Tài liệu Báo cáo khoa học: Single phosphorylation of Tyr304 in the cytoplasmic tail of ephrin B2 confers high-affinity and bifunctional binding to both the SH2 domain of Grb4 and the PDZ domain of the PDZ-RGS3 protein ppt

Single phosphorylation of Tyr304 in the cytoplasmic tail of ephrin B2 confers high-affinity and bifunctional binding to both the SH2 domain of Grb4 and the PDZ domain of the PDZ-RGS3 pro

Single phosphorylation of Tyr304 in the cytoplasmic tail of ephrin B2 confers high-affinity and bifunctional binding to both the SH2 domain

of Grb4 and the PDZ domain of the PDZ-RGS3 protein

Zhengding Su, Ping Xu and Feng Ni

Biomolecular NMR and Protein Research Group, Biotechnology Research Institute, National Research Council of Canada, Montreal, Quebec, Canada

The B class cell-attached ephrins mediate contact-dependent

cell–cell communications and transduce the contact signals

to the host cells through the binding interactions of their

cytoplasmic domains Two classes of intracellular effectors

of B ephrins have been identified: one contains the PSD-95/

Dlg/ZO-1 (PDZ) domain (for example PDZ-RGS3), and the

second the Src homology 2 (SH2) domain (e.g the Grb4

adaptor protein) The interaction with Grb4 requires

phos-phorylation of tyrosine residues on the conserved

cytoplas-mic C-terminal region of B ephrins, while binding to the

PDZ domain is independent of tyrosine phosphorylation

However, the exact phosphorylation site(s) required for

signaling remained obscure and it is also unknown whether

the two classes of effectors can bind to B ephrins

simulta-neously or if the binding of one affects the binding of the

other We report here that phosphorylation of Tyr304 in the

functional C-terminal region (residues 301–333) of ephrin B2

confers high-affinity binding to the SH2 domain of the

Grb4 protein Tyrosine phosphorylation at other candi-date sites resulted in only minor change of the binding

of Tyr304-phosphorylated ephrin B peptide (i.e eph-rinB2(301–333)-pY304) with the SH2 domain 1H-15N NMR HSQC experiments show that only the eph-rinB2(301–333)-pY304 peptide forms a stable and specific binding complex with the SH2 domain of Grb4 The SH2 and PDZ domains were found to bind to the Tyr304 phosphopeptide both independently and at the same time, forming a three-component molecular complex Taken together, our studies identify a novel SH2 domain binding motif, PHpY304EKV, on the cytoplasmic domains of B ephrins that may be essential for reverse signaling via the Grb4 adaptor protein alone or in concert with proteins containing PDZ domains

Keywords: ephrin B; Grb4; SH2; PDZ; phosphorylation

The ephrin-Eph signaling systems play critical roles in

multiple cell functions including cell migration, tissue border

formation in vascular development and angiogenesis, and

cell–cell communications for axonal guidance and at the

synaptic junction [1] The Eph molecules resemble classical

receptor tyrosine kinases (PTKs) in that they are

transmem-brane proteins with kinase domains and other binding

motifs projecting into the cytoplasmic space The plasma

membrane-bound ephrins, however, orchestrate cell

move-ments and morphogenesis by transducing bidirectional

signals into cells expressing the Eph molecules as well as

cells expressing the ephrins [2–6] The unique capacity of

reverse signaling, by the B-class cell-attached ephrins in particular, is to communicate the cell contact signals to the host cells through the association of their cytoplasmic domains with intracellular effector proteins So far, two types of intracellular targets for B ephrins have been identified Proteins containing PSD-95/Dlg/ZO-1 (PDZ) domains have been shown to bind to the C-termini of the

B ephrins [7–10] At least one protein containing an Src homology 2 (SH2) domain, Grb4, is recruited to the cytoplasmic tails of B ephrins upon phosphorylation [11] Although details of these signaling events have yet to be investigated, many of the molecular interactions down-stream of the Eph and ephrin molecules appear to lead to the regulation of the cytoskeleton of the interacting cells [1,2,6] Phosphorylation of the well-conserved cytoplasmic domains of B ephrins has been a subject of significant interest [11–16] as it is required for reverse signaling into the ephrin B-expressing cells The short cytoplasmic tails of the B ephrins contain five tyrosine residues, all of which are located within the extreme C-terminal 33-residue region [12,13] Three tyrosine residues are contained in a 22-residue

for ephrin B2, which was shown to be responsible for Grb4 binding and is identical in both ephrin B2 and ephrin B1 [11] Strikingly, the Grb4 SH2 domain shows strong binding to tyrosine-phosphorylated ephrin B1, whereas one

Correspondence to F Ni, Biotechnology Research Institute,

National Research Council of Canada, 6100 Royalmount Avenue,

Montreal, Quebec, H4P 2R2, Canada.

Fax: + 514 4965143, Tel.: + 514 4966729,

E-mail: fengni@bri.nrc.ca

Abbreviations: FGF, fibroblast growth factor; HSQC, heteronuclear

single quantum coherence; NOESY, nuclear Overhauser effect

spectroscopy; PDZ, PSD-95/Dlg/ZO-1; SFK, Src family kinase;

SH2, Src homology 2; TOCSY, total correlation spectroscopy.

