Human blood samples

Patients with NMOSD (n = 15) and healthy controls (n = 15) were recruited from Tianjin Medical University General Hospital (Tianjin, China). Patients with NMOSD were diagnosed on the basis of the 2015 International Panel for Neuromyelitis Optica Diagnosis criteria [25]. Blood samples for flow cytometry were collected from patients during the acute phase of NMOSD. The participants in this study were matched for age and sex. Patient studies were approved by the Ethics Committees of Tianjin Medical University General Hospital.

Isolation of exosomes from human plasma

Exosomes in plasma were extracted following established protocols [26,27,28]. Peripheral blood from patients in the acute stage of NMOSD was collected and centrifuged at 2000 × g for 20 min, followed by additional centrifugation at 10,000 × g for 20 min to remove cell debris. The supernatant was carefully aspirated. Subsequently, the plasma was diluted with PBS, and an exosome extraction solution (Invitrogen, total exosome precipitation reagent from plasma) was added prior to 10 min of incubation at room temperature. The plasma was centrifuged again at 10,000 × g for 5 min. Thereafter, the supernatant was discarded, and the mixture was then centrifuged at 10,000 × g for 1 min. The supernatant was aspirated, leaving the precipitated exosomes at the bottom. Exosomes were resuspended in PBS and prepared for fluorescence staining for 1 h. The following antibodies were used: TMEM119 (Abcam, AF647) and CD22 (BioLegend, AF488). Exosomes were detected via an ImageStream®X MKII imaging flow cytometer (ISx, EMD Millipore, Seattle, WA, USA). Data analyses were performed via Image Data Exploration and Analysis Software (IDEAS®, Version 6.2, 2015).

Single-cell RNA sequencing

Peripheral blood samples were collected from 5 NMOSD patients and 5 healthy controls. Peripheral blood mononuclear cells were extracted and subjected to single-nucleus sequencing via the 10x Genomics platform. The following procedures were used: (1) Quantification of each gene for cells was performed via the default parameters in the Cell Ranger pipeline. The filtered gene‒barcode matrix was processed via the Seurat package in R, and cells whose expression of genes was greater than 200 and whose percentage of mitochondria was less than 5% were selected for downstream analysis. (2) Data normalization (NormalizeData), identification of highly variable genes (FindVariableFeatures), standardization (ScaleData), and principal component analysis (RunPCA) were conducted via relevant functions in the Seurat package. A two-dimensional representation of overall cell populations was subsequently obtained via the t-distributed stochastic neighbor embedding (tSNE) method (RunTSNE). Further clustering (FindNeighbors, FindClusters) grouped cells with similar gene expression profiles, where the distance between cells represented the similarity between cell subgroups. (3) Cell identities were determined on the basis of the expression of subtype-specific genes for T cells (CD3D and TRAC), NK cells (KLRF1, NKG7, and KLRD1), B cells (CD19, CD79A, and CD79B), plasma cells (CD19, CD79A, TNFRSF17, and CD38), monocytes (CD14), and neutrophils (FCGR3B and NAMPT). The expression of CD22 was visualized via the FeaturePlot and Dotplot functions in the Seurat R package.

Mice

Six- to eight-week-old C57BL/6 mice were purchased from Spafford Laboratories (Beijing, China). All the mice were housed in a pathogen-free environment with ad libitum access to food and water under a standardized light‒dark cycle. All the mice were maintained at a standardized temperature (21 ± 1 °C) and humidity (50-60%). The mice were randomly assigned to experimental or control groups. All experimental procedures were approved by the Animal Care and Use Committees of Tianjin Neurological Institute.

Isolation of AQP4-IgG from patient serum

Serum samples were collected from AQP4-IgG seropositive NMOSD patients at Tianjin Medical University General Hospital (Tianjin, China). The serum IgG was purified according to a previously described protocol [29, 30] and diluted with Tris-buffered saline (pH = 7.0). Patient serum IgG was purified with agarose protein-A (Beyotime Biotechnology). The IgG on agarose protein-A was eluted with glycolic acid (pH = 2.5), and then, the acidic eluate was dialyzed to neutrality (pH = 7.0) in Tris-buffered saline for 8 h. The samples were concentrated in Amicon ultracentrifugal filter units (30 kDa, Merck Millipore) to a concentration of 8 mg/ml. Control IgGs were collected in the same way from healthy volunteers at a concentration of 8 mg/ml.

NMOSD model

Mice were injected with AQP4-IgG and human complement in the intraparenchyma as previously described [30, 31]. The mice were anesthetized via inhalation of 3% isoflurane and continuously inhaled 1% isoflurane to maintain anesthesia during surgery. Satisfactorily anesthetized mice were fixed on a stereotactic frame (RWD Life Science), then the scalps of the mice were sterilized, and an incision was made along the median line to fully expose the fontanel. A burr hole was made in the skull 2.5 mm to the right of the bregma. A 50-µL gas-tight glass syringe (Hamilton) attached to a 34-gauge needle was inserted into the brain parenchyma to a depth of 3 mm, and 6 µL of AQP4-IgG and 4 µL of HC were injected into the parenchymal tissue at a rate of 0.8 µL/min. After the injection was completed, surgical sutures were applied to the scalps of the mice. A small animal heating pad was used to maintain the rectal temperature at a body temperature of 37 °C during surgery. Mice that bled during surgery were excluded from the analysis. After injection, the mice were provided adequate water and food.

Drug administration in NMOSD mice

AQP4-IgG and human complement were used to induce NMOSD in mice. Sham mice received the same volume of injected IgG. Three days after surgery, the mice were sacrificed, and single brain cells were collected for flow cytometry analysis. An anti-CD22 monoclonal antibody (clone ID: CY34.1; BioXcell, West Lebanon, NH) was given to NMOSD mice via intrastriatal injection at a dose of 100 µg/mouse to deplete CD22-expressing cells [24].

