Molecular Cell, Vol. 20, 21–32, October 7, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.molcel.2005.08.020
A Membrane Binding Domain in the Ste5 Scaffold
Synergizes with G?? Binding to Control
Localization and Signaling in Pheromone Response
Matthew J. Winters,1Rachel E. Lamson,1
Hideki Nakanishi,2Aaron M. Neiman,2
and Peter M. Pryciak1,*
1Department of Molecular Genetics and Microbiology
University of Massachusetts Medical School
Worcester, Massachusetts 01605
2Department of Biochemistry and Cell Biology
The State University of New York at Stony Brook
Stony Brook, New York 11794
Activation of mitogen-activated protein (MAP) kinase
cascade signaling by yeast mating pheromones in-
volves recruitment of the Ste5 scaffold protein to the
plasma membrane by the receptor-activated G?? di-
mer. Here, we identify a putative amphipathic ?-heli-
cal domain in Ste5 that binds directly to phospholipid
membranes and is required for membrane recruitment
by G??. Thus, Ste5 signaling requires synergistic Ste5-
G?? and Ste5-membrane interactions, with neither alone
being sufficient. Remarkably, the Ste5 membrane bind-
ing domain is a dual-function motif that also mediates
nuclear import. Separation-of-function mutations show
that signaling requires the membrane-targeting activity
of this domain, not its nuclear-targeting activity, and
heterologous lipid binding domains can substitute for
its function. This domain also contains imperfections
that reduce membrane affinity, and their elimination
results in constitutive signaling, explaining some pre-
vious hyperactive Ste5 mutants. Therefore, weak mem-
brane affinity is advantageous, ensuring a normal level
of signaling quiescence in the absence of stimulus and
imposing a requirement for G?? binding.
Eukaryotic cells respond to a wide variety of external
signals that influence decisions regarding division,
gene expression, and behavior. Cellular responses are
often initiated at the plasma membrane when stim-
ulated receptors create membrane binding sites for cy-
toplasmic proteins, leading to their activation. The reg-
ulated recruitment of proteins to the plasma membrane
can involve interactions with either membrane proteins
or membrane lipids (Pawson and Scott, 1997; Hunter,
2000; Teruel and Meyer, 2000). Subsequent disassem-
bly of these membrane-based signaling complexes can
allow for dissemination of the signal to other parts of
the cell or termination of signaling.
The mating pathway of the yeast Saccharomyces
cerevisiae (Dohlman and Thorner, 2001) provides a model
for the regulated assembly of signaling complexes at
the plasma membrane. Mating of two haploid cells is
triggered by pheromones, termed a factor and α factor,
which activate a signal transduction pathway con-
sisting of two main modules: (1) a seven transmem-
brane domain receptor coupled to a heterotrimeric G
protein (Gαβγ) and (2) a cascade of protein kinases re-
lated to mammalian MAP kinase cascades. Another
protein, Ste5, is critical for communication between
these two modules. Ste5 is termed a “scaffold” protein
for its ability to bind all three kinases in the MAP kinase
cascade, which promotes their mutual interactions and
thereby facilitates signal transmission (Elion, 2001).
Ste5 also serves an adaptor function, in that it binds to
the Gβγ dimer and thereby connects the MAP kinase
cascade to the stimulus-sensing receptor/G protein
module. Analogs of Ste5 act as scaffolds for other MAP
kinase cascades in both yeast and mammalian cells
(Whitmarsh and Davis, 1998; Morrison and Davis, 2003),
and thus, Ste5 is a prototype for understanding the role
of scaffolding in signal transduction.
Mating pheromones trigger the recruitment of Ste5 to
the plasma membrane (Pryciak and Huntress, 1998),
and similar translocations occur for other yeast and
mammalian scaffold proteins (Reiser et al., 2000; Muller
et al., 2001). This event clearly has functional signifi-
cance, because artificially targeting Ste5 to the mem-
brane causes constitutive signaling (Pryciak and Hunt-
ress, 1998). Membrane recruitment of Ste5 is thought
to initiate activation of the Ste5-associated kinase
cascade by allowing the MAP kinase kinase kinase
(MAPKKK) Ste11 to become phosphorylated by its
membrane-localized activator, the PAK family kinase
Ste20 (Pryciak and Huntress, 1998; van Drogen et al.,
2000). The kinases downstream of Ste11, namely Ste7
and Fus3, are also brought to the membrane by Ste5,
which thereby serves as an activation platform for the
entire MAP kinase cascade (van Drogen et al., 2001).
Photobleaching studies suggest that the activated
MAPK Fus3 dissociates from Ste5 and then transits to
the nucleus to regulate cell cycle arrest and transcrip-
tional induction (van Drogen et al., 2001).
It is clear that membrane recruitment of Ste5 is pro-
moted by interaction with the receptor-activated Gβγ
dimer (Pryciak and Huntress, 1998; Mahanty et al.,
1999). Nevertheless, several aspects of Ste5 regulation
remain poorly understood. For example, Ste5 also has
an affinity for the nucleus, and one study suggested
that shuttling through the nucleus allows Ste5 to be-
come competent for membrane recruitment, though by
an undefined mechanism (Mahanty et al., 1999). It has
also been suggested that Gβγ may induce changes in
the conformation or oligomerization state of Ste5 as
part of the signaling activation mechanism (Yablonski
et al., 1996; Inouye et al., 1997; Sette et al., 2000; Wang
and Elion, 2003). These notions are based in part on
the existence of hyperactive mutants of Ste5 that have
been hypothesized to mimic an active conformation,
but whether such conformational changes actually oc-
cur and whether they are induced by Gβγ is unknown.
In this study, we identify a domain in Ste5 that binds
directly to phospholipid membranes. This domain is re-
quired for pheromone response in a manner that sug-
gests it acts cooperatively with the Gβγ binding do-
Figure 1. A Cortical Localization Domain in the Ste5 N Terminus
See Table S2 for plasmids.
(A) Behavior of the Ste5-NT fragment. Ste5 and Ste5?N are from
Pryciak and Huntress (1998).
(B) Top, cortical localization of GFP-Ste5-NT fusions in ste4D ste5D
cells (PPY886). Neither GFP alone nor GFP-Ste5-NT-?NLS are ex-
cluded from the nucleus, but they lack the strong nuclear enrich-
ment of GFP-Ste5-NT. Bottom, dominant negativity. Wild-type (wt)
cells (PPY1368) harboring plasmids as above were plated on –His/
Raff/Gal plates and overlaid with filter disks containing α factor (10
or 2 nmol).
(C) Mutant Ste5-NT fragments were tested for localization to nu-
cleus (nuc) and cortex (cort), for dominant negativity (dom neg),
and for binding to Ste4 (Gβ). Mutations were also moved to full-
length Ste5 and tested for α factor response. Results are summa-
rized; see Figure S1 and Figure 4 for supporting data. #, localization
results with C180A were ambiguous.
(D) Ste5 residues 37–76 are sufficient for plasma membrane local-
ization in PPY886. Top, GFP fusions to Ste5 residues 1–214, 1–125,
main, with neither interaction alone being sufficiently
strong to allow functional recruitment of Ste5. Remark-
ably, the Ste5 membrane binding domain is identical to
a previously characterized nuclear localization se-
quence, suggesting a revised view of the role of nuclear
shuttling in Ste5 function. Moreover, we find that the
normal level of signaling quiescence in the absence of
Gβγ activation depends on the imperfect nature of the
Ste5 membrane binding domain and that at least some
previous hyperactive Ste5 mutants are explained by an
increased affinity for membranes.
