Plasma was obtained from healthy controls and NMOSD patients. AQP4-IgG was assessed by a cell-based indirect immunofluorescence assay. The plasma was obtained from acute NMOSD patients before the administration of any corticosteroids or other immunosuppressive therapy. NMO-IgG and control-IgG were purified using Melon Gel IgG Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA), and further concentrated using Amicon Ultra-15 centrifugal filters (Merck Millipore, KGaA, Darmstadt, Germany). The protein concentration was determined by bicinchoninic acid assay (BCA) (Thermo Fisher Scientific), and the final concentration of IgG was adjusted to 20 mg/mL.

C57BL/6 female mice (18–20 g, 6–8 weeks old, Shanghai Model Organisms Center, Shanghai, China) were used. The NMOSD mouse model was constructed as previously reported [8, 9]. Mice were injected (s.c.) with an emulsion containing 4 mg/mL complete Freund’s adjuvant (BD Difco, Franklin Lakes, NJ, USA) at 4 sites. Then, they were injected (i.p.) with 200 ng pertussis toxin (List Biological Laboratories, Campbell, CA, USA) on days 4 and 7 after immunization. Mice started to receive injections (i.p.) of NMO-IgG on day 0 for 8 consecutive days (5 mg NMO-IgG/day). The control group received injections of adjuvant, pertussis toxin, and the same amount of control-IgG.

Motor impairment was graded by experimental allergic encephalomyelitis (EAE) score as previously described: stiff tail, 0.5; tail paralysis, 1; limp tail and waddling gait, 1.5; paralysis of one limb, 2; paralysis of one limb and weakness of another limb, 2.5; bilateral hind limb paralysis, 3; hind limb paralysis with forelimb weakness, 4; death, 5 [10]. Subtle locomotor abnormalities were assessed in the open field test (OFT). The OFT apparatus consisted of a plastic cage (420 $\times$ 420 $\times$ 420 mm) and a digital camera (Shanghai Soft maze Information Technology Co., Ltd., Shanghai, China). One hour before testing, mice were taken to the room for acclimation. The mouse was placed at the center of open field, and movement was recorded by the camera for 5 mins in order to examine distance traveled and duration of immobility. Before proceeding to the next trial, the open field was cleaned with 75 % ethanol.

Mice were euthanized and perfused with phosphate buffer saline (PBS) and 4% paraformaldehyde. The spinal cords, optic nerves and brains were harvested. The tissue was cryosectioned and incubated with primary antibodies overnight at 4 °C: mouse anti-glial fibrillary acidic protein (Cell Signaling Technology, Danvers, MA, USA), rabbit anti-AQP4 (Sigma-Aldrich, St. Louis, MO, USA), rabbit anti-Iba-1 (Wako, Osaka, Japan). Sections were then washed with PBS and incubated with appropriate secondary antibody (Thermo Fisher Scientific) for 1 h at room temperature. Sections were stained with 4${}^{\prime }$,6-diamidino-2-phenylindole (DAPI, Biolegend, Inc., San Diego, CA, USA) and visualized with a Leica microscope (Leica, Wetzlar, Germany). AQP4 and mouse anti-glial fibrillary acidic protein (GFAP) immunonegative areas were quantified using ImageJ software (1.8.0, National Institutes of Health, Bethesda, MD, USA).

