Autoantibodies against N-methyl-d-aspartate receptor 1 in health and disease

Abstract

Purpose of review

Humoral autoimmunity has gained highest interest in neurology and psychiatry. Despite numerous recent articles on this hot topic, however, the biological significance of natural autoantibodies (AB) and the normal autoimmune repertoire of mammals remained quite obscure. AB may contribute to disorder-relevant phenotypes and are even believed to induce diseases themselves, but the circumstances under which AB become pathogenic are not fully understood. This review will focus on the highly frequent AB against the N-methyl-d-aspartate receptor 1 (NMDAR1-AB) as an illustrating example and provide a critical overview of current work (please note that the new nomenclature, GluN1, is disregarded here for consistency with the AB literature). In particular, it will demonstrate how little is known at this point and how many conclusions are drawn based on small numbers of individuals, fragmentary experimental approaches or missing controls.

Recent findings

NMDAR1-AB were investigated by clinicians world-wide with numerous small studies and case reports appearing yearly. Many publications were on ‘anti-NMDAR encephalitis’ cases or tried to separate those from other NMDAR1-AB associated conditions. Original exclusivity claims (e.g. electroencephalogram, EEG or functional magnetic resonance imaging, fMRI findings) turned out not to be exclusive for ‘anti-NMDAR encephalitis’. Systematic analyses of representative NMDAR1-AB positive sera of all immunoglobulin (Ig) classes showed comparable distribution of different epitopes, often polyspecific/polyclonal, across health and disease. Sophisticated imaging tools provided findings on synapse trafficking changes induced by NMDAR1-AB from psychotic subjects but still lack epitope data to support any claimed disorder link. Persistently high titers of NMDAR1-AB (IgG) in immunized mice with open blood–brain barrier (BBB)-induced psychosis-like symptoms but failed to induce inflammation in the brain. Knowledge on peripheral NMDAR, for example in the immune system, and on potential inducers of NMDAR1-AB is only slowly increasing.

Summary

The present knowledge on the (patho) physiological role of NMDAR1-AB is very limited and still characterized by adamant rumors. Much more experimental work and more solid and informative clinical reports, including large numbers of subjects and adequate control groups, follow-up investigations and interdisciplinary approaches will be necessary to obtain a better understanding of the significance of humoral autoimmunity in general (in focus here: NMDAR1-AB) and its disease-relevance in particular.

INTRODUCTION

The biological significance of natural AB and the normal autoimmune repertoire of mammals – even though principally recognized for many decades – has to a large degree remained mysterious. Natural AB of different classes are produced from early development on in the absence of exposure to foreign antigens. Their physiological role is only partly understood, and little is known about the circumstances and mechanisms that may potentially turn these natural ‘harmless’ AB of the normal autoimmune repertoire into symptomatic or disease-relevant AB.

According to the present concept, natural AB (IgM, IgA or IgG) are made by B1 cells and marginal zone splenic B cells, independent of T cell help. In contrast, immune IgM and subsequently IgG or IgA are produced in response to foreign antigens by B2 cells that require antigen binding to the B-cell receptor and additional T-cell support to generate AB. The IgM secreting B2 cells migrate to B-cell follicles where, with the help of T cells, they undergo isotype switching and somatic hypermutation. Resulting are long-lived memory B cells, capable of differentiating into plasma cells that produce high-affinity binding IgG or IgA [1–4]. Whether, where and how these two general models intermingle, leading to autoimmune disease, is still quite nebulous.

Based on protein arrays and immunostaining of brain sections, human subjects as well as other mammals were reported to carry up to 1000 different serum AB, many directed against yet to be identified brain antigens, with a pattern highly variable among individuals [5,6]. This astonishing finding indicates that these AB might be important mediators, for instance in regulating immune responses, including immune maturation and tolerance induction, dampening inflammation, or in apoptotic cell clearance and debris removal [1,2,6]. Of potential diagnostic relevance is their new application as biomarkers revealing or excluding specific disease signatures [7,8^▪,9,10].