(Received 23 December 2003, revised 27 February 2004,

accepted 9 March 2004)

of its closest known relatives, the Nck SH2 domain, has little

or no binding [11] The phosphorylations of Tyr311 and

Tyr316 were detected in both embryonic retinal tissues

and 293 cells stimulated with a multimeric EphB2 [15]

In Xenopus embryos, the activated fibroblast growth factor

(FGF) receptor was found to associate with and induce the

phosphorylation of ephrin B1 on equivalent positions of

Tyr311 and Tyr316, but not on Tyr304 [16] The region

from residues 308–316 containing two tyrosine residues was

found to be critical for binding interactions with the FGF

receptor Two other tyrosine residues are at the extreme

carboxyl terminus or residues YY(330–331)KV,

constitu-ting a PDZ-domain binding element [8] In neural tissues,

Tyr330 was found to be the major phosphorylation site of

ephrin B1 after binding to EphB2, while the

phosphoryla-tion of Tyr331 could not be detected [15] However, this

phosphorylation of ephrin B1 did not significantly affect

intracellular protein–protein interactions [8] More recently,

the Src family kinases (SFKs) and the PTP-BL phosphatase

were proposed to be mediators of the phosphorylation state

of the cytoplasmic domain of ephrin B1 whereby the SFKs

may create the docking sites for Grb4 and the phosphatase

may disengage Grb4 binding from B ephrins [17]

Our previous work suggested that the well-conserved

33-residue C-terminal region of residues 301–333 of ephrin

B2 might encode a latent three-dimensional structure [18],

which may be activated through phosphorylation for Grb4

binding More interestingly, the 22-residue region, i.e the

ephrinB2(301–322) peptide fragment encoding the Grb4

binding site [11], appears to assume an autonomously folded

structure, independently of the PDZ-binding extreme

C-terminal region, i.e residues IY330YKV [18] As well,

cellular functions ascribed to the Grb4 and PDZ-binding

events can work synergistically to coordinate cell migration

and morphogenesis in establishing defined patterns of cell

assemblies [1,2,6] It is therefore possible, upon

phosphory-lation, for the PDZ and SH2 domains to bind

simulta-neously to distinct regions of the already very short (33

residues) ephrin B cytoplasmic tails In the current work, we

first identify the phosphorylation site in ephrin B2 that

creates high-affinity binding to the Grb4 SH2 domain using

synthetic ephrinB2(301–333) fragments containing single or

double phosphorylations at five candidate tyrosine residues,

Tyr304, Tyr311, Tyr316, Tyr330 and Tyr331 We then

assess whether the binding of the PDZ domain of

PDZ-RGS3 affects binding of the Grb4 SH2 domain to

phosphorylated peptides, through studies of ternary

inter-actions among the PDZ domain, the phosphotyrosine

peptide and the Grb4 SH2 domain We also show that

phosphorylations of the ephrinB2(301–333) peptide alter

only the binding interactions and the three-dimensional

structure of the Grb4-binding region, i.e residues 301–322,

while leaving the PDZ-binding C-terminus essentially

unperturbed

Experimental procedures

Construction of expression vectors for the Grb4 SH2

and RGS3 PDZ domains

The DNA sequences encoding the Grb4 SH2 and the RGS3

PDZ domains were individually deduced from the amino

acid sequences of murine Grb4 protein and murine PDZ-RGS3 protein as published previously [7,19] using the codon preference of Escherichia coli The synthetic genes were amplified by PCR from six pairs of overlapping synthetic primers containing the two restriction sites of NcoI and BamHI for the SH2 domain and the two restriction sites

of BamHI and EcoRI for the PDZ domain at their two ends, respectively The double-digested DNA fragment of SH2 was subcloned into the pET3215 expression vector, which was modified from pET32 and pET15 vectors (Novagen, Madison, WI, USA), removing the original fusion carrier in the pET32 vector In order to facilitate protein purification, a His-tag with six histidine residues was placed at the N-terminus of the SH2 domain linked with a thrombin cleavage sequence The same insert was also subcloned into the pGFN GST fusion vector, which is

a derivative of the pGEX-4T-1 expression vector (Amer-sham Biosciences, Piscataway, NJ, USA), replacing the LVPR thrombin cleavage site with the FNPR sequence [20] The double digested DNA fragment of the PDZ domain was subcloned into the pGFN vector All these expression constructs were confirmed by DNA sequencing and trans-formed into the E coli BL21(DE3) expression host

