Contingent Phosphorylation/Dephosphorylation Provides a Mechanism of Molecular Memory in WASP
Cells can retain information about previous stimuli to produce distinct future responses. The biochemical mechanisms by which this is achieved are not well understood. The Wiskott-Aldrich syndrome protein (WASP) is an effector of the Rho-family GTPase Cdc42, whose activation leads to stimulation of the actin nucleating assembly, Arp2/3 complex. We demonstrate that efficient phosphorylation and dephosphorylation of WASP at Y291 are both contingent on binding to activated Cdc42. Y291 phosphorylation increases the basal activity of WASP toward Arp2/3 complex and enables WASP activation by new stimuli, SH2 domains of Src-family kinases. The requirement for contingency in both phosphorylation and dephosphorylation enables long-term storage of information by WASP following decay of GTPase signals. This biochemical circuitry allows WASP to respond to the levels and timing of GTPase and kinase signals. It provides mechanisms to specifically achieve transient or persistent actin remodeling, as well as long-lasting potentiation of actin-based responses to kinases.
Molecular Cell, Vol. 11, 1215–1227, May, 2003, Copyright 2003 by Cell Press
Provides a Mechanism of Molecular Memory in WASP
rise to morphological features such as filopodia and
lamellipodia (Nobes and Hall, 1995). Both Cdc42 and
Rac elicit actin filament formation through activation of
and Michael K. Rosen
Department of Biochemistry
University of Texas Southwestern Medical Center
proteins in the Wiskott-Aldrich syndrome protein (WASP)
5323 Harry Hines Boulevard
family (Caron, 2002). These proteins, which include
Dallas, Texas 75205
WASP, its widely expressed homolog N-WASP, the
Department of Physiology and Biophysics
Scar/WAVE proteins, and Bee1, link Cdc42 and Rac
Weill Medical College of Cornell University
directly to the cellular actin nucleation machine, the
New York, New York 10021
Arp2/3 complex (Higgs and Pollard, 2001; Pantaloni et
al., 2001). WASP and N-WASP, are effectors of Cdc42
and when bound to the activated GTPase can stimulate
the actin nucleating activity of Arp2/3 complex, leading
to rapid formation of new filaments (Higgs and Pollard,
Cells can retain information about previous stimuli to
2001; Pantaloni et al., 2001). The WAVE proteins function
produce distinct future responses. The biochemical
analogously to stimulate Arp2/3 complex in response to
mechanisms by which this is achieved are not well
Rac but are activated through an alternative mechanism
understood. The Wiskott-Aldrich syndrome protein
involving GTPase-mediated destabilization of an inhib-
(WASP) is an effector of the Rho-family GTPase Cdc42,
ited WAVE/Pir121/Nap1 complex rather than direct
whose activation leads to stimulation of the actin nu-
binding (Eden and Kirschner, 2002). The dynamics and
cleating assembly, Arp2/3 complex. We demonstrate
localization of Cdc42/Rac signals, and their coordina-
that efficient phosphorylation and dephosphorylation
tion with other pathways, are important determinants of
of WASP at Y291 are both contingent on binding to
the structure and dynamics of actin filament networks,
activated Cdc42. Y291 phosphorylation increases the
and their resultant functions in biological processes.
basal activity of WASP toward Arp2/3 complex and
Moreover, these signals likely exert their influence, at
enables WASP activation by new stimuli, SH2 domains
least in part, through modulating the timing and intensity
of Src-family kinases. The requirement for contin-
of WASP/WAVE signals to Arp2/3 complex. However,
gency in both phosphorylation and dephosphorylation
the relationship between WASP/WAVE activation dy-
enables long-term storage of information by WASP
namics and the nature of upstream signals is not well
following decay of GTPase signals. This biochemical
circuitry allows WASP to respond to the levels and
WASP activity toward Arp2/3 complex can be stimu-
timing of GTPase and kinase signals. It provides mech-
lated by a variety of other ligands in addition to Cdc42,
anisms to specifically achieve transient or persistent
, Nck, Grb2, WISH, and profilin (Caron,
actin remodeling, as well as long-lasting potentiation
2002). Of these, PIP
and Grb2 have been shown to act
of actin-based responses to kinases.
cooperatively with Cdc42 in vitro, suggesting that WASP
could act in an approximately switch-like manner in vivo
in response to low-levels of multiple stimuli (Carlier et
al., 2000; Prehoda et al., 2000). WASP is also phosphory-
Dynamic regulation of the actin cytoskeleton is an inte-
lated on tyrosine in response to a number of different
gral feature of many eukaryotic cellular processes, in-
signals, including overexpression of Hck, and stimula-
cluding migration, adhesion, and division (Borisy and
tion of B cell antigen, collagen, IgE, chemokine, and Trk
Svitkina, 2000; Pantaloni et al., 2001). The actin remodel-
receptors in appropriate cell types (Baba et al., 1999;
ing signals that drive these processes must be con-
Cory et al., 2002; Guinamard et al., 1998; Oda et al.,
trolled in their timing, duration, amplitude, and localiza-
1998; Okabe et al., 2002; Suetsugu et al., 2002). In a
tion. These parameters, often integrated through multiple
mast cell line, phosphorylation by Btk or Lyn is greatly
signaling pathways (Higgs and Pollard, 2001), cause de-
potentiated by coexpression of an activated mutant of
fined alterations in the basic biochemical properties of
Cdc42, indicating cooperativity between GTPase and
actin filaments, such as polymerization/depolymeriza-
kinase pathways (Guinamard et al., 1998). In B cells,
tion, branching, and bundling, resulting in functionally
macrophages, and PC12 cells, the site of WASP/
specific cytoskeletal structures and dynamics.
N-WASP phosphorylation has been mapped to Y291/
GTPases in the Rho family, including Cdc42, Rac, and
Y256, which lies in the conserved sequence IYDFI (Baba
Rho, are key components of signaling pathways that
et al., 1999; Cory et al., 2002; Suetsugu et al., 2002).
control actin filament networks (Etienne-Manneville and
This sequence is a consensus substrate for Src family
Hall, 2002). These proteins act as molecular switches in
tyrosine kinases (Songyang et al., 1995), suggesting that
vivo, alternating between an inactive GDP-bound state
Y291 is likely to be the site of phosphorylation by kinases
and an active GTP-bound state in response to upstream
in this group. Y291 lies in the region of WASP termed
signals (Etienne-Manneville and Hall, 2002). Cdc42 and
the GTPase binding domain (GBD, residues 230–310;
Rac act in many processes to induce polymerization of
Figure 1A), which mediates interaction with Cdc42
new actin filaments at the cell periphery, which give
(Abdul-Manan et al., 1999; Rudolph et al., 1998). In free
WASP, the GBD is bound in intramolecular fashion to a
C-terminal segment termed the VCA (for verprolin ho-
Figure 1. Structural Domains of WASP and
(A) Domain structure of WASP and N-WASP.
Thick and thin black bars indicate homolo-
gous regions and sequence gaps, respec-
tively. Layers 1 (WASP residues 230–276) and
2 (WASP residues 277–310) of the autoinhib-
ited fold, comprising the GBD, are boxed in
yellow and blue, respectively. The central re-
gion (WASP residues 466–476) of the VCA
segment is boxed in red.
(B) Molecular surface of the autoinhibited do-
main of WASP. Structural layers are colored
as in (A). The Tyr 291 side chain is shown in
purple as a stick model. Figure was created
with GRASP (Nicholls et al., 1991).
mology, central hydrophobic, and acidic), resulting in addition, p-WASP can be activated by SH2 domains of
Src family kinases due to steric incompatibility betweenautoinhibition toward Arp2/3 complex (Kim et al., 2000;
Miki et al., 1998; Rohatgi et al., 1999). The structures of the autoinhibited fold of WASP and SH2-pTyr interac-
tions. This biochemical scheme has important implica-the GBD-VCA and GBD-Cdc42 complexes are mutually
incompatible (Kim et al., 2000), and binding of Cdc42 tions for the role of WASP phosphorylation in signaling
pathways that connect Src family kinases throughto the GBD causes global destabilization of the autoin-
hibited GBD-VCA domain (Buck et al., 2001; M. Buck et Cdc42 to the actin machinery. It provides mechanisms
to achieve differential actin-based outputs based on theal., submitted), leading to release of the VCA, facilitating
its activation of Arp2/3 complex. In free WASP, Y291 relative timing, duration, and amplitude of GTPase and
kinase signals, and suggests a means by which the cellis ⬎90% buried in the fold of the autoinhibited domain
(Figure 1B). On this basis, we have suggested that desta- could retain a record of previous stimulation in order to
respond differently in the future.bilization by Cdc42 may be necessary to provide effi-
cient access to Y291 by kinases (Kim et al., 2000).
