transmembrane signaling protocols

Compartsons of SH2-target sequences m tyrosme-phosphorylated proteins such as platelet-derived growth-factor PDGF receptor and the polyoma-virus middle-T antigen indicated that residues

Peptide Recognition Mechanisms

Chi-Hon Lee, David Cowburn, and John Kuriyan

1 Introduction

The formation of specific protem-protein interactions is one of the key mechanisms for signal transduction mediated by tyrosme phosphorylation These intermolecular mteracttons target signaling proteins to particular cellu- lar locations and modulate the enzymatic activities that further propagate the signal A dtstmctive characteristic of the pathways that are mitiated by tyrosme phosphorylation is that target recognition and catalytic activity are usually functions of separate domains within the signaling molecules that participate

m these pathways Each of the signalmg molecules contains one or more of a set of modular peptide-bmdmg domains that are responsible for generating protein-protein interactions Such peptide-recognition domains are modular in both structural and functtonal respects: They are capable of folding correctly when removed from the parent protein, and they can usually recognize their targets even when isolated

The first peptide-recognition modules to be identified were the Src homol- ogy 2 and 3 domains (SH2 and SH3 domains), so named because they share sequence similarity with two separate noncatalytic regions of the Src family tyrosme kmases (1,2) SH2 and SH3 domains are now well-known for their crittcal roles m eukaryotic signal transduction, and they function by recogmz- ing sites that contam phosphotyrosyl residues (for SH2) and prolme-rich sequences (for SH3) (reviewed in refs 3-5)

Several other peptide-bmdmg domains have been discovered recently, and the determinatton of their three-dimensional structures have provided some surprtses The phosphotyrosme bindmg/phosphotyrosine interaction (PTB/PI)

From Methods m Molecular Bfology, Vol 84 Transmembrane Slgnahng Protocols

Edlted by D Bar-Sag1 0 Humana Press Inc , Totowa, NJ

3

domain bmds to phosphopeptides containmg NPXY* motifs (Y*, phosphoty- rosme) (6,7) The architecture and mode of peptide recogmtton of the PTB domains is unrelated to that of the SH2 domains, although both recognize phosphotyrosme Most strikmgly, the architecture and topology of the PTB domams resemble closely that of another signalmg module, the plekstrin homology (PH) domain, although there IS no sequence similartty between these domams (8-10) Furthermore, the newly discovered PDZ domains, which rec- ognize non-phosphorylated peptide sequences at the carboxyl-termmus of ron- channel proteins, have a core topology and peptide-binding mechanism with elements m common with the PTB domams (II) The WW domains, whose structure has been determined recently, represent an alternative mode of recog- nizing prolme-containing motifs when compared to the well-known SH3 domains (12) Again, the SH3 and WW domains are unrelated m sequence or structure

In this chapter, we focus on the structural aspects of these peptide-bmdmg domains, with emphasis on the sequence-specific recogmtton of targets Much

of the discussion is focused on the SH2 and SH3 domains, because more is known about them The PTB and PDZ domains are discussed briefly m the context of their structural resemblance to PH domains Newly characterized domams, such as the WW domam and the 14-3-3 protein, are not discussed

2 SH2 Domains

The SH2 domain was first recognized as a phosphotyrosme-binding module during studies of the mechanisms of viral oncogenes that interfere with cellular signaling (1,13,14) Subsequent experiments demonstrated that an individual SH2 domain binds to specific regions of tyrosme-phosphorylated proteins, such

as particular sequences m the cytoplasmic regions of activated receptor tyro- sme kinases (reviewed m ref 15) The first three-dimensional structures of SH2 domains confirmed that the module corresponds to a well-folded domain with a defined peptide-binding surface (16-18) In addition, the crystal struc- ture of the Src tyrosme kmase SH2 domain complexed with low-affinity phosphotyrosyl peptides revealed the mechamsm of phosphotyrosme recogm- tion that has subsequently been found to be conserved in general terms among all SH2 domains of known structure (18)

Compartsons of SH2-target sequences m tyrosme-phosphorylated proteins such as platelet-derived growth-factor (PDGF) receptor and the polyoma-virus middle-T antigen indicated that residues immediately surrounding the phosphotyrosme determme the binding specificity of SH2 domains (19-22) However, a general picture of SH2-target specificity did not emerge until an exhaustive investigation was carried out using a peptide library approach

Peptide Recognition Mechamsms 5

(23,24) This established that the three residues immediately C-terminal to the phosphotyrosme are the key determinants of specificity The determination of the structures of high-affinity peptide complexes of Src and the closely related Lck-SH2 domains provided the fu-st view of sequence-specific peptide recog- rutron (25,26) By combmmg the structural information with selecttvity data from the pepttde-library study, the sequence preference can be correlated with particular residues in the SH2 domain (23,27) Subsequently, the structures of peptrde complexes of the SH2 domams of the tyrosme phosphatase SH-PTP2 (28), phospholipase C-~(29) and the adapter protems GRB2 (30) and She (31) have further clarified the mechanism of peptide recognition and have extended our understanding of SH2 specificity

An additional level of complextty was added when the brochemtcal and structural analysts extended toward larger components of signaling molecules, containing more than one domain Structures of the adapter-protein GRB2 (32) and the regulatory unit of Abl tyrosme kinase (33) have provided insights into spatial arrangements of multiple domains Furthermore, structural analysis of multi-domain constructs of ZAP-70 (34), Lck tyrosine kmase (35), and the tyrosine phosphatase SH-PTP2 (36) revealed the cooperatrve recogmtton

of peptides by larger-signaling molecules of which these domains are compo- nent parts

2.1 General Architecture

The SH2 domain is a compact a-@-structure comprised of around 100 residues (see Fig 1 for a sequence alignment) The central scaffold is an anti- parallel P-sheet formed by strands A, B, C, D, and G Two a-helices, aA and

aB, flank the central P-sheet (see Fig 2 for a schematic diagram and the nota- tion used) This P-sheet runs perpendicular to the peptide-binding surface, and divides the domain mto two functronally distinct regions One region, com- prrsmg helix aA, loop BC (the phosphate-binding loop), and the adjacent face

of the central P-sheet, provides resrdues that interact with the phosphotyrosme The other region includes helix aB, loops EF, BG, and the other face of the central P-sheet, and interacts with pepttde resrdues immediately followmg the phosphotyrosme; this regton accounts for the sequence-specific recognition The peptide ligand lies across the surface of the domain approximately orthogonal to the central P-sheet (Fig 2) The peptide ligands are usually in an extended conformatron and do not participate m secondary-structure forma- tion with the domain The phosphotyrosine residue appears to be the main anchor point of the SH2-peptlde complex, allowing the domain to read out the three to six residues immediately followmg the phosphotyrosme The peptide residues N-terminal to the phosphotyrosine make limited and nonspectftc mter-

AB pB BC PC CD PD PD’

EG3

II

HGQLKE KNGDVI

WC G IDVYIIGG IRRFIS lsLsDLIGYVsHVl SCLL KGE KLL IYP I

Fig 1 Alignment of SH2 sequences and defmmon of the residue notation The sequences of different SH2 domains are aligned, based on the secondary-structure definitions of Src and Lck (26) The boundaries of the secondary structural elements

of Src are shown by solid boxes, and the notation for these elements is shown sche- matically at the bottom The important residues are mdrcated by vertical lines at the top (Adapted with permrssron from ref 28.)

6

Peptide Recognition Mechamsms 7

actions with the domain, and therefore most likely contribute little to the bmd-

mg specificity The N- and C-termmi of the SH2 domain are located on the side of the domain opposite to the peptide-bmdmg surface For this reason, the domain can be readily inserted into different molecular contexts without affecting the peptide-binding ability

2.2 Peptide-Binding Specificity and Affinity

Several lines of evidence indicate that different SH2 domains bind to distinct phosphotyrosme contammg sites of their target proteins m vivo, and that the linear sequence surrounding the specific phosphotyrosine determines the binding specificity (19-22) To illustrate, a point mutation (Tyr 739 to Phe) in the PDGF receptor selectively elimmates the binding of the Ras GTPase activating protem (GAP) to the activated receptors, but the bmdmg of other SH2-containing proteins (such as PLC-y and PI-3 kmase) remains intact (37) It appears that the local sequence, rather than the ter- tiary structure, of the SH2-targets dominates the binding specificity Tyrosine-phosphorylated peptides that contain sequences resembling the local sequence of the target protein (the Tyr 739 of PDGF receptor in this case) compete efficiently for the bmdmg of the target protein (PDGFR) to a particular SH2 domain (GAP) (37) In addition, the observation that a mutant PDGF receptor contammg a deletion near the GAP-SH2 binding site binds to the GAP-SH2 domain with nearly the same affinity as the wild- type PDGF receptor suggested that the tertiary structure is not a primary factor m determmmg bmdmg affmlty (38) These observations establish the relevance of studies usmg isolated peptides

A systematic search for optimal peptide sequences for SH2 domains had been carried out by screening a random phosphopeptide library (23,24) Of over 20 different SH2 domains tested, each showed distinct selectivity m the three residues immediately C-terminal to phosphotyrosme in the peptide ligand Such sequence preference could be correlated with the side-chains of residues

at several critical positions of the SH2 domain (24) The clearest example of this correlation is provided for the residue at the PDS position of the SH2 domain, which contacts the peptide side chains at position +l and +3 Certam SH2 domains, including Src-family tyrosme kmases as well as GAP and the adapter proteins GRB2 and Nck, have aromatic residues at pD5, and preferen- tially bmd to pepttdes contammg polar side chains at +l In contrast, other SH2 domains (~85, phosphohpase C-y, the tyrosme phosphatases) contam hydro- phobic side chains at pD5, and select for hydrophobic residues at +l

Quantitative analysis using isothermal-titration calorimetry and surface- plasma resonance (39) mdicated that the SH2-peptide mteraction is of only moderate strength (Kd -0.1-3 0 pM> compared with strong mteractions

