GRP78 autoantibodies initiate the breakdown of the blood-brain barrier in neuromyelitis optica

Antibody-based therapy offers vast potential in the treatment of several neurodegenerative diseases. Many drug companies are now engineering antibodies targeting β-amyloid protein for Alzheimer disease and α-synuclein protein for Parkinson disease. However, the development of antibody-based therapies against the central nervous system (CNS) diseases has lagged, partly due to the poor blood-brain barrier (BBB) permeability of large molecules [1]. In a recent study published in Science Translational Medicine [2], we searched for targets of anti-endothelial cell antibodies that can modulate BBB permeability in patients with neuromyelitis optica (NMO). NMO is a disabling autoimmune CNS disease. The main clinical manifestations of NMO are recurrent optic neuritis and longitudinal transverse myelitis [3]; however, widespread lesions are frequently observed throughout the brain and brain stem. The discovery of serum autoantibodies for the water channel aquaporin-4 (AQP4) expressed on astrocyte endfeet (known as NMO-IgG or AQP4-IgG) has highlighted NMO as a distinct clinical entity from multiple sclerosis [4]. The perivascular binding pattern of NMO-IgG on rodent brain tissue sections was consistent with reactivity against AQP4, which is predominantly found on astrocyte foot processes forming the glia limitans of the BBB. The presence of serum AQP4-IgG has facilitated the clinical diagnosis and early treatment of NMO patients worldwide [4]. Antibody-depleting therapy and B-cell targeting therapy for the removal of AQP4-IgG are effective for reducing the disease activity in NMO patients. Several lines of experimental evidence support that the binding of AQP4-IgG to AQP4 on astrocytes initiates complementand antibody-dependent cellular cytotoxicity in target astrocytes. Thus, NMO is considered to be the first CNS autoimmune disease for which a specific tissue target molecule has been identified. AQP4-IgG is mostly produced by peripheral plasmablasts rather than intrathecal plasmablasts, which suggests that CSF AQP4-IgG is derived from serum. However, the presence of serum AQP-IgG alone is not sufficient to cause NMO disease [5]. Although the peripheral injection of AQP4-IgG did not induce NMOlike histopathology when injected into naive animals, it caused disease in experimental animals with pre-existing T-cell mediated CNS inflammation. AQP4-IgG was detectable in the sera of some NMO patients for many years before the onset of the disease, which suggests that AQP4-IgG can persist in the circulation without causing any evident disease. Thus, the BBB breakdown that allows for the subsequent penetration of AQP4-IgG into the brain seems a prerequisite for the development of NMO. Clinical observations demonstrating the Gdenhanced lesions on MRI, increased albumin leakage in the cerebrospinal fluid (CSF) in the acute phase of NMO, and diffuse perivascular lesions along with pathological evidence of the disruption of the BBB would support the significance of BBB disruption in the pathogenesis of NMO. How circulating AQP4-IgG can access the AQP4 on the astrocyte endfeet behind the BBB in NMO is poorly understood. One possible route is through circumventricular organs such as the area postrema, where the endothelial cells lack tight junctions and the AQP4 expression is enriched [6], as indicated by the observation that NMO lesions are frequently observed in the hypothalamus and the periaqueductal brainstem surrounding the circumventricular system. After AQP4IgG enters the CSF space, it can freely access the target astrocytes and cause BBB dysfunction via intrathecal inflammation. The other possible route is via direct BBB transit. However, serum AQP4-IgG cannot open the BBB itself, as the BBB-endothelial cells do not express the AQP4 protein. We recently identified a distinct endothelial cell targeted-autoantibody that enhances the BBB transit of AQP4-IgG in NMO [2]. First, we identified two Editorial