To date, the phosphorylation of WASP has been de- Results
scribed only phenomenologically in nearly all instances,
and the physiological significance of this modification Efficient Phosphorylation of WASP Proteins Is
Dependent on Binding to Activated Cdc42remains largely unexplored. However, very recently it
has been reported that N-WASP/WASP phosphorylation The autoinhibited GBD-VCA domain of WASP is com-
posed of a ␤ hairpin and five helices, which form aby Src family kinases increases the basal activity of
the proteins toward Arp2/3 complex (Cory et al., 2002; structure that can be considered in three layers (Kim et
al., 2000). The first layer, residues 230–276, consists ofSuetsugu et al., 2002). Moreover, Suetsugu et al. have
shown that this modification is necessary for NGF- the ␤ hairpin and ␣1 helix, which are the minimal ele-
ments necessary to bind Cdc42 with high affinity. Theinduced neurite outgrowth in PC12 cells (Suetsugu et
al., 2002). Interestingly, in this system, N-WASP phos- second layer, residues 277–310, consists of helices
␣2–␣4 and contains the remainder of the GBD. Thesephorylation persists well beyond the lifetime of activated
Src, suggesting that mechanisms exist in vivo to prolong elements are disordered in GBD-Cdc42 complexes
(Abdul-Manan et al., 1999). Layer 3 consists of a singlephosphorylation for extended periods.
Here we have examined the control of WASP phos- helix, ␣5, formed by the central region of the VCA (resi-
dues 466–479). Y291 is located near the center of helixphorylation at Y291 by Cdc42, kinases, and phospha-
tases, as well as the functional significance of this cova- ␣3 in the second layer of structure. It is an integral part
of the interface between layers 1 and 2, and is nearlylent modification. We demonstrate that WASP can be
efficiently phosphorylated at Y291 only in the presence buried by surrounding side chains of the ␤ hairpin and
the ␣3 helix (Figure 1B). Less than 10% of the residueof activated Cdc42. Moreover, isolated p-Y291 WASP
is highly protected from tyrosine phosphatases. Phos- surface area is accessible to solvent due to packing
interactions with the hydrophobic portions of the V249,phorylation can thus provide a long-lasting biochemical
record of coincident GTPase and kinase signals. This W252, K288, and D292 side chains. This organization
suggested that Y291 should be a poor substrate forrecord can be erased by a second round of Cdc42 acti-
vation, since dephosphorylation occurs readily in the tyrosine kinases in the autoinhibited state of WASP.
However, since binding to Cdc42 destabilizes the GBD-presence of GTPase. Phosphorylation at Y291 increases
the basal activity of WASP toward Arp2/3 complex. In VCA domain (Buck et al., 2001; M. Buck et al., submit-
Molecular Memory in WASP
Figure 2. Active Cdc42 Increases the Phos-
phorylation of Autoinhibited WASP and
N-WASP Constructs but Not of the GBD
In all figures, upper panel is Coomassie blue
stained SDS-PAGE gel, lower panel is anti-
phosphotyrosine Western blot. Identical re-
sults were obtained in all cases with Abl, Lck,
and Lyn; selected data are shown.
(A) Phosphorylation of 5 M GBD by the Abl
kinase domain alone (lane 1), or in the pres-
ence of 5 M Cdc42-GMPPCP (lane 2) or
Cdc42-GDP (lane 3).
(B) Phosphorylation of 4 M GBD by Lck alone
(lane 1), or in the presence of 50 M VCA (lane
2), 50 MVCA⫹ 8 M Cdc42-GMPPCP (lane
3),orVCA⫹ 8 M Cdc42-GDP (lane 4).
(C) Phosphorylation of 5 M tWASP by Lck
alone (lane 1), or in the presence of 40 M
Cdc42-GMPPCP (lane 2) or Cdc42-GDP (lane
3). Identical results were obtained with phos-
phorylation by Abl or Lyn.
(D) Phosphorylation of 20 M N-GBD-VCA by
the Lyn kinase domain alone (lane 1), or in
the presence of 15 M Cd42-GMPPCP (lane
2) or Cdc42-GDP (lane 3).
ted), the GTPase should increase the accessibility of 2D). In all cases, mass spectrometric analyses demon-
strated that the WASP proteins are singly phosphory-Y291 to kinases, facilitating its phosphorylation.
In order to test these hypotheses, we examined phos- lated at Y291 (Y256 in N-WASP), the only tyrosine resi-
due in each construct (data not shown). These dataphorylation of the WASP GBD by the Lck, Lyn, and Abl
kinases in the presence and absence of the VCA and indicate that in WASP/N-WASP, the autoinhibited fold
protects Y291/Y256 against covalent modification byCdc42. Identical results were obtained with all kinases,
indicating the effects of VCA and Cdc42 are due to kinases. Destabilization of the fold by active Cdc42 en-
ables ready phosphorylation. Thus, WASP and N-WASPchanges in Y291 accessibility rather than changes in
specificity of the GBD substrate. As shown in Figure could function as logical AND gates through their phos-
phorylation at Y291/Y256, with modification possible2A, the GBD peptide, which has no persistent tertiary
structure on its own, can be readily phosphorylated in only in the presence of both activated Cdc42 and acti-
vated kinase. In this way, the proteins could act as coin-isolation. Phosphorylation is unaffected by Cdc42 in ei-
ther the GMPPCP- or GDP-bound states (the former cidence detectors of GTPase and kinase signals, provid-
ing a biochemical explanation for the cooperativemimics the GTP-bound state). However, as shown in
Figure 2B, addition of the VCA, which induces folding behavior observed in vivo (Guinamard et al., 1998).
of the GBD (Kim et al., 2000), results in a dramatic de-
crease in phosphorylation efficiency. This decrease can Efficient Dephosphorylation Is Also Dependent
on Activated Cdc42be reversed by addition of Cdc42-GMPPCP, but is only
slightly affected by Cdc42-GDP. Thus, induction of the We next examined the effect of the autoinhibited fold
on the ability of tyrosine phosphatases to dephosphory-autoinhibited fold results in protection against phos-
phorylation, and destabilization of this structure by late p-Y291-WASP proteins. Since the phosphatase(s)
that act on WASP in vivo are not known, we used twoCdc42 enables covalent modification. This behavior is
reproduced in a C-terminal fragment of WASP, tWASP enzymes in these studies: the catalytic domain of the
mammalian leukocyte antigen-related protein tyrosine(residues 230–502), where the GBD and VCA elements
are tethered by the natural proline-rich linker. As shown phosphatase, LAR-PTP, a typical eukaryotic phospha-
tase, and the highly active bacterial phosphatase, YOPin Figure 2C, this protein is phosphorylated poorly in
isolation and in the presence of Cdc42-GDP. But it is (Dixon, 1995). Similar results were obtained in both
cases. As shown in Figure 3A, p-GBD is rapidly dephos-phosphorylated well in the presence of Cdc42-GMPPCP.
Similar behavior is also observed for an N-WASP con- phorylated in the presence of 5 nM LAR-PTP (lane 1).
In preliminary quantitative analyses (data not shown),struct, N-GBD-VCA, containing the GBD (residues 192–
275) linked directly to the VCA (residues 393–505; Figure LAR acts on p-GBD with k
⫽ 1.3 ⫻ 10
Figure 3. Active Cdc42 Increases the Rate of
Dephosphorylation of Autoinhibited WASP
In all figures, upper panel is Coomassie blue-
stained SDS-PAGE gel, lower panel is anti-
phosphotyrosine Western blot.
(A) Time course of the dephosphorylation of
5M p-GBD by 5 nM LAR phosphatase alone
(lane 1), with 15 M VCA (lane 2), with 15 M
VCA ⫹ 5 M Cdc42-GMPPCP (lane 3), and with
15 M VCA ⫹ 5 M Cdc42-GDP (lane 4).
(B) Dephosphorylation (30 min) of 5 M
p-tWASP by 5 nM LAR phosphatase alone
(lane 1), with 20 M Cdc42-GMPPCP (lane 2),
and with 20 M Cdc42-GDP (lane 3).