A

Fig 2 (see also facing page) Schematic diagram of two SH2-peptide complexes (A) The Src-YEEI complex and (B) the N-terminal SH2 domain of SH-PTP2 complexed with a peptide derived from Tyr 895 of IRS-l, The view is from the pep- tide-binding surface and illustrates the secondary-structure elements and the notation used The peptide is shown in a ball-and-stick representation and comprises phosphotyrosine (p-Tyr), residue +l, residue +2, and so on ol-helices and P-strands are shown as ribbons and arrows, respectively

such as those between transcription factors and their specific DNA targets (Kd ~1 nil4) The phosphotyrosine is absolutely required for binding to SH2 domains (40)

Peptide Recognition Mechanisms

Fig 2

Peptide residues immediately following the phosphotyrosine (+ 1 to +6) are the critical determinants for binding to individual SH2 domains; however, a varying range of amino acids are tolerated at each site Although the selectivity

of individual SH2 domains is not sharply defined, the specificity and affinity can increase dramatically when cooperative binding interactions occur (see discussion Subheading 5.1 for tandem SH2 domains of ZAP-70) Kinetic analysis of SH2-peptide interaction has shown that the association and disso- ciation rates (k,,, and I&) are both very rapid even for high-affinity peptide ligands (41) Fast turnover rates could allow the rapid sampling of different binding sites and are observed for many protein-protein interactions involved

in signal transduction

2.3 Recognition of Phosphotyrosine

The recognition of phosphotyrosme 1s the defmmg feature of the SH2- pepttde interface Although the details vary slightly from one SH2 complex to another, the overall features of the mteractron are strikmgly conserved Rest- dues from aA, PB, PD, and the BC loop form the phosphotyrosine-bmdmg pocket and provide hydrophobtc mteractions with the phenoltc ring of phosphotyrosme and hydrogen-bonding interactions with the phosphate group

(Figs 3A,B)

The most critical interaction wtth the phosphotyrosine is provided by Arg PBS, which forms a bidentate-tome interaction with the phosphate group This arginine is located at the bottom of the binding pocket and becomes completely maccessible to solvent upon binding Arg PB5 is strictly conserved m all SH2 domams, and even the conservative mutation of this residue to lysine abolishes bmdmg (42) With the backbone of the phosphotyrosme residue held m post- non by the outer strand of the central P-sheet (PD), the ionic mteraction between the phosphate group and Arg PBS provtdes a stereochemical “ruler” that appears to

be the key for discrimmatmg between phosphotyrosme and other restdues The location of Arg PBS is such that, in a fully extended conformation, this side chain IS Just long enough to interact with the phosphate group of a fully extended phosphotyrosme side chain, thus excludmg phosphoserme or phosphothreomne

An interesting feature often observed in the SH2-peptide complexes is the presence of an ammo-aromatic interaction between an ammo nitrogen of Arg

aA and the phosphotyrosine rmg (18) Ammo-aromatic interactions have been observed in a number of protein structures as well as m some small molecules structures (43) The ammo mtrogen of this argmme hydrogen bonds with the phosphate group and the backbone-carbonyl group of the peptide These mter- actions mediated by Arg aA appears to be optimal for phosphotyrosme and were first identified m the Src (18) and Lck structures (26) and later m other SH2-peptide structures, including ZAP-70 (34) However, the SH2 domains of the tyrosine phosphatases do not have Arg aA (it is replaced by glycme) In the SH2 structure of the phosphatase SH-PTP2 (28), the phosphate group rotates by -180”, facmg toward the BC loop (Fig 3B), and the number of hydrogen bond with the phosphate group is almost the same as m Src or Lck The ammo-aromatic mteraction is also not seen m other SH2 structures (such

as p85 ref 44) even when Arg aA 1s present

2.4 Peptide Recognition

The structures of the closely-related Src and Lck SH2 domams (25,26) m complex with a high-affmtty pepttde containmg the Tyr-Glu-Glu-Ile (YEEI) motif provided the first piece of structural mformation on sequence-specific

Peptide Recognition Mechanisms 11

Fig 3 Stereoviews of the phosphotyrosine binding sites of (A) Src and (B) N-terminal SH2 domains of SH-PTP2 The polypeptide backbone of the peptide is shown as a tube and the phosphotyrosine side chain is shown in black Hydrogen bonds are indicated by dashed lines (Adapted with permission from ref 28.)

recognition by the SH2 domain (Fig 3A) In these structures, the peptide binds

to the SH2 domain in an extended conformation and the interaction resembles

a two-pronged plug (the peptide) engaging a two-holed socket (the SH2 domain) The two prongs refer to the phosphotyrosine and the Ile +3 residue of the, peptide, which fit into the corresponding pockets on the SH2 surface This type

of interaction is also observed in several other SH2-peptide complexes

The most important feature m the Src and Lck structures is that the Ile +3 residue of the YEEI peptlde engages a well-defined hydrophobic pocket of the SH2 domain This interaction 1s responsible for the selection of large hydro- phobic residues at +3 posltlon The residues lmmg this pocket (which arise from PD, DE, loops EF and BG) are rather divergent In particular, the two variable loops, BG and EF, shape the surface topography of this pocket Muta- tions as well as large msertlons/deletlons are often found in this region, and these have been shown to be important for binding specificity The glutamate residues at +l and +2 do not form extensive interactions with the SH2 domain, but are m the vlcmlty of basic residues that may account for the moderate selectivity against basic residues at these positions in the peptides (Fig 2A) The prototypical two-pronged Interaction 1s also observed in two X-ray structures of the N-terminal SH2 domain of the ~85 subunit of PI-3 kmase (p85N), complexed separately with two high-affinity peptldes containing the optimal Y-M/V-X-M motif (44) Although m p85N SH2 the position of Met +3 shifts shghtly toward the central P-sheet, the interaction between this residue and the hydrophobic pocket 1s similar to that seen m Src and Lck This resem- blance is expected, because both SH2 domains favor large hydrophobic residues at this posltlon (although p85N shows a higher preference for Met) A unique feature of the p85N SH2 domain 1s that this hydrophobic pocket 1s blocked by the side chain of Tyr BG5 m the absence of hgand, and this side chain has to move by 8 8, to open up the pocket for hgand binding The large movement of the Tyr side chain might account for the changes in circular dlchroism and fluorescence spectra that had been noted upon peptide bmdmg, because no other large-conformatlonal change induced by peptide binding was found

A notable difference in the binding speciflcltles of p85N and Src 1s that, at the +1 position, p85N SH2 prefers hydrophobtc residues, whereas Src SH2 favors nonbasic polar residues Ile PDS (m p85N) appears to be the major determinant for this difference, since replacement of this residue by Tyr (found

in Src) shifts the selectlvlty toward that of the Src-family SH2 domains (27) In the p85N structure, the less bulky side chain of Ile pD5 opens up a shallow-

Peptide Recognition Mechanisms 13 hydrophobic pocket for housing a hydrophobic residue such as Met or Val at the +1 position Likewise, the SH-PTP2 and PLC-y SH2 domains (which have Ile and Cys at PDS, respectively) also prefer a hydrophobic residue at the +l position In contrast, the bulky side chain of Tyr pD5 closes up this hydropho- bic pocket m Src and Lck

2.4.2 Type 2: SH-PTP2 and PLC-y

The crystal structures of the N-terminal SH2 domains of the SH-PTP2 tyrosine phosphatase have been determined in separate complexes with two high-affinity peptides A distinctive feature in these structures is that five resi- dues following the phosphotyrosine of the peptide run through a hydrophobic groove on the SH2 domain (28) Mutagenesis studies confirmed the strong selectivity for hydrophobic residues at the +5 position (28); truncating the pep- tide or replacing the residue at +5 with a hydrophilic residue completely abol- ishes its interaction with the SH-PTP2 SH2 domain (45) This selectivity was unexpected because this residue had not been randomized in the peptide-library study, and consequently this enhanced selectivity had not been predicted The differences m surface topography for the SH2 domains of Src and SH-PTP2 arise from the opening of pockets for housing peptide residues +l and +5 The presence of a shallow-hydrophobic pocket for +l m SH-PTP2 is primarily owing

to the less bulky residue (Ile) at the PBS position, The +5 binding site is flanked

by the two variable loops, EF and BG, and these are opened up relative to their positions m Src and Lck Such significant changes could perhaps have been anticipated by examining the primary sequences of the SH2 domains, but the precise structural details would be difficult to model without mformation from crystallography or nuclear magnetic resonance (NMR)

The binding of peptide to the C-terminal SH2 domain of phospholipase C-y1 (PLC-y) also mvolves extended interactions in a surface groove (29) The solution structure of this SH2 domain has been determined, revealing an interaction involving six peptide residues following the phosphotyrosine Sur- prismgly, a bmdmg study mdicated that residues at the +2 to +6 positrons con- tribute little binding energy, although they make extensive contacts with the PLC-y-SH2 domain; truncation of the peptide to just the three residue DYI resulted in only a 15-fold reduction m binding affinity (46) The discrepancy between the structural results and the binding data has been further investi- gated by examining the changes m the dynamic properties of the SH2 domain upon peptide binding (46) This analysis demonstrates that the residues con- tacting the phosphotyrosine (which contribute to binding energy) undergo a sigmficant restriction m dynamic flexibility upon binding, whereas the resi- dues interacting with C-terminal end of the peptide (which contribute little

binding energy) do not The results have general implications for studying molecular interactions It has been suggested that analysis of dynamic behav- ior m response to bmding could be used to distinguish the residues that contrib- ute most to bmdmg energy (46)

