen,H., Raingeaud,J. and Davis,R.J. (1998) Selective activation of

p38 MAP kinase isoforms by the MAP kinase kinases MKK3 and

MKK6. J. Biol. Chem., 273, 1741–1748.

Gille,H., Strahl,T. and Shaw,P.E. (1995) Activation of ternary complex

factor Elk-1 by stress-activated protein kinases. Curr. Biol., 5, 1191–

1200.

Goedert,M., Cuenda,A., Craxton,M., Jakes,R. and Cohen,P. (1997)

Activation of the novel stress-activated protein kinase SAPK4 by

cytokines and cellular stresses is mediated by SKK3 (MKK6);

comparison of its substrate speciﬁcity with that of other SAP kinases.

EMBO J., 16, 3563–3571.

Gupta,S. and Davis,R.J. (1994) MAP kinase binds to the NH

-terminal

activation domain of c-Myc. FEBS Lett., 353, 281–285.

Gupta,S., Campbell,D., De

rijard,B. and Davis,R.J. (1995) Transcription

factor ATF2 regulation by the JNK signal transduction pathway.

Science, 267, 389–393.

Gupta,S., Barrett,T., Whitmarsh,A.J., Cavanagh,J., Sluss,H.K.,

rijard,B. and Davis,R.J. (1996) Selective interaction of JNK protein

kinase isoforms with transcription factors. EMBO J., 15, 2760–2770.

Han,J., Jiang,Y., Li.,Z., Kravchenko,V.V. and Ulevitch,R.J. (1997)

Activation of the transcription factor MEF2C by the MAP kinase p38

in inﬂammation. Nature, 386, 296–299.

Hibi,M., Lin,A., Smeal,T., Minden,A. and Karin,M. (1993) Identiﬁcation

of an oncoprotein- and UV-responsive protein kinase that binds and

potentiates the c-Jun activation domain. Genes Dev., 7, 2135–2148.

Janknecht,R. and Hunter,T. (1997a) Convergence of MAP kinase

pathways on the ternary complex factor SAP-1a. EMBO J., 16,

1620–1627.

Janknecht,R. and Hunter,T. (1997b) Activation of the SAP-1a

transcription factor by the c-Jun N-terminal kinase (JNK) mitogen-

activated protein kinase. J. Biol. Chem., 272, 4219–4224.

Jiang,Y., Chen,C., Li,Z., Guo,W., Gegner,J.A., Lin,S. and Han,J. (1996)

Characterization of the structure and function of a new mitogen-

activated protein kinase (p38) J. Biol. Chem., 271, 17920–17926.

Kallunki,T., Su,B. Tsigelny,I., Sluss,H.K., De

rijard,B., Moore,G.,

Davis,R.J. and Karin,M. (1994) JNK2 contains a speciﬁcity-

determining region responsible for efﬁcient c-jun binding and

phosphorylation. Genes Dev., 8, 2996–3007.

Kallunki,T., Deng,T., Hibi,M. and Karin,M. (1996) c-Jun can recruit

JNK to phosphorylate dimerization partners via speciﬁc docking

interactions. Cell, 87, 929–939.

Kyriakis,J.M., BanerjeeP., Nikolakaki,E., Dai,T., Rubie,E.A.,

Ahmad,M.F., Avruch,J. and Woodgett,J.R. (1994) The stress-activated

protein kinase subfamily of c-Jun kinases. Nature, 369, 156–160.

Lechner,C., Zahalka,M.A., Giot,J.-F., Møller,N.P.H. and Ullrich,A.

(1996) ERK6, a mitogen-activated protein kinase involved in C2C12

myoblast differentiation. Proc. Natl Acad. Sci. USA, 93, 4355–4359.

Livingstone,C., Patel,G., and Jones,N. (1995) ATF-2 contains a

phosphorylation-dependent transcriptional activation domain. EMBO

J., 14, 1785–1797.

1749

Mao,C., Ray-Gallet,D., Tavitian,A. and Moreau-Gachelin,F. (1996)

Differential phosphorylations of Spi-B and Spi-1 transcription factors.

Oncogene, 12, 863–873.

Price,M.A., Cruzalegui,F.H. and Treisman,R. (1996) The p38 and ERK

MAP kinase pathways co-operate to activate ternary complex factors

and c-fos transcription in response to UV light. EMBO J., 15,

6552–6563.