(C) Dephosphorylation (90 min) of 2 M
p-N-GBD-VCA by 5 nM LAR phosphatase
alone (lane 1), with 20 M Cdc42-GMPPCP
(lane 2), and with 20 M Cdc42-GDP (lane 3).
indicating catalytic efficiency only slightly below that biochemical record, or molecular memory, of previous
coincident signaling events.found with other substrates (k
⫽ 2.5 ⫺ 10 ⫻ 10
[Cho et al., 1992]). Addition of VCA decreases
activity of LAR toward p-GBD substantially (lane 2). Ad- Phosphorylation Increases the Basal Activity of WASP
Proteins toward Arp2/3 Complexdition of Cdc42-GMPPCP but not Cdc42-GDP, reverses
this inhibition and returns p-GBD to an efficient phos- We originally speculated, based on the autoinhibited
GBD-VCA structure, that phosphorylation at Y291 mightphatase substrate (lanes 3 and 4, respectively). Since
NMR data (see below) indicate that VCA peptides induce lead to constitutive activation of WASP through destabi-
lization of the autoinhibited fold (Kim et al., 2000). Tofolded structure in the phosphorylated GBD, these re-
sults indicate that the autoinhibited fold protects p-WASP test this, we examined the effects of phosphorylation
on affinity of GBD peptides for VCA peptides, and onagainst tyrosine phosphatases. These effects are also
reproduced in the tethered proteins tWASP and N-GBD- WASP protein activities in Arp2/3 complex-stimulated
pyrene-actin polymerization assays. Affinity was mea-VCA, which are dephosphorylated poorly in isolation,
but readily in the presence of Cdc42-GMPPCP (Figures sured using isothermal titration calorimetry (ITC). Figure
4A shows calorimetric titrations of the N-WASP VCA3B and 3C, respectively).
The combined data on kinase and phosphatase ac- into pGBD and GBD. Fitting of the heats evolved as a
function of molar ratio yields a K
value of 2.8 ⫾ 0.4 Mtions indicate that the phosphorylation status of WASP
at Y291 can only be efficiently altered in the presence for the GBD, and 16.0 ⫾ 0.2 M for p-GBD. Thus, while
p-GBD can bind the VCA, its affinity is decreased nearlyof activated Cdc42. This property results directly from
the structural and thermodynamic mechanisms of 5-fold compared to the nonmodified protein.
To determine the functional significance of this de-WASP regulation. These give rise to steric occlusion of
the (p)Y291 side chain in the free protein, and increased crease in affinity, we first examined the ability of GBD
peptides to block the activation of Arp2/3 complex by VCAaccessibility in the Cdc42-bound state. In the context
of signaling, following one round of GTPase and kinase peptides. Figure 4B shows polymerization of G-actin over
time (monitored by the increase in fluorescence of py-activation, these properties would allow WASP to remain
phosphorylated long after the initial stimulus has sub- rene-labeled [5%] actin). Polymerization of actin alone
is characterized by a significant lag phase and slowsided. Phospho-Y291 could thus serve as a long-lasting
Molecular Memory in WASP
Figure 4. Phosphorylation Decreases Affinity of GBD for VCA and Increases Basal Activity of WASP Constructs toward Arp2/3 Complex
(A) ITC trace of the titration of N-WASP VCA into p-GBD (left panel). Curve fits for the VCA into p-GBD and GBD titrations are shown in the
center and right panels, respectively. For p-GBD, K
⫽ 16.0 ⫾ 0.2 M, with stoichiometry of 0.97 ⫾ 0.02: 1. For GBD, K
⫽ 2.8 ⫾ 0.4 M,
with stoichiometry of 0.99 ⫾ 0.01: 1.
(B) Polymerization of 4 M G-actin (5% pyrene labeled) in the presence of 10 nM Arp2/3 complex. N-WASP VCA peptides were added to 25
nM. N-WASP GBD peptides were added to 5 M. Similar effects were observed with WASP VCA and GBD peptides.
(C) Comparison of the ability of tWASP (dashed lines) and p-tWASP (solid lines) to stimulate polymerization of 2 M G-actin (5% pyrene
labeled) by 10 nM Arp2/3. The tWASP proteins were added to 10, 50, and 250 nM (green, blue, and red, respectively). Effects of 250 nM VCA
are shown in black for comparison.
rate. These parameters are not significantly altered by In contrast, p-tWASP shows significant activity toward
Arp2/3 complex that increases in a dose-dependentaddition of isolated Arp2/3 complex, consistent with
previous reports that the highly purified assembly is manner in this same range.
The apparent discrepancy between resistance toinactive (Higgs and Pollard, 2001). Addition of 25 nM
VCA peptide causes a dramatic decrease in lag time phosphatases (Figure 3; suggesting a stable GBD-VCA
complex) and activity toward Arp2/3 complex (Figuresand an increase in overall polymerization rate. This effect
is greatly suppressed by addition of 5 M GBD, which 4B and 4C; suggesting an unstable GBD-VCA complex)
can be reconciled by our recent observation that theshould sequester 64% of the VCA, based on a 2.8 M
(excluding effects of Arp2/3 complex on the GBD/ autoinhibited domain is modular in structure and stabil-
ity (Buck et al., 2001; M. Buck et al., submitted). ThatVCA binding equilibrium). In contrast, addition of 5 M
p-GBD (24% calculated VCA sequestration) has a much is, small decreases in stability of the core helices (␣1–␣4)
of the domain result in appreciably larger decreases inweaker inhibitory effect on VCA-stimulated Arp2/3 com-
plex activity. At higher concentrations (data not shown) stability of the VCA helix, ␣5. Thus, while phosphoryla-
tion may have only modest effects on the stability of ␣3,p-GBD can block activity to a degree similar to GBD,
indicating that the effect of phosphorylation is largely enabling protection from phosphatases, it could have a
larger destabilizing effect on ␣5, enabling Arp2/3 com-on affinity for VCA, rather than intrinsic ability to block
Arp2/3 complex activity. We also examined these effects plex activation. To examine structural aspects of this
model, we determined the backbone
C chemical shiftsin tWASP. Figure 4C shows that, as previously reported,
tWASP is a poor activator of Arp2/3 complex at concen- of
C-labeled p-GBD and GBD in complex (sepa-
rately) with VCA at 25⬚C (data not shown). Values aretrations between 10 and 250 nM (Devriendt et al., 2001).
nearly identical (rmsd, C␣⫽0.22 ppm; rmsd, C␤⫽0.22 tively). The poor chemical shift dispersion and broad
lines indicate that both proteins lack persistent tertiary
ppm), indicating that helices ␣1–␣4 are preserved after
structure in isolation. In both cases, addition of excess
phosphorylation (Wishart and Sykes, 1994). Thermody-
unlabeled VCA peptide (not observed in the spectra)
namic aspects of the model are also consistent with our
causes a dramatic increase in chemical shift dispersion
observations (data not shown) that while nonphosphory-
and narrowing of lines, indicating induction of the folded
lated GBD-VCA proteins from WASP and N-WASP show
autoinhibited structure (Figures 5D and 5E). Addition of
a single, high-temperature thermal unfolding transition
excess SH2 domain to the GBD-VCA complex does not
by circular dichroism (T
⫽ 60⬚C–80⬚C, depending on
cause changes in chemical shifts or line widths of the
linker length), the phosphorylated proteins show an ad-
GBD signals (Figure 5F). However, the SH2 domain
ditional transition at approximately 35⬚C that may repre-
causes a dramatic collapse of p-GBD signals in the
sent dissociation of the VCA helix from the ␣1–␣4 core.
p-GBD-VCA spectrum, with a large decrease in chemical
In summary, the combined data show that phosphory-
shift dispersion and changes in line widths, indicating
lation of Y291/Y256 in the WASP/N-WASP GBD de-
disruption of the autoinhibited domain (Figure 5G). The
creases affinity for the VCA, and consequently increases
combined biochemical and NMR data demonstrate that
the basal activity toward Arp2/3 complex. Thus, an im-
the Src SH2 domain is capable of binding p-GBD and
portant functional outcome of coincident Cdc42 and
displacing it from VCA.
kinase signals is conversion of transient WASP/Arp2/
To determine if these effects are sufficient to activate
3 complex activation, with lifetime dependent on the
WASP toward Arp2/3 complex, we used pyrene-actin
nucleotide cycle of the GTPase, to persistent, limited-
polymerization assays. At low concentrations (1–2 M),
the Src SH2 domain has little effect on the ability of
p-GBD to inhibit VCA-mediated Arp2/3 complex activa-
Src Family SH2 Domains Can Activate p-WASP
tion (data not shown), consistent with decreased affinity
Proteins toward Arp2/3 Complex
of the SH2 domain for p-GBD when the latter is in the
The sequence surrounding Y291 is a consensus binding
autoinhibited fold. The same sequestration of p-GBD by
site for Src family SH2 domains (Yaffe et al., 2001). In
VCA prevents catalytic concentrations of phosphatases
the structures of pTyr substrates bound to SH2 domains
from acting on p-Y291 in autoinhibited WASP proteins.
of this class, the peptide backbone is in an extended
However, as shown in Figure 6A, at high concentrations
conformation, with side chains of the pTyr and I⫹3 resi-
(⬎10 M), the Src SH2 domain causes substantial acti-
dues bound into two pockets that are well separated
vation of a p-GBD-VCA complex, demonstrating that
on the protein surface (Waksman et al., 1993). In autoin-
the physical displacement observed above leads to
hibited WASP, the Y291 motif is located in helix ␣3of
functional activation. The same concentrations of SH2
the domain, with the edge of the Y291 ring exposed to
domain have no effect on the nonphosphorylated GBD.