2.4 3 Type 3 GRl32/Sem-5

Most SH2 domains show strong selectivity for particular side chains at the +l and +3 positions of the peptide, and the reason for this can be readily under- stood from the structural data High-affinity peptides usually bmd to SH2 domains m an extended conformation and, as a result, significant contacts are only made by odd-numbered residues, including those at the +l, +3, and +5 positions The GRB2 SH2 domain is unique m that it selects most strongly for

an Asn at the +2 position, with the optimal motif being YVNX (24) This unusual target specificity is owmg to the presence of a Trp residue at the EFl

position; strikmgly, replacement of Thr EFl by Trp m the Src SH2 domam shifts the binding specificity of Src SH2 toward that of GRB2 (47) The recently determined structure of a GRB2 SH2-peptide complex reveals an unusual mode

of peptide recognmon and explains the binding selectivity (30)

In the GRB2-peptide structure, the side chain of Trp EFl fills up the +3-bmdmg site and thus prevents the peptide from binding in an extended con- formation Instead, the peptide residues YVNV adopt a p-turn conformation, which is stabilized by a hydrogen bond between the carbonyl group of the phosphotyrosine residue and the backbone-amide group of Val+3 of the pep- tide In addition, the carboxamide oxygen of the Asn +2 side chain forms a hydrogen bond with the backbone-amide group of Lys PD6, whereas the mtro- gen atom of the Asn side chain is hydrogen bonded to the carbonyl groups of Lys PD6 and Leu PE4 The selection for Asn at +2 1s because of the formation

of these hydrogen bonds that are specific for asparagine, and replacement

of the Asn at the +2 position by Glu completely abolishes the bmdmg (30)

3 PTB, PH, and PDZ Domains

The PTB/PI interaction domain is a component of the Src homology 2/collagen homology (SHC) adapter protem, shown to bind phosphotyrosyl peptides in a manner different from that of the SH2 domains (Fig 4) (6,7,48)

This domain m SHC is specific for NPXY motifs, and the N-terminal selectrv- ity is mconsistent with all known modes of SH2/phosphopeptide mteractions The solution structure of the SHC-PTB domain m complex with a ligand first revealed the unique nature of the PTB-peptide mteractron (IO), and this was further illustrated by the solution and crystal structures of the PTB domam of msulm-receptor substrate- 1 (IRS- 1) in complex with hgands (8,9) The general

Peptide Recognition Mechanisms

Fig 4 Altgnment of PTB domain and PH domam sequences The sequences of SHC-like PTB domains (mcludmg SHC, Xl 1, dNumb, mP96) and IRS-l PTB domam are aligned according to the secondary structure The sequences of PH domains (PLCG and Dynamm) are listed for reference The boundaries of the secondary structural elements of

Xl 1 are shown by solid boxes, and the notation for these elements is shown on the top The secondary structures for experimentally determined domains are Indicated by shadows Noted there is no detectable sequence homology among the three groups The residues m SHC and IRS-l that interact with phosphotyrosme are marked with asterisks at the top (Adapted with permission from ref 8.)

architecture of the two domains is similar, with the SHC-PTB domain some- what larger and more complex (Fig SA) In both cases, the core architecture and topology of the protein fold is similar to that of the PH domain, a signaling module with various functions (4) In this PH-domain superfamily, two medium-size P-sheets pack against each other with inter-strand angles of about 60”, and a C-terminal a-helix lies along one edge of these sheets PTB domains mcorporate their target ligands into the structure by extending one of the P-sheets, using antiparallel hydrogen-bonding mteractions The NPXY motif, characteristic of the PTB-domain ligands, is at the C-terminus of this P-strand, and forms a p-turn

In the case of SH2 domains, it had been relatively easy to identify sequence homology, and the recognmon of phosphotyrosme proceeds by a mechanism that is common to all SH2 domains of known structure The PTB domains do not, however, have such a conservation of sequence, even though the same structural mechanism for peptide recognition is used For example, m the SHC- and IRS-l-PTB domains, different sets of residues recognize the phosphotyrosine In addition, the Xl 1- and FE65PTB domains bmd to NPXY

the phosphotyrosine represented in a ball-and-stick model The figure is generated using MOLSCRIPT (81) (B) Schematic diagram indicating the similarity in topology and peptide binding between FTB domains and PDZ domains (Adapted with permis- sion from ref II.)

16

Peptide Recognition Mechanisms 17

motifs in the Alzheimer-precursor protein in the absence of phosphorylation

A crystal structure of this hgand bound to the X 11 -PTB domain reveals a bind- ing mode essentially similar to that of the phosphorylated hgands to the SHC- and IRS-l-PTB domams (Zhang et al , to be submitted) Some PH domains have clear protein-protem interactions, mapped to a similar area of that molecule by fragment-expression methods, and identified as part of the extension of the highly positively charged sheet/helix interface by NMR (Fushman et al., to be submitted)

In general, the peptide-recognmon mechanisms used by the SH2 and PTB domains differ in two major ways First, the phosphotyrosine is deeply buried and tightly coordmated m the SH2 domains This appears to be less so for the PTB domains Second, the N-terminal region of the PTB ligands forms extensive backbone contacts with the PTB domains No such interactions are observed in the SH2-ligand complexes

Recent studies of the PDZ domain have revealed a similar core architecture These small (-100 residues) domains are components of several protems that are involved m synaptic junctions, and they bmd to short nonphosphorylated sequences at the C-termml of Shaker-type potassium channels, and N-methyl-o-aspartate (NMDA) receptor-ion channels The crystal structure of the peptide complex of one PDZ domain has recently been reported (II) The peptide ligand is bound m a manner similar to that seen m the PTB domains, with the formation of antrparallel hydrogen bonds with the peptide, which packs against a C-terminal a-helix of the domain (Fig 5B)

analyses of peptide complexes revealed that the peptide can bind to the SH3 surface in two orientations, depending on the particular sequence of the ligand, providing an additional level of specificity (56-59) Biochemical and struc- tural analyses of the interaction between the HIV- 1 Nef protein and Src family SH3 domains has revealed how tertiary interactions can further augment the binding affinity and specificity of SH3 domains (60,61)

4.1 PxxP Peptide Recognition

The helical PP-II conformation adopted by SH3 ligands exhibits threefold pseudosymmetry in cross-section When bound to the SH3 domain, two of its three edges provide six peptide residues (P-s, P-*, Pa, P-t, P,,, and P+3) that fit into corresponding binding pockets on the SH3 surface (see Fig 6 for the notion used) The interface is pseudosymmetrical (P-t and Pa are equivalent, and so are P,,, and P+3), and an interesting consequence is that the peptide can bind in two opposite orientations, referred to as plus and minus (5657) (Fig SA) The particular orientation utilized is determined primarily by an ionic interaction between a conserved-acidic residue of the SH3 domain (labeled g in Fig 6) and a basic residue (usually an Arg) at the P-, position of the peptide Peptide resi- dues at the P-r, P,, P+2, and P,, positions interact with the hydrophobic-binding surface of the SH3 domain and are usually proline or other hydrophobic residues The pseudosymmetry of the PP-II helix breaks down in the presence of nonproline residues in the helix (Fig SB) Nonproline residues at one of the two edges can pack tightly against the SH3 surface, whereas nonproline resides

at the other edge cannot The selection of proline residues over nonproline at certain positions is linked to the orientation of the bound peptide To illustrate, when a peptide binds to the SH3 domain in the minus orientation, nonproline residues are tolerated only at one edge (corresponding to positions, P-i and P+*), because at the other edge (PO and P+j) the side chain of the nonproline residue would extend away from the binding surface (Fig SB) Thus, proline residues are required at one edge in one orientation but at the opposite edge in the reverse orientation, leading to the PxxP motif Peptides containing the motif

“PxxPxR” are likely to bind to SH3 domains in the “minus” orientation, whereas peptides containing the motif “RxxPxxP” will bind in the “plus” orientation (5657)

4.2 Binding Affinity and Specificity

The binding of PxxP-containing peptides to SH3 domains is rather weak in general, with dissociation constants around 2-50 pM (53) This may be because

of the relatively small interface area (typically -400 A*) between the peptide ligand and the SH3 domain Residues of the SH3 domain that interact with the

Peptide Recognition Mechanisms 19

distal loop

KDE NTE

ES SSE YNH KPE GSL(

STN KCS KKG ELE VLE

DPH

GD

GE

GD NGE

EQ :14)EIG

KD

DG QQG

As discussed above, one reason for a lack of specificity in many SH3-peptide interactions is owing to limited variation in the interaction sur- face Selections with longer peptide segments and nonnatural analogs have demonstrated that higher affinity and selectivity can be obtained by exploiting

n

Fig 7 SH3-peptide interaction, for the GRB2/Sem-5 SH3 domain complexed with

a peptide bound in the minus orientation (57) The bound-PxxP peptide is shown in a ball-and-stick representation and the peptide residues are labeled according to Fig 8A The conserved hydrophobic residues that form the peptide-binding surface and the acidic residue that interacts with the Arg at P-3 of the peptide are shown in light- gray stick

nonconserved regions of SH3 (66) However, it is likely that many SH3 domains can only achieve high affinity and high specificity by cooperative and/or tertiary interactions

4.3 Tertiary interactions Between HIV-1 Nef

and Src-Family SH3 Domains

HIV-l Nef is an early gene product of immunodeficiency viruses (including HIV-l, HIV-2, and SIV) and is essential for AIDS pathogene- sis (reviewed in ref 67) The molecular mechanism by which Nef promotes disease progress is not known in detail, although several lines of evidence suggest that Nef might function by interacting with cellular-signaling pro- teins Nef contains an invariant PxxP motif which is critical for optimal viral replication and has been shown to mediate specific interaction with certain Src familySH3 domains (68)