Protein expression and purification The SH2 protein was expressed at 37C The cells were harvested four hours after induction with isopropyl thio-b-D-galactoside at D600¼ 0.8 Labeling with the 15N isotope was accomplished using M9 media containing 1.0 gÆL)1of [15N]ammonium sulfate, 5 gÆL)1of glucose plus

a supplement of trace levels of metal ions and vitamins Uniform15N/13C-labeling was accomplished by substitution

of unlabelled glucose with 2.0 gÆL)1of [13C6]glucose Protein purification was performed under denaturing conditions with Ni-nitriloacetic acid agarose beads (Qiagen) in the presence of 20 mM2-mercaptoethanol at pH values of 8.0, 6.3, 5.9 and 4.5 for the binding, two washing, and eluting steps, respectively Protein fractions were analyzed using 20% Phast gels (Amersham Biosciences) The fractions containing the SH2 domain were collected and refolded by dialyzing 2–3 times against a large volume of 50 mMsodium phosphate buffer plus 20 mM2-mercaptoethanol (pH 6.8)

at 4C The pellet was removed by centrifugation and the supernatant was concentrated by ultrafiltration (Millipore, Bedford, MA, USA) The protein concentration was determined at A280with a calculated extinction coefficient

of 12 210M )1Æcm)1[21]

The expression and purification of SH2 and GST-PDZ proteins were carried out with standard protocols provided by the supplier (Amersham Biosciences) Throm-bin cleavage was performed at room temperature for 2 h in

1· NaCl/Pibuffer The concentration of the PDZ protein was determined at A280 with a calculated extinction coefficient of 12 620M )1Æcm)1[21]

Peptide synthesis and purification All the peptides, either with or without phosphotyrosine (pY), were synthesized chemically using standard Fmoc solid-phase chemistry at the Biotechnology Research Insti-tute’s Peptide Facility The synthetic peptides were purified

by HPLC using a reverse-phase C18 semipreparative

column with an acetonitrile gradient of 10–35% The purity

of the peptides was verified by a reversed-phase analytical

HPLC column and the identity of the final products was

verified by mass spectral analysis and NMR assignments

Fluorescence polarization measurements

Fluorescein-labeled peptides were prepared through

reac-tion of the ephrinB2(301–333) peptide and its variants

containing a single phosphotyrosine residue (Fig 3), or

double phosphotyrosine residues with

5-iodoacetamido-fluorescein (Molecular Probes, Eugene, OR, USA) Upon

completion of labeling, an excess of 2-mercaptoethanol was

added to consume the excess 5-iodoacetamidofluorescein

followed by the removal of small organic compounds using

a C18 September-Pak column The eluate containing

fluorescein-labeled peptide was further purified by HPLC

The authenticity of fluorescein labeling was confirmed by

mass spectroscopy

Fluorescence polarization was performed at 25C on

a Perkin-Elmer fluorescence polarization instrument, the

EnVisionTM multilabel plate reader, which was equipped

with two fluorescence polarizers All the polarization values

are expressed in mili-polarization units (mP) Fluorescence

measurements were carried out with an excitation

wave-length of 490 nm and an emission wavewave-length of 520 nm

For binding studies, each fluorescein-labeled peptide was

dissolved in 50 mMphosphate buffer (pH 6.8, with 20 mM

2-mercaptoethanol) to a concentration of 25 nM The

dissociation constants were obtained by fitting the binding

curves using the computer program ORIGINTM 6.0

(Nor-thampton, MA, USA) based on the following equations:

Lþ P ()Kon

Koff LP where L and P denote the ligand and protein, respectively

The fluorescence polarization (DmP) is related to the

dissociation constant Kdas follows:

DmP¼C0½LT½P

Kdþ ½P

where C0 is a constant dependent on the properties of

the ligand, [L]Tis the total ligand concentration, [P] is the

concentration of free SH2 protein, and Kdis the equilibrium

dissociation constant Average Kdvalues were determined

from multiple independent measurements

NMR spectroscopy and resonance assignments

All NMR spectra were collected at 15C on a Bruker

Avance 500 MHz or 800 MHz spectrometer using

triple-resonance probes equipped with pulse field gradients

Protein samples for NMR analysis contained 0.5 mMof

uniformly15N- or15N-/13C-labeled protein Assignments of

the H/15N/13C NMR signals from the main-chain atoms

were obtained from a combined analysis of the 1H-15N

HSQC, CBCA(CO)NH and HNCACB experiments [22]

Proton NMR experiments with the peptides, which included

data acquisition, processing and analysis, were the same as

described previously [18,23]

GST pull-down assays

A GST pull-down experiment was employed to recons-titute the three-component molecular complexes formed

by the ephrin B peptides, the SH2 domain and the PDZ protein GST or GST-PDZ proteins bound to glutathione agarose beads were incubated with 5 lL of the SH2 protein in the binding buffer (50 mM sodium phosphate,

pH 6.8) for 2 h at 4C in the absence or presence of ephrin B2 peptides in excess amount The beads were washed extensively with the binding buffer and the samples were boiled for 10 min in the SDS/PAGE sample buffer and analyzed by SDS/PAGE Phast gel (Amersham Biosciences)