solvent and I294 buried in the hydrophobic core of the
We also examined the effect of the SH2 domain on
domain. Incompatibility between these conformations
the tethered tWASP protein. As in the intermolecular
suggested that SH2 domain binding to p-Y291 could,
system, low concentrations of SH2 domain had no effect
analogous to Cdc42, destabilize the autoinhibited struc-
on the activity of p-tWASP, while high concentrations
ture, leading to physical separation of the GBD from the
were stimulatory (data not shown). However, the magni-
VCA and activation toward Arp2/3 complex. We note
tude of this increase was modest, and required higher
that the low affinity of phosphatases toward p-GBD
SH2 domain concentrations than with the unjoined p-GBD
⫽ 155 M for LAR, see above) would prevent cata-
and VCA peptides, due to the increased stability of the
lytic concentrations of these enzymes from analogously
autoinhibited fold imparted by the tether. The SH3 do-
destabilizing the autoinhibited domain and accessing
mains of Src family kinases have been shown to bind the
proline-rich sequences of WASP and N-WASP, which
To test this hypothesis, we first examined the ability
separate the GBD and VCA elements (Banin et al., 1996;
of SH2 domains to bind p-GBD and displace it from VCA
Scott et al., 2002). Thus, we reasoned that Src constructs
peptides. Immobilized Src SH2 domain can bind p-GBD
containing the SH3-SH2 module should be capable of
with reasonable affinity, but is unable to bind nonphos-
making bivalent interactions with p-tWASP. This multi-
phorylated GBD. As described in Experimental Proce-
valency should afford such constructs appreciably
dures, this difference formed the basis of our purification
higher affinity, and increased potency as p-tWASP acti-
of p-WASP proteins throughout this work. Preliminary
vators than the SH2 domain alone. Viewed alternatively,
isothermal titration calorimetry measurements of the af-
the SH3-proline interactions of an SH3-SH2 construct
finity of p-GBD for the Src SH2 domain indicate a K
should greatly increase the local concentration of the
approximately 2 M (data not shown). To determine if
SH2 domain toward p-Y291, enabling the latter to acti-
this interaction is incompatible with binding of p-GBD
vate. As shown in Figure 6C, these predictions were
to VCA, we used two assays. Figure 5A shows that both
correct, and an Src SH3-SH2 protein at 1 M concentra-
GBD and p-GBD can be retained by immobilized GST-
tion activated p-tWASP to levels comparable to the free
VCA. However, incubation of the resin with excess solu-
VCA. This ability required multivalent interactions with
ble SH2 domain resulted in substantial elution of p-GBD
WASP, since a nonlinked mixture of SH3 and SH2 do-
but not GBD, demonstrating that SH2 domains can de-
main proteins at the same concentration had no effect.
stabilize the WASP autoinhibitory interactions. We also
The actions of the SH3-SH2 protein are dependent on
examined this issue by NMR spectroscopy. Figure 5
phosphorylation since while tWASP is strongly activated
N HSQC spectra of
N-labeled GBD and
by Cdc42-GMPPCP, it is not affected by Src constructs
consisting of the SH2 domain, SH3 domain, or combinedp-GBD (Figures 5B, 5D, and 5F; 5C, 5E, and 5G, respec-
Molecular Memory in WASP
Figure 5. Src SH2 Domain Can Physically Disrupt the Interaction between p-GBD but Not GBD and VCA
(A) GST-VCA was immobilized on glutathione-Sepharose beads and incubated with the Src SH2 domain and either p-GBD (left panels) or
GBD (right panel). Proteins bound to the beads (lanes 1 and 3) and remaining in solution (lanes 2 and 4) were analyzed by Coomassie blue-
stained SDS-PAGE gels and anti-phosphotyrosine Western blotting.
(B and C)
N HSQC spectra of 100 M
N-labeled GBD and p-GBD, respectively.
(D and E)
N HSQC spectra of 100 M
N-labeled GBD and p-GBD in the presence of 250 M and 400 M unlabeled VCA, respectively.
(F and G)
N HSQC spectra recorded after 200 M unlabeled Src SH2 domain was added to samples from (D) and (E), respectively.
SH3-SH2 module (Figure 6B). Thus, steric incompatibil- (p)Y291 to modifying enzymes. Once phosphorylated at
Y291, WASP becomes partially activated toward Arp2/3ity between the SH2-bound and autoinhibited structures
of the p-Y291 motif results in physical displacement of complex, likely because of moderate destabilization of
the GBD-VCA contacts in the autoinhibited fold. In addi-p-GBD from VCA by SH2 domains, leading to activation
toward Arp2/3 complex. These data demonstrate that tion, since the helical conformation of the p-Y291 motif
in autoinhibited, WASP is incompatible with that in SH2-a second functional consequence of WASP phosphory-
lation at Y291 is sensitivity to a new input signal, binding peptide complexes, the autoinhibited domain of the
phosphorylated protein can also be disrupted by Srcof Src-family SH2 domains, which has no effect on the
nonphosphorylated protein. family SH2 domains, leading to activation.
These properties generate a potentially powerful cir-
cuit governing the cellular response to Cdc42 and SrcDiscussion
family kinase signals. The output of this circuit, manifest
by the level of phosphorylated WASP (p-WASP), in re-We have demonstrated here that both phosphorylation
and dephosphorylation of WASP at Y291 are dependent sponse to different GTPase and kinase inputs, assuming
constant phosphatase activity, is described schemati-on binding to activated Cdc42. The structure of the GBD-
VCA complex sterically protects the (p)Y291 side chain cally in Figure 7A. When Cdc42 or kinase is active inde-
pendently (periods a and b), WASP is not phosphory-from both kinases and phosphatases in the free state
of the protein. Binding to Cdc42 substantially destabi- lated (third trace), although it is active toward Arp2/3
complex for the duration of the GTPase signal (fourthlizes the autoinhibited domain, enabling access of
Figure 6. The Src SH2 Domain Can Activate p-WASP toward Arp2/3 Complex
(A) Polymerization of 4 M G-actin (5% pyrene labeled) in the presence of 10 nM Arp2/3 complex. GBD curves are dashed; p-GBD curves
(B and C) Polymerization of 2 M G-actin (5% pyrene labeled) in the presence of 10 nM Arp2/3 complex.
trace). When Cdc42 and kinase are active simultane- dence of GTPase and kinase signals. Moreover, the cir-
cuit can function as a memory device, maintaining aously, both in time and location (periods c and e), WASP
is rapidly phosphorylated in its GTPase complex. If the long-lasting record of this event even after the initiating
signals are terminated.kinase signal subsides prior to the Cdc42 signal, WASP
will be returned to its nonphosphorylated state by the Suetsugu et al. have reported that PIP
phosphorylation of N-WASP at Y256 by Fyn (Suetsuguaction of phosphatases, which act readily on the Cdc42-
WASP complex (period d). If, however, the Cdc42 signal et al., 2002), analogous to our findings with Cdc42. Al-
though these authors did not examine whether PIP
in-terminates prior to the kinase (period f), p-WASP will
fold to the autoinhibited, phosphatase-resistant confor- creases the intrinsic activity of Fyn, one possible expla-
nation of their data is that the phospholipid increasesmation and remain phosphorylated for potentially long
periods. During this time, partial destabilization of the accessibility of Y256 in the autoinhibited fold to kinases.
This commonality would indicate that deconstruction ofautoinhibited domain imparts increased basal activity
on the protein. The p-Y291 switch can ultimately be the autoinhibited domain is common to WASP activation
by both Cdc42 and PIP
. Thus, cooperativity betweenreset (period g) by subsequent activation of Cdc42 inde-
pendent of kinase. Decay of this second GTPase signal GTPase and phospholipid, and perhaps others other
activators as well, is likely to involve not only WASP(period h) enables WASP to return to its completely
inactive state. This circuit ultimately arises because of activity, as has been reported by several groups (Higgs
and Pollard, 2000; Prehoda et al., 2000; Rohatgi et al.,the structure and activation mechanism of WASP. These
dictate that the efficiency of WASP as a substrate for 2000), but also WASP phosphorylation. The existence
of multiple pathways that enhance phosphorylation mayboth kinases and phosphatases is dependent on binding
to activated Cdc42. Thus, the biochemical and structural explain the observation that in COS-7 cells, WASP phos-
phorylation induced by overexpression of Hck, whichproperties of WASP result in a circuit that senses coinci-
Molecular Memory in WASP
Figure 7. Dynamic Responses of WASP to
Cdc42 and Src Family Kinase Signals
(A) Phosphorylation status and activity of
WASP/N-WASP in response to Cdc42 and
(B) Dynamic features of feedforward loop be-
tween Src family kinases and WASP. Top
panel shows initial response of the system,
where GTPase and kinase must both be acti-
vated to achieve WASP phosphorylation.