A G??-Independent Membrane Localization
Domain Maps to the Ste5 NLS
Previous work showed that the pheromone-activated
Gβγ dimer recruits Ste5 to the plasma membrane (Pry-
ciak and Huntress, 1998). Unexpectedly, we later found
that an amino-terminal fragment of Ste5 (Ste5-NT; resi-
dues 1–214) could localize to the cell cortex even with-
out pheromone treatment (often asymmetrically; see
Discussion) when overexpressed from the GAL1 pro-
moter (Figures 1A and 1B). Although this fragment con-
tains the binding site for Gβγ, its cortical localization
was still observed in cells lacking the Ste4 (Gβ) subunit,
such as ste4D ste5D cells (Figure 1B), as well as in cells
lacking any of the other mating pathway components
(data not shown).
To identify the responsible sequences, we made a
series of small deletions (Figure 1C). A deletion referred
to as ?NLS (?48–67), which removes a previously de-
fined nuclear localization sequence (NLS) (Mahanty et
al., 1999), eliminated cortical localization of the Ste5-
NT fragment as well as its strong nuclear enrichment
(Figures 1B and 1C). Previously, the NLS was found to
be required for membrane recruitment of full-length
Ste5 in response to pheromone, leading to the proposal
that nuclear shuttling allows Ste5 to become compe-
tent for recruitment (Mahanty et al., 1999). Our findings
show that the NLS governs cortical localization in a way
that is inherent to the Ste5-NT fragment and, hence,
can be studied in the absence of Gβγ, pathway signal-
ing, and other Ste5 sequences.
Though it did not affect binding to Ste4 (Gβ), the
?NLS mutation disrupted the ability of the Ste5-NT
fragment to display dominant-negative activity (Figures
1B and 1C), which is thought to reflect inhibition of Gβγ
(Whiteway et al., 1995; Feng et al., 1998). Therefore, al-
though not part of the Gβγ binding site, the NLS is
nonetheless required for functional interaction of the
Ste5 N terminus with Gβγ. In contrast, interaction with
Gβγ in both assays was disrupted, but cortical localiza-
tion was unperturbed by two mutations flanking the
RING-H2 domain, ?152–173 and ?169–171, which re-
semble the RING-H2 mutation C180A (Inouye et al.,
1997; Feng et al., 1998) but have the advantage of dis-
rupting only Gβγ binding and not other interactions
or two copies of residues 1–125 expressed in tandem ([1–125]x2).
Bottom, GST is fused at the N terminus of GFP, yielding GST-GFP-
Ste5 fusions of Ste5 residues 1–125, 37–125, or 37–76.
Cooperative Interactions Control Ste5 Localization
such as oligomerization (Figures S1A–S1D available in
the Supplemental Data with this article online). Thus,
the ability of Ste5 to engage Gβγ in its normal signaling
environment at the plasma membrane requires the Gβγ
binding site plus separate cortical localization informa-
tion provided by the Ste5 NLS.
Further truncations defined a minimum cortical local-
ization domain (Figure 1D). Localization was sharply re-
duced when the RING-H2 domain was removed (e.g.,
fragments 1–176 [data not shown] and 1–125). This was
not due to loss of Gβγ binding, as the cells (ste4D
ste5D) already lacked Gβ. Because the Ste5-NT frag-
ment can interact with itself via the RING-H2 domain
(Figure S1C), we hypothesized that the NLS region
could be a weak membrane binding motif whose bind-
ing avidity is enhanced by oligomerization, as seen for
other domains (Levine and Munro, 2002; Stefan et al.,
2002; Szeto et al., 2003). Indeed, cortical localization
was restored (Figure 1D) when two copies of Ste5(1–
125) were expressed in tandem (Ste5[1–125]x2) or when
this sequence was fused to GST, which forms homodi-
mers and can enhance Ste5 function (Inouye et al.,
1997; Sette et al., 2000; Wang and Elion, 2003). Ulti-
mately, a 40 residue segment overlapping the NLS,
Ste5(37–76), proved sufficient to function as an autono-
mous cortical localization signal (Figure 1D). Because
this localization was consistent with plasma membrane
binding, we refer to this region of Ste5 hereafter as the
The PM/NLS Domain Binds Acidic
In theory, cortical localization could reflect binding to
plasma membrane proteins or lipids. A screen of multi-
ple mutant strains for defects in Ste5-NT localization
failed to reveal a candidate protein partner, whereas a
two-hybrid screen yielded several Ste5-NT interaction
partners, but all required the RING-H2 domain and not
the PM/NLS region (M.J.W. and P.M.P., unpublished
In contrast, we did find that the Ste5 PM/NLS domain
binds directly to lipid membranes in vitro. The minimal
Ste5 membrane-localization domain was purified from
E. coli and assessed for membrane binding by copellet-
ing with liposomes (Figure 2A). This GST-Ste5(37–76)
fusion protein only bound liposomes containing acidic
phospholipids such as phosphatidic acid (PA), phos-
phatidylinositol 4-phosphate (PI4P), and phosphatidyli-
nositol 4,5-bisphosphate (PIP2). Although lacking strong
selectivity, binding favored increased negative charge
(PA < PI4P < PIP2) and could be enhanced by raising
either the charge or molar percentage of the lipid (Fig-
ure 2C), suggesting a role for electrostatic interaction
involving basic residues in the Ste5(37–76) fragment.
These results were reinforced by altering specific
lipid pools in vivo (Figures 2D and 2E). Stt4 and Pik1
are PI 4-kinases that produce PI4P at either the plasma
membrane (Stt4) or internal membranes (Pik1), whereas
Mss4 is a PI4P 5-kinase that produces PIP2 at the
plasma membrane (Audhya et al., 2000; Stefan et al.,
2002). Ste5-NT localization was reduced only mildly in
mss4 mutants but severely in stt4 mutants where
plasma membrane pools of both PI4P and PIP2are de-
Figure 2. Membrane Binding by the Ste5 PM/NLS Domain
(A) Bacterially expressed GST-Ste5(37–76) was copelleted with li-
posomes containing 100% PC or 80% PC plus 20% supplemental
phospholipid. Top, protein in supernatant (S) and pellet (P) frac-
tions. Bottom, the fraction of input protein recovered in the pellet
(percentage bound); the mean ± SD is from two to five trials, de-
pending on the lipid.
(B) Liposome binding is not observed for GST alone.
(C) Binding of GST-Ste5(37–76) to PC liposomes containing
increasing molar percentage of PA, PI(4)P, or PIP2(mean ± SD, n =
(D) Localization of GFP-Ste5-NT (pPP1750) in cells bearing temper-
ature-sensitive alleles of STT4, PIK1, or MSS4. Localization was
normal at 23°C (data not shown); cells shown were incubated at
37°C for 60 min.
(E) Effect of increased PA synthesis. GST-GFP-Ste5(1–125) was ex-
pressed in PPY886 with a vector control (left) or a plasmid express-
ing diacylglycerol (DAG) kinase (right). Arrowheads show recruit-
ment of the Ste5 fragment to intracellular lamellae.
(F) Disruption of nuclear transport does not block membrane local-
ization of the Ste5 N terminus. Strain PPY1649 harboring plasmid
pPP1978 was incubated at 23°C or was shifted to 37°C for 60 min.
pleted. By contrast, pik1 mutants, in which PI4P is de-
pleted from internal membranes, had little effect. Levels
of PA can be increased by expression of E. coli diacyl-
glycerol (DAG) kinase (Nakanishi et al., 2004), and this
diverted the Ste5 fragment toward lamellar structures
presumed to be new PA-rich membranes (Figure 2E).