C-

- P-

21

The interaction between Nef and the SH3 domain of Hck (which is a member

of the Src family of kinases) is of high affinity and specificity (61) The affinity

is among the tightest known for SH3-ltgand mteractrons (Kd = 0.25 @t4), More- over, Nef is able to discriminate between the Hck-SH3 domam and the closely related Fyn-SH3 domain, with a selectivity of over loo-fold Mutagenesis indicates that the differential bmdmg to Nef IS mediated by a single ammo acid

m the RT-loop of the SH3 domain Interestingly, the high affnnty and high specificity are only evident for the folded-Nef protem, because a peptide corre- sponding to the Nef-PxxP motif binds to SH3 domains only weakly

The crystal structure of the conserved core of HIV-l Nef m complex with a mutant (R961) Fyn-SH3 domain (to which Nef bind tightly) has been deter- mmed recently (60) The Nef-PxxP motif adopts a PP-II-helical conformation, which interacts with the SH3 domain m a manner resembling closely the mter- actions between the SH3 domain and isolated peptides The Nef-PxxP motif forms a PP-II helix even in the absence of the SH3 domam, as revealed by the solution structure of Nef in an unliganded form (69) The SH3 domain thus Interacts with a preformed PP-II helix on Nef, which augments the bmdmg affinity by reducing the entropic penalty for forming the PP-II helix

A striking feature of the structure of the complex is that the interface of Nef with the SH3 domain includes elements that are distmct from the PxxP motif of Nef Most important of these is a hydrophobic pocket on the surface

of Nef that engages an isoleucine residue on the RT-loop of the SH3 domain

(Fig 9) It is this mteraction that allows Nef to distinguish between closely related SH3 domains The Ile-binding pocket is formed by the antiparallel arrangement of two a-helices that follow the PxxP motif m sequence and bracket it m the tertiary structure of Nef

The observation that the RT-loop of the SH3 domam contributes to bmdmg specificity is not entirely unexpected The RT-loop is very divergent among different SH3 domams, and it borders the peptide-bmdmg surface of the SH3 domain It has been shown that the RT-loop plays an important role m sub- strate binding and auto-inhibition of Src-family tyrosine kmases (70-72) It is likely that other SH3 domams also uttlize the drvergent regions, such as the RT- and n-Src loops, to enhance the binding specificity in combination with the conventional PxxP-SH3 interaction

5 Cooperative Interactions of Tandem SH2 Domains

5.1 Interaction of ZAP-70 with the ITAM Motif

The activation of the T-cell receptor complex inmates a series of signaling events that are critical for T-cell function (reviewed in ref 73) One such event mvolves the tyrosme-phosphorylation of the cytoplasmrc regions of the c-chain

of CD3 complexes that contain immunoreceptor tyrosme activation motifs

Peptide Recognition Mechanisms 23

(ITAM) The ITAM motifs contain the sequence Y-X-X-L/l-X,-s-Y-X-X-L/I,

and are phosphorylated on both tyrosmes upon receptor stimulatron, resulting

m the formation of two contiguous SH2-binding motifs The activated ITAM motifs serve as docking sites for several SH2-containing signalmg molecules, particularly the ZAP-70/Syk tyrosine kinase family

ZAP-70 consists of two SH2 domams connected by a 65-residue linker (the inter-SH2 region), followed by the catalytic kinase domain The interaction between the tandem SH2 domains of ZAP-70 (ZAP-NC) and an ITAM are cooperative and of high affinity, for which both SH2 domams are required (74) A peptide with an ITAM motif that is phosphorylated on only one of the two tyrosines binds to ZAP-NC with an affinity that is 100-1000 times weaker than that of a doubly phosphorylated peptide (75) The crystal structure of the tandem SH2 domains of ZAP-70 complexed with a peptide contaming a com- plete ITAM motrf has been determined, which reveals the molecular basis for the cooperative binding (34)

The ZAP-NC-ITAM complex IS Y-shaped, with the 65 residues of the inter-SH2 region forming a coiled coil structure that is the stem of the Y The two SH2 domains (ZAP-N and ZAP-C) are positioned side by side and form the two upper branches of the Y A contrguous peptide-bmdmg surface is formed by the adjacent SH2 domains at the tips of the Y The peptide binds to the ZAP-NC SH2 domains m a head-to-tail orientation with the N-terminal phosphotyrosine of the ITAM motif bound to the C-terminal SH2 domain (ZAP-C) Both SH2-docking sites of the ITAM motif bmd to the correspond- ing SH2 domains in a manner resembling the prototyptcal two-pronged mter- action, with one important difference: the binding site for the C-terminal phosphotyrosine of the ITAM motif is composed of resrdues from both SH2 domains This explains why the ZAP-N SH2 domains fails to bmd to phosphopeptides as an isolated domain, The linker region (7-8 residues) between the two SH2-binding motifs of the ITAM adopts a helical conforma- tion and provides an appropriate spacmg that is critical for specificity; msertion

or deletion of two or more residues in this region abolishes the cooperatrve Interaction

The total surface area burred between the ZAP-NC SH2 domains and the ITAM peptide is relatively large (around 1300 A*), which may account for the high affinity of the mteraction In contrast, the interface between the two SH2 domains is relatively small (-200 A2), suggestmg that they are held m position

by the bound peptide and the inter-SH2 coiled-coil

5.2 Tandem SH2 Domains of SH-PTP2-Tyrosine Phosphatase

The SH-PTP2 tyrosine phosphatase belongs to a group of nonreceptor tyrosine phosphatases that contam two tandem SH2 domains followed by a

disordered lo’ip

Fig 9 Nef-SH3 interaction The SH3 domain is shown in white, and the Nef protein is in gray The invariant prolines of the Nef-PxxP motif and the critical Ile in the RT-loop of the SH3 domain that determine the binding specificity are shown in ball-and-stick representation

phosphatase catalytic domain The SH2 domain(s) of SH-PTP2 have been shown to mediate the interaction of the phosphatase with activated receptor tyrosine kinases (76,77) and to downregulate the phosphatase enzymatic activ- ity (78-80) Occupation of either SH2 domain by phosphopeptides stimulates the phosphatase activity, with much more potent activation resulting from peptides that contain double SHZbinding sites The crystal structure of the tandem SH2 domains of SH-PTP2 complexed with phosphotyrosyl peptide has

Peptide Recognltlon Mechanisms 25 been determined (36) In contrast to the structure of the ZAP-70 tandem SH2 domains, m which the two peptlde-binding sites are in a linearly contmuous

arrangement, in the SH-PTP2 structure the two peptlde-binding sites are widely separated (by about 40 A) and are antlparallel to each other Phosphotyrosyl peptldes bmd to each SH2 domam as observed in the single SH2-peptide com- plex discussed above (28)

Although only a relatively small hydrophobic interface was found between two SH2 domains, the relative orientation of two SH2 domains appears to be

rigidly constrained by the presence of a buried dlsulflde bond that links two

domams Dlsulflde bonds m cytoplasmlc proteins are unusual, but not without

precedent A possible function of this fixed orientation may be to position the

tandem SH2 domains for mteractlon with phosphotyrosyl groups that are spaced appropriately in dlmerlc activated receptors

6 Conclusions

Each of the peptlde-bindmg modules described here binds to its target pep- tides utlllzmg a conserved mechanism Although tertiary interactions appear to

be important, the primary determinants of specificity appear to be the linear

sequence elements of the targets Thus, both the peptide-recogmtlon domains

and their ligands are modular and self-contained, which allows for the con- structlon of large slgnalmg molecules that integrate multiple domains and bmd-

mg sites within them The “modular design” is economical and efficient, and

might account for the wide utllizatlon of peptide-recogmtlon mechanisms in

various cell processes such as cell signaling and sorting

References

1 Sadowskl, I , Stone, J C , and Pawson, T (1986) A noncatalytlc domain con- served among cytoplasmlc protein-tyrosme kmases modifies the kmase function and transforming activity of fuJmam1 sarcoma virus p130@‘g-fps A4ol Cell Bd 6,

4396-4408

2 Moran, M F., Koch, C A., Anderson, D , Ellis, C , England, L., Martm, G S., and Pawson, T (1990) Src homology region 2 domams direct protein-protein mteractlons in signal transduction Proc Nut1 Acad SLJ USA 87,8622-8626

3 Pawson, T and Schlessmger, J (1993) SH2 and SH3 domains Curr Bzol 3,

7 Blarkre, P et al (1994) A region m She drstmct from the SH2 domain can

32,03 l-32,034

8 Eck, M J , Dhepaganon, S , Trub, T., Nolte, R T., and Shoelson, S E., (1996) Structure of the IRS-l PTB domain bound to the Juxtamembrane region of the Insulin receptor Cell 85,695-705

9 Zhou, M -M , Huang, B., GleJniczak, E T , Meadows, R P , Shuker, S B , Miyazakt, M., Trub, T., Shoelson, S E., and Fesrk, S W (1996) Structural basis for IL-4 receptor phosphopeptrde recogmtron by the IRS-l PTB domam Nature Struct Blol 3, 388-393

10 Zhou, M M., Ravichandran, K S , OleJnlczak, E F , Petros, A M , Meadows, R P , Sattler, M., Harlan, J E , Wade, W S , Burakoff, S J., and Fesrk, S W (1995) Struc- ture and hgand recognitron of the phosphotyrosme bmdmg domain of She Nature 378,584592

11 Doyle, D A , Lee, A., Lewis, J., Kim, E., Sheng, M , and MacKmnon, R (1996) Crystal structures of a complexed and pepttde-free membrane protein-bmdmg

1067-1076

12 Chen, H I and Sudol, M (1995) The WW domain of Yes-assoctated protein binds a prolme-rich hgand that differs from the consensus establtshed for Src homology 3-binding modules Proc Nat1 Acad Scz USA 92,7819-7823

13 Matsuda, M , Mayer, B J , Fukui, Y , and Hanafusa, H (1990) Bmdmg of Trans- forming Protein, P47gag-crk, to a Broad Range of Phosphotyrosme-Containing Proteins Sczence 248, 1537-1539

