Orthogonal array formation by human aquaporin-4: examination of neuromyelitis optica-associated aquaporin-4 polymorphisms

Abstract

Pathogenic autoantibodies target aquaporin-4 (AQP4) water channels in individuals with neuromyelitis optica (NMO). Recently, allelic mutations were reported at residue 19 of AQP4 in three cases of NMO, and it was suggested that polymorphisms may influence disease by altering AQP4 supramolecular assembly into orthogonal arrays of particles (OAPs). We analyzed the determinants of OAP formation by human AQP4 to investigate the possible role of polymorphisms in NMO pathogenesis. NMO-associated mutations R19I and R19T in AQP4 did not affect OAP assembly, palmitoylation-dependent regulation of assembly, or NMO autoantibody binding. Residue-19 polymorphisms in AQP4 are thus unlikely to be disease relevant.

Fig. 1

OAP-forming properties of human AQP4. A. Schematic diagram of the N-terminus ofM1 AQP4 showing the human sequence. Sites of cysteine palmitoylation are shown in red. Arg-19, which is mutated in a small subset of NMO patients, is shown in green. Hydrophobic residues that are required for OAP assembly by M23 AQP4 are shown in yellow. Transmembrane domains are shown in light blue. B. Representative single particle trajectories from quantum dot-labeled human AQP4 isoforms M23 (black) and M1 (red). Trajectories were acquired at 91 Hz for 6 s. Bar, 1 μm. C. Cumulative distribution of the diffusion range at 1 s for AQP4 isoforms M23 (black) and M1 (red). Both human (solid) and rat (dotted) diffusion profiles are shown for comparison.

Fig. 2

Arg-19 mutations in AQP4 do not affect AQP4 OAP assembly. A. Cumulative distribution of the range at 1 s for M1 Arg-19 mutants R19I (red), R19T (green) and double cysteine mutant CCA (blue). Top panel: diffusion of pure AQP4 isoforms or mutants. Middle panel: Diffusion of 1:1 mixtures of M23 and M1 (orange), or M23 and mutants (same colors as top). Bottom panel: Diffusion of 3:1 mixtures of M23 and M1 or mutants (same colors as middle). Diffusion of pure AQP4 isoforms M23 (black) and M1 (grey) are shown in each panel. B. Diffusion coefficients for pure AQP4 isoforms or M1 mutants and 3:1 mixtures of M23 and M1 or mutants (mean ± S.E., n> 16 cells). * P <0.01 when compared to non-mutated M1. C. AQP4 immunoblot following blue-native gel electrophoresis (top) or Tricine SDS-PAGE (bottom) of lysates from U87MG cells transfected with AQP4 isoforms, M1 mutants, or mixtures composed of 3:1 M23 to M1 (native or mutant).

Fig. 3

Arg-19 mutations in AQP4 do not affect NMO autoantibody binding. A. Fluorescence micrographs showing M1 and M23 expressing U87MG cells stained with 10% NMO serum (green), and with reference AQP4 antibody (red). B. Binding curves for NMO patient serum to M23 or M1 AQP4, and M1 Arg-19 mutants. Left panel: fraction of NMO-IgG bound to cells expressing pure M23 (black circles), M1 (open circles), R19I (red) or R19T (green). Right panel: fraction of NMO-IgG bound to cells expressing 3:1 mixtures of M23 to M1 (black), R19I (red) or R19T (green). Circles represent mean ± S.E., n =5). Curves represent fit to single-site binding model. C. Binding of NMO serum (measured at 10%) from three additional different NMO patients to M1, R19I and R19T alone and in combination with M23 (mean ± S.E., n =3, differences not significant).