Results

Characterization of the expressed SH2 and PDZ protein domains

The SH2 domain (Fig 1A) of murine Grb4 was expressed using a synthetic DNA fragment subcloned for protein production in the E coli host (Experimental procedures) The over-expressed SH2 protein was found

in inclusion bodies and purified with Ni-nitiloacetic acid agarose beads under denaturing conditions The quality

of the expressed SH2 domain is improved at each step of the purification procedure (Fig 1B) A synthetic DNA fragment encoding the PDZ domain of PDZ-RGS3 was prepared in the same way as for the SH2 domain based

on the published amino acid sequence of the RGS3 protein [20] The PDZ domain was over-expressed as a fusion to the GST carrier protein in a mostly soluble form and the intact domain can be purified after digesting the GST-PDZ fusion protein with thrombin (Fig 1C) The expressed PDZ domain is reasonably soluble in solution and is functionally active as it has the same binding affinity to the ephrin-B2 peptide as the GST-PDZ fusion protein does (see below)

The1H-15N HSQC spectrum of the SH2 domain shows

a good dispersion of the 1H-15N correlation peaks (Fig 2A), indicative of a well-folded protein Nearly complete assignments of the main chain 1HN, 15N, 13Ca, and 13Cb NMR signals of the SH2 domain were obtainable using triple-resonance heteronuclear NMR experiments Figure 2B shows the differences between the 13Ca and 13Cb chemical shifts of an 15N/13C-labeled sample of the SH2 domain The secondary structure elements were deduced from the positive or negative derivations of the13Caand13Cbchemical shift differences from random coil values over a number of consecutive amino acids [24–26] Overall, secondary structures of the Grb4 SH2 domain are very similar to these of the Src SH2 domain (Fig 1A) However, there are some signifi-cant differences in the structural regions determining the binding specificity towards phosphopeptides For example, the b-sheet structure after the EF3 segment in the Src SH2 does not exist in the Grb4 SH2 domain while the insertion after the BG2, BG3 and BG4 structures in the Grb4 SH2 domain forms one additional b-sheet structure (Fig 1A and Fig 2B)

Identification of the binding site(s) on ephrin B2

for the SH2 domain of Grb4

Six synthetic peptides including ephrinB2(301–333) and

its five individual tyrosine-phosphorylated derivatives,

ephrinB2(301–333)-pY304, ephrinB2(301–333)-pY311,

ephrinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and

ephrinB2(301–333)-pY331 (Fig 3), were labeled with

fluorescein through the Cys301 residue at their N-termini

The highly sensitive fluorescence polarization method

was used to detect the binding interactions between the

peptides and the Grb4 SH2 domain The affinity of

the binding interactions was evaluated by measuring the

changes of fluorescence polarization of the peptides at

each step of titration with the Grb4 SH2 domain, i.e the

binding isotherms (Fig 4A) The

ephrinB2(301–333)-pY304 peptide was found to have the strongest binding

to the Grb4 SH2 domain while the other five peptides,

ephrinB2(301–333), ephrinB2(301–333)-pY311, eph-rinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331, showed low or no binding affinity to the SH2 domain The dissociation constant of the ephrinB2(301–333)-pY304/Grb4 SH2 complex was estimated to be 0.2 lM from the titration curves (Table 1) The calculated dissociation constants of the complexes between Grb4 SH2 and each of the other five peptides (i.e ephrinB2(301–333), ephrinB2(301–333)-pY311, ephrinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331) were larger than

500 lM These results show that only phosphorylated Tyr304 can result in high-affinity and specific binding of ephrinB2(301–333) to the Grb4 SH2 domain The fluorescence binding experiments also indicate that the Cys301 residue may not be required for binding as it is labeled by the bulky fluorescein group in the peptide fragments

Fig 1 Comparison of amino acid sequences of SH2 domains from Src and the mouse Grb4, and expression and purification of the Grb4 SH2 and RGS3 PDZ domains (A) Secondary structural elements are from the X-ray crystallographic structure of the Src SH2 domain in complex with a pYEEI peptide [35] The notation used for the binding pockets of the phosphotyrosine peptide is as described previously [35] Residues determining the binding specificity are underlined (B) SDS/PAGE of each purification step of the expressed SH2 domain of Grb4 Lane 1: the soluble part of the cell lysate; Lane 2: the insoluble part of the cell lysate; Lane 3: the Ni-nitiloacetic acid agarose beads bound with the SH2 protein at pH 8.0; Lane 4: the Ni-nitiloacetic acid agarose beads with the bound SH2 protein at pH 6.3; Lane 5: the eluted protein at pH 4.5 and Lane M: molecular mass markers (Amersham Bioscience) The sample of the Ni-nitiloacetic acid agarose beads with the bound SH2 protein at pH 5.9 was not shown here as the protein was already pure The position of the SH2 protein is marked by the arrow (C) SDS/PAGE patterns following each purification step of the expressed GST-PDZ Lane 1: the soluble part of the cell lysate The high-density band at the bottom is lysozyme, which was used to lyse the cells Lane 2: the purified GST-PDZ fusion protein by the glutathione-agarose beads Lane 3: The digestion of the GST-PDZ fusion protein by thrombin

at room temperature for 1 h The fusion proteins are readily and completely digested within 30 min Lane 4: the purified PDZ protein by glutathione-agarose beads followed by ion-exchange chromatography Lane M: molecular mass markers (Amersham Bioscience).