14 Mayer, B J , Jackson, P K , and Baltimore, D (1991) The noncatalytic src homology region 2 segment of abl tyrosme kmase bmds to tyrosme-phosphorylated cellular proteins with high affinity Proc Nat1 Acad Scz USA S&627-63 1

15 Pawson, T (1992) Tyrosme kmases and their interactions with srgnallmg mol- ecules Curr Open Genet Dev 2,4-12

16 Overdum, M., Rros, C B., Mayer, B J , Baltimore, D., and Cowburn, D (1992) Three-dimensional solutton structure of the src homology 2 domain of c-abl Cell 70,697-704

17 Booker, G W., Breeze, A L., Downing, A K., Panayotou, G., Gout, I., Waterfield, M D., and Campbell, I D (1992) Structure of an SH2 domain of the p85a subunit of phosphatrdylmosrtol-3-OH kmase Nature 358,684-687

18 Waksman, G., Kommos, D , Robertson, S R., Pant, N , Baltimore, D., Barge, R B , Cowburn, D , Hanafusa, H., Mayer, B J , Overdum, M , Resh, M D., Rtos, C B., Silverman, L., and Kurryan, J (1992) Crystal structure of the phosphotyrosme recognmon domam SH2 of v-src complexed with tyrosine-phosphorylated peptides Nature 358,64&653

19 Cohen, B , Yoakrm, M , Prwmca-Worms, H , Roberts, T , and Schaffhausen, B S (1990) Tyrosme phosphorylatton 1s a signal for the traffickmg of pp85, a polypep- tide associated with phosphatrdylmosrtol kmase activity Proc Nat1 Acad SCL USA 87,4458-4462

Peptde Recogmtion Mechamsms 27

20 Kazlauskas, A., Kashishtan, A , Cooper, J A., and Valius, M (1992) GTPase activat-

mg protein and phosphattdylmositol 3-kmase bmd to a distinct region of the platelet-dertved growth factor receptor psubunit Mol Cell Btol 12,2534-2544

21 Talmadge, D A., Freund, R , Young, A T , Dahl, J., Dawe, C J., and Benjamin,

T L (1989) Phosphorylation of middle T by pp60c-src: a switch for bmdmg of phosphatidylmosttol3-kmase and opttmal tumorigenests Cell 59,5.5-65

22 Cantley, L C , Auger, K R , Carpenter, C , Duckworth, B., Graztam, A., Kapeller, R., and Soltoff, S (1991) Oncogenes and signal transduction Cell 64,281-302

23 Songyang, Z , Shoelson, S E , McGlade, J , Olivter, P , Pawson, T., Bustelo, X R , Barbactd, M., Sabe, H., Hanafusa, H , Yt, T., Ren, R., Baltimore, D., Ratnofsky, S., Feldman, R A., and Cantley, L C (1994) Specific motifs recognized by the SH2 domains of Csk, 3BP2, fps/fes, GRB-2, HCP, SHC, Syk and Vav Mol Cell Blol 14, 2777-2785

24 Songyang, Z , Shoelson, S E., Chaudhun, M., Gish, G., Pawson, T., Haser, W G., King, F., Roberts, T., Ratnofsky, S., Lechleider, R J , Neel, B G., Birge, R B , FaJardo, J E., Chou, M M., Hanafusa, H., Schaffhausen, B., and Cantley, L C (1993) SH2 domains recognize spectfic phosphopeptide sequences Cell 72,767-778

25 Waksman, G , Shoelson, S E , Pant, N , Cowburn, D., and Kuriyan, J (1993) Bmdmg of a high affinity phophotyrosyl peptide to the src SH2 domain: crystal structures of the complexed and pepttde-free forms Cell 72,779-790

26 Eck, M., Shoelson, S E., and Harrison, S C (1993) Recogmtton of a high affinity phosphotyrosyl peptide by the Src homology 2 domain of ~56’“~ Nature 362, 87-91,

27 Songyang, Z , Gtsh, G., Mbamulu, G., Pawson, T., and Cantley, L C (1995) A single point mutatton switches the specificity of group III src homology (SH) 2 domams to that of group I SH2 domains J Blol Chem 270,26,029-26,032

28 Lee, C.-H., Kommos, D , Jacques, S , Margohs, B., Schlessinger, J., Shoelson, S

E , and Kurtyan, J (1994) Crystal structures of peptide complexes of the amino- terminal SH2 domam of the Syp tyrosme phosphatase Structure 2,423-438

29 Pascal, S M., Singer, A U , Gish, G., Yamazaki, T., Shoelson, S E., Pawson, T , Kay, L E., and Forman-Kay, J D (1994) Nuclear magnetic resonance structure

of an SH2 domam of phospholipase C-y1 complexed with a high affimty bmdmg peptide Cell 77,461472

30 Rahuel, J , Gay, B., Erdmann, D., Strauss, A., Garcia-Echeverrta, C., Furet, P., Caravattr, G., Fretz, H., Schoepfer, J., and Grtitter, M (1996) Structural basis for specificity of GRB2-SH2 revealed by a novel ligand bmdmg mode Nat Struct Biol 3,586-589

31 Zhou, M M., Meadows, R P., Logan, T M., Yoon, H S., Wade, W S , Ravtchandran, K S., Burakoff, S J., and Fesik, S W (1995) Solution structure of the She SH2 domam complexed with a tyrosine-phosphorylated peptide from the T-cell receptor Proc Nat1 Acad Scl USA 92,7784-7788

32 Matgnan, S , Gurlloteau, J -P , Fromage, N., Arnoux, B , Becquart, J., and Ducruix, A (1995) Crystal structure of the mammahan GRB2 adaptor Sczence 268,291-293

33 Gosser, Y Q., Zheng, J., Overdum, M., Mayer, B J., and Cowburn, D (1995) The solutton structure of Abl SH3, and its relationship to SH2 m the SH(32) construct Structure 3, 1075-1086

34 Hatada, M H., Lu, X., Laird, E R., Green, J , Morgenstern, J P., Lou, M., Marr,

C S., Phdhps, T B., Ram, M K., Thertault, K., Zoller, M J., and Karas, J L (1995) Molecular basis for the interaction of the protein tyrosine kmase ZAP-70 with the T-cell receptor Nature 377, 32-38

35 Eck, M., Atwell, S K., Shoelson, S E , and Harrison, S C (1994) Crystal struc- ture of the regulatory domams of the Src-family tyrosme kmase lck Nature 368,

764-769

36 Eck, M J., Pluskey, S , Trub, T , Harrison, S C , and Shoelson, S E (1996) Spatial constraints on the recogmtion of phosphoprotems by the tandem SH2 domams of the phosphatase SH-PTP2 Nature 379,277-280

37 Fantl, W J , Escobedo, J A , Martm, G A , Turck, C W , Rosario, M ,

factor receptor bmd to specific molecules that medtate different stgnallmg path- ways Cell 69,413-423

38 Kashishian, A., Kazlauskas, A , and Cooper, J A (1992) Phosphorylation sites in the PDGF receptor with different spectfictties for bmdmg GAP and PI3 kmase m VIVO EMBO J 11, 1373-1381

39 Ladbury, J E., Lemmon, M A., Zhou, M., Green, J., Botfteld, M C., and Schlessmger, J (1995) Measurement of the bmdmg tyrosyl phosphopeptides to SH2 domains: a reappraisal Proc Nat1 Acad Scz USA 92,3199-3203

40 Lemmon, M A and Ladbury, J E., (1994) Thermodynamic studies of tyrosyl-phosphopeptide bmdmg to the SH2 domain of p56tCk Bzochemzstry 33,

5070-5076

41 Felder, S., Zhou, M , Hu, P., Urena, J , Ullrich, A , Chaudhuri, M , White, M., Shoelson, S E., and Schlessmger, J (1993) SH2 domams exhibit high-affinity bmdmg to tyrosme-phosphorylated pepttdes yet also exhibit rapid dtssoctation

42 Mayer, B J., Jackson, P K , Van Etten, R A , and Baltimore, D (1992) Point mutations m the abl SH2 domam coordmately impair phosphotyrosme bmdmg in vitro and transforming ability m vivo Mel Cell Bzol 12,609-618

FEBS Lett 203, 139

44 Nolte, R T., Eck, M J., Schlessmger, J , Shoelson, S E., and Harrison, S C (1996) Crystal structure of the PI 3-kmase p85 ammo-termmal SH2 domain and its phosphopeptide complexes Nature Struct Bzol 3, 364-313

45 Case, R D , Piccione, E., Wolf, G , Lechleider, R J., Chaudhuri, M , Neel, B G., and Shoelson, S E (1994) SH-PTP2 SH2 domain bmdmg specifictty IS defined

by direct mteracttons with PDGF P-receptor, EGF receptor, and IRS-l derived

46 Pascal, S M , Yamazakt, T., Singer, A U , Kay, L E , and Forman-Kay, J D (1995) Structural and dynamic characterizatton of the phosphotyrosme bmdmg

PeptIde Recognition Mechanisms 29

region of a Src homology 2 domain-phosphopeptide complex by NMR relaxation, proton exchange and chemical shift approaches Biochemzstry 34,11,353-l 1,362

47 Marengere, L E M , Songyang, Z., Gish, G D., Schaller, M D., Parsons, J T , Stem, M J., Cantley, L C , and Pawson, T (1994) SH2 domam specificity and activity modified by a single residue Nature 369,502-505

48 Gustafson, T A., He, W , Craparo, A., Schaub, C D., and O’Neill, T J (1995) Phosphotyrosme-dependent mteraction of SHC and insulin receptor substrate 1 with the NPEY motif of the msulm receptor via novel non-SH2 domain Mel Cell Bzol l&2500-2508

49 Cicchetti, P , Mayer, B J., Then, G., and Baltimore, D (1992) Identification of a protein that binds to the SH3 region of abl and is similar to Bcr and GAP-rho Science 257,803-806