Open arrowheads indicate indirect pathway
from kinase to WASP phosphorylation, filled
arrowhead indicates direct pathway. Inde-
pendent GAP activation (red arrow) results in
WASP phosphorylation that persists follow-
ing decay of the kinase signal. Bottom panel
shows response when WASP is initially phos-
phorylated by a previous round of coincident
GTPase and kinase signals.
activates diverse pathways, is not strongly affected by sis and T cell activation, strong evidence exists for colo-
calization of kinase, Cdc42, and WASP/N-WASP (Can-coexpression of dominant-negative or constitutively ac-
tive Cdc42 mutants. non et al., 2001; Castellano et al., 1999; Coppolino et
al., 2001; Greenberg and Grinstein, 2002; Krause et al.,Pathways connecting Src family kinases to Cdc42 to
downstream machineries have been described in a num- 2000; Stinchcombe et al., 2001). Thus, the biochemistry
we have described indicates it is highly likely that WASPber of biological processes. In HeLa cells and macro-
phages, constitutively activated Hck signals through will be locally phosphorylated on Y291 during these pro-
cesses. The biochemical interactions we have de-Cdc42 and Rac to stimulate the formation of actin-rich
membrane protrusions (Carreno et al., 2002). Macro- scribed, when present in these biological contexts,
would thus create the coherent feedforward circuitphage phagocytosis mediated by the Fc␥ receptor
(Fc␥R) appears to use an analogous pathway (Green- shown in Figure 7B. The indirect arm of the circuit be-
tween the Src family kinase and WASP (open arrow-berg and Grinstein, 2002). In this system, receptor stimu-
lation leads to activation of Hck and Lyn (Carreno et al., heads) would be delayed relative to the direct arm (filled
arrowheads) due to additional intermediates between2002; Kedzierska et al., 2001; Suzuki et al., 2000), and
recruitment (and presumably activation) of Cdc42 kinase and Cdc42 GEF, and because Cdc42 must disso-
ciate from RhoGDI and migrate to membranes during(Caron and Hall, 1998), WASP/N-WASP (Castellano et
al., 2001; Lorenzi et al., 2000; May et al., 2000), and activation (Olofsson, 1999). In addition, phosphorylation
of Vav at multiple sites appears to be necessary toArp2/3 complex (May et al., 2000) to the phagocytic cup,
providing actin rearrangements necessary for efficient achieve high-level activation of GEF activity (Bustelo,
2000), a feature that could give rise to hypersensitivityphagocytosis. Cdc42 activation at this site may proceed
through Hck-mediated phosphorylation and activation (Ferrell, 1999). Analogous coherent feedforward circuits
are highly abundant in both prokaryotic and eukaryoticof a guanine nucleotide exchange factor (GEF) in the
Vav family (Coppolino et al., 2001; English et al., 1997), transcriptional regulatory networks, as well as in the
neuronal synaptic connectivity network in C. eleganswhose members have been shown to act on Cdc42 in
vivo and in vitro (Abe et al., 2000). A similar pathway (Milo et al., 2002; Shen-Orr et al., 2002). Mathematical
modeling and experimental investigations have demon-has been delineated during T cell activation, where stim-
ulation of the T cell antigen receptor (TCR) leads to strated that such circuits act to reject short duration/
low amplitude input stimuli, producing positive outputactivation of the tyrosine kinases Lck, Fyn, and ZAP-
70, which phosphorylate Vav1 and SLP-76, leading to (i.e., WASP phosphorylation in the present case) only
upon persistent, strong stimulation (Shen-Orr et al.,localized activation of Cdc42, WASP, and Arp2/3 com-
plex (Cannon and Burkhardt, 2002; Cannon et al., 2001; 2002). These circuits also produce rapid decay of the
signal following input termination (Shen-Orr et al., 2002).Dustin and Cooper, 2000; Krause et al., 2000; Krawczyk
et al., 2002). These interactions appear to be critical to Although as described above, if the Cdc42 signal is
terminated prior to kinase, for example by independentthe polymerization and dynamics of actin during forma-
tion of the immunological synapse (Dustin and Cooper, activation of a GTPase-activating protein (GAP; red
arrow in Figure 7B), the memory function of the circuit2000). Finally, PDGF stimulation of NIH 3T3 fibroblasts
causes activation of Src, which phosphorylates Vav, is activated and WASP can remain phosphorylated for
long periods. The analysis indicates that the biochemicalleading to activation of Cdc42 and Rac, ultimately in-
creasing expression of the c-myc transcription factor interactions we have described could be used during
Cdc42/actin-dependent processes such as macro-(Chiariello et al., 2001). In both macrophage phagocyto-
phage phagocytosis or T cell activation, to reject low- a mechanism for potentiating the system toward ki-
nases. When WASP is not phosphorylated, high levelslevel or transient kinase signals, producing p-WASP only
in response to long-lasting, high-level stimulation. of kinase are necessary to initiate actin rearrangements
since the signal must pass through GEF and Cdc42Mechanisms leading to activation of most GEFs for
Rho GTPases, including those specific for Cdc42, are before reaching WASP (see above). In contrast, once
WASP is phosphorylated by an initial intense signal,not yet known. Many will certainly involve pathways that
are independent of Src family kinases. In these cases, actin polymerization can be initiated later by only a weak
stimulation of kinase pathways, since the SH3/SH2 mod-p-WASP could also function as the output of circuits
that detect the simultaneous activation of independent ule of the kinase can activate p-WASP directly without
the lags inherent in the GTPase pathway. Alternatively, ifCdc42 and kinase pathways. Experimental testing of
this idea must await a better understanding of signaling the phosphorylating and activating (i.e., SH2-containing)
kinase are distinct, modification of WASP could providemechanisms that control Cdc42.
We have shown that the functional consequences of a mechanism for actin rearrangements in response to
an entirely new stimulus. Thus, phosphorylation ofWASP phosphorylation at Y291 are 2-fold. First, p-WASP
has significantly higher basal activity than WASP. Sec- WASP can serve as a general mechanism to sensitize
a system to actin rearrangements in response to Srcond, phosphorylation enables WASP to respond to a
new signal, SH2 domain binding. Potential implications family kinase pathways. This behavior could allow the
cell to be maintained in an inert state, relatively resistantof these effects will be discussed separately, although
both could be important in particular systems. In sys- to stimuli, but converted to a responsive state by an
initial, committing, high-intensity burst of input signal.tems/processes where the enhanced basal activity of
p-WASP is sufficient to mediate signaling, phosphoryla- The biochemical process we have discovered for
WASP provides a mechanism for cells to retain informa-tion would effectively convert transient WASP activa-
tion, with a lifetime governed by the GTPase nucleotide tion for time periods of intermediate duration: longer
than the rapid rise and decay of many signaling events,cycle, to persistent activation. In a biological context,
this could serve either to impart new functionality to but likely shorter than times characteristic of large-scale
changes in gene expression observed in processes suchWASP signals, or to extend the lifetime of existing func-
tions. In the former case, the requirement of Cdc42/ as differentiation. Analogous hysteresis can be gener-
ated on a cellular basis by feedback elements withinkinase coincidence would enhance the specificity of
signals leading to the new functionality. The requirement multicomponent signaling networks, as demonstrated
both theoretically and experimentally (Bhalla et al., 2002;of N-WASP phosphorylation in neurite outgrowth could
be an example of such behavior (Suetsugu et al., 2002). Ferrell, 2002). The information storage described here
is achieved by a distinct mechanism, which is based onIn the latter case, the biochemistry we have described
could provide a mechanism for the cell to sample a the biophysical construction of WASP. The behavior of
WASP, and its ultimate causes are in many respectsWASP-dependent processes transiently, under control
of the GTPase clock, and then commit to that process similar to those of Ca
kinase II (CamKII), an extensively studied protein postu-when appropriate, independent kinase signals are re-
ceived. Interestingly, Wu
lfing and colleagues (C. Wu
lf- lated to function as a molecular memory device in neu-
rons (Hudmon and Schulman, 2002; Lisman et al., 2002).ing, personal communication) have recently demon-
strated that stimulation of primary T cells with limiting WASP and CamKII, both store information through long-
lasting phosphorylation events. In each case, phosphor-antigen (presented by antigen presenting cells, APCs)
in the presence of costimulatory blockade (antibodies ylation requires an initial conformational change to
transform the protein into an efficient kinase substrate.to the CD28 and LFA-1 ligands, ICAM-1 and B7) results
in only transient actin accumulation at the T cell-APC In WASP, binding of Cdc42 to the GBD destabilizes the
autoinhibited domain, enabling modification of Y291. Ininterface. In contrast, sustained actin accumulation,
which is necessary for accumulation and organization CamKII, which forms a dimer of hexameric rings, binding
/calmodulin to one kinase subunit reveals its siteof the signaling and adhesion receptors at the immuno-
logical synapse, requires high levels of antigen and co- of autophosphorylation (T286), enabling modification by
a neighboring subunit (Hudmon and Schulman, 2002). Instimulation of CD28. The biochemistry we have de-
scribed could, in principle, provide circuitry that both proteins, mechanisms also exist for the phosphate-
mediated activity to be retained for long periods. Incontributes to such behaviors.
