Crystal structure of the pleckstrin homology-phosphotyrosine binding (PH-PTB) targeting region of insulin receptor substrate 1

Abstract

We have determined the crystal structure at 2.3-A resolution of an amino-terminal segment of human insulin receptor substrate 1 that encompasses its pleckstrin homology (PH) and phosphotyrosine binding (PTB) domains. Both domains adopt the canonical seven-stranded beta-sandwich PH domain fold. The domains are closely associated, with a 720-A(2) contact surface buried between them that appears to be stabilized by ionic, hydrophobic, and hydrogen bonding interactions. The nonconserved 46-residue linker between the domains is disordered. The PTB domain peptide binding site is fully exposed on the molecular surface, as is a large cationic patch at the base of the PH domain that is a likely binding site for the head groups of phosphatidylinositol phosphates. Binding assays confirm that phosphatidylinositol phosphates bind the PH domain, but not the PTB domain. Ligand binding to the PH domain does not alter PTB domain interactions, and vice versa. The structural and accompanying functional data illustrate how the two binding domains might act cooperatively to effectively increase local insulin receptor substrate 1 concentration at the membrane and transiently fix the receptor and substrate, to allow multiple phosphorylation reactions to occur during each union.

Figure 1

Sequence alignment of the amino-terminal domains of human IRS-1 and IRS-2 and rat IRS-3 and IRS-4. Secondary structural elements of IRS-1 are shown above the alignments, and colored green (β-sheets) or turquoise (α-helices). Residues of the PTB domain that bind IR are labeled with red (phosphate binding) or black (all others) squares (13). PH domain residues forming the cationic patch at its base are labeled with black squares. Residues that are buried at the interface between domains are identified either as contributing to the hydrophobic patch (φ) or as a specific interaction [e.g., the PH domain residue labeled a (Arg-75) forms a salt bridge with PTB domain residue a (Glu-162), etc.]. Note the high degree of residue conservation at the interface and the great variability in length and composition of the interdomains of the four proteins.

Figure 2

Structure of the IRS-1 targeting domain. (A) Ribbon diagram of the PH-PTB structure, with β-sheets shaded green, α-helices in turquoise, 3₁₀ turns colored indigo, and intervening coils or loops in brown. Corey–Pauling–Koltun space-filling model oriented as the ribbon diagram; residues within the PTB domain binding site (facing away) and putative PH domain binding site are colored dark gray and numbered; residues at the interface between the two domains are colored according to percent buried: red, 50–100%; orange, 25–50%; and yellow, 1–25%. (B) The PH/PTB domain interface viewed as an open book. The PH and PTB domains each were rotated 90^o, relative to their orientations in A, but in opposite directions to expose the buried surface between them. Elements of secondary structure and contact residues are labeled.

Figure 3

Structure and function of individual domains. Ribbons (Upper) and surface potential (Lower) diagrams of the IRS-1 PTB and PH domains and the PLC-δ₁ PH domain are similarly oriented. The IRS-1 PTB domain is bound to the IR juxtamembrane NPXpY peptide; the PLC-δ₁ PH domain is bound to Ins(1,4,5)P₃. Solvent-accessible surfaces are shaded according to electrostatic potential, −5 kt/e, red to +7 kt/e, blue, by using the program grasp. The Ins(1,4,5)P₃ binding pocket at the base of the PLC-δ₁ domain is positively charged. An analogous pocket is at the base of the IRS-1 PH domain; the PTB domain has a distinct mode of binding.

Figure 4

Binding assays. (A) The PH domain binds [³H]dioctanoyl PI(3,4,5)P₃; the PTB domain does not. We are not reporting K_D values for interactions with the isolated PH domain protein because of its inherent instability. (B) The IRS-1 PH-PTB domain protein binds [³H]dioctanoyl PI(3,4,5)P₃ with a K_D value of 3.5 ± 1.0 μM. (C) Binding of phosphatidylinositides to the PH-PTB protein. Competition assays were used to determine relative affinities of PH-PTB toward phosphatidylinositides and phosphoinositides, by using [³H]dioctanoyl PI(3,4,5)P₃ as the tracer.

Figure 5

Model for insulin signaling through the IR/IRS axis. (A) Under basal conditions in the absence of insulin, IR (green) is not phosphorylated (TM, transmembrane domain; JM, juxtamembrane domain; IRK, IR kinase). IRS (blue) associates reversibly with PI(4,5)P₂ (red) in the plasma membrane. Phosphates are represented by the letter P within closed black circles. (B) Insulin stimulates IR activation, leading to phosphorylation of three Tyr residues in the activation loop of IRK and one in the JM NPXY motif. Membrane-bound IRS proteins associate with the receptor JM, which transiently fixes the two proteins for phosphorylation of multiple tyrosine residues in the IRS activation domain.