Moreover, disruption of nuclear transport by inactiva-
ting the Ran GTPase Gsp1 (Wong et al., 1997) did not
block membrane localization by the Ste5 N terminus
(Figure 2F). Collectively, these in vitro and in vivo results
suggest that plasma membrane targeting by the Ste5
PM/NLS domain reflects a direct affinity for acidic phos-
A Putative Amphipathic ? Helix: Separation of PM
and NLS Functions
To test if the signaling role of the Ste5 PM/NLS domain
requires the PM function, the NLS function, or both, we
wished to separate these overlapping signals. Resi-
dues 45–67 contain two clusters of basic residues,
forming a bipartite NLS (Mahanty et al., 1999). A helical
wheel plot revealed that a 21 residue segment (residues
45–65) might form an amphipathic α helix in which
charged and nonpolar residues lie on opposite faces
(Figure 3A). Some amphipathic α helices can insert into
membranes with the helical axis parallel to the plane of
the lipid bilayer, thus burying nonpolar side chains in
the hydrophobic core while leaving basic groups ex-
posed to interact with acidic phosphates (Hristova et
This model suggested that basic residues in the PM/
NLS region may be important for both nuclear and
membrane targeting, whereas nonpolar residues may
be required mostly for membrane targeting. Indeed,
substituting alanines for all seven basic residues in the
two outer clusters (generating the “NLSo” allele) dis-
rupted both nuclear and membrane targeting, resulting
in loss of dominant negativity of the Ste5-NT fragment
and defective signaling by full-length Ste5 (Figures 3B
and 3D). In contrast, replacement of four nonpolar resi-
dues with polar asparagines (“NLSb”) disrupted only
membrane targeting and not nuclear targeting, thereby
separating these two functions. Despite intact nuclear
targeting, the NLSb mutation disrupted both dominant
negativity of the Ste5-NT fragment and signaling by
full-length Ste5 (Figure 3D). Therefore, nuclear targeting
by the Ste5 PM/NLS domain is not sufficient for mem-
brane binding or signaling activity, which instead re-
quire both basic and hydrophobic residues.
A reciprocal separation phenotype was obtained by
substituting alanines for a group of residues between
the outer basic clusters (“NLSm”). This disrupted only
nuclear targeting and not membrane targeting (Figures
3B and 3D), possibly due to the removal of three basic
residues (out of ten total) while maintaining a nonpolar
helical face. Despite reduced nuclear targeting, the
NLSm mutation did not disrupt dominant negativity by
the Ste5-NT fragment or signaling by full-length Ste5
(Figure 3D). We conclude that the membrane binding
function of the PM/NLS domain is critical for Ste5 sig-
naling, whereas the nuclear-targeting function is insuffi-
cient and possibly even unnecessary.
These mutations also affected in vitro liposome bind-
ing in a way that correlated with the in vivo results, in
that the NLSo and NLSb mutations caused the strong-
est defects, whereas the NLSm mutant was closest to
wild-type (wt) (Figure 3C). This in vitro assay only partly
recapitulated the drastic in vivo differences between
NLSb and NLSm, which may be exacerbated in vivo by
differences in membrane phospholipid composition or
by the competitive effect of nuclear localization (which
is intact for NLSb but is disrupted for NLSm).
Membrane-Tethered Ste11 Distinguishes Membrane
Binding and G?? Binding Roles of Ste5
Whereas the isolated Ste5 PM/NLS domain could local-
ize in the absence of Gβγ, signaling by full-length Ste5
is normally initiated by Gβγ. To test if the signaling role
of the PM/NLS domain is separable from Gβγ binding,
we devised a new assay involving a membrane-teth-
ered form of the MAPKKK Ste11. As seen previously
with Ste5 (Pryciak and Huntress, 1998), fusion of Ste11
to a membrane-targeting sequence (Cpr, from yeast
Ras2) caused constitutive activation of the mating
pathway (Figure 4A). This signaling bypassed phero-
mone, receptor (Ste2), and Gβ (Ste4) but still required
Ste5, Ste20, and other downstream components. Com-
paring ste4D with ste4D ste5D cells showed unequivo-
cally that Ste5 mediates Ste11-Cpr signaling without
Ste4 (Gβ). Our preliminary studies suggest that Ste5 is
required to enhance signaling from transiently activated
Ste11-Cpr molecules through the remainder of the MAP
kinase cascade (R.E.L. and P.M.P., unpublished data).
For the purposes here, Ste11-Cpr acts as a replace-
ment for Gβγ by providing an alternative membrane-
localized binding partner for Ste5. The results (below)
suggest that in both settings the Ste5 PM/NLS domain
serves to bind the plasma membrane cooperatively
with binding of Ste5 to a membrane-localized protein,
which can be either Gβγ or Ste11-Cpr.
In contrast to α factor response (which requires Gβγ),
signaling by Ste11-Cpr was unaffected by Ste5 muta-
tions that impinged on the Gβγ binding site (?152–173,
?169–171, and C180A) but was still disrupted by the
?NLS mutation (Figure 4B). Therefore, Ste11-Cpr re-
veals stark differences between Ste5 mutations in the
membrane binding and Gβγ binding domains that are
indistinguishable in α factor response assays. Consis-
tent with previous studies (Feng et al., 1998; Mahanty
et al., 1999), neither mutation blocked signaling by a
preactivated derivative of Ste11, Ste11-4, which by-
passes the need for activation by Ste20 at the mem-
brane and requires only the “scaffolding” function of
Ste5. Among mutants with altered PM/NLS sequences
(NLSo, NLSb, and NLSm), those that disrupted re-
sponse to α factor also disrupted signaling by Ste11-
Cpr, but not by Ste11-4 (Figure 4B, bottom). Because
Ste11-Cpr signaling is initiated at the membrane and
depends on membrane-localized Ste20 to activate
Ste11, we conclude that the PM/NLS region of Ste5 is
required for membrane-initiated signaling in general
rather than Gβγ binding per se or any other Gβγ-specific
aspect of membrane recruitment.
Functional Replacement of the Ste5 PM/NLS Domain
with Lipid Binding Domains
To address whether there was a vital role either for the
specific type of membrane binding motif present in
Cooperative Interactions Control Ste5 Localization
Figure 3. Separation of Nuclear-Targeting and Membrane-Targeting Signals in the Ste5 PM/NLS Domain
(A) A potential amphipathic α helix, shown by helical wheel projection of residues 45–65 (underlined). Black circles, charged residues; gray
circles, uncharged polar residues; and white squares, nonpolar residues. At left are two basic residues at the C-terminal end, K66 and R67;
their inclusion in the helix would disrupt amphipathicity, and so they could either be uninvolved or terminate the α helix with a different
(B) Summary of mutations and their effects on localization of N-terminal Ste5 fragments.
(C) Top, effects of mutations on binding of GST-Ste5(37–76) to PC + 7% PIP2liposomes. Bottom, quantitation of multiple trials (mean ± SD;
n = 5–8).
(D) Signaling role of the Ste5 PM/NLS domain correlates with membrane-targeting activity, not nuclear-targeting activity. (i) Localization of
GST-GFP-Ste5(1–125) fragments in ste4D ste5D strain PPY886. (ii) Dominant negativity of Ste5-NT(1–214) fragments in wt strain PPY1368. (iii)
Basal localization of full-length Ste5-GFPx3expressed from the STE5 promoter in ste5D strain PPY858. (iv) Pheromone response of full-length
Ste5-myc13expressed from the STE5 promoter in PPY858. Note that the vector control expresses GFP in column (i), but not (iii).
