A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by D. Alessi (Dundee, U.K.), T. Cass (Imperial College London, U.K.),
T. Corfield (Bristol, U.K.), M. Cousin (Edinburgh, U.K.), A. Entwistle (Ludwig Institute for Cancer Research, London, U.K.), I. Fearnley
(Cambridge, U.K.), P. Haris (De Montfort, Leicester, U.K.), J. Mayer (Nottingham, U.K.) and M. Tuite (Canterbury, U.K.).
Pleckstrin homology domains: not just for
phosphoinositides
M.A. Lemmon1
Department of Biochemistry and Biophysics, University of Pennsylvania School of Medicine, 809C Stellar-Chance Laboratories, 422 Curie Blvd,
Philadelphia, PA 19104-6059, U.S.A.
Abstract
PH domains (pleckstrin homology domains) are the 11th most common domain in the human genome and
are best known for their ability to target cellular membranes by binding specifically to phosphoinositides.
Recent studies in yeast have shown that, in fact, this is a property of only a small fraction of the known
PH domains. Most PH domains are not capable of independent membrane targeting, and those capable of
doing so (approx. 33%) appear, most often, to require both phosphoinositide and non-phosphoinositide
determinants for their subcellular localization. Several recent studies have suggested that small GTPases
such as ARF family proteins play a role in defining PH domain localization. Some others have described a
signalling role for PH domains in regulating small GTPases, although phosphoinositides may also play a role.
These findings herald a change in our perspective of PH domain function, which will be significantly more
diverse than previously supposed.
Introduction
The PH domain (pleckstrin homology domain), as originally
defined, is the 11th most abundant domain class in the human
genome [1]. It was originally identified in 1993 [2,3] as a
stretch of 100–120 amino acids that appears twice in the
platelet protein pleckstrin [4] and is found in many molecules
involved in cellular signalling, cytoskeletal organization,
membrane trafficking and/or phospholipid modification
[5]. Structural characterization of several PH domains
quickly followed their identification [6]. PH domains have
a core seven-stranded β-sandwich structure, with one corner
capped off by a C-terminal α-helix and another by three
inter-strand loops that were implicated in potential binding
sites. Drawing analogies to the SH2 (Src homology-2) and
SH3 domains, which had been identified in the mid-1990s
as modules that drive key protein–protein interactions in
Key words: GTPase, membrane targeting, PH domain, phosphoinositide, yeast.
Abbreviations used: DAPP1, dual adaptor for phosphotyrosine and 3-phosphoinositides; FAPP1,
PtdIns-four-phosphate adaptor protein 1; GFP, green fluorescent protein; OSBP, oxysterolbinding protein; PH domain, pleckstrin homology domain; OSBP-PH, OSBP PH domain; PLCδ 1 , phospholipase C-δ 1 ; PtdIns(4,5)P2 , phosphatidylinositol 4,5-bisphosphate; SH domain, Src
homology domain; SPR, surface plasmon resonance.
1
email [email protected]
cellular signalling [7], many laboratories sought a common
protein (or peptide) target for PH domains [5]. Several
potential protein targets were identified, but none appeared to
represent an equivalent for PH domains of the SH2 and SH3
domain targets of phosphotyrosine-containing and prolinerich peptide sequences respectively. Instead, studies of PH
domains became focused on phosphoinositides as potential
binding targets after Harlan et al. [8] demonstrated that the
N-terminal PH domain from pleckstrin binds to PtdIns(4,5)P2 (phosphatidylinositol 4,5-bisphosphate).
