Male mice on the C57BL/6J background were purchased from the Jackson Laboratory (Bar Harbor, ME) and used for all experiments. EAE was induced in 8- to 10-week-old mice, and age-matched naïve mice were used as untreated controls. Animals were housed and treated in accordance with the National Institutes of Health (NIH) and the Institutional Animal Care and Use Committee (IACUC) guidelines of the Cleveland Clinic.
Induction and scoring of EAE
EAE was induced in 8- to 10-week-old male mice as previously described [47, 48] by subcutaneous injection of 50 µg MOG35-55 peptide emulsified in complete Freund’s adjuvant containing 125 µg desiccated Mycobacterium tuberculosis (Hooke Laboratories, Lawrence, MA). On days 0 and 2 of EAE, mice were injected intraperitoneally with 200 ng Bordetella pertussis toxin (Hooke Laboratories, Lawrence, MA) in 500µL phosphate-buffered saline (PBS). EAE symptoms were monitored daily and mice were assigned clinical scores as follows: 0 = healthy, no symptoms, 1 = loss of tail tone, 2 = flaccid tail, 3 = partial hind limb paralysis, 4 = complete hind limb paralysis, 5 = moribund animal (humanely euthanized), 6 = death. Subsets of mice were sacrificed at each time point for immunohistochemistry and electron microscopy.
Optical coherence tomography and scanning laser ophthalmoscopy
Optical coherence tomography (OCT) was performed using an ultra-high resolution spectral domain OCT system (Envisu R2210 UHR Leica Microsystems Inc.) for in vivo cross-sectional imaging in mice. Prior to imaging mice were anesthetized by intraperitoneal injection with either pentobarbital (75 mg/kg) or a cocktail of ketamine (80 mg/kg ketamine HCL, Zoetis) and xylazine (16 mg/kg, AnaSed Injection). The corneal surface was anesthetized with 1% proparacaine HCL, and mice were placed on a temperature-regulated heating pad. Pupils were dilated with 0.5% topical tropicamide/phenylephrine combination eyedrops (Santen Pharmaceuticals, Osaka, Japan). The scan parameters were 1.8 mm by 1.8 mm rectangular volume scan, 1000 a-scans/200 b-scans averaged 3 times per b-scan. For scanning laser ophthalmoscopy (SLO), 50 µL sodium fluorescein was injected intraperitoneally (1% in saline, AL-FLUOR 10%, Akron Inc.) and retinas were imaged using a SPECTRALIS ophthalmoscope (Heidelberg Engineering, Germany). Animals were imaged prior to induction of EAE to obtain baseline retinal thickness (for OCT) and retinal nerve fiber layer area (for SLO), and followed longitudinally at 15, 35, and/or 60 days post-EAE induction (dpi). In a second experiment, OCT and SLO were performed during chronic EAE in a subset of mice at 35 dpi and another subset at 60 dpi. Mice were euthanized following visual assessments by transcardiac perfusion with 4% paraformaldehyde (PFA). Retinal thickness was measured by caliper method using InVivoVue software. OCT measurements were performed 450 µm from the optic nerve head in both the nasal and temporal directions, and in superior and inferior regions for a total of four measurements per animal. Average thicknesses from both eyes per animal were calculated for analysis. SLO images were analyzed using ImageJ software (version 1.52p) by measuring the area filled by hyper-reflective RNFL. Average RNFL area for both eyes was calculated for analysis.
For human imaging, OCT was performed using spectral domain Cirrus HD-OCT (models 4000 or 5000; Carl Zeiss Meditec, Dublin, CA). All scans were performed without pupillary dilation by trained medical technicians at the Cleveland Clinic Mellen Center. Optic disc and macular scans were obtained using the Optic Disc Cube 200 × 200 and the Macular Cube 512 × 128 protocols, respectively. Software-extracted segmented regions or layers included for analysis include: peripapillary RNFL (pRNFL), macular RNFL, ganglion cell-inner plexiform layer thickness (GCIP), outer retina thickness (ORT), and total macular volume (TMV).
Visual evoked potentials
Visual evoked potentials (VEP) were recorded following OCT and SLO for baseline, 15 dpi, and 35 dpi according to previously described methods . Briefly, VEPs were recorded using an active electrode positioned subcutaneously along the midline of the visual cortex, with a reference needle electrode placed in the cheek. A third electrode was inserted in the tail to serve as the ground lead. Responses to achromatic strobe flash stimuli presented in a ganzfeld device (LKC Technologies, Gaithersburg, MD) were recorded under light-adapted conditions. The interstimulus interval ranged from 1.1 to 6 s, increasing with stimulus luminance from 0.4 to 1.9 log cd s/m2. The amplifier band-pass was set at 1–100 Hz and up to 60 successive responses were averaged to obtain single VEP waveforms. The mouse VEP is dominated by a negative component, N1 [50, 51] and the implicit time of the N1 component was measured at the negative peak. The amplitude of the VEP was measured to N1 from the preceding baseline, or positive peak (P1). Averages of VEP implicit time and amplitude for each stimulus condition was calculated per animal for analysis.
