A mouse model of seizures in anti-N-methyl-d-aspartate receptor encephalitis

Abstract

Objective: Seizures develop in 80% of patients with anti-N-methyl-d-aspartate receptor (NMDAR) encephalitis, and these represent a major cause of morbidity and mortality. Anti-NMDAR antibodies have been linked to memory loss in encephalitis; however, their role in seizures has not been established. We determined whether anti-NMDAR antibodies from autoimmune encephalitis patients are pathogenic for seizures.

Methods: We performed continuous intracerebroventricular infusion of cerebrospinal fluid (CSF) or purified immunoglobulin (IgG) from the CSF of patients with anti-NMDAR encephalitis or polyclonal rabbit anti-NMDAR IgG, in male C57BL/6 mice. Seizure status during a 2-week treatment was assessed with video-electroencephalography. We assessed memory, anxiety-related behavior, and motor function at the end of treatment and assessed the extent of neuronal damage and gliosis in the CA1 region of hippocampus. We also performed whole-cell patch recordings from the CA1 pyramidal neurons in hippocampal slices of mice with seizures.

Results: Prolonged exposure to rabbit anti-NMDAR IgG, patient CSF, or human IgG purified from the CSF of patients with encephalitis induced seizures in 33 of 36 mice. The median number of seizures recorded in 2 weeks was 13, 39, and 35 per mouse in these groups, respectively. We observed only 18 brief nonconvulsive seizures in 11 of 29 control mice (median seizure count of 0) infused with vehicle (n = 4), normal CSF obtained from patients with noninflammatory central nervous system (CNS) conditions (n = 12), polyclonal rabbit IgG (n = 7), albumin (n = 3), and normal human IgG (n = 3). We did not observe memory deficits, anxiety-related behavior, or motor impairment measured at 2 weeks in animals treated with CSF from affected patients or rabbit IgG. Furthermore, there was no evidence of hippocampal cell loss or astrocyte proliferation in the same mice.

Significance: Our findings indicate that autoantibodies can induce seizures in anti-NMDAR encephalitis and offer a model for testing novel therapies for refractory autoimmune seizures.

Keywords: animal model; anti-NMDAR antibodies; autoantibodies; autoimmune seizures; refractory seizures.

Figure 1.

Experimental protocol to assess the role of antibodies in seizure responses during the infusion, followed by behavioral and neurophysiological analysis.

Figure 2.

Mice treated with CSF from patients with anti-NMDAR encephalitis or rabbit anti-NMDAR antibody develop seizures. (A) Total seizure counts and (B) total time in seizures during video EEG monitoring of mice treated for 2 weeks with continuous icv infusion of anti-NMDAR antibody solution (open triangles) or human CSF (open and closed circles). Green, red, and blue closed circles represent PBS-, normal rabbit IgG, and normal CSF-treated mice, respectively. Horizontal bars indicate median values for combined control group (n=20), rabbit anti-NMDAR IgG (n=11), and patients’ CSF (n=17). (C) Behavioral expression of antibody-induced seizures. The number of seizures observed for each Racine score is indicated. (D) The log of the seizure duration in mice treated with patients’ CSF (n=17) histogram has a normal distribution (R² = 0. 995) with the peak at 1.77 corresponding to a median seizure duration of 58.9 sec. (E) Total seizure counts in mice treated with patient anti-NMDAR antibodies in the IgG fraction purified from the pooled CSF (final titer 1:1024), normal human CSF, human albumin and normal human serum IgG (blue, purple and orange closed circles, respectively). p = 0.02. *, p < 0.05; **; p < 0.01, ***; p < 0.001, ****, p<0.0001 vs. control, Dunns’ selected comparison tests (A, B) or Mann-Whitney test (E).

Figure 3.

Tracings of the nonconvulsive seizures in mice treated with continuous icv infusion of CSF from patients with anti-NMDAR encephalitis. (A) Seizure initiation with a single spike followed by irregular spike pattern. (B) Fast rhythmic activity that fluctuates in amplitude.

Figure 4.

Seizure burden in mice treated with CSF from patients with anti-NMDAR encephalitis. (A) Time-course of daily seizure counts (mean ± SEM) in mice that exhibited >39 or ≤39 seizures in 2 weeks (high or low seizure burden, respectively). (B) The log of the inter-seizure interval histogram has a normal distribution. A total of 1,694 seizure intervals were pooled from 16 mice perfused with patients’ CSF and recorded for 2 weeks. The log histogram is reasonably well fit by a Gaussian model (R = 0.944) with peak at 0.982 (= 9.6 min) and log SD = 1.11. (C) Seizure counts in CSF-treated mice with high seizure burden (mean + SEM) had minimal diurnal fluctuations. Daily seizures recorded in 2 weeks in 8 CSF-perfused mice with high seizure density were pooled in successive 2h intervals. (D) Daily seizure counts during the light and dark cycles in CSF-treated mice with high seizure burden were not different (p = 0.72, t-test).

Figure 5.

Effects of antibodies on the behavioral phenotype. Two-week treatment with rabbit anti-NMDAR IgG or CSF from patients with anti-NMDAR encephalitis did not affect memory (A), anxiety-related behavior (B) locomotor activity (C), or balance (D) in mice. Data are the means + S.E.M.

Figure 6:

Abundance of GFAP-positive astrocytes in the CA1 region of hippocampus of mice following 2 weeks’ administration of anti-NMDAR antibody solution or CSF of patients with anti-NMDAR encephalitis. Representative immunostaining images of the CA1 region of sham mouse (A,B), mouse treated with CSF from patients with anti-NMDAR encephalitis (C,D) and a mouse 4 days after pilocarpine-induced status epilepticus (E, F) at 10x and 40x, respectively.