Subsequent studies have shown that several PH domains
recognize particular phosphoinositides with high affinity and
specificity [9], and this has led to their use [fused with
GFP (green fluorescent protein)] as in vivo probes for phosphoinositide distribution and generation [10]. Several
relatively recent reviews have discussed the structural basis
for this specific phosphoinositide recognition [9,11] and its
physiological importance, particularly in phosphoinositide
3-kinase signalling [12]. Although phosphoinositide binding
of this nature has become the role for which PH domains
are best known, recent studies have shown that it is not a
property of most PH domains. On the contrary, most PH
domains that have been analysed bind phosphoinositides
C 2004
Biochemical Society
Structure Related to Function: Molecules and Cells
Structure Related to Function:
Molecules and Cells
707
708
Biochemical Society Transactions (2004) Volume 32, part 5
with only low affinities (if at all) and with little-to-no
specificity [13–15]. Moreover, most PH domains do not
appear to be independently capable of membrane targeting
in vivo [15]. These and other studies [16] argue that most
of the PH domains may have binding targets other than (or
in addition to) phosphoinositides and that their functions
are much more diverse than previously supposed. We will
restrict our consideration here to PH domains that do not
bind phosphoinositides with high affinity or specificity, and
discuss what is currently known about their functions.
Only one of the 33 Saccharomyces
cerevisiae PH domains binds specifically
to phosphoinositides
We recently reported a genome-wide analysis of PH domains
in S. cerevisiae [15]. The most surprising finding from that
study was that only one of the 33 yeast PH domains functions
as a high-affinity and specific phosphoinositide-binding
module. This is the PH domain from Num1p, which resembles the well-characterized PH domain from PLC-δ 1
(phospholipase C-δ 1 ) [17] in binding specifically to PtdIns(4,5)P2 (K D in the 1 µM range). Of the remaining 32 yeast PH
domains, only six bound to phosphoinositides sufficiently
strongly to be detectable using SPR (surface plasmon resonance) or other quantitative approaches. All six of these
PH domains were quite promiscuous in their interactions.
Apparent K D values ranged from 1 to 20 µM (for the monomeric PH domain), but none showed significant preference
for any one particular phosphoinositide, in sharp contrast
with the PH domains from Num1p and PLC-δ 1 or mammalian PH domains that bind phosphoinositide 3-kinase
products. A further 20 yeast PH domains showed signs of
phosphoinositide binding in a highly sensitive lipid-overlay
blot [13,15,18], but these interactions could not be detected
using SPR, indicating that their K D values were well in excess
of 100 µM. The final six yeast PH domains showed no
evidence of phosphoinositide binding in any assay [15].
Most yeast PH domains are
not membrane-targeted, and
phosphoinositides do not define
the location of those that are
To extend this to an in vivo context, we analysed the ability
of each yeast PH domain to bind cellular membranes using
a membrane-targeting assay in yeast and by observing the
localization of GFP–PH domain fusion proteins in yeast and
mammalian cells. Only 14 of the 33 yeast PH domains showed
evidence of membrane localization, of which only eight
were detectably membrane-targeted as GFP fusion proteins.
The membrane-targeted PH domains included almost all of
those that bound phosphoinositides detectably in SPR assays
and eight domains that bound phosphoinositides with
K D > 100 µM [15]. These findings argue that more than
one-half of all yeast PH domains are not capable of independently targeting to cellular membranes. Moreover, those
C 2004
Biochemical Society
PH domains that are membrane-targeted do not require highaffinity phosphoinositide binding. Studies on several yeast
mutants defective in generating particular phosphoinositides
indicated that these lipids do contribute to membrane targeting of most of the PH domains, but that they do
not define the specific subcellular localization [15]. For
example, two PH domains with identical phosphoinositidebinding specificities (from Osh1p and Skm1p) are found at
quite different intracellular locations (the Golgi and plasma
membrane respectively). Similarly, two PH domains that
target the Golgi (from Osh1p and Osh2p) appear to recognize
different pools of the same phosphoinositide [PtdIns(4)P].
One yeast PH domain appeared to be membrane-targeted in
a phosphoinositide-independent manner. Phosphoinositide
binding appears to play an accessory, but not defining, role
in targeting these PH domains to membranes.
Membrane targeting of PH domains driven
by more than one binding partner
Levine and Munro [19] pointed out, in 2002, that the PH
domain from human OSBP (oxysterol-binding protein) is
targeted to the Golgi apparatus in yeast and mammalian cells.