The substantially growing interest in brain antigen-directed AB is reflected by numerous recent reviews (e.g. [11^▪,12–14,15^▪,16,17^▪]). The search for their impact on neuropsychiatric disorders has led to new unexpected findings that question hitherto existing dogmas. AB previously believed to be illness indicators turned out to be equally detectable in healthy individuals. Screening of > 4200 individuals revealed an identical seroprevalence of 25 brain antigen-directed AB across various different disease groups and healthy subjects. Also, distribution of immunoglobulin classes (IgM, IgG and IgA) and AB titers turned out comparable, thus challenging an unambiguous causal relationship of any of these AB with brain disease [18]. A somewhat perplexing observation in this study on 25 brain antigen-directed AB measured in serum was that the location of the epitope (intracellular versus extracellular) obviously determines the Ig class, with intracellular epitopes predisposing to IgG. This may at least question the widely believed exclusive association of IgG with pathology [18]. 

Box 1.

no caption available

HIGH SEROPREVALENCE OF N-METHYL-d-ASPARTATE RECEPTOR 1-AB ACROSS HEALTH AND DISEASE

An interesting example in this regard and therefore in the focus of the present article are AB against the N-methyl-d-aspartate receptor 1 (NMDAR1-AB; new nomenclature, GluN1, disregarded here for consistency with AB literature). NMDAR exist as di-heteromeric or tri-heteromeric complexes, with the composition of the four subunits determining receptor properties [19–21]. As NMDAR1 is an obligatory partner in all possible constellations, AB against this subunit may potentially affect all NMDAR in brain and periphery.

NMDAR1-AB of the IgG class were originally reported in a condition called ‘anti-NMDAR encephalitis’ and claimed to be disease-pathognomonic [22–24]. The array of symptoms described in this condition is somewhat variable but essentially consistent with NMDAR1 antagonism as inducible by for example the NMDAR antagonist ketamine: Psychosis, epileptic seizures, extrapyramidal movement disorders, cognitive decline, reduced consciousness and autonomic dysfunction. In none of the earlier reports on ‘anti-NMDAR encephalitis’, healthy controls were analyzed to support the claim of disease-specificity. Therefore, the age-dependent seroprevalence of up to >20% NMDAR1-AB in healthy subjects as well as across all investigated disease groups (neuropsychiatric as well as other medical conditions like hypertension and diabetes) came as surprise [25–28]. Compared with other brain-antigen directed AB (mostly <2%), NMDAR1-AB have the highest disease-independent seroprevalence [18]. Serum titers as well as the distribution of immunoglobulin classes (mainly IgM and IgA; IgG rarest; IgE not detected so far) are similar in health and any to date investigated disease [27–29,30^▪▪]. Many reports, including recent ones, describing either higher seroprevalence in certain disorders and/or absence/scarcity in a particular disease/health group are based on much too small numbers of subjects for a firm conclusion to be drawn (e.g. [31^▪,32^▪]). Among them are case reports, often without mentioning immunoglobulin classes, titers, or whether serum or cerebrospinal fluid (CSF) was analyzed. Inadequate or uncritical citation of the literature is prevailing [33^▪,34^▪].

In a heterogeneous group of 323 cancer patients, the seroprevalence of neuronal surface AB (IgA, IgM) was reported to be high [32^▪], but is actually in the expected range of the investigated age (NMDAR1-AB 16.7%; mean age 62 years) [18,27,28]. The number of controls in this study (N = 65 neurological disease controls and N = 40 healthy blood donors) is too low for solid comparative conclusions. Cognitive deficits described by this study in AB-positive patients with chronic BBB dysfunction (pathological albumin quotient) are an interesting finding, even though not unexpected (chronic ketamine-like effects), but unlikely due to specific features of any particular cancer [32^▪]. On the other hand, some cancers may well lead – via as yet widely unknown mediators – to a persistently perturbed BBB and thus allow greater access of circulating AB to the brain.

FUNCTIONALITY AND EPITOPES OF N-METHYL-d-ASPARTATE RECEPTOR 1-AB IN HEALTH AND DISEASE

All NMDAR1-AB, independent of the immunoglobulin class, proved to be functional in human and mouse, both in vivo[27,28] and in vitro[27,30^▪▪,35], next raising the question of target epitopes. Published work on NMDAR1-AB epitopes had been limited to IgG recognizing NTD and NTD-G7 domain (N368/G369), originally deemed pathognomonic for ‘anti-NMDAR encephalitis’ [22,36]. Thus, for the first time, NMDAR1-AB of three immunoglobulin classes (IgM, IgG, IgA), derived from randomly selected individuals (out of thousands) of different age, sex and medical condition, were tested regarding in-vitro functionality and epitope location [30^▪▪]. All NMDAR1-AB positive sera led to NMDAR1 internalization in inducible pluripotent stem cell (IPSC)-derived human cortical neurons and to reduced glutamate-evoked response in NR1–1b/NR2A-expressing oocytes. Several different epitopes recognized by NMDAR1-AB (all classes) were identified, located in the extracellular part of NMDAR1 (NTD, LBD) or intracellular, C-terminal (CTD) and in the extralarge pore domain (xlp). No consistent functional or epitope pattern was detectable regarding immunoglobulin class or health/disease state. Half of the screened sera showed polyclonality/polyspecificity, the other half was monoclonal or oligoclonal/oligospecific (mainly IgG) [30^▪▪]. Factors predisposing young women to neuropsychiatric manifestations of NMDAR-associated autoimmunity, for example in lupus erythematosus [37] or anti-NMDAR encephalitis [22,36], may indeed be related to NTD or NTD-G7 epitopes.