The ability of Src family SH2 domains to bind and WASP this occurs through inherent protection of p-Y291
from phosphatases. In CamKII, slow dissociation of cal-activate p-WASP could also have several functional con-
sequences. First, it could extend the lifetime of the actin modulin from phosphorylated subunits (Schulman et al.,
1992), coupled with the proposed ability of one activatedpolymerization signal. That is, since Src proteins both
phosphorylate and bind WASP, it seems likely that after subunit to phosphorylate its neighbor in the presence
of basal Ca
levels (Lisman et al., 2002; Lisman andphosphorylation, the kinase will remain bound to
p-WASP through bivalent interactions of its SH2 and Zhabotinsky, 2001; Zhabotinsky, 2000), both provide
mechanisms to prolong activity and duration of phos-SH3 domains. This would lead to high-level activation
of WASP in the absence of Cdc42. Second, SH2 domain phorylation. Finally, phosphorylation can potentiate the
activation of both WASP and CamKII by later stimuli bybinding would also extend the lifetime of the p-Y291
itself, since the phosphate group would be protected lowering the threshold levels of Src family kinase (see
above) and Ca
signals, respectively, needed for activa-from phosphatase both when bound to SH2 and when
free. Third, in the context of the circuit in Figure 7B, tion (De Koninck and Schulman, 1998; Hanson et al.,
1994; Meyer et al., 1992; Shen et al., 2000). In bothinteractions of p-WASP with SH2 domains could provide
Molecular Memory in WASP
proteins, autoinhibitory regulation is central to these
Experiments were performed in an Omega titration calorimeter (Mi-
biochemical functions (Hudmon and Schulman, 2002;
croCal, Inc). Titrations were performed by injecting 25 ⫻ 10 lof
Kim et al., 2000). Large structural rearrangements, in-
1.6 mM VCA at 3 min intervals into a 1.4 ml solution of 100 M GBD
volving domain separation and extensive exposure of
or p-GBD at 25⬚C in 25 mM Tris-HCl (pH 7.5), 100 mM NaCl, 2 mM
buried surface area, are inherent in such designs. These
EDTA, and 2 mM DTT.
features make autoinhibited proteins ideally suited to
Actin Polymerization Assays
integrate multiple signals through contingent covalent
Detailed procedures are described in Higgs et al. (1999). Reactions
modification. In fact, such behavior has now been ob-
contained 2 or 4 M Mg-ATP actin (5% pyrene labeled), 10 mM
served directly and indirectly in other autoregulated
imidazole (pH 7.0), 50 mM KCl, 1 mM EGTA, and 1 mM MgCl
GTPase effectors, including the Pak and Raf kinases
buffer-Mg (2 mM Tris-HCl [pH 8.0], 0.5 mM DTT, 0.1 mM ATP, 0.1
(Renkema et al., 2002; Stokoe et al., 1994). Our studies
). Pyrene fluorescence (
⫽ 365 nm,
⫽ 407 nm) was
with WASP, and the large body of biochemical work on
monitored with a Fluorolog-3 spectrafluorimeter (JY Horiba). Fluo-
rescence was normalized to the final value observed at steady state
CamKII, show that when autoregulation appears in an
within each experiment.
appropriate structural and thermodynamic context, it
can also lead to mechanisms of biochemical memory. In
CamKII, this functionality is thought to make important,
GST-VCA was immobilized on glutathione Sepharose beads at a
perhaps defining, contributions to induction and mainte-
concentration of 5 mg/ml resin. Beads (20 l) were incubated 10
nance of long-term potentiation, a neuronal property
min at 25⬚C with 100 lof100M GBD or p-GBD, plus 20 lof170
that plays important roles in learning and memory (Lis-
M Src SH2 domain in PBS buffer. Supernatant was removed and
beads were washed three times with 120 l of PBS. Proteins bound
man et al., 2002). However, in other systems, including
to beads were eluted with SDS sample buffer. Supernatant and
WASP, the biological uses of such circuits remain largely
bound samples were analyzed by SDS-PAGE.
speculative. As our understanding of these potentially
powerful biochemistries develops, discovery of how
their functionalities control specific biological processes
NMR experiments were carried at 25⬚C on a Varian Inova 600 MHz
will represent a critical avenue for future research.
spectrometer using a gradient and sensitivity enhanced
HSQC pulse sequence (Kay et al., 1992). All protein samples were
in 25 mM phosphate (pH 7.0), 100 mM NaCl, 2 mM EDTA, 2 mM
DTT, and 10% D
Human N-WASP GBD amino acids 192–275, human WASP GBD
amino acids 230–310, Src SH2 amino acids 144–249, and human
We thank Dr. Annette Kim for first recognizing the potential impor-
Cdc42 amino acids 1–179 were cloned into pET11a. N-WASP VCA
tance of WASP phosphorylation at Y291. We thank Drs. Harry Higgs,
amino acids 393–505, WASP VCA amino acids 420–502, Src SH3
Don Kaiser, and Tom Pollard for advice regarding purification of
amino acids 83–144, and Src SH3SH2 amino acids 83–249 were
bovine Arp2/3 complex and actin; Dr. Gaya Amarasinghe for Lck
cloned into pGEX-2T. Truncated WASP (tWASP), amino acids 230–
expression in Sf9 cells; Ivan Haller and Bikash Pramanik for per-
502, was cloned into pET16b. N-WASP GBD-VCA (N-GBD-VCA),
forming mass spectrometry analyses on phosphorylated WASP con-
amino acids 192–275 linked to the VCA region by a (GGS)
structs; Dr. William Lowry for performing preliminary WASP phos-
was cloned into pET11a. All proteins were expressed in E. coli strain
phorylation experiments; Drs. Rama Ranganathan, Uri Alon, and Jan
BL21(DE3) except for tWASP, which was expressed in BL21-Co-
Burkhardt for exciting discussions regarding biochemical circuits
donPlus (Stratagene). Lck tyrosine kinase was overexpressed in Sf9
and lymphocyte signaling; Dr. Yuh Min Chook for critical reading of
insect cells. Expression and purification protocols for all proteins
the manuscript; and Dr. Christoph Wu
lfing for communicating re-
are available from the authors. Cdc42 was loaded with GMPPCP
sults prior to publication. This work was supported by grants from
as described (Abdul-Manan et al., 1999). The Abl kinase domain,
the NIH (RO1 GM56322) and the Welch Foundation (Chemical Re-
and the LAR and YOP phosphatases were purchased from NEB.
search Grant #I1544). Preliminary studies were performed at the
LynA was purchased from Biomol.
Howard Hughes Medical Institute at the Memorial Sloan-Kettering
Bovine Arp2/3 complex was purified from calf thymus as de-
scribed (Higgs et al., 1999). Actin was purified from rabbit muscle
and labeled with pyrene as described (Pollard and Cooper, 1984;
Received: November 22, 2002
Spudich and Watt, 1971).
Revised: February 7, 2003
N-WASP GBD, WASP GBD, N-GBD-VCA, and tWASP were phos-
Accepted: March 6, 2003
phorylated on preparative scale by Lck or Abl kinase (New England
Published: May 22, 2003
BioLabs) in the presence of a 5-fold excess of Cdc42-GMPPCP.
Phosphorylation was monitored by anti-p-Tyr Western blotting.
Phosphoproteins were purified on a GST-Src SH2 domain affinity
column eluted with 10 mM phospho-tyrosine in 50 mM Tris-HCl, pH
Abdul-Manan, N., Aghazadeh, B., Liu, G.A., Majumdar, A., Ouerfelli,
7.0. Single-site phosphorylation of p-GBD of WASP and N-WASP,
O., Siminovitch, K.A., and Rosen, M.K. (1999). Structure of Cdc42
p-N-GBD-VCA, and p-tWASP was confirmed by mass spectrometry.
in complex with the GTPase binding domain of the “Wiskott-Aldrich
Each construct contains only a single tyrosine, Y291/Y256 for
syndrome” protein. Nature 399, 379–383.
Abe, K., Rossman, K.L., Liu, B., Ritola, K.D., Chiang, D., Campbell,
S.L., Burridge, K., and Der, C.J. (2000). Vav2 is an activator of Cdc42,
Rac1, and RhoA. J. Biol. Chem. 275, 10141–10149.
Reactions were performed in 50 mM Tris-HCl (pH 8.0), 50 mM NaCl,
Baba, Y., Nonoyama, S., Matsushita, M., Yamadori, T., Hashimoto,
2 mM MgCl
, 2 mM DTT, 5 mM ATP, and 0.01% Brij 35 at 30⬚C.