50 Ren, R., Mayer, B J., Cicchetti, P., and Baltimore, D (1993) Identification of a ten-ammo acid prolme-rich SH3 bmdmg sate Sczence 259, 1157-l 161

51 Yu, H , Rosen, M K , Shin, T B , Seidel-Duggan, C , Brugge, J S., and Schreiber,

S L (1992) Solution structure of the SH3 domam of Src and Identification of Its Ligand-Bmdmg Site Sczence 258, 1665-1668

52 Musacchio, A., Noble, M , Pauptit, R , Wierenga, R , and Saraste, M (1992) Crystal structure of a Src-homology 3 (SH3) domain Nature 359, 851-855

53 Yu, H., Chen, J K , Feng, S , Dalgamo, D C , Brauer, A W , and Schreiber, S L (1994) Structural basis for the bmdmg of prolme-rich peptides to SH3 domains Cell 76,933-945

54 Musacchio, A., Saraste, M , and Wilmanns, M (1994) High-resolution crystal structures of tyrosme kmase SH3 domains complexed with prolme-rich peptides Nature Struct Blol 1,546-55 1

55 Lim, W A and Richards, F M (1994) Crttical residues m an SH3 domam from Sem-5 suggest a mechanism for prolme-rich peptide recognition Nature Struct Blol 1,221-225

56 Feng, S , Chen, J K., Yu, H., Simon, J A., and Schreiber, S A (1994) Two binding orientations for peptides to the Src SH3 domain development of a gen- eral model for SH3-hgand mteractions Sczence 266, 1241-1247

57 Lim, W A , Richards, F M , and Fox, R 0 (1994) Structural determinants of peptide-bindmg orientation and of sequence specificity in SH3 domains Nature 372,375-379

58 Goudreau, N., Camille, F., Duchesne, M., Parker, F., Tocque, B., Garbay, C , and Roques, B P (1994) NMR structure of the N-terminal SH3 domam of GRB2 and its complex with a proline rich peptide from SOS Nature Strucr Blol 1,898-907

59 Terasawa, H , Kohda, D., Hatanaka, H , Tsuchiya, S , Ogura, K , Nagata, K , Ishn, S., Mandiyan, V., Ullrich, A , Schlessmger, J., and Inagaki, F (1994) Structure of the N-terminal SH3 domain of GRB2 complexed with a peptide from the guamne nucleotide releasing factor SOS Nature Struct Biol 1, 891-897

60 Lee, C.-H , Saksela, K., Mirza, U A., Chait, B T , and Kuriyan, J (1996) Crystal structure of the conserved core of HIV-l Nef complexed with a Src family SH3 domain Cell 85,93 l-942

61 Lee, C.-H., Leung, B., Lemmon, M A , Zheng, J , Cowburn, D , Kuriyan, J , and Saksela, K (1995) A single ammo acid m the SH3 domain of Hck determines tts high affinity and specificity m binding to HIV-l Nef protein EMBO J 14, 5006-5015

62 Cheng, G , Ye, Z S , and Baltimore, D (1994) Bmdmg of Bruton’s tyrosme kmase

to Fyn, Lyn, or Hck through a Src homology 3 domain-mediated mteraction Proc Nat1 Acad Scz USA 91,8152-8155

63 Rtckles, R., Botfield, M C., Weng, Z., Taylor, J., Green, 0 M., Brugge, J., and Zoller, M J (1994) Identification of Src, Fyn, Lyn, PI3K, and Abl SH3 domain hgands using phage display libraries EMBO J X3,5598-5604

64 Knudsen, B , Zheng, J., Feller, S M., Mayer, J P., Burrell, S K., Cowburn, D , and Hanafusa, H (1995) Affinity and specificity requirements for the first Src homology 3 domain of the Crk protein EMBO J 14, 2191-2198

65 Wu, X , Knudsen, B., Feller, S M., Zheng, J., Sah, A., Cowburn, D , Hanafusa,

H , and Kuriyan, J (1995) Structural basis for the specific interaction of lysme- contaming prolme-rich peptides with the N-terminal SH3 domain of c-Crk Struc- ture 3,2 15-226

66 Feng, S., Kasahara, C , Rtckles, R J., and Schretber, S L (1995) Specific mterac- tions outside the prolme-rich core of two classes of Src homology 3 hgands Proc: Natl Acad Scl USA 92, 12,408-12,415

67 Trono, D (1995) HIV accessory proterns leading roles for the supportmg cast Cell 82, 189-192

68 Saksela, K , Cheng, G., and Baltimore, D (1995) Prolme-rich (PxxP) motifs m HIV- 1 Nef bmd to SH3 domains of a subset of Src kmases and are required for the enhanced growth of Nef+ vu-uses but not for down-regulation of CD4 EMBO J 14,484-49 1

69 Grzesiek, S , Bax, A., Clore, G M., Gronenborn, A M., Hu, J -S., Kaufman, J , Palmer, I., Stahl, S J., and Wmgfield, P T (1996) The solution structure of HIV-

1 Nef reveals an unexpected fold and permits dehneation of the bmdmg surface for the SH3 domain of Hck tyrosme protem kmase Nature Struct Blol 3, 340-345

70 Alexandropoulos, K and Baltimore, D (1996) Coordinate activation of c-Src by SH3- and SH2-binding sites on a novel pl30Cas-related protein, Sin Genes Dev 10,1341-1355

71 Abrams, C S and Zhao, W (1995) SH3 domains spectfically regulate kmase activity of expressed Src family proteins J Bzol Chem 270,333-339

72 Superti-Furga, G and Courtnetdge, S A (1995) Structure-function relationships

in Src family and related protem kmases Bzoessays 17, 321-330

73 Weiss, A (1993) T Cell Antigen Receptor Signal Transduction A Tale of Tads

and Cytoplasmic Protein-Tyrosme Kmases Cell 73,209-212

74 Wange, R L , Malek, S N , Desiderio, S., and Samelson, L E (1993) Tandem SH2 domains of ZAP-70 bmd to T cell antigen receptor zeta and CD3 epsilon from activated Jurkat T cells J Bzol Chem 268, 19,797-19,801

Peptide Recognition Mechanisms 31

75 Isakov, N , Wange, R L , Burgess, W H , Watts, J D., Aebersold, R., and Samelson, L E (1995) ZAP-70 binding specificity to T cell receptor tyrosine- based activation motifs the tandem SH2 domains of ZAP-70 bind distinct ty- rosme-based activation motifs with varying affinity J Exp Med 181,375-380

76 Feng, G.-S., Hut, C.-C., and Pawson, T (1993) SHZcontammg phosphotyrosme phosphatases as a target of protein-tyrosine kmases Sczence 259, 1607-1614

77 Wolfgang-Vogel, Lammers, R., Huang, J., and Ullrich, A (1993) Activation of

1611-1614

78 Dechert, U , Adam, M., Harder, K W., Clark-Lewis, I., and Jrrrk, F (1994) Char- acterization of protein tyrosme phosphatase SH-PTP2 Study of phosphopepttde substrates and possible regulatory role of SH2 domams J Bml Chew 25, 5602-5611

79 Sugimoto, S , Lechleider, R J , Shoelson, S E., Neel, B G., and Walsh, C T (1994) Expression, purtficatton and characterization of SH2-contammg protein tyrosine phosphatase, SH-PTP2 J Biol Chem 268,22,77 l-22,776

80 Lechletder, R J., Sugtmoto, S , Bennett, A M., Kashtshran, A S , Cooper, J A., Shoelson, S E , Walsh, C T., and Neel, B G (1993) Acttvatron of the SH2-

phosphotyrosine 1009, on the PDGF receptor J Biol Chem 268,21,478-21,481

8 1 Krauhs, P (1991) MOLSCRIPT: A program to produce both detailed and sche- matic plots of protein structures J Appl Crystallogr 24,946-950

82 Noble, M E M., Musacchto, A., Saraste, M., Courtneidge, S A , and Wterenga,

R K (1993) Crystal structure of the SH3 domam m human Fyn; compartson of the three-dimensronal structures of SH3 domains m tyrosine kmases and spectrm EMBO J 12,2617-2624

In the space of a few short years, the way m which we look at these processes has changed in a fundamental way from an emphasis on the regulation of enzymes and their substrate specificities to a new emphasis on the regulation and specificity of protein-bmdmg surfaces We now appreciate that the cell is less like an aqueous solution m a test tube and more like a dense gel of interact- ing proteins, where the actual activity of an enzyme is as dependent on its binding partners and subcellular localization as it is on the kinetic parameters

of its catalytic activity

Early biochemical work on metabolic pathways had emphasized concepts

of pathways and cascades, m which one step leads to subsequent steps m a relatively linear fashion, often with amplification of a signal These concepts often proved inadequate, however, when applied to the mechanisms of signal transduction A good illustration is the case of receptor tyrosine kmases In the early 1980s it was discovered that the receptors for many mitogenic growth factors, such as the epidermal growth factor (EGF), were transmembrane pro- tein-tyrosme kinases It seemed obvious that the key to understanding signal transmission would be to find and identify the substrate proteins phosphory- lated by the liganded receptors, which must surely be the effecters responsible for stimulating the cell to proliferate When lysates of growth-factor stimu-