We next address the questions of whether secondary

tyrosine phosphorylations would affect the Grb4 SH2

domain binding to the ephrinB2(301–333)-pY304 peptides

and more specifically whether the combined

lation of Tyr311 and Tyr316 would replace

phosphory-lation at Tyr304 The choice of Tyr311 and Tyr316 was

because these two residues were reported to be the major

detectable phosphorylation sites in vivo [15] and a peptide

derived from residues 301–322 of ephrinB2 or the

N-terminal 22 residues of ephrinB2(301–333) was found

to contain all the phosphorylation sites for high-affinity

binding [11] As shown in Table 1, the phosphorylation of

Tyr311 or Tyr316 does not significantly affect the binding

of the ephrinB2(301–333)-pY304 peptide to the Grb4

SH2 domain Our data also show that the peptide with

double phosphorylations at Tyr311 and Tyr316 has no

significant binding to the Grb4 SH2 domain Therefore,

the high-affinity Grb4 SH2 binding of the ephrinB2(301–

333) fragment conferred by the Tyr304 phosphorylation is

independent of phosphorylations at the other two tyrosine residues, Tyr311 and Tyr316

To further assess the binding specificity of the Grb4 SH2 domain, we synthesized the short phosphorylated peptide, ephrin B2(301–309) or CPHpY304EKVSG with a number

of substitutions at the KV positions The single amino acid substitutions successively transformed the pYEKV sequence of ephrin B2 to the pYEEI sequence specific for the Src SH2 domain As shown in Fig 4B, only the CPHpY304EKVSG peptide exhibited high-affinity binding

to the Grb4 SH2 domain The dissociation constant of the CPHpY304EKVSG/Grb4 SH2 complex was estimated to

be 0.23 lMfrom the fluorescence titration curve, whereas all amino acid substitutions C-terminal to the pTyr304 led to dramatically reduced binding (Table 1)

Two-dimensional NMR spectroscopy was employed to investigate the specificity of binding of phosphorylated ephrinB2(301–333) peptides to the Grb4 SH2 domain The

1H-15N HSQC spectrum of the Grb4 SH2 domain (Fig 2A)

Fig 2 Characterization of the Grb4 SH2 by heteronuclear NMR (A) The1H-15N HSQC spectrum of the SH2 domain of Grb4 with the assignment of the amide 1 H- 15 N correlations to specific residues The measurement was performed at 15 C and pH 6.8 Sequence specific assignments of the backbone 15 N and 13 C resonances were achieved using triple-resonance NMR experiments with a 15 N/ 13 C-labeled SH2 sample (B) Secondary structure of the Grb4 SH2 protein deduced from the differences of the13C a and13C b chemical shifts of the

15

N/13C-labeled SH2 domain.

responded to the addition of the ephrinB2 peptide phos-phorylated at Tyr304, or ephrinB2(301–333)-pY304 (Fig 5A) Almost twice the numbers of HSQC peaks were found at lower peptide concentrations and most of the peaks redistributed at a molar ratio of 2 : 1 for the peptide-SH2 concentrations (Fig 5A, red spectrum) More signifi-cantly, in a stepwise titration experiment, the HSQC peaks did not undergo gradual shifting with increased concentra-tions of the peptide (data not shown), indicating that the free and peptide-bound SH2 must exchange slowly as observed for other SH2 domains In contrast, the nonphos-phorylated peptide, ephrinB2(301–333), did not induce any significant changes to the HSQC peaks up to a molar concentration ratio of 2 : 1 (Fig 5B, red spectrum) As well,

no significant changes in the HSQC peaks were observed in the presence of two other single tyrosine-phosphorylated peptides, ephrinB2(301–333)-pY311 and ephrinB2(301– 333)-pY316, similar to that of the unphosphorylated ephrinB2(301–333) (spectra not shown) Taken together, both the NMR and fluorescence polarization experiments

Fig 4 The effects of tyrosine phosphorylation on ephrin B2 binding to

the SH2 domain of Grb4 analyzed by fluorescence polarization.

(A) Fluorescence binding curves were collected for the unmodified

peptide ephrinB2(301–333) (j) and the three singly phosphorylated

peptides, ephrinB2(301–333)-pY304 (d), ephrinB2(301–333)-pY311

(m) and ephrinB2(301–333)-pY316 (.) Data for two other singly

phosphorylated peptides including ephrinB2(301–333)-pY330 and

ephrinB2(301–333)-pY331 are not shown, only their experimentally

determined binding constants are listed in Table 1 for comparison.

(B) Binding curves were collected for YEKV-pY304 (d),

YEKI-pY304 (j), YEKI-YEKI-pY304 (m) and YEEI-YEKI-pY304 (r) (Fig 3).