Its targeting to this organelle is specifically dependent on the
production of PtdIns(4)P at the Golgi; this is despite the fact
that the OSBP-PH (OSBP PH domain) shows no selectivity
for this phosphoinositide in vitro and does not appear to
bind the pool of PtdIns(4)P at the plasma membrane. These
observations suggested that OSBP-PH may have a second
binding partner that is Golgi-specific and that its targeting to
the Golgi is driven by a simultaneous binding to this other
factor and to phosphoinositides [19]. Hence, the apparent
PtdIns(4)P dependence of OSBP-PH localization to the
Golgi might simply reflect the abundance of this particular
phospholipid at the locations where this ‘second factor’ is
found. Using mutated forms of OSBP-PH in which the
predicted phosphoinositide-binding site was mutated, Levine
and Munro [19] detected a residual (non-phosphoinositidedependent) OSBP-PH localization to the Golgi that was
diminished in yeast cells lacking the small GTPase, Arf1p.
A recent study [20] on the sequence-related FAPP1 (PtdInsfour-phosphate adaptor protein 1) PH domain supported
the suggestion that a simultaneous binding to two targets
is essential for specific targeting of this PH domain to the
trans-Golgi. This study also documented pull-down assays
showing that the FAPP1 (and OSBP) PH domains can bind
directly to ARF (ADP-ribosylation factor) in vitro [20].
Our genome-wide analysis in yeast [15] argues that a
similar ‘dual-key strategy’ [16] may also be employed by
numerous other PH domains. Two yeast PH domains (from
Osh1p and Osh2p) were Golgi-targeted and they could
possibly bind Arf1p in addition to phosphoinositides. However, several PH domains are strongly plasma-membranelocalized despite very weak phosphoinositide binding, and
must recognize ‘second factors’ distinct from Arf1p that are
localized at the plasma membrane. Identifying these factors
is an important challenge for the future.
Structure Related to Function: Molecules and Cells
PH domain binding to small G-proteins
is a burgeoning theme
Although many protein targets have been reported for
PH domains since their initial description, physiological
relevance has not been established in many cases [5]. One
important counterexample is the interaction of the βadrenergic kinase PH domain with βγ subunits of heterotrimeric G-proteins, which has now been visualized crystallographically [21]. More recently, a series of studies, including
those outlined above, have implicated small GTPases as PH
domain targets in numerous contexts. These findings lead
to the speculation that there may be a common connection
between PH domains and activated small GTPases, although
it is clear that much more work is required to establish this.
In addition to the reported interaction of the PH domains
from FAPP1 and OSBP with ARF family of proteins (in the
context of Golgi phosphoinositides), two other reports have
provided compelling evidence for GTP-regulated interaction
between PH domains and small GTPase targets. Jaffe et al.
[22] recently reported that the PH domain of hCNK, the
human homologue of connector enhancer of ksr (kinase suppressor of ras), interacts with Rho. This interaction is rather
weak (and may require co-operation with other domains
in CNK or, alternatively, with phosphoinositides), but appears to be selective for the GTP-bound form of Rho. Kim
et al. [23] have also presented evidence to suggest that Etk, a
tyrosine kinase from the Btk family, interacts specifically with
GTP-bound RhoA through the N-terminal Etk PH domain.
A more quantitative analysis has been performed for the
interaction between the N-terminal PH domain of PLC-β 2
and Rac GTPases [24]. Sondek and co-workers [24] analysed
PLC-β 2 binding by 17 different GTP-bound members of the
Rho family, and found that only the active forms of the three
Rac GTPases (Rac1, Rac2 and Rac3) bound significantly to
(and activated) PLC-β 2 . The binding site for GTP-bound Rac
in PLC-β 2 was localized to the PH domain, and the isolated
PLC-β 2 PH domain had essentially the same Rac-binding
affinity (less than 2-fold different) as full-length PLC-β 2
(K D in the range 5–10 µM). These studies add substantial
credibility to the increasing number of reports indicating that
certain PH domains bind small GTPases. By binding only
to the GTP-bound states of the small GTPases, it is possible
that PH domains from CNK, Etk, PLC-β 2 , FAPP1 etc. may
serve as Rho, Rac or ARF effectors. One structural study
has provided a glimpse of how this may be achieved. The
first Ran-binding domain from Ran-binding protein-2 was
found to resemble PH domains very closely in structure [25].