S., Imai, K., Arai, S., Kunikata, T., Kurimoto, M., Kurosaki, T., et al.
(1999). Involvement of Wiskott-Aldrich syndrome protein in B-cell
cytoplasmic tyrosine kinase pathway. Blood 93, 2003–2012.
Reactions were carried out at room temperature in 25 mM Tris-HCl, Banin, S., Truong, O., Katz, D.R., Waterfield, M.D., Brickell, P.M.,
and Gout, I. (1996). Wiskott-Aldrich syndrome protein (WASp) is a2 mM MgCl
, 100 mM NaCl, and 2 mM DTT.
binding partner for c-Src family protein-tyrosine kinases. Curr. Biol. and the actin cytoskeleton: molecular hardware for T cell signaling.
Nat. Immunol. 1, 23–29.6, 981–988.
Eden, S., and Kirschner, M.W. (2002). Inhibition of Scar/WAVE pro-
Bhalla, U.S., Ram, P.T., and Iyengar, R. (2002). MAP kinase phospha-
teins by the Pir121/NAP-1 dimer is relieved by Rac. Nature, in press.
tase as a locus of flexibility in a mitogen-activated protein kinase
signaling network. Science 297, 1018–1023.
Eden, S., Rohatgi, R., Podtelejnikov, A.V., Mann, M., and Kirschner,
M.W. (2002). Mechanism of regulation of WAVE1-induced actin nu-
Borisy, G.G., and Svitkina, T.M. (2000). Actin machinery: pushing
cleation by Rac1 and Nck. Nature 418, 790–793.
the envelope. Curr. Opin. Cell Biol. 12, 104–112.
English, B.K., Orlicek, S.L., Mei, Z., and Meals, E.A. (1997). Bacterial
Buck, M., Xu, W., and Rosen, M.K. (2001). Global disruption of the
LPS and IFN-gamma trigger the tyrosine phosphorylation of vav in
WASP autoinhibited fold on Cdc42 binding. Ligand displacement
macrophages: evidence for involvement of the hck tyrosine kinase.
as a novel method to monitor amide hydrogen exchange. Biochem-
J. Leukoc. Biol. 62, 859–864.
istry 40, 14115–14122.
Etienne-Manneville, S., and Hall, A. (2002). Rho GTPases in cell
Bustelo, X.R. (2000). Regulatory and signaling properties of the Vav
biology. Nature 420, 629–635.
family. Mol. Cell. Biol. 20, 1461–1477.
Ferrell, J.E., Jr. (1999). Building a cellular switch: more lessons from
Cannon, J.L., and Burkhardt, J.K. (2002). The regulation of actin
a good egg. Bioessays 21, 866–870.
remodeling during T cell-APC conjugate formation. Immunol. Rev.,
Ferrell, J.E., Jr. (2002). Self-perpetuating states in signal transduc-
tion: positive feedback, double-negative feedback and bistability.
Cannon, J.L., Labno, C.M., Bosco, G., Seth, A., McGavin, M.H.,
Curr. Opin. Cell Biol. 14, 140–148.
Siminovitch, K.A., Rosen, M.K., and Burkhardt, J.K. (2001). Wasp
Greenberg, S., and Grinstein, S. (2002). Phagocytosis and innate
recruitment to the T cell:APC contact site occurs independently of
immunity. Curr. Opin. Immunol. 14, 136–145.
Cdc42 activation. Immunity 15, 249–259.
Guinamard, R., Aspenstrom, P., Fougereau, M., Chavrier, P., and
Carlier, M.F., Nioche, P., Broutin-L’Hermite, I., Boujemaa, R., Le
Guillemot, J. (1998). Tyrosine phosphorylation of the Wiskott-Aldrich
Clainche, C., Egile, C., Garbay, C., Ducruix, A., Sansonetti, P., and
syndrome protein by Lyn and Btk regulated by CDC42. FEBS Lett.
Pantaloni, D. (2000). GRB2 links signaling to actin assembly by en-
hancing interaction of neural Wiskott-Aldrich syndrome protein
(N-WASp) with actin-related protein (ARP2/3) complex. J. Biol.
Hanson, P.I., Meyer, T., Stryer, L., and Schulman, H. (1994). Dual
Chem. 275, 21946–21952.
role of calmodulin in autophosphorylation of multifunctional CaM
kinase may underlie decoding of calcium signals. Neuron 12,
Caron, E. (2002). Regulation of Wiskott-Aldrich syndrome protein
and related molecules. Curr. Opin. Cell Biol. 14, 82–87.
Higgs, H.N., and Pollard, T.D. (2000). Activation by Cdc42 and PIP(2)
Caron, E., and Hall, A. (1998). Identification of two distinct mecha-
of Wiskott-Aldrich syndrome protein (WASp) stimulates actin nucle-
nisms of phagocytosis controlled by different Rho GTPases. Science
ation by Arp2/3 complex. J. Cell Biol. 150, 1311–1320.
Higgs, H.N., and Pollard, T.D. (2001). Regulation of actin filament
Carreno, S., Caron, E., Cougoule, C., Emorine, L.J., and Maridon-
network formation through Arp2/3 complex: activation by a diverse
neau-Parini, I. (2002). p59Hck isoform induces F-actin reorganization
array of proteins. Annu. Rev. Biochem. 70, 649–676.
to form protrusions of the plasma membrane in a Cdc42- and Rac-
Higgs, H.N., Blanchoin, L., and Pollard, T.D. (1999). Influence of the
dependent manner. J. Biol. Chem. 277, 21007–21016.
C terminus of Wiskott-Aldrich syndrome protein (WASp) and the
Castellano, F., Montcourrier, P., Guillemot, J.C., Gouin, E., Ma-
Arp2/3 complex on actin polymerization. Biochemistry 38, 15212–
chesky, L., Cossart, P., and Chavrier, P. (1999). Inducible recruitment
of Cdc42 or WASP to a cell-surface receptor triggers actin polymer-
Hudmon, A., and Schulman, H. (2002). Neuronal Ca2⫹/calmodulin-
ization and filopodium formation. Curr. Biol. 9, 351–360.
dependent protein kinase II: the role of structure and autoregulation
Castellano, F., Le Clainche, C., Patin, D., Carlier, M.F., and Chavrier,
in cellular function. Annu. Rev. Biochem. 71, 473–510.
P. (2001). A WASp-VASP complex regulates actin polymerization at
Kay, L.E., Keifer, P., and Saarinen, T. (1992). Pure absorption gradi-
the plasma membrane. EMBO J. 20, 5603–5614.
ent enhanced heteronuclear single quantum correlation spectros-
Chiariello, M., Marinissen, M.J., and Gutkind, J.S. (2001). Regulation
copy with improved sensitivity. J. Am. Chem. Soc. 114, 10663–
of c-myc expression by PDGF through Rho GTPases. Nat. Cell Biol.
Kedzierska, K., Vardaxis, N.J., Jaworowski, A., and Crowe, S.M.
Cho, H., Ramer, S.E., Itoh, M., Kitas, E., Bannwarth, W., Burn, P.,
(2001). FcgammaR-mediated phagocytosis by human macrophages
Saito, H., and Walsh, C.T. (1992). Catalytic domains of the LAR and
involves Hck, Syk, and Pyk2 and is augmented by GM-CSF. J. Leu-
CD45 protein tyrosine phosphatases from Escherichia coli expres-
koc. Biol. 70, 322–328.
sion systems: purification and characterization for specificity and
Kim, A.S., Kakalis, L.T., Abdul-Manan, N., Liu, G.A., and Rosen, M.K.
mechanism. Biochemistry 31, 133–138.
(2000). Autoinhibition and activation mechanisms of the Wiskott-
Coppolino, M.G., Krause, M., Hagendorff, P., Monner, D.A., Trimble,
Aldrich syndrome protein. Nature 404, 151–158.
W., Grinstein, S., Wehland, J., and Sechi, A.S. (2001). Evidence for
Krause, M., Sechi, A.S., Konradt, M., Monner, D., Gertler, F.B., and
a molecular complex consisting of Fyb/SLAP, SLP-76, Nck, VASP
Wehland, J. (2000). Fyn-binding protein (Fyb)/SLP-76-associated
and WASP that links the actin cytoskeleton to Fcgamma receptor
protein (SLAP), Ena/vasodilator-stimulated phosphoprotein (VASP)
signalling during phagocytosis. J. Cell Sci. 114, 4307–4318.
proteins and the Arp2/3 complex link T cell receptor (TCR) signaling
Cory, G.O., Garg, R., Cramer, R., and Ridley, A.J. (2002). Phosphory-
to the actin cytoskeleton. J. Cell Biol. 149, 181–194.
lation of tyrosine 291 enhances the ability of WASp to stimulate actin
Krawczyk, C., Oliveira-dos-Santos, A., Sasaki, T., Griffiths, E.,
polymerization and filopodium formation. Wiskott-Aldrich Syndrome
Ohashi, P.S., Snapper, S., Alt, F., and Penninger, J.M. (2002). Vav1
protein. J. Biol. Chem. 277, 45115–45121.
controls integrin clustering and MHC/peptide-specific cell adhesion
De Koninck, P., and Schulman, H. (1998). Sensitivity of CaM kinase
to antigen-presenting cells. Immunity 16, 331–343.