Mice were anesthetized with tribromoethanol (M2940, Aibei biotechnology, Nanjing, Jiangsu, China) and perfused with cold PBS. The spinal cord and spleen were isolated and stored in hanks’ balanced salt solution (HBSS) on ice until processing. The spinal cord was cut with sharp scissors and digested in a water bath at 37 °C for 30 min. The tissue was then homogenized with a plunger in fluorescence-activating cell sorter (FACS) buffer and centrifuged at 600 $\times$g for 5 min. The cell pellet was re-suspended in 40% Percoll and centrifuged at 650 $\times$g for 30 min. The cells were washed twice and prepared for FACS using anti-CD11b, anti-CD45.2, anti-iNOS, anti-CD206 (Biolegend, Inc., San Diego, CA, USA) antibodies. The spleen was homogenized and centrifuged at 350 $\times$g for 5 min. The supernatant was aspired and the pellet was re-suspended in the red-cell lysis buffer for 5 min on ice. FACS buffer was used to stop the lysis reaction. The cells were acquired and prepared for FACS. For intracellular interferon (IFN)-$\gamma$, interleukin (IL)-17, and IL-10 staining, cells were stimulated with phorbol myristate acetate (PMA, Sigma-Aldrich, St. Louis, MO, USA), ionomycin (Sigma-Aldrich, St. Louis, MO, USA), and monensin (Thermo Fisher Scientific, Waltham, MA, USA) for 4 h. Flow cytometry was performed using a FACS celesta (BD Bioscience, San Jose, CA, USA), and data were analyzed by FlowJo software (Tree Star, Ashland, OR, USA).

Variables were checked for normality before analysis. Differences between 2 groups were analyzed by a 2-tailed unpaired Student’s t test or Mann-Whitney U test. Data were presented as mean $\pm$ standard deviation (SD). Calculation was performed using GraphPad Prism software (GraphPad, La Jolla, CA, USA, v. 8.0). A p 0.05 was regarded as statistically significant.

To determine the effects of NMO-IgG on motor function, the EAE scoring system was used. All NMOSD mice showed stiff tails on day 8, with only one NMOSD mouse showing a waddling gait on day 5, but the motor dysfunction resolved on day 6. Subtle locomotor abnormalities in NMOSD mice were measured in the OFT. As shown in Fig. 1, NMOSD mice had significantly less total distance traveled (p 0.001) and longer immobility duration than did the control group (p 0.001).

Fig. 1.

Subtle motor impairments of NMOSD mice detected by open field test. (A) Graphical representation of the walking paths of control and NMOSD mice. (B,C) Total distance traveled and immobility time of mice on OFT. ***p 0.001. Data are presented as mean $\pm$ standard deviation (SD). n = 6/gp. NMOSD, neuromyelitis optica spectrum disorders; OFT, open field test.

Human IgG staining revealed that NMO-IgG infiltrated the spinal cord parenchyma; the deposition of IgG was mainly restricted to the perivascular space, pericentral canal, and subpial space (Fig. 2A). Furthermore, IgG/GFAP double staining showed that the NMO-IgG bound to astrocytes in the spinal cord parenchyma (Fig. 2B). These results confirmed that systemically administrated NMO-IgG infiltrated the spinal cord parenchyma via the breached BBB.

Fig. 2.

NMO-IgG infiltrated the mouse spinal cord and bound to astrocytes. (A) Representative photomicrographs of NMO-IgG-infiltrated spinal cord parenchyma. (a) Deposition of NMO-IgG in the spinal cord parenchyma around perivascular space; (b) deposition of NMO-IgG in the pericentral canal; (c) deposition of NMO-IgG in the subpial space. (B) NMO-IgG bound to astrocytes in NMOSD mice. Scale bar = 50 µm. NMO-IgG, neuromyelitis optica immunoglobulin G antibodies; DAPI, 4${}^{\prime }$,6-diamidino-2-phenylindole; GFAP, glial fibrillary acidic protein.

To examine the effect of NMO-IgG on the expression of AQP4 and GFAP, immunofluorescence analyses were performed. Results showed that GFAP (p 0.0001) and AQP4 (p 0.0001) were significantly lower in the spinal cord white matter of NMOSD mice than in control mice (Fig. 3A,B). Furthermore, we evaluated NMO-IgG-induced astrocytopathy in the optic nerve. As shown in Fig. 3C,D, GFAP (p 0.01) and AQP4 (p 0.01) were also significantly lower in the optic nerve of NMOSD mice than in control mice.

Fig. 3.