(E) Anti-myc immunoblot showing Ste5-myc13levels in PPY858.
Ste5 or for binding to a particular phospholipid, we re-
placed the PM/NLS domain with heterologous lipid
binding motifs (Figure 5A). We used pleckstrin homol-
ogy (PH) domains of various phospholipid preferences
from mammalian PLCδ, mammalian FAPP1, and yeast
Osh1 (Kavran et al., 1998; Dowler et al., 2000; Levine
and Munro, 2002; Stefan et al., 2002; Yu et al., 2004),
plus a basic-rich (B) motif from yeast Spo20 that binds
PA-rich membranes (Nakanishi et al., 2004). The chime-
ras were tagged at the Ste5 C terminus with three tan-
dem copies of GFP and expressed from the STE5 pro-
moter (Figure 5B).
Each of the chimeras dramatically reversed the sig-
naling defect of the Ste5?NLS mutant (Figure 5C). No-
tably, signaling and plasma membrane localization still
required pheromone treatment (Figures 5C and 5D) and
Gβγ (data not shown), indicating that the foreign do-
mains did not bypass the role of Gβγ but instead coop-
erated with Ste5-Gβγ binding to mediate pheromone re-
sponse and recruitment. Signaling by Ste11-Cpr was
also restored (Figure S2). Even when the FAPP1 PH do-
main caused localization to Golgi membranes in un-
stimulated cells, signaling and plasma membrane re-
cruitment were still inducible by pheromone (Figures 5C
and 5D). None of the PH domains promoted nuclear
localization (even in msn5D mutants [data not shown]),
and although the Spo20 motif did cause some nuclear
localization, a variant motif (spo20*) carrying a mutation
that disrupts membrane binding (L67P; [Nakanishi et
al., 2004]) did not rescue signaling despite the nuclear
Figure 4. A Membrane-Targeted Ste11 Derivative Reveals a Func-
tional Role for the Ste5 PM/NLS Domain Independent of Gβγ
(A) Membrane-targeted Ste11 causes constitutive mating pathway
signaling that requires Ste5, but not Ste4 (Gβ). The indicated strains
harbored PGAL1-driven forms of Ste11 fused to the membrane-tar-
geting prenylation/palmitoylation domain from Ras2 (Cpr) or a con-
trol sequence (Cpr-SS) (Pryciak and Huntress, 1998). Growth arrest
was assessed by spotting onto −His/Glu (GLU) or −His/Raff/Gal
(GAL) plates. Transcriptional activation was measured by FUS1-
lacZ assay (mean ± SD; n = 6).
(B) Ste5 mutants were compared for their ability to mediate signal-
ing initiated by α factor, Ste11-Cpr, or Ste11-4 in strain PPY858
(left) or PPY886 (middle, right). Bars, FUS1-lacZ activation (mean ±
SD; n = 6).
localization. Therefore, the mere presence of basic-
rich or nuclear-targeting sequences was not sufficient
to restore function. Similarly, we failed to rescue the
Ste5?NLS mutant by promoting nuclear shuttling with
heterologous nuclear import and export sequences
(Figure S3). We conclude that the membrane binding
function of the Ste5 PM/NLS domain is sufficient to ex-
plain its critical positive role in pheromone response
and that this role does not require a unique structural
motif or lipid preference.
Hyperactive Ste5 Mutations Increase Membrane
Binding by the PM/NLS Domain
The native PM/NLS domain behaves as a weak mem-
brane binding motif that ordinarily cannot bring full-
length Ste5 to the membrane in the absence of Gβγ.
We wondered if mutations that improve or expose this
domain might cause Gβγ-independent membrane local-
ization and, hence, constitutive signaling. In fact, four
hyperactive Ste5 mutants were identified previously
(Hasson et al., 1994; Sette et al., 2000), but the molecu-
lar explanation for their hyperactivity has remained elu-
sive. Two of these mutations (P44L and T52M) lie within
the PM/NLS region (Figure 6A), possibly altering its
Indeed, the P44L and T52M mutations caused mem-
brane localization of the Ste5(1–125) fragment (Figure
6D), which ordinarily requires dimerization (see Figure
1D). Even when Ste5(1–125) was dimerized with GST,
the P44L and T52M mutations caused increased mem-
brane partitioning and reduced nuclear abundance
(Figure 6D). Enhanced membrane binding was not sim-
ply a result of nuclear depletion, because the T52M mu-
tation caused membrane localization in settings where
the wt sequence showed little nuclear enrichment (Fig-
ure 6E), such as when using the 37–76 fragment (panels
i and ii) or when nuclear import of the 1–125 fragment
was blocked in gsp1-1tscells (panels iii and iv). Rather,
nuclear depletion appeared to be a secondary result
of increased membrane binding, because both effects
were blocked when T52M was combined with the NLSb
mutations (Figure 6E, panels v–vi).
These mutations also affected localization of full-
length Ste5 in a way that can explain their constitutive
signaling. Because several mutants were isolated as
Ste5-GST fusions (Sette et al., 2000), we checked their
localization as both GFP-Ste5 and GFP-Ste5-GST fu-
sions. The T52M mutation caused clear repartitioning
toward the plasma membrane in either context,
whereas P44L only altered localization in the GFP-Ste5-
GST context (Figure 6F). Thus, T52M has a stronger ef-
fect than P44L, whose weaker phenotype is revealed
by fusion to GST (possibly by the avidity effect of di-
merization, as seen for Ste5-NT fragments). This dis-
tinction correlated with their signaling phenotypes (Fig-
ure 6B) and with differing abilities to enhance binding
to liposomes in vitro (Figure 6C). The dependence of
the P44L mutant on GST also explains why no localiza-
tion effect was seen previously (Sette et al., 2000). The
other two mutants, C226Y and S770N, were the least
hyperactive in signaling assays, and neither had a
strong localization effect (data not shown; see also Fig-
ure S4). Importantly, membrane localization behaved as
a cause and not an effect of constitutive signaling, as it
was still observed when signaling was blocked in ste4D
ste7D cells (Figure 6F).
Creation of New Hyperactive Ste5 Mutants
by Replacement of Suboptimal Residues
Modeling the PM/NLS domain as an amphipathic α he-
lix revealed that the P44L and T52M mutations add hy-
Cooperative Interactions Control Ste5 Localization
Figure 5. Functional Replacement of the Ste5 PM/NLS Domain by Heterologous Membrane Binding Domains
(A) Schematic description of chimeric proteins.
(B) Anti-GFP immunoblot showing levels of Ste5-GFPx3chimeras.
(C) Restoration of pheromone response. FUS1-lacZ levels were measured (mean ± SD; n = 4) with and without α factor (αF).
(D) Subcellular localization of chimeric Ste5-GFPx3fusions before and after α factor treatment.
In all panels, strain PPY858 (ste5D) harbored plasmids described in Table S2.
drophobic side chains to the predicted nonpolar face
(Figure 7A). This nonpolar face includes some polar res-
idues (see Figure 3A) that could interfere with optimal
membrane binding, two of which—Thr52 and Gln59—
are found in other Ste5 orthologs (Figure 7B), perhaps
implying that weak membrane binding is beneficial.
This view led us to predict that: (1) mutations at Gln59
might also cause hyperactivity, (2) hyperactivity should
require increased hydrophobicity, and (3) the effects of
increased hydrophobicity should require placement on
the nonpolar face of the amphipathic helix. These pre-
dictions were tested by new mutations (Figure 7C).
Indeed, at the original positions Pro44 and Thr52,
constitutive signaling required replacement with bulky
hydrophobic residues such as Leu or Phe (Figure 7D).