Perfusion and preparation of mouse spinal cords, optic nerves, and retinas for immunohistochemistry
Spinal cords, optic nerves, brains, and eyes were removed after transcardiac perfusion with 4% PFA in PBS. Left eyes were post-fixed in PFA for 2 h and whole retinas were subsequently removed and prepared for immunofluorescent staining (see method below). Spinal cords, optic nerves, brains, and right eyes were fixed overnight in 4% PFA and cryoprotected in 30% sucrose. These were then embedded in Optimal Cutting Temperature freezing medium (Fisher HealthCare, Waltham, MA) and snap-frozen in 2-methylbutane on dry ice. Spinal cords were cut into 6 segments for embedding. All frozen tissue was cut into 16 µm-thick sections on a cryostat (Leica Biosystems, Buffalo Grove, IL) for immunohistochemistry.
Preparation for electron microscopy
Mice were perfused with 4% PFA and 2% glutaraldehyde in 0.1 M sodium cacodylate buffer at pH 7.4. The lumbar regions of spinal cords and whole optic nerves were processed by the Cleveland Clinic Electron Microscopy Core. Tissue was post-fixed in 1% osmium tetroxide in water, stained with 1% uranyl acetate in maleate buffer (pH 5.1), and dehydrated with ethanol and propylene oxide before being embedded in Pure Eponate 12 resin (Ted Pella Inc., Redding, CA). Ultrathin 85-nm sections were cut with a diamond knife, stained with uranyl acetate and lead citrate, and observed with a transmission electron microscope (FEI Company, Hillsboro, OR).
Immunohistochemistry with 3,3'-Diaminobenzidine (DAB)
Immunohistochemistry with DAB was performed to stain for myelin basic protein (MBP) as previously published  in spinal cords and optic nerves. Antigen retrieval was performed using 10 mM citrate buffer (pH 3.0) at 37 °C for 30 min. Sections were blocked with 5% goat serum and 0.3% Triton X-100 in PBS for one hour at room temperature (RT). Rat anti-MBP (1:500; Ab7349, Abcam, Cambridge, UK) primary antibody was diluted in blocking buffer, added to sections, and incubated overnight at 4 °C. A goat anti-rat biotinylated secondary antibody (1:200, BA-9400, Vector Laboratories, Burlingame, CA) was diluted in blocking buffer and added to sections, incubating for one hour at RT. Endogenous peroxidase activity was blocked by incubating slides in 0.3% hydrogen peroxide in methanol for 10 min at RT. Antibodies were visualized using the avidin–biotin-immunoperoxidase complex (ABC) method using the VECTASTAIN ABC Kit (PK-4000, Vector Laboratories, Burlingame, CA) and DAB Peroxidase (horseradish peroxidase) Substrate Kit (SK-4100, Vector Laboratories, Burlingame, CA). Slides were serially dehydrated and mounted in Permount (SP15, Thermo Fisher Scientific, Waltham, MA).
Antigen retrieval was performed for anti-NeuN immunofluorescence using 10 mM citrate buffer (pH 3.0) at 37 °C for 30 min. For all stains, sections were blocked with 5% serum corresponding to the host of the secondary antibody and 0.3% Triton X-100 in PBS for one hour at RT. Primary antibodies were diluted in blocking buffer and incubated on sections overnight at 4 °C. For anti-NeuN immunofluorescence, biotinylated secondary antibodies were diluted in blocking buffer and incubated on slides for one hour at RT. Following incubation with biotinylated secondary antibodies, slides were incubated in fluorescent-conjugated streptavidin diluted in PBS wash buffer for 30 min. For all other immunofluorescence staining, fluorescent-conjugated secondary antibodies were diluted in blocking buffer and incubated on sections for one hour at RT. Slides were mounted with Fluoromount-G (0100–01, SouthernBiotech, Birmingham, AL) containing bisbenzimide (1:1000; H3569, Invitrogen, Carlsbad, CA).