From Methods m Molecular Bfology, Vol 84 Transmembrane Sgnalmg Protocols

Edlted by D Bar-Sagt 0 Humana Press Inc , Totowa, NJ

33

lated cells were analyzed with phosphotyrosme-specific antibodies, however,

a problem arose By far, the most promment tyrosme-phosphorylated protem was found to be the receptor itself Clearly this was mconslstent with models in which the receptor mmates a signalmg cascade by phosphorylatmg many sub- strate proteins What has become apparent is that the key to signal transmission

m this case is the creation of bmdmg sites on the receptor, via autophosphory- lation, for proteins containing Src homology 2 (SH2) domains; it is unclear whether the receptor needs to phosphorylate any protein other than itself to initiate mitogemc signalmg From such studies, a new paradigm emerged m which an enzyme’s predominant function can be to alter its bmdmg activities

in response to ligand

When closely exammed even “classical” signaling pathways reveal the crltl-

cal importance of stable and regulated protein-protein interactions Among the best understood signalmg cascades are those mediated by heterotrimeric G pro- teins (3) In the P-adrenergic pathway, for example, an agonist-stimulated receptor activates many molecules of a heterotrimeric G protein, each of which can then activate a molecule of the enzyme adenyl cyclase The resultmg rise

m mtracellular cychc adenosine monophosphate (CAMP) in turn activates many molecules of protein kinase A, which then phosphorylate many intracellular

proteins on serine and threomne residues The details of this relatively simple signaling apparatus reveal at least five critical protein-protein Interactions: The heterotrimeric G protein binds to the hganded (but not the unliganded) recep- tor; conformational changes brought about by receptor bmdmg and concomi- tant guanosine diphosphate (GDP) release and guanosine triphosphate (GTP) bmdmg induce the dissociation of the a-subunit of the G protein (G,) from its

p and y-subunits (Gay), and from the hganded receptor; the released G, subunit binds to and activates the cyclase; meanwhile Gpy binds to and relocahzes the P-adrenergic receptor kmase (P-ARK), leading to receptor phosphorylation and desensmzatron; and finally, CAMP binding causes the drssocration of the regu- latory subunit of PKA, thereby releasing the active catalytic subunit Although enzymes (kinases, GTPases) are involved, it is obvious that changes m pro- tem-protein interactions play a central role in signal transmission

Much of this volume is devoted to the many techniques now used to analyze protein-protein interactions Such mteractions are now appreciated to be so important to understandmg the function of signalmg proteins that often one of the first experiments performed on a newly identified protein is a search for interaction partners In this chapter, I briefly consider specificity and regula- tion of binding mteractions, review some of the classes of well-known pro- tein-protein interactions known to be mvolved m intracellular signaling, and discuss how the significance of a particular interaction can be assessed

interactions in Signaling Cascades 35

2 Specificity

Two of the defining parameters of protein-protem mteractions are specific- ity and whether that specificity can be regulated Specificity is, of course, a function of both the affinity for target sites and the affinity for “nonspeclftc” sites In cases where specificity is very high for a single target molecule (for example, the regulatory subunit of PKA for its catalytic subunit), we might term the two proteins subunits of a holoenzyme Clearly, however, there is no fundamental difference between such an interaction and one that is somewhat less specific, for example, the bmdmg of the same heterotrimeric G-protem P-subunit to several different a- and ‘y- subumts, or one that is much less spe- cific, for instance, the binding of an Src homology 3 (SH3) domain to prolme- rich sites m tens to hundreds of different protems

Specificity is usually thought of either m terms of dissociation constants or,

in a more practical sense, of signal-to-background (e.g., a specific association gives a dark plaque or a blue colony in a sea of light plaques or colonies) It is worth thmkmg of specificity a bit more carefully m terms of concentrations of protems in a cell A protein that represents l/10,000 of total-cell protein is present in the cytosol at a concentration on the order of 10M7 M, simplistically, for two Interacting proteins at this level of abundance, the dissociation con- stant for the complex would have to be submicromolar for a significant amount

of the complex to exist m VIVO Dissociation constants for known complexes are usually in this range, for example 10d9 M for the assoctatton of the regula- tory and catalytic subunits of PKA, and 1O-8-1O-7 M for complexes of SH2 domains with tyrosine-phosphorylated targets Significant mteractions can certainly have less impressive dissociation constants, however; mdividual SH3 domain-peptide mteractions usually have affinities m the range of 10” to 10e5 M, but the presence of multiple-bmdmg sites and multtple SH3 domains

in many actual bmdmg partners probably raises the overall affinity by mcreas- ing the avidity of binding An extreme example is actin, for which the Kd for binding of monomers to the end of a filament is -10m5 M, but complex forma- tion (polymerization) is favored because the total intracellular concentration of actm is very high

Regulation of specificity is often (but not always) critical if the complexes are to be important to signalmg Whereas specific, unregulated protein com- plexes might be important for function, and are certamly worth knowing about,

it is changes m binding that drive signal transduction This is of practical impor- tance because it can provide an experimental handle to identify mteractions involved in signaling (for example, proteins that bind to a G protem only when

it is bound to GTP, and not to GDP, would be candidate-effector molecules)

Changes m bmdmg specificity can be due to allosteric alterations m one of the binding partners, dependent for example on whether GDP or GTP is bound to

a G protein, or to direct changes in the bmdmg site, for example dependent on tyrosine phosphorylatton to create an SH2-binding site

Obviously there are many cases m which such distmcttons are blurred; in one example, phosphorylation of the p47phox protein results m the dissolution

of an intramolecular SH3-prolme-rich mteraction, thereby freeing both the SH3 domains and the proline-rich SH3-bmdmg site of p47phox for mteraction with other proteins m trans (4,5) This cisltrans swatch is critical for generating the oxidative burst m phagocytes

3 Protein-Binding Modules Involved in Signal Transduction

It is now clear that not only are protein-protem mteractions important for signaling, but that many signalmg proteins contam recogmzable modules that confer bmdmg activity (Table 1) This 1s indeed fortunate, because it allows us

m many cases to predict what type of bindmg mteracttons to expect based on the amino acid sequence of a novel protein Such a modular system makes sense from an evolutionary point of view, in that domams can be shuffled and existing interaction pairs fine-tuned during evolution so that specific-binding surfaces don’t need to arose independently for each pan of interacting proteins

In this chapter, some relatively well-characterrzed intracellular-binding mod- ules will be summarized

Protein-interaction modules that are recognizable by sequence similarity fall into two overlapping classes First there are those such as the SH2 and SH3 domains, which are independently folding units that confer a characteristic and specialized type of binding interaction (tyrosine-phosphorylated peptides, for example) In these modules, the most conserved residues are those that are directly mvolved in bindmg to ligands The other broad class are those m which the sequence similarity is owing to a common folded structure but does not necessarily predict the specific type of bmdmg mteraction, for example, WD repeats Often these motifs are repeated many times in proteins containing them, and may assemble with other repeats mto higher order structures Such motifs most hkely represent an evolutionary solution to the design problem of small, stable folded domams that can evolve to display variable-surface resi- dues mvolved m specific-bmdmg interactions An example, which will not be discussed further, is the variety of zinc-bmdmg “fmgers” that mediate protem- protein and protem-DNA interactions, m which metal binding stabilizes com- pact-folded structures It should be noted that the list m Table 1 is far from comprehensive, and as our ability to analyze sequence mformatton improves and as three-dtmensronal structural information accumulates, it is likely that many other binding modules will emerge

Interactions in Signaling Cascades 37

Table 1

Protein-Binding Modules Implicated in Signaling

No Repeats/ Core Module Sizea protein binding site” Regulated? 3-D structure?C

Armadillo -42 7-13 Unknown Unknown; tyrosme None

phosphorylatton7 ONumber of ammo acrds m module (not mcludmg spacers between repeats)

bMnnmal bindmg site requtred for recogmtron (other residues might be involved m bmdmg to specific examples) Y(P), phosphotyrosme; x, any ammo acid, n, variable residue involved m specificity; a’, hydrophobic ammo acid

‘Number of different high-resolution three-dimensional structures of module available

1 Many proteins rmphcated m signaling contained SH2 domams;

2 These proteins could often be shown to bind tightly to hgand-activated growth-

factor receptors; and

3 Bacterially expressed SH2 domams could be shown to bind to tyrosme- phosphorylated proteins, including activated receptors (7-11)

We now know that these domains serve a general role in signaling m com- plex eukaryotes, mediating the relocalization or assembly of SH2-containing proteins m response to changes m tyrosine phosphorylatron (because they are

both lacking m yeast, SH2 domains and true tyrosine kinases must be rela- tively recent evolutionary innovations, presumably for dealing with the greater signaling demands of multicellular life)

A great deal is known about the structure and binding interactions of these domains, and only a brief summary will be given here Bmdmg to tyrosme- phosphorylated sites is quite tight, with measured affinities in the range of 1O-8-1O-7 A4 (12), and is absolutely dependent on phosphorylation, because bmdmg to unphosphorylated hgands 1s undetectable Bmdmg 1s dependent only

on short peptide sequences and can be mtmtcked using synthetic peptides, so the interaction is largely independent of the larger protein containing a phos- phorylated site There is considerable speciftcity among SH2 domains for dif- ferent phosphorylated peptide sites, and a degenerate peptlde-library approach allowed the bmdmg specificities of a number of SH2 domams to be determmed (13,14) Specificity was found to be dependent on the three (in rare cases up to five or six) ammo acids C-terminal to the phosphorylated tyrosme, with resi- dues N-terminal to the phosphotyrosme having httle or no effect on binding However, it should be remembered that all SH2 domains have a detectable affinity for phosphotyrosme itself (indeed, this can be used as a purificatron scheme to isolate SH2 domains), so specificity is relative rather than absolute Which SH2 domains will bind to a particular site in vivo will depend on the local concentration, as well as the relative affinities of potential bmdmg part- ners It has recently been shown that some SH2 domains can also bmd with high affinity to mosnol lipids phosphorylated on the 3’ position (IS), so it 1s worth remembering that protein-binding domains might have hitherto unap- preciated activmes that will affect then behavior m vivo

3.2 PTB Domains

A less common domain that also binds tyrosme-phosphorylated sites was found during analysis of the She adaptor protein She contains an SH2 domam and was known to bind to tyrosine-phosphorylated proteins, but it became apparent that many She-bmdmg sites consisted of an NPxY(P) motif (where Y(P) represents phosphotyrosme) quite different from known SH2-bmdmg sites It was ultimately shown that binding to these sites mapped to an approx