As before, mutations at Thr52 had stronger effects than
those at Pro44, which required the GST moiety. Muta-
tion of Ser43 (S43A) had no effect, showing that the Pro44
phenotypes do not result from loss of an SP phosphoryla-
tion site. Confirming our prediction, Ste5 could also be
made hyperactive by substitutions at Gln59 (Figure 7D),
yielding three phenotypic classes—strongly hyperactive
(Leu and Phe), partially hyperactive (Ala and Gly), and
fully quiescent (His and Gln)—that agree with the ten-
dency of each residue to partition into a hydrophobic
membrane environment (Wimley and White, 1996;
Hessa et al., 2005). Substitution of Leu for Ser62 (S62L),
which is also predicted to lie on the nonpolar face,
caused measurable hyperactivity in the Ste5-GST con-
text, though this was quite weak. Therefore, the side
chains at Ste5 positions 52 and 59 play an especially
critical role in enforcing signaling quiescence. By con-
trast, placing Leu residues at positions expected to lie
on the opposite (polar) face of the amphipathic helix
(i.e., E53L, R60L, and S61L) did not cause constitutive
signaling (Figure 7D). Other mutations that delete or
disrupt the PM/NLS region also had no effect (Figure
7D, bottom), and thus, hyperactivity is not a trivial by-
product of reduced nuclear localization or conforma-
Subcellular localization of these Ste5 mutants agreed
with their signaling phenotypes, as constitutive mem-
brane localization was shown only by those mutants
with strong constitutive signaling (Figure 7E and data
not shown). As noted above for T52M and P44L, mem-
brane localization of these new hyperactive mutants
was a cause rather than an effect of signaling, as it was
maintained in ste4D ste7D cells (shown for Ste5-Q59L;
Figure 7E). Furthermore, the effects of these mutations
are clearly attributable to direct changes in membrane
binding by the PM/NLS motif, because they caused
parallel effects on membrane localization by the mini-
mal Ste5(37–76) domain, including a distinction be-
tween stronger and weaker hyperactive mutants (Fig-
ure S5A). Also, only mutants with enhanced membrane
localization in vivo showed increased liposome binding
in vitro (Figure S5B). In sum, the “imperfections” in the
Ste5 PM/NLS domain prevent promiscuous association
with the plasma membrane in the absence of Gβγ acti-
Ste5-Membrane and Ste5-G?? Affinities Set
Thresholds for Productive Signaling
We also analyzed some of the strongest hyperactive
mutants when expressed at normal levels from the na-
tive STE5 promoter. The T52L and Q59L mutants showed
a 10- to 30-fold increase inbasal FUS1-lacZ activity(Figure
S5C), which was sufficient to cause slowed growth and
increased duration of G1 phase (data not shown),
though signaling was still increased substantially by
Figure 6. Increased Membrane Binding by Hyperactive Ste5 Mutants
(A) Sites of mutations.
(B) Constitutive signaling by Ste5 mutants, expressed from the GAL1 promoter with or without a C-terminal GST tag, in PPY886 (ste4D ste5D).
Bars, mean ± SD (n = 4). See Figure S4 for additional data.
(C) Liposome binding by wt, P44L, and T52M forms of GST-Ste5(37–76). Top, representative result using 5% PIP2. Bottom, results of multiple
trials using 5% or 7% PIP2were normalized to wt binding from each trial and combined (mean ± SD, n = 5–8).
(D) The P44L and T52M mutations enhance membrane localization of GFP-Ste5(1–125), expressed with or without an N-terminal GST tag
(E) Relationship between increased membrane binding and reduced nuclear localization. (i and ii) T52M increases membrane localization of
Ste5(37–76), which shows minimal nuclear enrichment; strain, PPY886. (iii and iv) T52M increases membrane localization of Ste5(1–125) even
when nuclear localization is disrupted by incubation of gsp1-1tscells at 37°C; strain, PPY1649. (v and vi) T52M does not cause nuclear
depletion when membrane binding is disrupted by the NLSb mutation; strain, PPY886.
(F) The P44L and T52M mutations cause constitutive membrane localization of full-length Ste5. GFP-Ste5 or GFP-Ste5-GST derivatives were
expressed in PPY886 (ste4D ste5D) or PPY1215 (ste4D ste7D).
pheromone addition. These results reinforce the finding
that the mating pathway cannot maintain appropriate
dependence on pheromone if Ste5 binds membranes
too strongly and illustrate the benefit of maintaining
residues that decrease membrane affinity (e.g., Thr52
and Gln59). They further imply that the low native con-
centration of Ste5 also helps minimize promiscuous
The submaximal activation when these alleles were
expressed at normal levels allowed us to show that in-
creased Ste5-membrane affinity can compensate for
reduced Ste5-Gβγ affinity. Two mutations in Ste4 (Gβ),
Ste4-K55E and Ste4-D62G, cause moderate defects in
binding Ste5 (Figure S5E) and signaling defects (Figure
7F). Pheromone response by these Gβ mutants was re-
stored dramatically when Ste5 harbored the T52L or
Q59L mutations (Figure 7F). This suggests that the ef-
fective affinity of Gβγ for Ste5 is determined by integ-
ration of separate Ste5-membrane and Ste5-Gβγ bind-
Finally, although increased Ste5-membrane affinity
can reduce or even bypass the need for Ste5-Gβγ bind-
ing, we wondered if the converse was also true. With-
out an obvious way to directly increase Ste5-Gβγ bind-
ing affinity, we instead increased Ste5 abundance.
Indeed, expression from a multicopy plasmid (Figure
7G) or from the GAL1 promoter (data not shown) par-
tially restored function to Ste5 derivatives that lack the
PM/NLS domain (?NLS) or a larger region (?26–138). In
effect, the PM/NLS domain reduces the concentration
of Ste5 needed for functional interaction with Gβγ. This
supports the notion of cooperative binding between
separate Ste5-membrane and Ste5-Gβγ interactions
that alone are normally too weak to be productive.
We identify a motif in the Ste5 N terminus (the PM/NLS
domain) that binds acidic phospholipid membranes,
which we argue reflects a propensity to form an amphi-
pathic α helix. Although distinct from the Gβγ binding
domain, the PM/NLS domain is required for Ste5 to en-
gage Gβγ at the plasma membrane and, hence, to me-
diate signaling responses to mating pheromone. There-
fore, we propose that membrane recruitment of Ste5
requires the cooperative effect of two separate, weak
interactions: a Ste5-Gβγ interaction and a Ste5-mem-
brane interaction (Figure 8).
Weak membrane binding is clearly advantageous, as
stronger binding by the Ste5 PM/NLS domain causes
deregulated signaling (Figures 6 and 7). Combining
Cooperative Interactions Control Ste5 Localization
Figure 7. Creation of New Hyperactive Ste5 Mutants
(A) The P44L and T52M mutations add hydrophobic side chains to the predicted nonpolar face of the amphipathic α helix. Pro44 was not
shown on the helical wheel in Figure 3A because of its helix-breaking nature, but P44L is predicted to lie on the nonpolar face.
(B) Alignment of PM/NLS sequences in Ste5 orthologs from S. cerevisiae, S. bayanus, S. castelli, and S. kluyveri. Thr52 and Gln59 (arrowheads)
are polar residues that are predicted to lie on the nonpolar face of the helix (see Figure 3A) and thus impede membrane binding.
(C) New mutations created in the PM/NLS domain.
