Monoclonal Antibodies From Anti-NMDA Receptor Encephalitis Patient as a Tool to Study Autoimmune Seizures

Abstract

Anti-N-methyl-D-aspartate (NMDA) receptor encephalitis manifests with precipitous cognitive decline, abnormal movements, and severe seizures that can be challenging to control with conventional anti-seizure medications. We previously demonstrated that intracerebroventricular (i.c.v.) administration of cerebrospinal fluid from affected patients, or purified NMDA receptor antibodies from encephalitis patients to mice precipitated seizures, thereby confirming that antibodies are directly pathogenic for seizures. Although different repertoires of anti-NMDA receptor antibodies could contribute to the distinct clinical manifestations in encephalitis patients, the role of specific antibodies in the expression of seizure, motor, and cognitive phenotypes remains unclear. Using three different patient-derived monoclonal antibodies with distinct epitopes within the N-terminal domain (NTD) of the NMDA receptor, we characterized the seizure burden, motor activity and anxiety-related behavior in mice. We found that continuous administration of 5F5, 2G6 or 3C11 antibodies for 2 weeks precipitated seizures, as measured with continuous EEG using cortical screw electrodes. The seizure burden was comparable in all three antibody-treated groups. The seizures were accompanied by increased hippocampal C-C chemokine ligand 2 (CCL2) mRNA expression 3 days after antibody infusion had stopped. Antibodies did not affect the motor performance or anxiety scores in mice. These findings suggest that neuronal antibodies targeting different epitopes within the NMDA receptor may result in a similar seizure phenotype.

Keywords: anti-NMDA receptor encephalitis; autoimmune seizures; cytokines; monoclonal antibodies; neuronal autoantibodies.

FIGURE 1

Administration of 5F5, 2G6, or 3C11 monoclonal anti-N-methyl-D-aspartate receptor (NMDAR) antibodies precipitated seizures in mice. (A) Seizure burden was similar in 5F5 (n = 15, red circles), 2G6 (n = 12, blue circles) and 3C11 (n = 12, green circles) antibody-treated groups. Data are median total seizure counts in 2 weeks. **p < 0.01, ANOVA with Dunnett’s multiple comparisons tests; open cycles, 1X antibodies; filled cycles, 10X antibodies. (B) The distribution of cumulative seizure counts in all anti-NMDAR antibody-treated groups. ***p < 0.001, t-test. (C) Time to the first seizure was not different between the four treatment groups (p = 0.39, Kruskal-Wallis test). Data are median latencies. Note log scale on Y axis.

FIGURE 2

Representative EEG tracings of electrographic seizures recorded from the frontal cortex of mice infused with 1X (A) and 10X (B) 2G6 monoclonal anti-NMDAR antibodies.

FIGURE 3

Administration of 5F5, 2G6, or 3C11 monoclonal anti-NMDAR antibodies did not affect mouse behavior. (A) Locomotor activity was similar in mice treated with different monoclonal antibodies and control 6A IgG (p = 0.08, ANOVA). N = 12 (control 6A IgG, black circles), n = 15 (5F5, red circles), n = 12 (2G6, blue circles), and n = 11 (3C11, green circles). (B) The distribution of total distance traveled in all anti-NMDAR antibody treated groups. (C) Despite the difference in the overall analysis, the anxiety scores in mice treated with either anti-NMDAR antibody were similar to those in control mice (p = 0.01, ANOVA; p = 0.1 vs. 5F5; p = 0.99 vs. 2G6; p = 0.99 vs. 3C11, Tukey tests). N = 13 (control 6A IgG, black circles), n = 15 (5F5, red circles), n = 11 (2G6, blue circles) and n = 12 (3C11, green circles). (D) The distribution of cumulative anxiety scores in anti-NMDAR antibody-treated groups. Data are medians (horizontal bars), 25–75% interquartile ranges (IQR, boxes), and the minimum and maximum values (whiskers, p = 0.43, t-test).

FIGURE 4

Expression of astrocytic and microglial markers of inflammation in the CA1 region of hippocampus of mice with anti-NMDAR antibody-induced seizures. The expression of GFAP in 1X and 10 × 5F5, 2G6, or 3C11 antibody-treated mice (A) and combined antibody-treated mice (B) was similar to that in control 6A IgG-infused mice. The expression of Iba1 in 5F5, 2G6 or 3C11- antibody treated mice (C) and combined antibody-treated mice (D) was comparable in all groups and similar to that in control 6A IgG-infused mice. GFAP: n = 13 (control 6A IgG, black circles), n = 13 (5F5, red circles), n = 10 (2G6, blue circles), and n = 13 (3C11, green circles). Iba1: n = 13 (control 6A IgG), n = 14 (5F5), n = 10 (2G6), and n = 12 (3C11). The abundance of GFAP or Iba1 labeling in the CA1 region was determined as the sum of the products of mean pixel intensity (gray scale units, gsu) and area of each event (μ²) in a fixed scan area. The data are mean ± SEM.

FIGURE 5

Expression of inflammatory mediators in the hippocampus of mice with antibody-induced seizures. (A) C-C chemokine ligand 2 (CCL2) mRNA level was increased in mice that developed seizures during the infusion of antibodies. Data are median fold change in mRNA expression (horizontal bars) in mice treated with 1X and 10X concentrations of 5F5, 2G6 and 3C11 antibody protein (closed symbols) or 6A control IgG protein (open symbols), relative to the corresponding values in mice that were not infused with antibodies. N = 34–36 (5F5, 2G6 and 3C11 antibodies), n = 11–12 (6A antibodies). The boxes are interquartile ranges (IQR, 25–75%), the whiskers are the minimum and maximum values. ***p < 0.001, Kruskal-Wallis test with Dunn’s multiple comparison tests. COX-2, cyclooxygenase-2; IL-6, interleukin-6 (IL-6); TNF, tumor necrosis factor (B) The increase of hippocampal CCL2 mRNA was related to 5F5, 2G6, and 3C11 monoclonal antibody protein (mAb) concentration. Data are medians (horizontal bars), 25–75% interquartile ranges (IQR, boxes). Note log scale on Y axis. (C) Increase in CCL2 mRNA expression in hippocampus of mice treated with 10 × 5F5, 2G6 and 3C11 monoclonal antibody protein was related to the number of antibody-induced seizures. 1x: n = 14 (mAb, blue diamonds) and n = 5 (control 6A IgG, black diamonds); 10X: n = 18 (mAB, red diamonds) and n = 6 (control 6A IgG, green diamonds). Note log scale on Y axis. **p < 0.01, and *p < 0.05, Kruskal-Wallis test with Dunn’s correction.