To deplete microglia, the mice were fed a diet containing a CSF-1R inhibitor (PLX5622, Selleckchem, Houston, TX) [32]. Six-week-old mice were fed a diet containing PLX5622 or a control diet for two weeks before NMOSD induction. To deplete Gr-1+ myeloid cells, the mice were intraperitoneally (I.P.) injected with 250 µg of anti-Ly6G/Ly6C (Gr-1) mAb (Clone ID: RB6-8C5; BioLegend, San Diego, CA) one day before and one day after NMOSD induction. To deplete B cells, an anti-CD20 mAb was administered by intravenous (I.V.) injection at a dose of 10 mg/kg three days prior to NMOSD induction. The mouse spleen tyrosine kinase (SYK) inhibitor R406 (Selleckchem, Houston, TX) was administered by intraperitoneal injection at a dose of 10 mg/kg after anti-CD22 mAb (anti-CD22 mAb) injection [33].

Magnetic resonance imaging

As previously described [34], lesion volume was quantified on day 3 after NMOSD induction using 9.4T MRI. The mice were anesthetized with 3% isoflurane and maintained at 1–2% isoflurane. A T2-weighted image scan was used to measure lesion volume. The scanning parameters were set as follows: TR = 5100 ms, effective TE = 31.71 ms, inversion time = 2506.097 ms, and slice thickness = 0.50 mm. Areas of lesions were manually outlined on each slice. Medical image processing was used to sum the lesion volumes on each slice and multiply them by the thickness of the slice to obtain the final lesion volume. The lesion volumes were calculated by two blinded investigators.

Immunostaining

Brain tissue was collected from the mice and incubated overnight at 4 °C with the following primary antibodies: anti-GFAP (1:500, Abcam), anti-AQP4 (1:200, Boster), anti-MBP (1:500, CST), and anti-Iba1 (1:500, Abcam). The tissue was then incubated with the appropriate fluorochrome-conjugated secondary antibodies: donkey anti-goat 546 (1:1000, Invitrogen) and donkey anti-rabbit 488 (1:1000, Invitrogen) at room temperature for 1.5 h. Finally, all slices were incubated with Fluoroshield mounting medium with 4’6-diamidino-2-phenylindole (DAPI) (Abcam). Images were captured via fluorescence microscopy (Biotek Cytation 5, USA) and analyzed via ImageJ software.

Flow cytometry

Peripheral blood samples were collected from NMOSD patients in the acute phase and healthy volunteers. Mononuclear cells were isolated from whole blood samples. Human blood mononuclear cells were extracted and stained with fluorescently labeled antibodies. The following antibodies were used to quantify the differential expression of CD22 on mononuclear cells: CD45, CD11b, CD3, CD4, CD14, CD8, CD16, CD56, and CD19. Mouse brain tissue was prepared as single-cell suspensions for flow-staining analysis as previously described [24]. On day 3 after NMOSD induction, the mice were satisfactorily anesthetized with 3% isoflurane and then sacrificed, and the brains were removed after perfusion with cold PBS. The brain was minced and incubated with papain and DNase I at 37 °C for 30 min on a constant-temperature shaker. Single-cell suspensions were resuspended in 1% BSA after myelin debris was removed via centrifugation in 30% Percoll (GE Healthcare Bio-Science AB, Uppsala, Sweden), followed by staining. Single cells were stimulated for 4 h in a 1x cell stimulation cocktail (BioLegend) for intracellular cytokine staining. Then, the cells were harvested and prepared for surface and intracellular staining according to the manufacturer’s instructions. The following antibodies were used: CD45, CD11b, CD3, CD4, CD19, Ly6C, Ly6G, CD86, CD206, IL-10, IL-1β, TGF-β, TNF-α, and NK1.1. Simultaneously, the fluorescence minus one (FMO) control was also stained. All the antibodies were obtained from BioLegend or eBioscience. Flow cytometry data were collected on a FACS Aria III flow cytometer (BD Bioscience) and analyzed by Flow Jo version V10 (flowjo.com).

Western blots

The mice were sacrificed after anesthesia and perfused with cold PBS, and the ipsilateral hemisphere was harvested for protein extraction. Unless otherwise noted, 20 µg of total protein was separated on 8–12% acrylamide gels via SDS‒PAGE. Protein was transferred to PVDF membranes (Merck KGaA, Darmstadt, Germany). After being blocked, the membranes were incubated with primary antibodies at 4 °C overnight. Primary antibodies for western blotting were diluted as follows: Syk (1:1000, Cell Signaling Technology, Danvers, MA), phospho-Syk (1:1000, Cell Signaling Technology, Danvers, MA), and β-actin (1:1000, Cell Signaling Technology, Danvers, MA). After incubation at 4 °C overnight, the PVDF membranes were incubated with species-appropriate horseradish peroxidase (HRP)-labeled secondary antibodies (1:500, Transgene Biotech, Beijing, China) for 1 h at room temperature. The protein-specific signals were detected via a Bio-Rad 721BR08844 Gel Doc Imager (Bio-Rad, Hercules, CA). The gray values of the IB bands were quantified via ImageJ software.

Statistical analysis

All the data were analyzed via GraphPad Prism 8 software (GraphPad, Inc., San Diego, CA). The experimental design was based on our previously published similar mechanistic studies [30, 34]. All the animals were randomly assigned to the treatment groups or the control groups, and all the results were analyzed by investigators who were blinded to the experimental conditions. Two-tailed unpaired t tests were used to compare two groups, and one-way analysis of variance (ANOVA) was used for comparisons of data from multiple groups. The data are expressed as the mean ± SEM. A p value < 0.05 was considered statistically significant.