Differentially Altered NMDAR Dependent and Independent Long-Term Potentiation in the CA3 Subfield in a Model of Anti-NMDAR Encephalitis

Abstract

Purpose: Autoantibodies against NMDA receptors (NMDAR) in the cerebrospinal fluid (CSF) from anti-NMDAR encephalitis patients have been suggested to be pathogenic since in previous studies using patient CSF, NMDAR-dependent processes such as long-term potentiation (LTP) were compromised. However, autoantibodies may represent a family of antibodies targeted against different epitopes, and CSF may contain further autoantibodies. Here, we tested the specificity of the autoantibody by comparing NMDAR-dependent and NMDAR-independent LTP within the same hippocampal subfield, CA3, using CSF samples from four anti-NMDAR encephalitis patients and three control patients. Methods: We performed a stereotactic injection of patient-derived cell-free CSF with proven presence or absence of NMDAR-antibodies into the rat hippocampus in vivo. Hippocampal brain slices were prepared 1-8 days after intrahippocampal injection, and NMDAR-dependent LTP at the associational-commissural (A/C) fiber-CA3 synapse was compared to NMDAR-independent LTP at the mossy fiber (MF)-CA3 synapse. Results: The LTP magnitude at A/C fiber-CA3 synapses in slices from control-CSF-treated animals (168 ± 8% n = 54) was significantly higher than LTP in slices from NMDAR-CSF-treated animals (139 ± 9%, n = 40; P = 0.015), although there was some variation between the individual CSF samples. We found residual LTP in NMDAR-CSF-treated tissue which could be abolished by the NMDAR inhibitor D-AP5. Moreover, the CA3 field excitatory postsynaptic potential (fEPSP) was followed by epileptiform afterpotentials in 5% of slices (4/78) from control-CSF-treated animals, but in 26% of slices (12/46) from NMDAR-CSF-treated animals (P = 0.002). Application of the LTP-inducing paradigm increased the proportion of slices with epileptiform afterpotentials, but D-AP5 significantly reduced the occurrence of epileptiform afterpotentials only in NMDAR-CSF-treated, but not in control tissue. At the MF synapse, no significant difference in LTP values of control-CSF and in NMDAR-CSF-treated tissue was observed indicating that NMDAR-independent MF-LTP is intact in NMDAR-CSF-treated tissue. Conclusion: These findings indicate that anti-NMDAR containing CSF impairs LTP at the A/C fiber-CA3 synapse, although there is substantial variation among CSF samples suggesting different epitopes among patient-derived antibodies. The differential inhibition of LTP at this synapse in contrast to the MF-CA3 synapse suggests the specificity and underlines the pathophysiological role of the NMDAR-antibody.

Keywords: LTP; NMDA receptor; associational-commissural fibers; epileptiform afterpotentials; mossy fibers.

Figure 1

Experimental design to study associational-commissural (A/C) and mossy fiber (MF) input in CA3. (A) Microphotographs showing the ink dispersion in the hippocampus 1 h after injection into CA3 stratum radiatum (s.r.; denoted by an asterisk). Magnification 40×. Leftmost panel: native slice (350 μm). Middle panel: air-dried slice stained with toluidine blue and hematoxylin in order to demonstrate hippocampal cell layers. DG, dentate gyrus; sub, subiculum; EC, entorhinal cortex. Rightmost panel: air-dried slice indicating diffusion of ink along the vessels towards CA3 (see arrows). The scale bar indicates 1000 μm. (B₁) Localization of stimulation and recording electrodes in the hippocampus. A/C fiber responses were evoked by stimulation placed in s.r. at the border between CA2 and CA3 and registered within CA3 s.r. MF responses were evoked by stimulation placed in stratum lucidum (s.l.) close to the dentate gyrus and registered within CA3 s.l. (B₂) Typical responses of A/C fiber and MF stimulation showing the characteristic paired-pulse ratios (PPR; indicated by dotted lines). Note that MF responses typically contain a fiber volley (*) which does not express paired-pulse plasticity. (C) Input-output (I/O) curves showing no significant difference between control-CSF and NMDAR-CSF slices (P = 0.062, two-way-ANOVA). The I/O curve of naive, non-operated animals is indicated by a gray line. (D) Paired-pulse ratio did not differ between control-CSF and NMDAR-CSF treated groups (P value calculated by using Mann-Whitney U-test).

