Apolipoprotein E4 carrier status plus circulating anti-NMDAR1 autoantibodies: association with schizoaffective disorder

In a Letter to the Editor, Titulaer and Dalmau¹ commented on our recent publication,² calling for a direct response for scientific accuracy. We reported that autoantibodies against the NR1 subunit of the N-methyl-D-aspartate receptor (NMDAR-AB) are highly seroprevalent in healthy and in neuropsychiatrically ill subjects.² This unexpected finding has now been fully reproduced by Steiner et al.³who provided an addendum to their previous paper, cited by Titulaer and Dalmau,¹ revising their earlier results on healthy individuals.⁴ Any further comments on this topic are thus outdated.

More importantly, Titulaer and Dalmau¹ question the validity of a widely used cell biological technique for measuring the internalization of membrane proteins from the cell surface. In fact, endocytosis duration determines the endocytic compartment, with early endosome appearing after 5–10 min, already recycling endosomes at 30 min, and late endosomes at 30–60 min. Thus, after ⩾30 min one would expect recycling to occur.^5,6 Titulaer and Dalmau¹ cite a paper that does not investigate internalization.⁷ It shows in cultured neurons consequences on surface NMDAR1 expression of an exposure for 1–7 days to human anti-NMDAR1 immunoglobulin (Ig)G. Although this time frame is certainly long enough to include endocytosis, days of serum contact allow for many secondary adaptations, including well-known toxicity effects that have not been distinguished by the authors from the internalization of NR1. In contrast, we did analyze endocytosis and found exactly what was predicted, notably for all investigated Ig subclasses, IgG, IgA and IgM. We are adding representative confocal images, before and after permeabilisation of neurons, reflecting our experimental conditions (Supplementary Figure S1).

We also provided experimental evidence that circulating NMDAR-AB induced behavioral abnormalities only in ApoE^−/− mice with their known leaky blood-brain barrier (BBB),⁸ but not in respective controls. It was our specific intention to investigate behavior in this mouse model as a translational readout of psychiatric symptoms. Moreover, in a retrospective, hypothesis-driven analysis, we found more pronounced neurological symptoms in well-characterized NMDAR-AB seropositive patients with likely (but not proven) compromised BBB due to birth complications or past neurotrauma,² in line with earlier studies (for review see ref. 9). To further support the hypothesis of a permeable BBB as prerequisite for circulating NMDAR-AB to influence brain function, we now provide additional data.

Here, we determined the apolipoprotein E (APOE) carrier status in a total of 2492 individuals tested twice independently for serum NMDAR-AB.^2,10 DNA was isolated from blood using the JETQUICK Blood and Cell Culture Kit (Genomed, Loehe, Germany). The two single-nucleotide polymorphisms (SNPs) rs7412 and rs429358 were genotyped using the KASP genotyping system (LGC Genomics, Berlin, Germany). These two SNPs allow determination of the APOE allele status (APOE2, APOE3 and APOE4). Failed samples (3.7%) were Sanger-sequenced using primers flanking both SNPs (forward: 5′-AACAACTGACCCCGGTGGCG-3′, reverse: 5′-TCCGGCTGCCCATCTCCTCC-3′). Group differences in categorical variables were assessed using binary logistic regression, including NMDAR-AB and APOE4 carrier status, and age as covariates. Pearson’s χ² P-values are displayed in Table 1. In case of N<5 in one cell, Fisher’s exact test was employed.

Table 1 Combined presence of APOE4 carrier status and NMDAR-AB in psychiatric patients

APOE4 carrier frequency was not significantly different in neuropsychiatric disease groups compared with controls (24.3%) (Supplementary Table S1). The higher frequency in bipolar patients (33.3%) did not reach statistical significance. NMDAR-AB were present in 71 individuals with at least one APOE4 allele (2.8% of 2492 subjects). Whereas χ² statistics did not show significant deviations from the theoretical distribution in case of schizophrenia, bipolar disorder, unipolar depressive patients and health, an excess of schizoaffective patients carried both APOE4 and NMDAR-AB (P=0.001, relative odds (OR)=4.929) (Supplementary Table S2). Compared with controls, the interaction of APOE4 carrier status and NMDAR-AB seropositivity was significantly associated with schizoaffective disorder (P=0.001, OR=6.109), corrected for APOE4 and NMDAR-AB main effects, and age (Table 1). Although the percentage of bipolar patients carrying both NMDAR-AB and APOE4 exceeded that of schizoaffective patients (6.7 vs 5.5%), there was no significant association, possibly owing to low statistical power (N=60). No difference was found with respect to disease phenotypes, or age at prodrome or disease onset after correction for multiple testing (Supplementary Tables S3a and b).

In summary, these novel results again highlight the importance of the BBB for NMDAR-AB-mediated pathology. APOE4 is associated with BBB leakage in mouse and man.^8,11 Delusions of grandiosity and mania are common psychiatric symptoms in anti-NMDAR encephalitis.¹² Thus, NMDAR-AB may cause or boost these symptoms in neuropsychiatrically ill APOE4 carriers, which are then more likely diagnosed schizoaffective. Independent replication is necessary, but future studies investigating NMDAR-AB in neuropsychiatric disorders should consider factors determining BBB integrity.