(D) Constitutive signaling by the new Ste5 mutants in PPY886 (ste4D ste5D). Also tested in the same vectors were the ?NLS, NLSo, NLSb,
and NLSm mutants. Bars, mean ± SD (n = 4). Numerical values are shown where signaling was significantly greater than wt Ste5 but was not
evident in the bar graphs.
(E) Localization of some GFP-Ste5 mutants (without GST) in PPY886 or in ste4D ste7D strain PPY1215 (bottom right).
(F) Increased Ste5-membrane affinity compensates for decreased Ste5-Gβγ affinity. Ste5 variants were expressed from the STE5 promoter in
PPY657 (ste4D ste5D) along with wt Ste4 (Gβ) or Ste4 mutants (K55E and D62G) with moderate Ste5 binding defects (Figure S5C). Bars,
mean ± SD (n = 4).
(G) Overexpression partially rescues Ste5 mutants lacking the PM/NLS domain. Ste5 derivatives expressed from a CEN ARS plasmid (low
copy) or a 2 ?m plasmid (high copy) were tested for mating and FUS1-lacZ induction (mean ± SD; n = 3) by α factor (in PPY858).
multiple weak interactions may be generally beneficial
when the assembly of membrane-associated protein
complexes is dynamic and/or regulated (Teruel and
Meyer, 2000). For example, the actin regulatory protein
N-WASP is activated cooperatively by PIP2and Cdc42
via separate domains (Prehoda and Lim, 2002; Papayan-
nopoulos et al., 2005). Even individual PH domains can
have separate lipid binding and protein binding inter-
faces that act in concert to govern subcellular localiza-
tion (Lodowski et al., 2003; Godi et al., 2004; Roy and
Levine, 2004). These combinatorial targeting strategies
could be undermined if individual interactions had
strong affinities, perhaps explaining the feeble lipid
binding and localization of many PH domains (Yu et
The level of flexibility we observe is remarkable, as
surrogate interactions can substitute for either the
Ste5-membrane interaction (e.g., using PH domains;
Figure 5) or the Ste5-Gβγ interaction (e.g., using Ste11-
Cpr; Figure 4). Even simultaneous substitution of both
native interactions can yield robust signaling (Figure
S2). This illustrates the functional plasticity inherent to
Figure 8. Model for the Role of Membrane Binding in Ste5 Localization and Signaling
For clarity, some pathway components are omitted from panels (B)–(D) in order to emphasize factors governing Ste5 localization.
(A) Original model for Ste5 membrane recruitment triggered by Gβγ, and its role in promoting the Ste20/Ste11 step.
(B) Revised model, in which membrane recruitment of Ste5 requires cooperativity between a weak Ste5-Gβγ interaction and a weak Ste5-
membrane interaction. The cylinder denotes a putative amphipathic α helix formed by the PM/NLS domain.
(C) The Ste5-membrane interaction can also synergize with the Ste5-Ste11 interaction in cells expressing membrane-tethered Ste11 (Ste11-
(D) Hyperactive Ste5 mutants have PM/NLS domains with increased membrane affinity, causing membrane localization and signaling in the
absence of Gβγ.
synergistic schemes, which could aid the wiring of sig-
naling pathways during development or evolution
(Pawson and Nash, 2003; Dueber et al., 2004). In prin-
ciple, nothing limits this strategy to two interactions.
Indeed, a separate region of Ste5 with similarity to PH
domains may also contribute to Ste5 membrane re-
cruitment (L.S. Garrenton and J. Thorner, personal
Ste5 Activation Mechanisms
Regulation of Ste5 signaling has been proposed to in-
volve changes in Ste5 localization, conformation, and
oligomerization (Yablonski et al., 1996; Inouye et al.,
1997; Pryciak and Huntress, 1998; Mahanty et al., 1999;
Sette et al., 2000; Wang and Elion, 2003). Although hy-
peractive mutants have been used to infer conforma-
tional changes in Ste5, we find that some of these mu-
tations in fact alter Ste5 localization. The membrane
association induced by activating mutations in the PM/
NLS domain (Figures 6 and 7) is sufficient to explain
their signaling phenotypes, as this would promote acti-
vation of Ste11 by membrane-localized Ste20 (see Fig-
ure 8). Conceivably, however, changes in Ste5 confor-
mation or oligomerization could further affect the
exposure or binding avidity of the PM/NLS domain. In-
deed, the fact that the Ste5-NT fragment, but not full-
length Ste5, shows constitutive membrane localization
implies that the PM/NLS domain may be masked in full-
length Ste5. This masking effect might be relieved in
other Ste5 hyperactive mutants (Sette et al., 2000) or
by Gβγ binding.
Whereas some signaling pathways can be triggered
by increased phospholipid synthesis (Hurley and Meyer,
2001), the Ste5 PM/NLS domain binds several different
acidic phospholipids and can be functionally replaced
with foreign domains of various lipid preferences. There-
fore, we suggest that pheromone activation of Gβγ
dimers is sufficient to explain the regulated activation
of Ste5 signaling, with the Ste5-membrane interaction
playing a role that is passive but crucial. Membrane
targeting by the PM/NLS domain tends to be asymmet-
ric, either in full-length Ste5 (Figure 6F) or fragments
(Figure 1D). A similar tendency was noted for the Osh2
PH domain (Roy and Levine, 2004), and so both may
reflect an underappreciated asymmetry in the distribu-
tion of acidic phospholipids. Although intriguing, this
bias is not obviously required for polarized recruitment
of Ste5 by Gβγ, which is still seen with Ste5 chimeras
Our findings also help constrain other models for
Ste5 regulation. For example, those invoking a role for
oligomerization via the RING-H2 domain must consider
that this domain appears to be dispensible (e.g., using
the C180A mutant) when Gβγ is bypassed (e.g., using
Ste11-Cpr; Figure 4B). Also, given the level of function
shown by the PH domain chimeras (Figure 5C and Fig-
ure S2), a signaling-proficient conformation postulated
for Ste5 (Sette et al., 2000) seems unlikely to absolutely
require the PM/NLS domain, though a subtle role re-
Dual-Function Targeting Domains
We show that the Ste5 PM/NLS domain governs tar-
geting to both plasma membrane and nucleus. A prior
study proposed that nuclear shuttling of Ste5 is re-
quired for its membrane recruitment (Mahanty et al.,
1999), based largely on the phenotypes of cis-acting
mutations that disrupt nuclear targeting but which in
retrospect seem likely to also disrupt direct membrane
binding by Ste5. Other data in that study involving
trans-acting nuclear import/export factors indicated
that Ste5 was poorly recruited to the membrane if se-
questered in the nucleus, which is consistent with our
results. Our findings argue that membrane binding, and
Cooperative Interactions Control Ste5 Localization
not nuclear targeting, best explains the role of the PM/
NLS domain in Ste5 function. These include: (1) direct
membrane binding by the PM/NLS domain, (2) separa-
tion-of-function mutations in the PM/NLS domain, and
(3) functional replacement of the PM/NLS domain with
lipid binding, but not nuclear-targeting, domains.
Is there a benefit to having the PM domain also pos-
sess NLS activity? Nuclear targeting could help mini-
mize promiscuous membrane contact by reducing
cytoplasmic concentrations of Ste5, and several obser-
vations suggest that nuclear sequestration of Ste5 is
inhibitory (Mahanty et al., 1999; Kunzler et al., 2001).
Competition between membrane and nuclear targeting
could also sharpen the transition between on and off
states (Ferrell, 1998). Notably, nuclear localization of
Ste5 is not required to propagate signal into the nu-
cleus, because trapping Ste5 at the plasma membrane
does not disrupt signaling (Pryciak and Huntress,
1998). Instead, Fus3 likely carries signal into the nu-
cleus after activation on membrane-localized Ste5 (van
Drogen et al., 2001).