II to the frequency of Ca2⫹ oscillations. Science 279, 227–230.
Lisman, J.E., and Zhabotinsky, A.M. (2001). A model of synaptic
Devriendt, K., Kim, A.S., Mathijs, G., Frints, S.G., Schwartz, M., Van
memory: a CaMKII/PP1 switch that potentiates transmission by or-
Den Oord, J.J., Verhoef, G.E., Boogaerts, M.A., Fryns, J.P., You, D.,
ganizing an AMPA receptor anchoring assembly. Neuron 31,
et al. (2001). Constitutively activating mutation in WASP causes
X-linked severe congenital neutropenia. Nat. Genet. 27, 313–317.
Lisman, J., Schulman, H., and Cline, H. (2002). The molecular basis
Dixon, J.E. (1995). Structure and catalytic properties of protein tyro-
of CaMKII function in synaptic and behavioural memory. Nat. Rev.
sine phosphatases. Ann. N Y Acad. Sci. 766, 18–22.
Neurosci. 3, 175–190.
Lorenzi, R., Brickell, P.M., Katz, D.R., Kinnon, C., and Thrasher, A.J.Dustin, M.L., and Cooper, J.A. (2000). The immunological synapse
Molecular Memory in WASP
(2000). Wiskott-Aldrich syndrome protein is necessary for efficient Spudich, J.A., and Watt, S. (1971). The regulation of rabbit skeletal
muscle contraction. I. Biochemical studies of the interaction of theIgG-mediated phagocytosis. Blood 95, 2943–2946.
tropomyosin-troponin complex with actin and the proteolytic frag-
May, R.C., Caron, E., Hall, A., and Machesky, L.M. (2000). Involve-
ments of myosin. J. Biol. Chem. 246, 4866–4871.
ment of the Arp2/3 complex in phagocytosis mediated by Fcgam-
maR or CR3. Nat. Cell Biol. 2, 246–248. Stinchcombe, J.C., Bossi, G., Booth, S., and Griffiths, G.M. (2001).
The immunological synapse of CTL contains a secretory domain
Meyer, T., Hanson, P.I., Stryer, L., and Schulman, H. (1992). Calmod-
and membrane bridges. Immunity 15, 751–761.
ulin trapping by calcium-calmodulin-dependent protein kinase. Sci-
ence 256, 1199–1202. Stokoe, D., Macdonald, S.G., Cadwallader, K., Symons, M., and
Hancock, J.F. (1994). Activation of Raf as a result of recruitment to
Miki, H., Sasaki, T., Takai, Y., and Takenawa, T. (1998). Induction
the plasma membrane. Science 264, 1463–1467.
of filopodium formation by a WASP-related actin-depolymerizing
protein N-WASP. Nature 391, 93–96. Suetsugu, S., Hattori, M., Miki, H., Tezuka, T., Yamamoto, T., Miko-
shiba, K., and Takenawa, T. (2002). Sustained activation of N-WASP
Milo, R., Shen-Orr, S., Itzkovitz, S., Kashtan, N., Chklovskii, D., and
through phosphorylation is essential for neurite extension. Dev. Cell
Alon, U. (2002). Network motifs: simple building blocks of complex
networks. Science 298, 824–827.
Suzuki, T., Kono, H., Hirose, N., Okada, M., Yamamoto, T., Yama-
Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and
moto, K., and Honda, Z. (2000). Differential involvement of Src family
association: insights from the interfacial and thermodynamic prop-
kinases in Fc gamma receptor-mediated phagocytosis. J. Immunol.
erties of hydrocarbons. Proteins 11, 281–296.
Nobes, C.D., and Hall, A. (1995). Rho, rac, and cdc42 GTPases
Waksman, G., Shoelson, S.E., Pant, N., Cowburn, D., and Kuriyan,
regulate the assembly of multimolecular focal complexes associated
J. (1993). Binding of a high affinity phosphotyrosyl peptide to the
with actin stress fibers, lamellipodia, and filopodia. Cell 81, 53–62.
Src SH2 domain: crystal structures of the complexed and peptide-
Oda, A., Ochs, H.D., Druker, B.J., Ozaki, K., Watanabe, C., Handa,
free forms. Cell 72, 779–790.
M., Miyakawa, Y., and Ikeda, Y. (1998). Collagen induces tyrosine
Wishart, D.S., and Sykes, B.D. (1994). The
C chemical-shift index:
phosphorylation of Wiskott-Aldrich syndrome protein in human
a simple method for the identification of protein secondary structure
platelets. Blood 92, 1852–1858.
C chemical-shift data. J. Biomol. NMR 4, 171–180.
Okabe, S., Fukuda, S., and Broxmeyer, H.E. (2002). Activation of
Yaffe, M.B., Leparc, G.G., Lai, J., Obata, T., Volinia, S., and Cantley,
Wiskott-Aldrich syndrome protein and its association with other
L.C. (2001). A motif-based profile scanning approach for genome-
proteins by stromal cell-derived factor-1alpha is associated with
wide prediction of signaling pathways. Nat. Biotechnol. 19, 348–353.
cell migration in a T-lymphocyte line. Exp. Hematol. 30, 761–766.
Zhabotinsky, A.M. (2000). Bistability in the Ca(2⫹)/calmodulin-
Olofsson, B. (1999). Rho guanine dissociation inhibitors: pivotal mol-
dependent protein kinase-phosphatase system. Biophys. J. 79,
ecules in cellular signalling. Cell. Signal. 11, 545–554.
Pantaloni, D., Le Clainche, C., and Carlier, M.F. (2001). Mechanism
of actin-based motility. Science 292, 1502–1506.
Pollard, T.D., and Cooper, J.A. (1984). Quantitative analysis of the
effect of Acanthamoeba profilin on actin filament nucleation and
elongation. Biochemistry 23, 6631–6641.
Prehoda, K.E., Scott, J.A., Mullins, R.D., and Lim, W.A. (2000). Inte-
gration of multiple signals through cooperative regulation of the
N-WASP-Arp2/3 complex. Science 290, 801–806.
Renkema, G.H., Pulkkinen, K., and Saksela, K. (2002). Cdc42/Rac1-
mediated activation primes PAK2 for superactivation by tyrosine
phosphorylation. Mol. Cell. Biol. 22, 6719–6725.
Rohatgi, R., Ma, L., Miki, H., Lopez, M., Kirchhausen, T., Takenawa,
T., and Kirschner, M.W. (1999). The interaction between N-WASP
and the Arp2/3 complex links Cdc42-dependent signals to actin
assembly. Cell 97, 221–231.
Rohatgi, R., Ho, H.Y., and Kirschner, M.W. (2000). Mechanism of
N-WASP activation by CDC42 and phosphatidylinositol 4, 5- bispho-
sphate. J. Cell Biol. 150, 1299–1310.
Rudolph, M., Bayer, P., Abo, A., Kuhlmann, J., Vetter, I., and Wittin-
ghofer, A. (1998). The Cdc42/Rac interactive binding region motif
of the Wiskott Aldrich syndrome protein (WASP) is necessary but
not sufficient for tight binding to Cdc42 and structure formation. J.
Biol. Chem. 273, 18067–18076.
Schulman, H., Hanson, P.I., and Meyer, T. (1992). Decoding calcium
signals by multifunctional CaM kinase. Cell Calcium 13, 401–411.
Scott, M.P., Zappacosta, F., Kim, E.Y., Annan, R.S., and Miller, W.T.
(2002). Identification of novel SH3 domain ligands for the Src family
kinase Hck. Wiskott-Aldrich syndrome protein (WASP), WASP-inter-
acting protein (WIP), and ELMO1. J. Biol. Chem. 277, 28238–28246.
Shen, K., Teruel, M.N., Connor, J.H., Shenolikar, S., and Meyer, T.
(2000). Molecular memory by reversible translocation of calcium/
calmodulin-dependent protein kinase II. Nat. Neurosci. 3, 881–886.
Shen-Orr, S.S., Milo, R., Mangan, S., and Alon, U. (2002). Network
motifs in the transcriptional regulation network of Escherichia coli.
Nat. Genet. 31, 64–68.
Songyang, Z., Carraway, K.L., Eck, M.J., Harrison, S.C., Feldman,
R.A., Mohammadi, M., Schlessinger, J., Hubbard, S.R., Smith, D.P.,
Eng, C., et al. (1995). Catalytic specificity of protein-tyrosine kinases
is critical for selective signalling. Nature 373, 536–539.