Fig 3 Schematic representation of peptides for fluorescence polarization assays The 33-residue peptide, ephrinB2(301–333), is derived from the extreme C-terminal sequence of the cytoplasmic domain of ephrin B2 The phosphorylated tyrosine residues are labeled with pY at the five individual sites, Tyr304, Tyr311, Tyr316, Tyr330 or Tyr331, for the five tyrosine-phosphorylated peptides, ephrinB2(301–333)-pY304, eph-rinB2(301–333)-pY311, ephrinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331 The consensus binding sequence of ephrin B2 for the Grb4 SH2 is minimized to a nine-residue segment, ephrinB2(301–309) or CPHpYEKVSG (referred to as YEKV-pY304) Three variants of the short peptide are generated to determine the binding specificity of the Grb4 SH2 in relation to the closest consensus binding sequence

of the Src SH2 domain (i.e pYEEI).

Table 1 The effect of tyrosine phosphorylations on the binding affinity

of ephrinB2(301–333) to the SH2 domain of Grb4 and to the PDZ domain of PDZ-RGS3 In the combination experiments, the two C terminal tyrosine residues were ignored due to their distant positions.

ND, not determined.

pY position K d (SH2) (l M ) K d (PDZ) (l M ) EphrinB2(301–333) series

EphrinB2(301–309) series

show that the phosphorylation of Tyr304 confers

high-affinity and specific binding of ephrinB2(301–333) to the

Grb4 SH2 domain

The ephrinB2(301–333)-pY304 peptide can bind to the

Grb4 SH2 and RGS3 PDZ domains simultaneously

It has been shown that the cytoplasmic tail of B-class

ephrins constitutes a PDZ domain binding motif, YYKV,

whose binding to PDZ domains is

phosphorylation-inde-pendent [8] Figure 6A shows that the GST-PDZ fusion

protein binds to the fluorescein-labeled ephrinB2(301–333)

peptide and this binding has an affinity of Kd¼ 3.0 lM

(Table 1) The binding affinities of the GST-PDZ fusion

protein to two single tyrosine phosphorylated peptides at

the extreme C-terminus (i.e the PDZ binding motif), i.e

ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331,

are essentially the same as that of the ephrinB2(301–333) peptide (Table 1) Therefore, the fluorescein-labeled eph-rinB2(301–333) peptides with single tyrosine phosphoryla-tion at Tyr304, Tyr311, Tyr316, Tyr330 or Tyr331 had essentially the same binding affinities to the GST-PDZ protein in agreement with previous observations [8] Binding experiments with the isolated PDZ domain showed similar binding affinities to the ephrin B peptides as that of the GST-PDZ protein Therefore, the GST-PDZ fusion protein was used for all other binding experiments including the GST pull-down experiments (see below)

On the other hand, the SH2 domain of Grb4 binds to the cytoplasmic tail of ephrin B2 in a phosphorylation-depend-ent manner It is therefore interesting to know whether Grb4 and PDZ-RGS3 can bind to the cytoplasmic tail

of ephrin B2 simultaneously or whether Grb4 binding can affect or even exclude the binding of the PDZ-RGS3

Fig 5 1 H- 15 N-HSQC spectra of the SH2

domain of Grb4 titrated by ephrinB2(301–333).

The spectra of the free Grb4-SH2 domain

(black) were overlaid on those in the presence

of ephrinB(301–333) peptides (red) (A)

Titration with ephrinB2(301–333)-pY304.

(B) Titration with ephrinB2(301–333) The

protein concentration was 0.5 m M and the

peptide concentration was 1 m M

protein We set out to address this question by use of the

SH2 domain of Grb4 and the PDZ domain of the RGS3

protein We titrated the SH2 protein into the

fluorescein-labeled ephrinB2(301–333)-pY304 after saturation by the

GST-PDZ fusion protein Consequently, the fluorescence

polarization increased further from that elicited by PDZ

binding with increasing concentrations of the SH2 protein

It was found that after normalization, the polarization

changes had similar trends as the titration experiments in

the absence of the GST-PDZ protein (Fig 6B) Therefore,

it appears that binding of the PDZ domain to the

ephrinB2(301–333)-pY304 peptide has no impact on the

binding of the Grb4 SH2 domain

Formation of three-component complexes among the

SH2 domain, the PDZ domain and ephrinB2(301–333)

peptides shown by GST pull-down experiments

In vitroGST pull-down experiments were also carried out to

verify the ternary interactions among the Grb4 SH2 and

RGS3 PDZ domains and the ephrinB2(301–333) peptides First, the Grb4 SH2 domain-binding peptide, eph-rinB2(301–333)-pY304, was used for the formation of the three-component complex with the Grb4 SH2 and RGS3 PDZ domains The purified GST-PDZ protein, which was bound to the glutathione-agarose beads, was mixed with the Grb4 SH2 domain and the ephrinB2(301–333)-pY304 peptide The beads were washed extensively to remove the unbound proteins The proteins bound to the bead were visualized by use of SDS/PAGE (Fig 7) As indicated by this kind of pull-down data, both the Grb4 SH2 and the RGS3 PDZ domains showed strong binding to the ephrinB2(301–333)-pY304 peptide in a three-component molecular complex Control experiments showed the lack of binding of the Grb4 SH2 domain to the GST-PDZ protein nor to the GST carrier protein Second, pull-down experi-ments were also carried out for the other five peptides including ephrinB2(301–333), ephrinB2(301–333)-pY311, ephrinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331 None of these peptides can link the Grb4 SH2 and the RGS3 PDZ domains to form a ternary complex (Fig 7)