Figure 2

A/C fiber-CA3 long-term potentiation (LTP). (A) Time course of the mean relative field excitatory postsynaptic potential (fEPSP) slopes (expressed as percentage of the baseline fEPSP slope) of slices from control-CSF-treated animals. Note that robust LTP was obtained in all three subgroups tested (C1–C3). In addition, LTP from naive non-operated rats is indicated as a black line. The representative sample traces were taken from the subgroup C1 at the timepoints ① (baseline) and ② (end of the experiment). Modified delta burst stimulation (mdBS) was applied at timepoint “0” (indicated by an arrow). (B) Time course of the mean relative fEPSP slopes of slices from NMDAR-CSF-treated animals. Note that subgroups N1 and N2 only showed little residual potentiation, while the LTP magnitude of subgroup N3 was indistinguishable from control tissues. For the sake of clarity, LTP from naive non-operated rats is again indicated as a black line. The representative sample traces were taken from the subgroup N1 at the timepoints ① and ②, and they also illustrate the epileptiform afterpotentials (indicated by full circles). (C) Box-whisker plots of the LTP magnitude (mean relative fEPSP slope during the last 5 min of the experiment). P values for unpaired comparisons (with vs. without D-AP5 or control vs. NMDAR) were calculated using the Mann-Whitney U-test. The multiple test for NMDAR subgroups was performed using ANOVA followed by Student-Newman-Keuls posthoc test. The diamonds (◊) indicate significance of LTP (Wilcoxon signed rank test). The circles colored like the box-whisker plots indicate outliers.

Figure 3

Epileptiform afterpotentials are associated with A/C fiber LTP. (A) Box-whisker plots of the number of afterpotentials following the primary negative deflection referred to as the fEPSP for all subgroups (see Table 1 for CSF data). The lowercase letters indicate that the number of afterpotentials was counted before (“b”) or after (“a”) application of the mdBS paradigm. P values for paired comparisons (before vs. after mdBS) were obtained by using the Wilcoxon rank sum test, P values for unpaired comparisons (with vs. without D-AP5) were obtained by using the Mann-Whitney U-test. Circles colored like the box-whisker plots indicate outliers. (B) Correlation between the number of afterpotentials and the LTP magnitude (fEPSP slope as percentage of baseline value) for all subgroups. There was a positive correlation for control-CSF-treated subgroups, but not for NMDAR-CSF-treated tissue (Pearson correlation coefficient, t-test).

Figure 4

MF LTP is intact in anti-NMDAR tissue. (A) Frequency facilitation (1 Hz) shows no significant difference between NMDAR-CSF-treated and control-CSF-treated tissue. (A1) Examples of original recordings. (A2) Summary plot. The asterisk (*) in panel (A₁) indicates the presynaptic fiber volley which remained constant during the 1 Hz stimulation paradigm. (B) Residual fEPSP amplitude following DCG-IV (expressed as the percentage of the baseline response). Note that the sensitivity towards the group II metabotropic glutamate receptor agonist DCG-IV was not different between control and anti-NMDAR tissue (P value calculated by using Mann-Whitney U-test). (C) The paired-pulse ratio did not significantly differ between both experimental groups (P value calculated by using Mann-Whitney U-test). (D) LTP at the MF synapse showed almost equal magnitudes at the end of the experiment for both NMDAR-CSF- and control-CSF-treated tissue. Note that DCG-IV significantly reduced the potentiated fEPSPs in both groups. Insets show typical recordings before and after tetanic stimulation as well as after DCG-IV.