Interestingly, dual-function targeting motifs are not
unique to Ste5, as two related examples were reported
recently: one in the S. cerevisiae transcription factor
Opi1 (Loewen et al., 2004) and another in the S. pombe
cytokinesis factor Mid1 (Celton-Morizur et al., 2004).
Given the diverse functions of these proteins, this strat-
egy may ultimately prove to be widespread.
Strains and Plasmids
Yeast strains are listed in Table S1. Plasmids are listed in Table S2;
DNA regions amplified by PCR or subjected to mutagenesis were
sequenced to ensure the absence of spurious mutations.
GFP fusions to full-length Ste5 or Ste5 fragments were visualized
without fixation. Results representative of multiple repeated experi-
ments are shown. Fusions expressed from the GAL1 promoter were
induced with 2% galactose for 60–180 min., depending on expres-
sion strength. Nuclear localization was verified in some experi-
ments by DAPI staining (data not shown). Ste5 expressed at native
levels was visualized as a Ste5-GFPx3fusion to increase signal
levels; cells were grown in selective medium either with or without
α factor (10 ?M for 60 min).
Assays of growth arrest, mating, and FUS1-lacZ induction were
performed as described previously (Pryciak and Huntress, 1998;
Lamson et al., 2002). Halo assays of growth arrest used 10 nmol
and 2 nmol α factor (Figure 1B) or 10 nmol α factor (Figure 3D). To
assay growth arrest by galactose-inducible constructs (Figure 4A),
cell suspensions were spotted onto selective GLU (2% glucose) or
GAL (2% raffinose + 2% galactose) plates. Patch mating assays
used strain PT2α as the mating partner. To measure induction of
FUS1-lacZ by pheromone, cells were incubated with and without 5
?M α factor for 2 hr. FUS1-lacZ in response to galactose-inducible
constructs was measured 3 hr after addition of 2% galactose to
cultures in 2% raffinose medium. Levels of FUS1-lacZ activity were
quantified by β-galactosidase assay as described previously (Pry-
ciak and Huntress, 1998; Lamson et al., 2002).
Interactions were measured by using the two-hybrid reporter strain
PPY760 and were quantified by liquid β-galactosidase assays
(Lamson et al., 2002).
Wt and mutant derivatives of GST-Ste5(37–76) were expressed
from pGEX-6P-1 and purified from E. coli strain BL21-RIPL (Strata-
gene) by using glutathione Sepharose as described previously (Na-
kanishi et al., 2004); solutions included 0.2 mM PMSF and 0.1 mM
benzamidine to inhibit proteolysis. Purified GST fusions (or GST
alone) were pelleted with sucrose-laden liposomes in 20 mM
NaPO4(pH 7.5), 100 mM NaCl. After electrophoresis and Coomas-
sie blue staining, the fraction of input protein recovered in the pellet
(percentage bound) was quantified (Nakanishi et al., 2004). Lipo-
somes contained phosphatidylcholine (PC) alone or mixed with
other lipids; to maximize sensitivity to changes in binding affinity,
tests of Ste5 mutants used 5% or 7% PIP2, based on the half-
maximal binding point in Figure 2C.
Yeast cell extracts were prepared by glass bead lysis as described
previously (Lamson et al., 2002). Epitope-tagged Ste5 was immu-
noprecipitated with mouse anti-myc (9E10; Santa Cruz Biotech) or
mouse anti-GFP (clones 7.1 and 13.1; Roche) antibodies and de-
tected by immunoblotting with rabbit anti-myc (A-14; Santa Cruz
Biotech) or mouse anti-GFP (B34; Covance) antibodies using meth-
ods described previously (Lamson et al., 2002).
Supplemental Data include five figures and two tables and are
available with this article online at http://www.molecule.org/cgi/
We thank the following for gifts of plasmids and yeast strains: A.
Audhya and S. Emr, N. Edgington and B. Futcher, D. Jenness, D.
Lambright, M. Peter, P. Silver, and J. Thorner. We also thank D.
McCollum and C. Goutte for comments on the manuscript and J.
Thorner for communicating unpublished results. This work was
supported by grants from the National Institutes of Health to P.M.P
(GM57769) and A.M.N. (GM62184).
Received: April 4, 2005
Revised: July 27, 2005
Accepted: August 17, 2005
Published: October 6, 2005
Audhya, A., Foti, M., and Emr, S.D. (2000). Distinct roles for the
yeast phosphatidylinositol 4-kinases, Stt4p and Pik1p, in secretion,
cell growth, and organelle membrane dynamics. Mol. Biol. Cell 11,
Celton-Morizur, S., Bordes, N., Fraisier, V., Tran, P.T., and Paoletti,
A. (2004). C-terminal anchoring of mid1p to membranes stabilizes
cytokinetic ring position in early mitosis in fission yeast. Mol. Cell.
Biol. 24, 10621–10635.
Dohlman, H.G., and Thorner, J.W. (2001). Regulation of G protein-
initiated signal transduction in yeast: paradigms and principles.
Annu. Rev. Biochem. 70, 703–754.
Dowler, S., Currie, R.A., Campbell, D.G., Deak, M., Kular, G.,
Downes, C.P., and Alessi, D.R. (2000). Identification of pleckstrin-
homology-domain-containing proteins with novel phosphoinosi-
tide-binding specificities. Biochem. J. 351, 19–31.
Dueber, J.E., Yeh, B.J., Bhattacharyya, R.P., and Lim, W.A. (2004).
Rewiring cell signaling: the logic and plasticity of eukaryotic protein
circuitry. Curr. Opin. Struct. Biol. 14, 690–699.
Elion, E.A. (2001). The Ste5p scaffold. J. Cell Sci. 114, 3967–3978.
Feng, Y., Song, L.Y., Kincaid, E., Mahanty, S.K., and Elion, E.A.
(1998). Functional binding between Gβ and the LIM domain of Ste5
is required to activate the MEKK Ste11. Curr. Biol. 8, 267–278.
Ferrell, J.E., Jr. (1998). How regulated protein translocation can pro-
duce switch-like responses. Trends Biochem. Sci. 23, 461–465.
Godi, A., Di Campli, A., Konstantakopoulos, A., Di Tullio, G., Alessi,
D.R., Kular, G.S., Daniele, T., Marra, P., Lucocq, J.M., and De
Matteis, M.A. (2004). FAPPs control Golgi-to-cell-surface mem-
brane traffic by binding to ARF and PtdIns(4)P. Nat. Cell Biol. 6,
Hasson, M.S., Blinder, D., Thorner, J., and Jenness, D.D. (1994).
Mutational activation of the STE5 gene product bypasses the
requirement for G protein β and γ subunits in the yeast pheromone
response pathway. Mol. Cell. Biol. 14, 1054–1065.
Hessa, T., Kim, H., Bihlmaier, K., Lundin, C., Boekel, J., Andersson,
H., Nilsson, I., White, S.H., and von Heijne, G. (2005). Recognition of
transmembrane helices by the endoplasmic reticulum translocon.
Nature 433, 377–381.
Hristova, K., Wimley, W.C., Mishra, V.K., Anantharamiah, G.M., Se-
grest, J.P., and White, S.H. (1999). An amphipathic α-helix at a
membrane interface: a structural study using a novel X-ray diffrac-
tion method. J. Mol. Biol. 290, 99–117.
Hunter, T. (2000). Signaling—2000 and beyond. Cell 100, 113–127.
