We recently used our model of experimental neuromyelitis optica spectrum disorder in Lewis rats to show the existence of highly pathogenic aquaporin-4 (AQP4)-specific T cells in Lewis rats. These T cells recognize AQP4 268-285 as their specific antigen, are reactivated behind the blood-brain barrier and deeply infiltrate the central nervous system parenchyma of the optic nerves, the brain and the spinal cord. Thus, these cells differ from T cells with other AQP4 peptide specificities, which are essentially confined to the meninges, as a result of the low local availability of "their" antigen precluding further T cell activation and parenchymal immigration. Although AQP4 268-285-specific T cells are found throughout the entire neuraxis, they have neuromyelitis optica-typical "hotspots" for infiltration ; that is, periventricular and periaqueductal regions, hypothalamus, medulla, the dorsal horns of spinal cord, and the optic nerves. Most remarkably, in the presence of neuromyelitis optica immunoglobulin G, they initiate large astrocyte-destructive lesions that are located predominantly in spinal cord gray matter. We also show that AQP4 268-285-specific T cells can induce retinitis and subsequent damage to retinal axons and neurons, both in the presence and in the absence of neuromyelitis optica immunoglobulin G. Furthermore, within inflammatory lesions induced by AQP4 268-285-specific T cells, retinal astrocytes in the Retinal Nerve Fiber layer (RNFL)/gan-glionic cell layers are spared from the action of neuromyelitis optica immunoglobulin G, whereas M€ uller cells lose AQP4 reactivity. These data show that damage to retinal cells can be a primary event in neuromyelitis optica spectrum disorder. The anterior visual pathway (AVP), which includes the retinas and the pathway from the optic nerves to the lateral geniculate nuclei, is a frequent site of injury, as shown by the presence of optic neuritis during the course of multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD). The optic nerves have several distinguishing structures. First, the long, thin cylindrical dimension of the optic nerves with the cul-de-sac anatomy of the subarachnoid space might cause unique cere-brospinal fluid dynamics, including restricted diffusion of pro-inflammatory elements, retained inflammation, and limited clearance of myelin and axonal debris, followed by enhanced MS and NMOSD lesion formation. In recent study (Hokari et al. Ann Neurol 2016; 79: 605), we showed that the AVP involvement in NMOSD is characterized by the following, compared with MS: (i) longitudinally extensive optic neuritis; (ii) more severe visual impairment and worse prognosis for optic neuritis; (iii) unique aquaporin-4 dynamics; and (iv) more severe neurodegeneration. These data suggest that severe and widespread neuroaxonal damage , and unique dynamics of astrocytes/M€ uller cells with alterations of aquaporin-4 were prominent in the AVP, and might be associated with poor visual function and prognosis in NMOSD. We will discuss progressing ...