Raingeaud,J., Gupta,S., Rogers,J.S., Dickens,M., Han,J., Ulevitch,R.J.

and Davis,R.J. (1995) Pro-inﬂammatory cytokines and environmental

stress cause p38 mitogen-activated protein kinase activation by dual

phosphorylation on tyrosine and threonine. J. Biol. Chem., 270,

7420–7426.

Robinson,M.J. and Cobb,M.H. (1997) Mitogen-activated protein kinase

pathways. Curr. Opin. Cell Biol., 9, 180–186.

Sadowski,I. and Ptashne,M. (1989) A vector for expressing GAL4(1–

147) fusions in mammalian cells. Nucleic Acids Res., 17, 7539.

Seth,A., Gonzalez,F.A., Gupta,S., Raden,D.L. and Davis,R.J. (1992)

Signal transduction within the nucleus by mitogen-activated protein

kinase. J. Biol. Chem., 267, 24796–24804.

Shore,P. and Sharrocks,A.D. (1994) The transcription factors Elk-1 and

serum response factor interact by direct protein–protein contacts

mediated by a short region of Elk-1. Mol. Cell. Biol., 14, 3283–3291.

Shore,P., Bisset,L., Lakey,J., Waltho,J.P., Virden,R. and Sharrocks,A.D.

(1995) Characterization of the Elk-1 ETS DNA-binding domain.

J. Biol. Chem., 270, 5805–5811.

Shore,P., Whitmarsh,A.J., Bhaskaran,R., Waltho,J., Davis,R.J. and

Sharrocks,A.D. (1996) Determinants of DNA binding speciﬁcity of

ETS-domain transcription factors. Mol. Cell. Biol., 16, 3338–3349.

Sluss,H.K., Barrett,T., Derijard,B. and Davis,R.J. (1994) Signal

transduction by tumor necrosis factor mediated by JNK protein

kinases. Mol. Cell. Biol., 14, 8376–8384.

Stein,B., Yang,M.X., Young,D.B., Janknecht,R., Hunter,T., Murray,B.W.

and Barbosa,M.S. (1997) p38-2, a novel mitogen-activated protein

kinase with distinct properties. J. Biol. Chem., 272, 19509–19517.

Strahl,T., Gille,H. and Shaw,P.E. (1996) Selective response of ternary

complex factor Sap-1a to different mitogen-activated protein kinase

sub-groups. Proc. Natl Acad. Sci. USA, 93, 11563–11568.

Treisman,R. (1994) Ternary complex factors: growth regulated

transcriptional activators. Curr. Opin. Genet. Dev., 4, 96–101.

Treisman,R. (1996) Regulation of transcription by MAP kinase cascades.

Curr. Opin. Cell Biol., 8, 205–215.

Van Dam,H., Wilhelm,D., Herr,I., Steffen,A., Herrlich,P. and Angel,P.

(1995) ATF-2 is preferentially activated by stress-activated protein

kinases to mediate c-jun induction in response to genotoxic agents.

EMBO J., 14, 1798–1811.

Wang,X. and Ron,D. (1996) Stress-induced phosphorylation and

activation of the transcription factor CHOP (GADD153) by p38 MAP

kinase. Science, 272, 1347–1349.

Wang,X.S. et al. (1997) Molecular cloning and characterization of a

novel p38 mitogen-activated protein kinase. J. Biol. Chem., 272,

23668–23674.

Waskiewicz,A.J., Flynn,A., Proud,C.G. and Cooper,J.A. (1997) Mitogen-

activated protein kinases activate the serine-threonine kinases Mnk1

and Mnk2. EMBO J., 16, 1909–1920.

Whitmarsh,A.J. and Davis,R.J. (1996) Transcription factor AP-1

regulation by mitogen-activated protein kinase signal transduction

pathways. J. Mol. Med., 74, 589–607.

Whitmarsh,A.J., Shore,P., Sharrocks,A.D. and Davis,R.J. (1995)

Integration of MAP kinase signal transduction pathways at the serum

response element. Science, 269, 403–407.

Whitmarsh,A.J., Yang,S.-H., Su,M.S.-S., Sharrocks,A.D. and Davis,R.J.

(1997) Role of p38 and JNK MAP kinases in the activation of ternary

complex factors. Mol. Cell. Biol., 17, 2360–2371.

Yang,S.-H., Whitmarsh,A.J., Davis,R.J. and Sharrocks,A.D. (1998) The

Elk-1 ETS-domain transcription factor containsaMAP kinase targeting

motif. Mol. Cell. Biol., 18, 710–720.

Received November 25, 1997; revised January 19, 1998;

accepted January 20, 1998