Hurley, J.H., and Meyer, T. (2001). Subcellular targeting by mem-
brane lipids. Curr. Opin. Cell Biol. 13, 146–152.
Inouye, C., Dhillon, N., and Thorner, J. (1997). Ste5 RING-H2 do-
main: role in Ste4-promoted oligomerization for yeast pheromone
signaling. Science 278, 103–106.
Kavran, J.M., Klein, D.E., Lee, A., Falasca, M., Isakoff, S.J., Skolnik,
E.Y., and Lemmon, M.A. (1998). Specificity and promiscuity in phos-
phoinositide binding by pleckstrin homology domains. J. Biol.
Chem. 273, 30497–30508.
Kunzler, M., Trueheart, J., Sette, C., Hurt, E., and Thorner, J. (2001).
Mutations in the YRB1 gene encoding yeast ran-binding-protein-1
that impair nucleocytoplasmic transport and suppress yeast mat-
ing defects. Genetics 157, 1089–1105.
Lamson, R.E., Winters, M.J., and Pryciak, P.M. (2002). Cdc42 regu-
lation of kinase activity and signaling by the yeast p21-activated
kinase Ste20. Mol. Cell. Biol. 22, 2939–2951.
Levine, T.P., and Munro, S. (2002). Targeting of Golgi-specific pleck-
strin homology domains involves both PtdIns 4-kinase-dependent
and -independent components. Curr. Biol. 12, 695–704.
Lodowski, D.T., Pitcher, J.A., Capel, W.D., Lefkowitz, R.J., and
Tesmer, J.J. (2003). Keeping G proteins at bay: a complex between
G protein-coupled receptor kinase 2 and Gβγ. Science 300, 1256–
Loewen, C.J., Gaspar, M.L., Jesch, S.A., Delon, C., Ktistakis, N.T.,
Henry, S.A., and Levine, T.P. (2004). Phospholipid metabolism regu-
lated by a transcription factor sensing phosphatidic acid. Science
Mahanty, S.K., Wang, Y., Farley, F.W., and Elion, E.A. (1999). Nuclear
shuttling of yeast scaffold Ste5 is required for its recruitment to the
plasma membrane and activation of the mating MAPK cascade.
Cell 98, 501–512.
Morrison, D.K., and Davis, R.J. (2003). Regulation of MAP kinase
signaling modules by scaffold proteins in mammals. Annu. Rev. Cell
Dev. Biol. 19, 91–118.
Muller, J., Ory, S., Copeland, T., Piwnica-Worms, H., and Morrison,
D.K. (2001). C-TAK1 regulates Ras signaling by phosphorylating the
MAPK scaffold, KSR1. Mol. Cell 8, 983–993.
Nakanishi, H., de los Santos, P., and Neiman, A.M. (2004). Positive
and negative regulation of a SNARE protein by control of intracellu-
lar localization. Mol. Biol. Cell 15, 1802–1815.
Papayannopoulos, V., Co, C., Prehoda, K.E., Snapper, S., Taunton,
J., and Lim, W.A. (2005). A polybasic motif allows N-WASP to act
as a sensor of PIP2density. Mol. Cell 17, 181–191.
Pawson, T., and Scott, J.D. (1997). Signaling through scaffold, an-
choring, and adaptor proteins. Science 278, 2075–2080.
Pawson, T., and Nash, P. (2003). Assembly of cell regulatory sys-
tems through protein interaction domains. Science 300, 445–452.
Prehoda, K.E., and Lim, W.A. (2002). How signaling proteins integ-
rate multiple inputs: a comparison of N-WASP and Cdk2. Curr.
Opin. Cell Biol. 14, 149–154.
Pryciak, P.M., and Huntress, F.A. (1998). Membrane recruitment of
the kinase cascade scaffold protein Ste5 by the Gβγ complex un-
derlies activation of the yeast pheromone response pathway.
Genes Dev. 12, 2684–2697.
Reiser, V., Salah, S.M., and Ammerer, G. (2000). Polarized localiza-
tion of yeast Pbs2 depends on osmostress, the membrane protein
Sho1 and Cdc42. Nat. Cell Biol. 2, 620–627.
Roy, A., and Levine, T.P. (2004). Multiple pools of phosphatidylinosi-
tol 4-phosphate detected using the pleckstrin homology domain of
Osh2p. J. Biol. Chem. 279, 44683–44689.
Sette, C., Inouye, C.J., Stroschein, S.L., Iaquinta, P.J., and Thorner,
J. (2000). Mutational analysis suggests that activation of the yeast
pheromone response mitogen-activated protein kinase pathway in-
volves conformational changes in the Ste5 scaffold protein. Mol.
Biol. Cell 11, 4033–4049.
Stefan, C.J., Audhya, A., and Emr, S.D. (2002). The yeast synapto-
janin-like proteins control the cellular distribution of phosphatidyli-
nositol (4,5)-bisphosphate. Mol. Biol. Cell 13, 542–557.
Szeto, T.H., Rowland, S.L., Habrukowich, C.L., and King, G.F.
(2003). The MinD membrane targeting sequence is a transplantable
lipid-binding helix. J. Biol. Chem. 278, 40050–40056.
Teruel, M.N., and Meyer, T. (2000). Translocation and reversible lo-
calization of signaling proteins: a dynamic future for signal trans-
duction. Cell 103, 181–184.
van Drogen, F., O’Rourke, S.M., Stucke, V.M., Jaquenoud, M., Nei-
man, A.M., and Peter, M. (2000). Phosphorylation of the MEKK
Ste11p by the PAK-like kinase Ste20p is required for MAP kinase
signaling in vivo. Curr. Biol. 10, 630–639.
van Drogen, F., Stucke, V.M., Jorritsma, G., and Peter, M. (2001).
MAP kinase dynamics in response to pheromones in budding
yeast. Nat. Cell Biol. 3, 1051–1059.
Wang, Y., and Elion, E.A. (2003). Nuclear export and plasma mem-
brane recruitment of the Ste5 scaffold are coordinated with oligo-
merization and association with signal transduction components.
Mol. Biol. Cell 14, 2543–2558.
Whiteway, M.S., Wu, C., Leeuw, T., Clark, K., Fourest-Lieuvin, A.,
Thomas, D.Y., and Leberer, E. (1995). Association of the yeast pher-
omone response G protein βγ subunits with the MAP kinase scaf-
fold Ste5p. Science 269, 1572–1575.
Whitmarsh, A.J., and Davis, R.J. (1998). Structural organization of
MAP-kinase signaling modules by scaffold proteins in yeast and
mammals. Trends Biochem. Sci. 23, 481–485.
Wimley, W.C., and White, S.H. (1996). Experimentally determined
hydrophobicity scale for proteins at membrane interfaces. Nat.
Struct. Biol. 3, 842–848.
Wong, D.H., Corbett, A.H., Kent, H.M., Stewart, M., and Silver, P.A.
(1997). Interaction between the small GTPase Ran/Gsp1p and
Ntf2p is required for nuclear transport. Mol. Cell. Biol. 17, 3755–
Yablonski, D., Marbach, I., and Levitzki, A. (1996). Dimerization of
Ste5, a mitogen-activated protein kinase cascade scaffold protein,
is required for signal transduction. Proc. Natl. Acad. Sci. USA 93,
Yu, J.W., Mendrola, J.M., Audhya, A., Singh, S., Keleti, D., DeWald,
D.B., Murray, D., Emr, S.D., and Lemmon, M.A. (2004). Genome-
wide analysis of membrane targeting by S. cerevisiae pleckstrin
homology domains. Mol. Cell 13, 677–688.