Supramolecular aggregation of aquaporin‐4 is different in muscle and brain: correlation with tissue susceptibility in neuromyelitis optica

Abstract

Neuromyelitis optica (NMO) is an autoimmune demyelinating disease of the central nervous system (CNS) caused by autoantibodies (NMO‐IgG) against the water channel aquaporin‐4 (AQP4). Though AQP4 is also expressed outside the CNS, for example in skeletal muscle, patients with NMO generally do not show clinical/diagnostic evidence of skeletal muscle damage. Here, we have evaluated whether AQP4 supramolecular organization is at the basis of the different tissue susceptibility. Using immunofluorescence we found that while the sera of our cohort of patients with NMO gave typical perivascular staining in the CNS, they were largely negative in the skeletal muscle. This conclusion was obtained using human, rat and mouse skeletal muscle including the AQP4‐KO mouse. A biochemical analysis using a new size exclusion chromatography approach for AQP4 suprastructure fractionation revealed substantial differences in supramolecular AQP4 assemblies and isoform abundance between brain and skeletal muscle matching a lower binding affinity of NMO‐IgG to muscle compared to the brain. Super‐resolution microscopy analysis with g‐STED revealed different AQP4 organization in native tissues, while in the brain perivascular astrocyte endfoot membrane AQP4 was mainly organized in large interconnected and raft‐like clusters, in the sarcolemma of fast‐twitch fibres AQP4 aggregates often appeared as small, relatively isolated linear entities. In conclusion, our results provide evidence that AQP4 supramolecular structure is different in brain and skeletal muscle, which is likely to result in different tissues susceptibility to the NMO disease.

The evolving mystery of why skeletal muscle is spared in seropositive neuromyelitis optica

“…In the light of this, the data in Figure 1 of Rosito et al [1] showing minimal or non-specific AQP4-IgG binding to skeletal muscle in seropositive NMO are perplexing, as other studies have shown that systemically administered AQP4-IgG rapidly binds to rodent skeletal muscle [8] and that IgG binding is found on skeletal muscle in humans with NMO myositis [9]. One potential explanation for this discrepancy might be loss of AQP4 antigenicity for autoantibody binding in the non-fixed frozen muscle used by Rosito et al [1].If AQP4 cluster size does not account for sparing of skeletal muscle in NMO, then what does? One possibility is complement inhibitor protein CD59, which is strongly expressed in skeletal muscle and other AQP4-expressing tissues in the periphery such as kidney and stomach.…”

“…Rosito et al [1] report evidence that differences in the supramolecular organization of AQP4 in skeletal muscle versus brain astrocytes are responsible for the sparing of skeletal muscle in NMO, arguing that the size of AQP4 supramolecular clusters (called orthogonal arrays of particles, OAPs) is smaller in skeletal muscle than in astrocytes resulting in reduced AQP4-IgG binding. This mechanism is motivated by the observation that some AQP4-IgG autoantibodies bind better to AQP4 OAPs than to separated AQP4 tetramers [2] and that AQP4 OAPs are required for C1q binding and complement activation [3].…”

Response to ‘The evolving mystery of why skeletal muscle is spared in seropositive neuromyelitis optica’

“…The use of g‐STED super‐resolution microscopy versus freeze‐fracture electron microscopy (FFEM) to analyse skeletal muscle and brain AQP4 supramolecular assemblies (OAPs) used in our study 1 has been disputed by Verkman et al . While we agree that FFEM is the gold standard to visualise OAPs and also measure their size, we also are aware that the very small amount of the plasma membrane that can be suitable for analysis represents a major limit to obtaining statistically significant data (such as the OAP dimension) representative of the entire tissue.…”

“…

We would like to thank you for the opportunity to respond to the questions raised in Dr Verkman's letter and to elucidate related aspects. We also thank Dr Verkman and colleagues for their attention to our study.The use of g-STED super-resolution microscopy versus freezefracture electron microscopy (FFEM) to analyse skeletal muscle and brain AQP4 supramolecular assemblies (OAPs) used in our study [1] has been disputed by Verkman et al While we agree that FFEM is the gold standard to visualise OAPs and also measure their size, we also are aware that the very small amount of the plasma membrane that can be suitable for analysis represents a major limit to obtaining statistically significant data (such as the OAP dimension) representative of the entire tissue. In contrast, STED microscopy has the enormous advantage of analysing, in real time, very large portions of the plasma membrane, with a resolution that in our setup can reach approximately 30 nm, providing the possibility to have a more complete vision of the entire tissue and handle a large amount of data.Considering the 'contradiction of available data', Verkman et al refer to an FFEM study on OAP structure and organisation performed before the identification of AQP4 (or MIWC) as the molecular determinant of OAPs [2].

…”

“…The use of g-STED super-resolution microscopy versus freezefracture electron microscopy (FFEM) to analyse skeletal muscle and brain AQP4 supramolecular assemblies (OAPs) used in our study [1] has been disputed by Verkman et al While we agree that FFEM is the gold standard to visualise OAPs and also measure their size, we also are aware that the very small amount of the plasma membrane that can be suitable for analysis represents a major limit to obtaining statistically significant data (such as the OAP dimension) representative of the entire tissue. In contrast, STED microscopy has the enormous advantage of analysing, in real time, very large portions of the plasma membrane, with a resolution that in our setup can reach approximately 30 nm, providing the possibility to have a more complete vision of the entire tissue and handle a large amount of data.…”