160 ammo acid regton of She (termed the PTB [phosphotyrosine binding] domain) with no sequence homology to the SH2 domain (16-19) Apparent affinity for NPxY(P) peptides is m the range of 10e6 M (19,20), but because very few PTB domains have been Identified, the range of target specifrcmes and affinities is unknown The degree of sequence similarity among PTB domains is weak, making tdenttfication from sequence problematrc; the IRS-l PTB was only identified by virtue of its binding activity (18,20) The tertiary structure of the She PTB is virtually identical to that of the PH domain (21)

interactions m Signaling Cascades 39

(dtscussed in Subheading 3.4.), so it is possible that the few known cases are actually a specialized subset of the larger PH-domain family

3.3 SH3 Domains

These small interaction modules were first identified in signaling proteins

as a region of homology to the Src kinase, as m the case of the SH2 domain They have subsequently been found in a wide range of protems including in yeast, in contrast to the SH2 (revtewed m ref 22) They are often found in the same proteins as SH2 domains, but this does not reflect any structural or func- tional similarity in the domams themselves but is more likely related to the frequent involvement of these domains in signal-transduction complexes Indeed, there is a class of proteins termed the SH2/SH3 adaptors that consist entirely of these two domams, and thereby serve as molecular “crosslinkers” to assemble complexes of signalmg protems

SH3 domams bind to short, prolme-rich bindmg sites in proteins (23-25) From structural studies and work using peptide libraries, it is known that the binding site consists of three turns of a left-handed proline- helix, and that SH3 domains can bind ligands m either and N-C or a C-N terminal orienta- tion, owing to the pseudosymmetry of the proline- helix (reviewed in ref 26)

Most SH3 domains bmd to core sites with the consensus +X@PX@P (class 1)

or @PXQPX+ (class 2) where + represents a basic residue, 0 represents a hydrophobic residue, and X can be any amino acid There is considerable speci- ficity among different SH3 domains for different binding sites, but as in the case of SH2 domams the differences m affinity between high- and low-affinity sites can be quite small, so it is difficult to predict a priori which specific sites might bind in VIVO Affinities are generally quite modest, with Kds m the range

of 10-6-10-5 Mfor specific SH3-peptide interacttons

The best characterized role for the SH3 domain is m recruitmg the Ras exchange-factor SOS to the membrane leading to the activation of Ras (27) In flies and nematodes, this has been shown to be mediated by the SH2/SH3- adaptor protein Grb2, whtch contains one SH2 and two SH3 domains It is thought that SOS and Grb2 exist as a preformed complex in the cytoplasm, and that this complex is recruited to the membrane by binding of the Grb2/SH2 domain to phosphorylated sites generated by activated growth-factor recep- tors As in this case, SH3-mediated binding in general has not been shown to

be directly regulated It is more likely that in most cases these domains bind constttutively, functioning as an intracellular adhesive and not as a switch As mentioned m the previous section, however, there are examples such as p47phox where phosphorylation of a protein can allosterically regulate the availability

of its SH3 domains and/or proline-rich target peptides

3.4 PH Domains

This widely distributed and diverse class of protein modules was first tden- tified in the platelet protein pleckstrin as a repeated segment, and was subse- quently found by sequence comparison in a number of other proteins (28-30) Sequence identity among different pleckstrin homology (PH) domains is quite low (in the 20% range in many cases), making identification from primary sequence difficult The size of the domain ranges up from -100 amino acids and varies considerably, owing to msertions m variable-loop regions The three-dimensional structures of a number of PH domains have now been solved revealing that then overall folds are very similar (31), although the variable loops are likely to make the surface properties of different PH domains quite variable All PH-domain structures currently available reveal a highly polar- ized electrostatic potential that may favor binding to membranes via the posi- tively charged portion of the domain

The jury is still out on whether PH domains as a class mediate protein- protein mteractions The PH domain of P-adrenergic receptor kinase (P-ARK) has been shown to mediate binding to the P-r subunits of heterotrimeric

G proteins (hence relocahzing P-ARK to the membrane in proximity to its substrate, the P-adrenergic receptor) However mutagenesis has shown that Gpy binding is confined to a long C-terminal alpha helix of the PH domain, and that most of the domain is, in fact, dispensable for bmdmg (32) On the other hand, several PH domains have been shown to bmd to polyphosphorylated mositols and mositol lipids with moderate affinity, and this may prove to be the more general role for PH domains in signalmg (33) In one case, the PH domain of PLC-6, the affinity for Ins (1,4,5) Pa is very high (&=210 nM) (34), but this appears to be an exception The wide diversity of PH domains, and the fact that PTB domains (above) have a virtually identical fold but quite different binding specificity, suggest that the PH domain might be more properly described as a folding scaffold that has been adapted for many uses Like different immunoglubulins, different PH domains might therefore bmd widely diver- gent ligands, which include both proteins and nonprotem molecules

3.5 Ankyrin Repeats

Ankyrm repeats were first identified as a repeating motif m the membrane- matrix protein ankyrm, and have subsequently been identified by sequence similarity m a wide variety of proteins, including a few prokaryotic and extra- cellular examples (35) The repeat itself consists of 33 residues and is always present m at least 4 (and up to 24) tandemly repeated copies This small size and the presence of multiple copies suggest that mdividual repeats are relatively unstable and that multiple repeats fold mto a more stable higher order structure

Interactions in Signaling Cascades 41

Many ankyrm-repeat proteins are known to participate m protein-protem mteractions, with perhaps some of the best examples being ankyrin itself (which binds to the anion transporter, Na/K adenosine triphosphatase [ATPase], tubulin, and the sodium channel) (36) and the inhibitory subunits of the NF-KB family of transcrtption factors that bmd to and inhibit the activity of the DNA-binding subunits (37) Phosphorylation has been shown to diminish binding of the inhibitor I-KB to NF-KB (38), but it is not known whether this is

a general property of ankyrin repeat-mediated interactions Because no sequence or functional similarity 1s apparent when known bmdmg proteins are compared, it is likely that the binding specificity of ankyrm repeats is con- ferred by variable-surface residues A high-resolution structure of an ankyrin repeat-containing protein would be extremely useful to identify residues involved in binding mteractions

3.6 WD Repeats

WD repeats (so named for the characteristics tryptophan-aspartate [WD] dipeptide often found at then C-terminal border) were origmally noted m the P-subunits of heterotrtmertc G protems, and have subsequently been found m a wide spectrum of eukaryotic proteins mvolved in signaling, vesicle traffic, RNA processmg, and many other functions (39) A common thread is that most,

if not all, of these protems are likely to be involved in the assembly of protein- protein complexes The repeats each consist of approx 3 1 residues with a vart- able spacer between repeats, and are present between four and eight times in all known examples

The recent solution of the three-dimensional structure of the heterotrimeric G-protein P-subunit (together with its y-subunit in the presence and absence of the GTP-binding a-subunit) has revealed the structural organization of the repeat (4042) Each repeat forms a compact beta-sheet structure that forms a blade of a so-called P-propeller, which in the case of the Ga consists of seven blades The conserved residues of the WD repeat are involved m inter- and intrablade interactions, so it is likely that all other WD-repeat proteins will form similar P-propeller structures As might be expected, the residues involved in binding to the a-subunit or to effector proteins are localized on the surface in positions that are not conserved among disparate WD repeats The dissociated Pr subunits of heterotrimeric G protems can function as effecters by binding to and modulating the activity of downstream stgnalmg proteins (3) This binding activity is regulated by the a-subunit because the effector-binding regions are sterically blocked by G, bmdmg; GDP release and GTP binding induce drastic conformational changes in the a-subunit, lead-

mg to the dissociation of the p r subunits, and thereby making them able to bind their effecters (40-42) This is an excellent example of the wealth of func-

tronal mformation about whole classes of proteins that can be gleaned from a single three-dimensional structure

3.7 Armadillo Repeats

Another repeating motif imphcated in protein-protein association is the armadillo or “Arm” repeat, origmally identified in the armadillo protein impli- cated m the wingless-signaling pathway m Drosophzla (43) These repeats are found m mtercellular lunctton components such as p-catemn and plakoglobm,

as well as several other proteins including the product of the tumor suppressor gene APC, nuclear pore protein SRPl, and smgGDS, a guanine-nucleotide exchange factor for small GTPases (44,45) The repeat consists of approx 42 residues, and is present m 7-13 copies in known examples As m the case

of the ankyrm repeat, it is likely that the Arm repeat encodes a structural scaf- fold that assembles together with other repeats, but confirmation of this awaits

a three-dimensional structure It IS known that several Arm-containing pro- teins can be tyrosine phosphorylated, raising the possibility that phosphoryla- tion may directly or mdn-ectly affect their binding activity

4 Is a Binding Interaction Significant?

Perhaps the most vexing question facing those of us working on signaling pathways 1s whether a potential interaction is significant Because sequence inspection leads to predictions about potential interaction partners, and because the techniques for detecting potential interactions are so sensitive, there are often not one or two but hundreds of candidate-binding proteins for any given protem of interest In some cases, this may actually reflect the messy reality that the protein of interest partitions among many different complexes m the cell, each of which might be important to some aspect of that protein’s function But how can we evaluate the significance of any single proposed interaction? The problem 1s one of establishing the relationship between in vitro- (or in the case of two-hybrid screening, m yeast-) bmdmg data to the biological prop- erties of the proteins m their normal cellular environment Specific controls for different methods of detectmg potential interactions wrll be detailed m the fol- lowing chapters, but it is worth considering some criteria at this time At the very least, the two proteins should be present m the cell m the same subcellular compartment at a suitable concentratton for the interaction to occur This would seem to require some detailed knowledge of the dissociation constant and the concentratron m various compartments, but in fact it can be quite easy to get the rough estimates of these parameters needed to evaluate an mteraction For example, if two interacting proteins are of very low abundance (a few thousand molecules per cell), and the apparent dissociation constant from simple m vitro-