Primary antibodies used included rat anti-CD3 (1:500; 16–0032-85, Invitrogen, Carlsbad, CA), mouse anti-GFAP (1:1000; 835,301, Biolegend, San Diego, CA), rabbit anti-Iba1 (1:500; 019–19,741, Wako, Richmond, VA), mouse anti-NeuN (1:500; MAB377, Millipore, Billerica, MA), rabbit anti-NeuN (1:1000; Ab177487, Abcam, Cambridge, UK), and rabbit anti-VGLUT2 (1:200; ab216463, Abcam, Cambridge, UK). Biotinylated secondary antibodies were used at 1:200 dilution and included horse anti-mouse IgG (BA-2000, Vector laboratories, Burlingame, CA) and goat anti-rabbit IgG (BA-1000, Vector Laboratories, Burlingame, CA). Alexa Fluor 647-conjugated streptavidin was used at 1:1000 dilution (lyophilized stock diluted in 1 mg/mL PBS, S21374, Invitrogen, Carlsbad, CA). All other fluorescent-conjugated secondary antibodies were used at 1:500 or 1:1000 (all purchased from Invitrogen, Carlsbad, CA) and included Alexa Fluor 488 goat anti-rat IgG (A11006), Alexa Fluor 488 goat anti-rabbit IgG (A11034), Alexa Fluor 647 goat anti-mouse IgG (A21235), Alexa Fluor 647 goat anti-rabbit IgG (A21245), and Alexa Fluor 594 goat anti-rabbit IgG (A11037).
Staining and preparation of retinal whole mounts
Isolated fixed retinas were permeabilized for 30 min in 0.5% Triton X-100 in PBS at RT. Retinas were then blocked in 10% goat serum, 1% bovine serum albumin, and 0.5% Triton X-100 in PBS for one hour at RT. Subsequently, retinas were incubated in goat anti-mouse F(ab) fragment (1:2000;ab6668, Abcam, Cambridge, UK) for one hour at RT. Mouse monoclonal anti-Brn3a (1:100; SC-8429, Santa Cruz Biotechnology, Dallas, TX) was diluted in blocking buffer and incubated on retinas overnight at 4 °C. Fluorescent-conjugated secondary antibody (1:500; Alexa Fluor 647, A21235, Invitrogen, Carlsbad, CA) was diluted in blocking buffer and incubated on retinas for two hours at RT. Bisbenzimide (1:100; Invitrogen H3569, Carlsbad, CA) was diluted in wash buffer and added to retinas for 20 min at RT. Four incisions were made in each retina cup to allow them to lie flat on a slide, and retinas were then mounted with Fluoromount-G (0100–01, SouthernBiotech, Birmingham, AL).
Images for DAB immunohistochemistry of MBP in the spinal cord and optic nerve were acquired using a Leica SCN400 slide scanner at 40 × magnification. Fluorescent images were acquired using a Nikon C2 confocal microscope system (Nikon, Melville, NY) with NIS-Elements software (Nikon, Melville, NY) at 40x (for sectioned retinas, optic nerve, and spinal cord white matter), 20x (for thalamic regions-dorsal lateral geniculate nucleus (dLGN) and ventral posterolateral nucleus (VPL), and spinal cord gray matter), or 10x (for NeuN in the thalamus and spinal cord gray matter) magnification. For retinal whole mounts, confocal z-stacks of Brn3a+ cells were captured in the retinal ganglion cell layer at 20 × magnification with 2.50 µm thickness and 0.5 µm step size. For qualitative images of EAE lesions in the spinal cord and optic nerve, 10 µm-thick confocal z stacks were captured at 20 × and 40x, and 3D reconstructions were generated for representative images. For electron microscopy, grids were examined on a Tecnai G2 SpiritBT transmission electron microscope (FEI Company, Hillsboro, OR) operated at 60 kV. NIS-Elements software (Nikon, Melville, NY) and Adobe Photoshop (for contrast, brightness, and color adjustments) were used to create figures and process images.
Quantification of T cell infiltration
CD3+ T cells were quantified blindly using the free NIH ImageJ software (version 1.52p). T cells were identified by CD3+ staining around bisbenzimide+ nuclei in every fifth or tenth 16 µm-thick serial section in the spinal cord, optic nerve, retina, dLGN, and VPL. Average number of T cells per square millimeter per animal were calculated and used for analysis.
Quantification of reactive gliosis
Percent area staining of GFAP and Iba1 antibodies was quantified blindly using NIS-Elements software (Nikon, Melville, NY) in every fifth or tenth 16 µm-thick serial section in the spinal cord, optic nerve, retina, dLGN, and VPL. Background subtraction was performed using the rolling ball correction (set at 5.0) or background subtraction from a region of interest (ROI) and advanced denoising for each image. Thresholds were set to the same baseline and adjusted manually as necessary for each image to assess the percent area stained. In the spinal cord, optic nerve, dLGN, and VPL, percent area was quantified per whole image area. For retinas, each retinal layer was traced using bisbenzimide to define each boundary and area fractions were measured per layer and in the total retina (sum of all layers). Average areas stained per animal were calculated and used for analysis.