Effects of tyrosine phosphorylation on the conformation

of ephrinB2(301–333) The unphosphorylated peptide, ephrinB2(301–333), was shown to form a well-folded b-hairpin structure for the putative SH2-binding region of residue 301–322 along with

a very flexible C-terminal tail from residues 323–333 [18] Phosphorylation at Tyr304, Tyr311 or Tyr316 does not affect the conformational characteristics of these PDZ-binding C-terminal tail residues For example, the typical NOE contact between the aCH proton of Ala327 and the

NH proton of Ile329, and the consecutive HN-HNNOEs from Lys329 to Tyr331, existed in all the peptides independent of phosphorylation (Fig 8A,B) Thus, the conformation of the PDZ binding domain of eph-rinB2(301–333) is essentially independent of tyrosine phosphorylation within the SH2-binding element of eph-rinB2(301–333)

On the other hand, tyrosine phosphorylation appears to have a profound effect on the b-hairpin conformation of the SH2-binding region First, all three kinds of single phos-phorylation at the Tyr304, Tyr311 or Tyr316 site caused perturbation to the loop region in the b-hairpin structure as the unique NOEs detected previously [18] completely disappeared in the NOESY spectrum of the phosphorylated ephrinB2(301–333) peptides, as shown in Fig 8A for ephrinB2(301–333)-pY304 Second, the consecutive back-bone HN-HNNOEs from residues Tyr316 to Gln319 were observed for ephrinB2(301–333)-pY304 and ephrinB2(301– 333)-pY316 (Fig 8B) but not for ephrinB2(301–333)-pY311 Third, there is a dramatic reduction of NOE contacts among the sidechains of aromatic residues pur-porting to side-chain packing interactions of a b-hairpin structure (Fig 8C) Many previously identified long-range NOEs such as those between Val318 and His303, Gln319 and His303, Lys306 and Tyr316 were absent in the phosphorylated peptides as shown in Fig 8C for eph-rinB2(301–333)-pY304 Interestingly, the long-range NOE contacts involving residues Lys306 and Tyr316, Glu305 and

Fig 6 Effect of PDZ binding on the interactions of the Grb4-SH2

domain with the ephrinB2(301–333)-pY304 peptide (A) The

PDZ-binding curves are for the fluorescein-labeled peptides including

eph-rinB2(301–333) (j), epheph-rinB2(301–333)-pY304 (d), ephrinB2(301–

333)-pY311 (m), ephrinB2(301–333)-pY316 (.) Corresponding data

for the phosphorylated peptides at Tyr330 and Tyr331 are listed in

Table 1 (B) Binding experiments performed with the

fluorescein-labeled ephrinB2(301–333)-pY304 peptide in the absence of the

GST-PDZ protein (d) or with 75 l M of the GST-PDZ protein (s).

Fig 7 Formation of a three-component complex through GST pull-down Lane M: molecular mass markers The molecular sizes of each band are 97.0, 66.0, 45.0, 30.0, 20.1 and 14.4 kDa from the top to the bottom Lane 1: the three-component complex of GST-PDZ, ephrinB2(301–333)-pY304 and the SH2 protein Lane 2: sample prepared similarly as that in Lane 1 except for the absence of ephrinB2(301–333)-ephrinB2(301–333)-pY304 Lane 3: sample prepared the similarly as that in Lane 2 except the GST was used instead of GST-PDZ Lane 4–8: samples prepared the same as that in Lane 1 except that the peptides are ephrinB2(301–333), ephrinB2(301–333)-pY311, ephrinB2(301–333)-pY316, ephrinB2(301–333)-pY330 and ephrinB2(301–333)-pY331, respectively.

Fig 8 Homonuclear NOESY spectra of the ephrinB(301–333)-pY304 peptide (A) NH-aH region of the NOESY spectrum (red) with a mixing time

of 200 ms in H 2 O overlaid with a TOCSY spectrum (dark) The cross indicates the missing medium-range NOEs in ephrinB(301–333)-pY304 compared with those in the unphosphorylated ephrinB(301–333) [18] (B) The NH-NH NOE connectivities for residues Tyr316-Gln319 and Lys329-Tyr331 (C) The aromatic region of the NOESY spectrum of the ephrinB(301–333)-pY304 peptide The NOE experiments were carried out with a mixing time of 200 ms in H 2 O at pH 6.8 and 288 K Some key medium- and long-range NOEs are still present, but many long-range NOEs are absent compared to the unphosphorylated ephrinB2(301–333) peptide [18] Interestingly, the number of side-chain NOE contacts in the two other phosphorylated peptides, ephrinB2(301–333)-pY311 and ephrinB2(301–333)-pY316, was further reduced and almost all NOEs have disappeared.