Ran-binding protein-2 binds specifically to the GTP-bound
form of the small GTPase Ran (involved in the regulation
of nucleocytoplasmic transport) using a binding site that
contains its three PH domain variable loops.
Studies on guanine nucleotide-exchange factors have also
shown PH domains binding specifically to nucleotide-free
small GTPases. In the crystal structure of the DH (Dbl
homology)/PH domain fragment from Dbs in complex with
its target, Cdc42 [26], direct contacts between the PH domain
and the DH domain-associated small GTPase are seen, and
disruption of these interactions by site-directed mutagenesis
prevents Cdc42 activation by Dbs [27]. Similarly, the PH
domain from ELMO binds to a Dock180–Rac complex, and
appears to stabilize the nucleotide-free transition state of
Rac, thus promoting exchange of GDP with GTP and Rac
activation [28].
Problems in identifying the protein targets
of PH domains as the ‘second factor’
As outlined above, there is good evidence, in several cases,
for physiologically relevant PH domain–small GTPase interactions. However, identifying and studying these and other
interactions presents several important challenges. As suggested by our analysis of yeast PH domains [15], up to
25–33% of PH domains may be membrane-targeted by
binding simultaneously to both phosphoinositides and small
GTPases (or some other component). In vitro analyses of
phosphoinositide binding alone by these PH domains suggests non-specific and weak (often very weak) interactions,
leading one to doubt its physiological significance. Only
by observing the influence on subcellular localization of
genetically manipulating phosphoinositide levels in yeast cells
could we be sure that these lipids played a part. It is probable
that binding to the ‘second factor’ is similarly weak (but
may be more specific), which will render it very difficult
to detect as a binary interaction in vitro. This has been our
experience with the Rho–CNK interaction, for example, and
may well explain why early yeast two-hybrid and expression cloning studies failed to identify small GTPases as PH
domain-binding partners. What is needed to resolve this
difficulty is a set of approaches in which simultaneous binding of PH domains to both protein (or other) targets and
phosphoinositides can be recapitulated. It may be possible to
achieve this in living yeast cells using the Ras rescue assay
pioneered by Skolnik and co-workers [29] for the analysis
of PH-domain membrane targeting. Doing so with purified
components will be far more cumbersome.
Sequence requirements for
phosphoinositide binding by PH domains
One outcome that we anticipated from our genome-wide
analysis of yeast PH domains was the identification of
examples with unique phosphoinositide-binding specificities.
We reasoned that some of these might differ structurally (and,
therefore, in sequence motifs) from the examples of highaffinity, specific PH domains that are already well understood
[9]. These expectations proved to be incorrect. Rather, we
found just one phosphoinositide-specific PH domain in
the yeast genome (from Num1p), and inspection of the
sequence in its β1/β2 loop region {known to determine phosphoinositide-binding specificity in the PLC-δ 1 , Grp1,
DAPP1 (dual adaptor for phosphotyrosine and 3-phosphoinositides) and other PH domains [9]} suggests that it
will resemble other known examples.
Interestingly, as shown in Figure 1, nearly all PH domains
that showed significant phosphoinositide binding in our
in vitro studies have a similar pattern of basic residues in
C 2004
Biochemical Society
709
710
Biochemical Society Transactions (2004) Volume 32, part 5
Figure 1 Sequence characteristics of PH domain phosphoinositide-binding sites
(A) Alignment of the β1/β2 loop region of all S. cerevisiae PH domains plus three mammalian PH domains of known structure (DAPP1, Grp1
and PLC-δ). Top: the positions of strands β1 and β2 of the seven-stranded sandwich are marked. The vertical red arrows indicate residues with
basic side chains that hydrogen-bond directly with phosphates in the bound inositol phosphate in the DAPP1, Grp1 and PLC-δ 1 PH domains [9].