Quantification of MBP immunostaining
Spinal cord and optic nerve sections were imaged at 40 × magnification and individual sections were traced for analysis. For spinal cords, only the white matter regions in lumbar segments were traced and measured. Percent area staining of MBP antibody was quantified blindly using the NIH ImageJ free software (version 1.52p). Thresholds were manually set for each image to obtain the percent area stained with MBP. Assessments were made in every fifth or tenth 16-µm serial section. The average percent MBP-negative area fraction (demyelinated area) was then calculated per animal for analysis.
Analysis of electron microscopy
Ultrathin 85 nm sections were imaged at × 2900 magnification in the ventral funiculus of the spinal cord, at × 11,000 in the optic nerve, and at × 18,500 in the dLGN. Axons were counted blindly in five to six images per animal for the spinal cord and optic nerve. Myelinated and unmyelinated axons were counted in each image and analyzed, which were then combined to provide a total axon count. Synapses and myelinated axons were counted blindly in four images per animal for the dLGN using the free NIH ImageJ software (version 1.52p). Average number of axons or synapses per square millimeter was calculated for each animal and used for analysis.
Quantification of retinal ganglion cells and neuronal nuclei
RGCs and neuronal nuclei were counted blindly using the NIH ImageJ free software (version 1.52p). For retinal whole mounts, 2.50 µm-thick z stacks were compressed into a two-dimensional maximum intensity projection to visualize the RGC layer. RGCs were identified as Brn3a+ cells colocalized with bisbenzimide+ nuclei and were counted blindly in 12 fields per retina as previously described . Briefly, four incisions were made in the retina to form four leaflets. Three images per leaflet were taken to represent the area at 1/6, 3/6, and 5/6 of the retinal radius from the center of each quadrant (central, medial, and peripheral positions). Total Brn3a+ cells per retina were used to calculate the number of RGCs per square millimeter. Neuronal nuclei were counted in the spinal cord gray matter, dLGN, and VPL from every tenth 16-µm serial section. For spinal cords, the ventral horns of the gray matter were traced and neurons were counted within the outlined region. The dLGN was traced using VGLUT2 as a marker, which is enriched in the dLGN, and neurons were counted in the entire dLGN for each section. Using anatomical landmarks and VGLUT2 staining, a single image was taken per VPL region in each serial section and neurons were counted in the whole image. Average number of NeuN+ neurons were calculated per animal and the number of neurons per square millimeter was used for analysis.
Multiple sclerosis patient population
Patients with MS at the Cleveland Clinic Mellen Center for Multiple Sclerosis Treatment and Research (Cleveland, Ohio, USA) undergo tablet based neuroperformance (e.g. patient determined disease steps, manual dexterity time in the dominant and non-dominant hands, and walking speed time) as part of their routine clinical visits (25,046,650, 31,054,035). Data were obtained from visits between September 2015 and July 2021 as part of routine clinical care, therefore informed consent was not required. Approval from an ethical standards committee (Cleveland Clinic Institutional Review Board) to conduct this study was received (#18–313). Neuroperformance data were merged with OCT and MRI data where applicable for longitudinal and cross-sectional analyses.
Magnetic resonance imaging (MRI) and image processing
MRI was acquired from patients with MS using a standard protocol on 3 T Siemens scanners and included 3D T1-weighted Magnetization Prepared RApid Gradient Echo (MPRAGE, echo time: 2.98 ms, repetition time: 2300 ms, inversion time: 900 ms, flip angle 9°, and 1mm3 isotropic. Data was transfer to analysis lab in the Lerner Research Institute for analysis. Thalamic subregions were segmented using MAGeT segmentation algorithm . Briefly, MAGeT pipeline is a multi-atlas label fusion algorithm where the gold standard atlas was generated with a comparison to histology (https://github.com/CoBrALab) . The atlas includes labels for lateral geniculate and ventral posterior subregions. MAGeT requires intermediate templates, and for this study, we randomly selected 30 intermediate images with varying degree of brain volume and measured thalamic subregional volumes from segmentation.
Statistical analyses for EAE data were performed using GraphPad Prism software version 9.0.0. (GraphPad Software, La Jolla, CA). Specific analyses performed including P- values are reported where indicated. BioRender was used to generate Fig. 1a. Naïve control groups in statistical analyses correspond to the age-matched control group for the latest EAE time point included in the analysis. Clinical MS data were analyzed using R Studio 2022.02.0 + 443/R 3.0.1 (R Core Team. R: A Language and Environment for Statistical Computing. 2013. http://www.r-project.org/). Analyses included linear mixed effects and multivariable models adjusting for MS disease characteristics and demographics and Spearman correlation coefficients; details provided in results.