Protein folding at the membrane interface, the structure of Nogo-66 requires interactions with a phosphocholine surface

Abstract

Repair of damage to the central nervous system (CNS) is inhibited by the presence of myelin proteins that prevent axonal regrowth. Consequently, growth inhibitors and their common receptor have been identified as targets in the treatment of injury to the CNS. Here we describe the structure of the extracellular domain of the neurite outgrowth inhibitor (Nogo) in a membrane-like environment. Isoforms of Nogo are expressed with a common C terminus containing two transmembrane (TM) helices. The ectodomain between the two TM helices, Nogo-66, is active in preventing axonal growth [GrandPre T, Nakamura F, Vartanian T, Strittmatter SM (2000) Nature 403:439-444]. We studied the structure of Nogo-66 alone and in the presence of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) vesicles and dodecylphosphocholine (DPC) micelles as membrane mimetics. We find that Nogo-66 is largely disordered when free in solution. However, when bound to a phosphocholine surface Nogo-66 adopts a unique, stable fold, even in the absence of TM anchors. Using paramagnetic probes and protein-DPC nuclear Overhauser effects (NOEs), we define portions of the growth inhibitor likely to be accessible on the cell surface. With these data we predict that residues (28-58) are available to bind the Nogo receptor, which is entirely consistent with functional assays. Moreover, the conformations and relative positions of side chains recognized by the receptor are now defined and provide a foundation for antagonist design.

Fig. 1.

(A) CD spectra of Nogo-66 (110 μM) in an aqueous environment (Red), in the presence of 100-nm diameter, unilamellar DMPC vesicles (10 mM, Black), and in the presence of DPC micelles (10 mM, Blue). (B) ¹⁵N HSQC spectra of Nogo-66 (1 mM ) in presence of 200 mM DPC, at 35 °C. (C) Thirty-three long-range NOEs (|i - j|≥5 Å) were used in structure determination of Nogo-66 in presence of DPC. (D) Eighteen additional long-range distance restraints (25 Å ≤ |i - j|≥15 Å) obtained as PREs from samples with the paramagnetic reagent MTSL attached at either position 31 or 42.

Fig. 2.

Solvent or lipid/DPC accessibility is defined using paramagnetic reagents and NOEs. Solvent accessibility of Nogo-66 in DPC was probed using paramagnetic manganese and gadolinium. Mn⁺² broadens signals of residues exposed to the solvent as well as those interacting within the head group region, whereas the compound gadodiamide will affect only those residues exposed to the solvent. Similarly, residues in contact with the paramagnetic lipid, 1-palmitoyl-2-stearoyl-(7-doxyl)-sn-glycero-3-phosphocholine are broadened by the 7-doxyl. The schematic (Left) demonstrates regions of accessibility of the various paramagnetic reagents including a few DPC molecules. In addition, protein groups interacting with DPC were identified by comparing ¹³C-edited NOESYHSQC spectra of Nogo-66 in presence of d-DPC to Nogo-66 in h-DPC. Spectra in h-DPC contained additional NOE peaks arising from specific protein-lipid interactions (Fig. S5). Residues affected by paramagnetic agents are shown on the structure of Nogo-66. Top panels depict affects of gadodiamide (Left) and Mn⁺² (Right). Bottom panels depict residues broadened by 7-doxyl (Left) and residues interacting with DPC based on NOEs (Right).

Fig. 3.

(A) Figure summarizing the main finding of the paper. NMR experiments reveal Nogo-66 to be a random coil in aqueous solution. However, upon association with a phosphochline surface, the protein adopts a unique, stable fold. The orientation of Nogo-66 in the lipid environment was guided by the observed NOEs between Nogo-66 and DPC and accessibility to three paramagnetic probes. Docking was based on the collective analysis of over 200 protein signals and reveals the conformation of Nogo groups available to bind receptor. (B) Aromatic residues, Phe and Tyr, are positioned to interact with lipid head groups. (C) Hydrophobic residues (Ile, Leu, Val, Phe, Met, Tyr) are depicted; such residues appear both within the protein core and exposed to lipid.

Fig. 4.

Relative dimensions of Nogo-66 (Dark Blue), the Nogo receptor (Light Blue), and POPC lipid bilayers (Gray, Left). Nogo-66 could easily bind within the concave surface of the Nogo receptor. The C-terminal domain (CTD) of the Nogo receptor is anchored to the axonal membrane through a glycophosphatidylinositol (GPI) group. Details of the molecular interactions of the Nogo receptor with coreceptors, p75 and LINGO, are not known. Within the bilayer, 15 Å of head group space is approximated (HG) on both sides of the acyl-chain (AC) region. Looking down on the cell membrane, panels on the right depict regions of Nogo exposed to the receptor; these form a tightly packed helical cluster. Residues 31–55, shown to have highest growth inhibition activity are highlighted in red (Top Right) (8). A putative antagonist (10) comprised of the first 40 residues of Nogo-66 are shown in green (Bottom Right). Most of the antagonist is actually either buried in lipid or occluded by Nogo-66 protein. The exposed region of putative antagonist consists of a short fragment (residues 28–40) which forms a helix-turn-helix.