Do All X-ray Structures of Protein-Ligand Complexes Represent Functional States? EPOR, a Case Study

Abstract

Based on differences between the x-ray crystal structures of ligand-bound and unbound forms, the activation of the erythropoietin receptor (EPOR) was initially proposed to involve a cross-action scissorlike motion. However, the validity of the motions involved in the scissorlike model has been recently challenged. Here, atomistic molecular dynamics simulations are used to examine the structure of the extracellular domain of the EPOR dimer in the presence and absence of erythropoietin and a series of agonistic or antagonistic mimetic peptides free in solution. The simulations suggest that in the absence of crystal packing effects, the EPOR chains in the different dimers adopt very similar conformations with no clear distinction between the agonist and antagonist-bound complexes. This questions whether the available x-ray crystal structures of EPOR truly represent active or inactive conformations. The study demonstrates the difficulty in using such structures to infer a mechanism of action, especially in the case of membrane receptors where just part of the structure has been considered in addition to potential confounding effects that arise from the comparison of structures in a crystal as opposed to a membrane environment. The work highlights the danger of assigning functional significance to small differences between structures of proteins bound to different ligands in a crystal environment without consideration of the effects of the crystal lattice and thermal motion.

Figure 1

Crystal structures of the soluble EPOR extracellular domain dimers. The extracellular domain is shown bound to EPO (A), in the apo form (B), bound to EMP1 (C), and bound to EMP33 (D) (PDBs: 1EER, 1ERN, 1EBP, and 1EBA, respectively) (3, 4, 5, 6). (A–D) (Green) Receptor chain A; (blue) receptor chain B; (red) ligand (A), (orange) ligand (C), and (purple) ligand (D). The four residues in the hinge region connecting the N-terminal D1 domain and the C-terminal D2 domain within a chain are represented (yellow and magenta spheres) in chain A and chain B, respectively. Note, the extracellular juxtamembrane region, and the transmembrane and cytosolic domains are depicted schematically in (A). (E) Top view (looking down onto the cell) of the two D1 domains in the EPOR dimer from the crystal structures of the EPOR bound to EPO (blue), EMP1 (red), EMP33 (green), and in the apo state (magenta). The structures in (B)–(E) have been aligned on the receptor chain A of (A) as a reference. To see this figure in color, go online.

Figure 2

The cross-action scissorlike motion for the EPOR proposed from the apo and EMP1-bound dimer. (Green) Receptor chain A; (blue) receptor chain B; (red) cytokine. (Orange spheres) Protein kinase domains of the JAK protein associated with the cytosolic domains of the receptor that require the correct alignment of two domains to transphosphorylate. To see this figure in color, go online.

Figure 3

Overlays of the individual sEPOR chain conformations obtained from all simulations of sEPOR₂–EPO (A), apo-sEPOR₂ (B), sEPOR₂–EMP1 (C), and sEPOR₂–EMP33 (D). In each case the chain has been superimposed onto the backbone atoms of the hinge residues Val¹¹⁸–Leu¹²¹. To see this figure in color, go online.

Figure 4

RMSD values for the sEPOR₂ dimer calculated over the backbone atoms after first performing a least-squares fit onto the backbone atoms of the chain A. sEPOR₂–EPO (A), apo-sEPOR₂ (B), sEPOR₂–EMP1 (C), and sEPOR₂–EMP33 (D). (Green, red, blue, orange, gray) Independent simulations 1–5, respectively. To see this figure in color, go online.

Figure 5

(A) The overall conformations of the sEPOR chains within the dimer were determined in terms of an angle between the chains (θ) and the distance between the center of mass of the D2 domains ($D{2}_d$). (B–E) Scatter plots of θ against $D{2}_d$ for simulations of the sEPOR₂–EPO (B), sEPOR₂ (C), sEPOR₂–EMP1 (D), and sEPOR₂–EMP33 (E) starting from the crystal structures. (Green, red, blue, orange, gray) Independent simulations 1–5, respectively. (Black dots) Values of the initial conformations. All simulations were performed for 50 ns, except for sEPOR₂–EPO simulations, which ran for 80 ns. To see this figure in color, go online.

Figure 6

The overall conformation of the sEPOR chains in terms of θ and $D{2}_d$ (see Fig. 5A) for the systems sEPOR₂-0 (A), sEPOR₂-20 (B), sEPOR₂–EPO R103A (C), sEPOR₂–EMP6 (D), sEPOR₂–EMP7 (E), sEPOR₂–EMP8 (F), and sEPOR₂–EMP16 (G). (Green, red, blue) Simulations 1–3, respectively. (Black dots) Values of the initial conformations. All simulations were performed for 50 ns. To see this figure in color, go online.

Figure 7

The overall conformation of the sEPOR chains in the sEPOR₂–EPO (A) and sEPOR₂–EMP1 (B) crystal lattice in terms of θ and $D{2}_d$ (see Fig. 5A). The values plotted are from the final 2.5 ns of a 15-ns unrestrained simulation after equilibration for the eight individual asymmetric units within the unit cell. (Purple, green, red, aqua, magenta, orange, blue, crimson) The asymmetric units, respectively. (Black circle) Initial values in the x-ray crystal structure; (gray square) values obtained from the structure of the average asymmetric unit. To see this figure in color, go online.