These residues are shaded red where conserved in yeast PH domains. The yeast PH domains are subdivided into six groups on the basis of the
findings of Yu et al. [15]. The Num1p PH domain binds specifically to PtdIns(4,5)P2 and contains basic residues that are probably arranged (and
may interact) in a manner similar to that seen for other specific PH domains. As described in the text, all the PH domains that showed significant
phosphoinositide-binding affinity by SPR preserve most or all of this pattern. All other yeast PH domains (except for Osh3p) neither bound strongly
to phosphoinositides nor retained this motif. (B) Inositol-(1,3,4,5)-tetrakisphosphate is shown bound to the Grp1 PH domain. Strands β1 and β2 and
the β1/β2 loop are marked. Labelled in red are the three residues marked by red arrows on the top of (A), illustrating their ability to hydrogen-bond
c (2001) Biochemical Society.
with the 1-, 3- or 4-phosphates of the bound headgroup. Adapted from [9] C 2004
Biochemical Society
Structure Related to Function: Molecules and Cells
their β1/β2 loop regions. Although the Cla4p and Skm1p
PH domains have a slightly different arrangement, they
retain a critical arginine residue in the middle of strand β2.
There is only one PH domain (from Osh3p) that contains
the β1/β2 loop sequence pattern coloured red for the
DAPP1 and Grp1 PH domains in Figure 1 yet does not
bind reasonably strongly to phosphoinositides. Thus the
occurrence of this simple motif appears to predict phosphoinositide binding rather well, as clearly indicated by
previous studies of mammalian PH domains [18,29]. The
genome-wide perspective in yeast further argues that all PH
domains with strong phosphoinositide binding use this basic
motif. There appear to be no alternative sequence patterns that
predict phosphoinositide binding or membrane targeting.
Why are PH domains so common?
As mentioned in the Introduction section, the PH domain
is the eleventh most populous domain family in the human
genome [1]. Our analysis of all the yeast PH domains,
together with the accumulated data on mammalian and other
examples, suggest that the PH domain may also be one of
the most functionally diverse domains, at least of signalling
domains. Whereas SH2, SH3, PX and other domains appear
to share a common (or at least closely related) function, this
does not appear to be true for PH domains [30]. The sequence
profiles that define PH domains primarily predict the core βsandwich structure and not any short stretch of sequence
that might define functional epitopes. The β-sandwich PH
domain superfold [6] is also seen in several other guises [5],
including PTB domains (phosphotyrosine-binding domain),
EVH (Ena/VASP homology) domains and the Ran-binding
domain, and was very recently seen in the transcription factor
TFIIH [31]. As argued previously [6,32–34], the core βsandwich/PH domain fold is a stable structure that can serve
as a scaffold for presenting one (or more) of several different
types of binding site. Many different classes of ligands for
these domains should therefore be expected. Conceptually, it
seems more appropriate to relate members of the PH domain
family to immunoglobulin-like domains than to SH2, SH3 or
PX domains. Accordingly, it seems reasonable to argue that
we have only begun to understand the functional properties
of this common domain class.
References
1 International Human Genome Sequencing Consortium (2001)
Nature (London) 409, 860–921