NMO-IgG caused AQP4 and GFAP loss. (A,C) Loss of AQP4 and GFAP in the spinal cord and optic nerve of NMOSD mice and control mice; (B,D) The statistical results of the AQP4 and GFAP loss between the two groups, n = 5–6/gp. **p 0.01; ****p 0.0001. Data are presented as mean $\pm$ SD. Scale bar = 50 µm. AQP4, aquaporin-4; GFAP, glial fibrillary acidic protein.

Immunostaining revealed markedly higher numbers of Iba-1${}^{+}$ cells in the spinal cord of NMOSD mice than in control mice (p 0.01) (Fig. 4A,B), whereas there was no significant difference in the number of Iba-1${}^{+}$ cells in the brain (Fig. 4C,D). Although inducible nitric oxide synthase (iNOS${}^{+}$) microglia numbers were also higher in the experimental group (p 0.01), no significant difference was found in CD206${}^{+}$ microglia between the two groups (p = 0.6943) (Fig. 4E–H). We also conducted hematoxylin and eosin (H&E) staining on spinal cord, brain and optic nerve slices and found on differences in inflammatory cells between NMOSD and control groups (Supplementary Fig. 1).

Fig. 4.

NMO-IgG modulated microglia polarization toward M1. (A,C) Immunofluorescence staining for Iba-1 in the spinal cord and brain of NMOSD and control mice, Scale bar = 50 µm; (B,D) The statistical results of the number of Iba-1${}^{+}$ cells between the two groups, n = 4–6/gp; (E,F) Histogram curves of the expression of the inducible nitric oxide synthase (iNOS) in the microglia, and the statistical results, n = 7–8/gp; (G,H) Histogram curves of the expression of the CD206 in the microglia, and statistical results, n = 7–8/gp. ns = no significant difference; Data are presented as the mean $\pm$ SD; **p 0.01.

Peripheral distribution of NMO-IgG was evaluated after 8 days of i.p. injections. We examined whether NMO-IgG bound to the peripheral organs that expressed AQP4 and induced changes. As shown in Fig. 5, AQP4 was expressed on the basolateral membrane of the kidney collecting duct and on the plasmalemma of skeletal muscle of naïve mice. IgG/AQP4 double immunostaining revealed that NMO-IgG bound to the basal side of kidney-collecting-duct cells and to the plasmalemma of skeletal muscle in NMOSD mice. However, there was no difference in the AQP4 immunoreactivities between the NMOSD and naïve mice, which indicated that tissue damage was not induced by NMO-IgG.

Fig. 5.

NMO-IgG localization to peripheral tissues that expressed AQP4. (A) IgG/AQP4 double immunostaining revealed NMO-IgG bound to the basal side of kidney collecting duct cells. (B) NMO-IgG localization to plasmalemma of skeletal muscle in NMOSD mice. Scale bar = 50 µm.

To determine the effect of NMO-IgG on peripheral immune cells, we performed flow cytometry on the spleen. As shown in Fig. 6, no significant differences were found in the percentage of macrophages/neutrophils of NMOSD and control mice (p = 0.3446 and p = 0.2128, respectively). Splenocytes were also stimulated ex vivo for 4 h, and stained with antibodies for the flow cytometric analysis of Th17/Th1 and IL-10-producing B cells. We found no significant differences in the percentage of IL-17 (p = 0.4990) and IFN-$\gamma$ secreting CD4${}^{+}$ T cells (p = 0.8824). Additionally, there was no significant difference in the percentage of IL-10-producing B cells between the two groups (p = 0.8674).

Fig. 6.

The effect of NMO-IgG on peripheral immune cells. (A–D) The percentage of CD11b${}^{+}$ly6G${}^{+}$ neutrophils and CD11b${}^{+}$F4/80${}^{+}$ macrophages and the statistical results. (E,F) Flow cytometric analysis of IFN-$\gamma$${}^{+}$ and IL-17A${}^{+}$ cells in CD4${}^{+}$ T cells. (G,H) The percentage of IL-10${}^{+}$ cells in CD19${}^{+}$ B cells and the statistical results. ns = no significant difference; Data are presented as the mean $\pm$ SD; n = 4–5/gp. IL, interleukin; IFN-$\gamma$, interferon-$\gamma$.