Tyr316 remained in ephrinB2(301–333)-pY304 but not in

the two other phosphorylated peptides

Figure 9 shows a structural model of

ephrinB2(301–333)-pY304, generated from the NOE data of Fig 8 Compared

to the solution conformation of the nonphosphorylated

peptide [18], phosphorylation at Tyr304 dramatically

increased the flexibility of the b-hairpin structure and

essentially unfolded it by phosphorylation at Tyr311 or

Tyr316 Instead, residues around Tyr316 were found to

form a helical structure with the ephrinB2(301–333)-pY304

peptide (Fig 9) Phosphorylation of Tyr316 destroys even

this helical structure as typical NOEs determining the short

helix were not observable in the ephrinB2(301–333)-pY316

peptide In contrast, the short helix found for the

PDZ-binding motif [18] remains unperturbed by any of the

phosphorylations Taken together, these results indicate

that phosphorylated ephrinB2(301–333) peptides become

much more flexible especially in the N-terminal region

around residue Tyr304 or the PHY304EKV sequence

region This kind of flexible conformations may be more

favorable for Grb4 SH2 domain binding, as almost all

SH2-binding phosphotyrosine peptides appear to assume

exten-ded conformations in the binding site of the complex [27]

Discussion

Using fluorescence polarization and NMR spectroscopy, we have identified a sequence segment around Tyr304 or PHpY304EKV on a C-terminal 33-residue peptide of the ephrin B2 cytoplasmic region as a high-affinity binding site for the Grb4 SH2 domain, in which tyrosine phosphory-lation is critical We also show that phosphoryphosphory-lation of two distant tyrosine residues, i.e Tyr311 or Tyr316, did not affect the binding of the pTyr304 motif to the Grb4 SH2 domain Some other SH2 domains, for instance, p85N-SH2 [28], were found to have higher affinity for doubly tyrosine-phosphorylated peptides Therefore, our results indicate that the phosphorylations of Tyr311 and Tyr316 identified

in vivomay have unknown alternative functions other than physical interactions with the Grb4 protein Earlier work already provided some evidence for low levels of in vivo phosphorylation at an equivalent position of residue Tyr304

in the Xenopus ephrin B1 [16] Other studies implicate Tyr311 and/or Tyr316 as the major phosphorylation sites [15,16] However, it is not known whether these two phosphorylated tyrosine residues contribute to high-affinity binding of the ephrin B cytoplasmic domain to the Grb4 adaptor protein [11] The equivalent Tyr304 residue of Xenopusephrin B1 was shown to be essential for binding to Grb4 in a recent work employing truncations and residue substitutions of ephrin B1 expressed in Xenopus oocytes cells [29] This latter work also did not find evidence for the involvement of phosphorylated Tyr311 or Tyr316 in ephrin B1 binding to the Grb4 protein, in sharp contrast to a previous claim otherwise [30] It is now clear that phos-phorylated Tyr304 alone determines the direct binding to Grb4 while sequence elements surrounding the Tyr311 site are important for the interactions of ephrin B with tyrosine kinases or phosphatases [29]

Structurally, the Grb4 SH2 domain binding motif on B ephrins, pYEKV, is somewhat different from other SH2 binding sequences [31,32], for example, pYEEI, pYDNV, pYTDM and pYTDL for Src family SH2 domains; pYENP, pYTEV and pYMDL for the Abl SH2 domain; pYDHP, pYKFL and pYNR for the CrK SH2 domain; pYDEP, pYDED and pYDEV for the Nck SH2 domain; pYLNV, pYLNV, pYIN and pYMN for the Sem5 SH2 domain; pYMXM, pYVXM, pYIXM and pYEXM for the N-terminal SH2 domain of p85; pYXXM for the C-terminal SH2 domain of p85; pYVIP, pYILI and pYILV for the C-terminal SH2 domain of PLC-r; pYELE, pYIDI and pYVDV for the N-terminal SH2 domain of PLC-r; and pYIXV, pYVXI, pYVXL and pYVXP for the N-terminal SH2 domain of SHPTP2 The different SH2 domain binding motifs provide diver-sity to signal transduction through a wide range of SH2 domain-containing proteins [31] Through the alignment

of the Grb4 and Src SH2 domains (Fig 1A), it is seen that the phosphotyrosine binding pocket of the Grb4 SH2 domain is almost identical to that of the Src SH2 domain, being composed of residues in the bA2, bB5, bD3, bD4, bD 6 and bD¢1 structural elements On the other hand, the local environment of the Grb4 SH2 domain binding pocket is different from that in the Src SH2 domain, which is composed of residues at the bD5, bE4, EF1, EF3, BG2, BG3 and BG4 sites This particular

Fig 9 Model of the flexible structure of ephrinB2(301–333)-pY304 in

solution A cluster of solution conformations was generated by use of

NOE data shown in Fig 7, specifying the secondary structure elements.

Residues for the Grb4 SH2 domain binding, i.e PHpY304EKV, are

colored in blue, while the tail residues IY330YKV for PDZ domain

binding are colored in red This representation of the peptide

confor-mation was prepared using INSIGHT II (Tripos, San Diego, CA, USA).