2 Haslam, R.J., Koide, H.B. and Hemmings, B.A. (1993) Nature (London)
363, 309–310
3 Mayer, B.J., Ren, R., Clark, K.L. and Baltimore, D. (1993)
Cell (Cambridge, Mass.) 73, 629–630
4 Tyers, M., Rachubinski, R.A., Stewart, M.I., Varrichio, A.M., Shorr, R.G.L.,
Haslam, R.J. and Harley, C.B. (1988) Nature (London) 333,
470–473
5 Lemmon, M.A. and Ferguson, K.M. (2000) Biochem. J. 350, 1–18
6 Blomberg, N., Baraldi, E., Nilges, M. and Saraste, M. (1999)
Trends Biochem. Sci. 24, 441–445
7 Schlessinger, J. (1994) Curr. Opin. Genet. Dev. 4, 25–30
8 Harlan, J.E., Hajduk, P.J., Yoon, H.S. and Fesik, S.W. (1994)
Nature (London) 371, 168–170
9 Lemmon, M.A. and Ferguson, K.M. (2001) Biochem. Soc. Trans. 29,
377–384
10 Balla, T. and Bondeva, T. (2000) Trends Pharmacol. Sci. 21, 238–241
11 DiNitto, J.P., Cronin, T.C. and Lambright, D.G. (2003) Science STKE Dec 16
(213):re16
12 Vanhaesebroeck, B., Leevers, S.J., Ahmadi, K., Timms, J., Katso, R.,
Driscoll, P.C., Woscholski, R., Parker, P.J. and Waterfield, M.D. (2001)
Annu. Rev. Biochem. 70, 535–602
13 Kavran, J.M., Klein, D.E., Lee, A., Falasca, M., Isakoff, S.J., Skolnik, E.Y. and
Lemmon, M.A. (1998) J. Biol. Chem. 273, 30497–30508
14 Takeuchi, H., Kanematsu, T., Misumi, Y., Sakane, F., Konishi, H.,
Kikkawa, U., Watanabe, Y., Katan, M. and Hirata, M. (1997)
Biochim. Biophys. Acta 1359, 275–285
15 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) Mol. Cell 13, 677–688
16 Itoh, T. and De Camilli, P. (2004) Nature (London) 429, 141–143
17 Ferguson, K.M., Lemmon, M.A., Schlessinger, J. and Sigler, P.B. (1995)
Cell (Cambridge, Mass.) 83, 1037–1046
18 Dowler, S., Currie, R.A., Campbell, D.G., Deak, M., Kular, G., Downes, C.P.
and Alessi, D.R. (2000) Biochem. J. 351, 19–31
19 Levine, T.P. and Munro, S. (2002) Curr. Biol. 12, 695–704
20 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) Nat. Cell Biol. 6, 393–404
21 Lodowski, D.T., Pitcher, J.A., Capel, W.D., Lefkowitz, R.J. and Tesmer, J.J.
(2003) Science 300, 1256–1262
22 Jaffe, A.B., Aspenstrom, P. and Hall, A. (2004) Mol. Cell. Biol. 24,
1736–1746
23 Kim, O., Yang, J. and Qiu, Y. (2002) J. Biol. Chem. 277, 30066–30071
24 Snyder, J.T., Singer, A.U., Wing, M.R., Harden, T.K. and Sondek, J. (2003)
J. Biol. Chem. 278, 21099–21104
25 Vetter, I.R., Nowak, C., Nishimotot, T., Kuhlmann, J. and Wittinghofer, A.
(1999) Nature (London) 398, 39–46
26 Rossman, K.L., Worthylake, D.K., Snyder, J.T., Siderovski, D.P., Campbell,
S.L. and Sondek, J. (2002) EMBO J. 21, 1315–1326
27 Rossman, K.L., Cheng, L., Mahon, G.M., Rojas, R.J., Snyder, J.T.,
Whitehead, I.P. and Sondek, J. (2003) J. Biol. Chem. 278, 18393–18400
28 Lu, M., Kinchen, J.M., Rossman, K.L., Grimsley, C., deBakker, C.,
Brugnera, E., Tosello-Trampont, A.C., Haney, L.B., Klingele, D.,
Sondek, J. et al. (2004) Nat. Struct. Mol. Biol. 11, 756–762
29 Isakoff, S.J., Cardozo, T., Andreev, J., Li, Z., Ferguson, K.M., Abagyan, R.,
Lemmon, M.A., Aronheim, A. and Skolnik, E.Y. (1998) EMBO J. 17,
5374–5387
30 Yu, J.W. and Lemmon, M.A. (2003) Curr. Opin. Chem. Biol. 7, 103–109
31 Gervais, V., Lamour, V., Jawhari, A., Frindel, F., Wasielewski, E.,
Dubaele, S., Egly, J.M., Thierry, J.C., Kieffer, B. and Poterszman, A. (2004)
Nat. Struct. Mol. Biol. 11, 616–622
32 Lemmon, M.A., Ferguson, K.M. and Schlessinger, J. (1996)
Cell (Cambridge, Mass.) 85, 621–624
33 Woolfson, D.N., Evans, P.A., Hutchinson, E.G. and Thornton, J.M. (1993)
Prot. Eng. 5, 461–470
34 Chothia, C. (1984) Annu. Rev. Biochem. 53, 537–572
Received 23 July 2004
C 2004
Biochemical Society
711