Different functionality of NMDAR1-AB (IgG) derived from schizophrenic as compared with healthy subjects has been reported by Jezequel et al.[38^▪▪]. Super-resolution imaging was used to provide a nanoscale surface organization map of NMDAR in hippocampal networks. The authors describe specific alterations of the NMDAR synaptic trafficking by NMDAR1-AB from four psychotic patients. According to their hypothesis, NMDAR hypofunction in schizophrenia might be induced by destabilizing synaptic NMDAR and its interacting partner EphB2R. In contrast, in their control conditions (1 AB− individual; 3 healthy AB+ subjects), NMDAR laterally diffuse and stabilize in nanometer-sized clusters within glutamatergic synapses [38^▪▪]. Another article of these authors, using the same method, finds one NMDAR1-AB carrying autistic individual not to differ from one healthy carrier [39^▪]. Even though highly interesting and based on a sophisticated method, more work is needed to confirm any psychosis-relation of these findings. NMDAR1-AB were not characterized by epitope mapping and questions of oligospecificity or polyspecificity or presence of other immunoglobulin classes (apart from IgG) remained open [38^▪▪]. What, if not the target epitope of AB, would explain the influence on synaptic anchoring in a standardized experiment?

Considering the low seroprevalence of only ∼1% IgG NMDAR1-AB based on more than 5000 individuals investigated across health and disease [40], the authors were incredibly lucky to find any IgG+ subject when screening only 48 schizophrenic [38^▪▪] or 24 autistic patients and 18 healthy controls [39^▪] for the presence of NMDAR-AB. Even when testing 104 healthy controls, three IgG+ individuals are unusual [38^▪▪]. In contrast, a recent report on 78 and 234 schizophrenic subjects, respectively, did not find a single IgG+ patient [31^▪]. These by chance findings underline the above mentioned problems of seroprevalence estimates based on too small numbers.

‘ANTI-NMDAR ENCEPHALITIS’ IS UNLIKELY A SEPARATE DISEASE ENTITY

Several current articles failed to confirm earlier ‘exclusivity claims’ of ‘anti-NMDAR encephalitis’. Extreme delta brush had been interpreted as unique EEG pattern in adults with ‘anti-NMDAR encephalitis’ [41], a statement now invalidated [42^▪]. Addressing the predisposition of AB carriers to epilepsy, high neurological AB prevalence (34.8% neurological AB; 3.6% NMDAR1-AB; pretreatment sera) was described in prospectively evaluated cases with epilepsy of unknown cause [43^▪▪]. AB+ patients showed greater likelihood of neuropsychiatric symptoms or autonomic dysfunction. Also in this study, immunoglobulin class and method of AB detection were not mentioned, BBB function not discussed, and the high overall AB seroprevalence in health and disease not considered.

Another recent claim is the ‘characteristic pattern of whole-brain functional connectivity alterations in anti-NMDAR encephalitis that is well suited to explain the major clinical symptoms of the disorder’ [44^▪▪]. The authors investigated at highly variable time points after the initial diagnosis 43 ‘anti-NMDAR encephalitis’ cases by resting state fMRI. A large proportion of subjects had at the day of fMRI negative NMDAR1-AB titers [44^▪▪]. Thus, any conclusion linking the described dysconnectivity to NMDAR1-AB is problematic. Long-term persisting effects of NMDAR1-AB once they are eliminated by e.g. immunosuppression or plasmapheresis, have not yet been documented anywhere. Experimental approaches rather point to a rapid loss of effects after AB vanishing [45]. Even more importantly, the study lacks a well matched group of encephalitis patients without history of NMDAR1-AB as adequate control. A connectivity disturbance is foreseeable for any kind of encephalitis [46^▪▪].

Along these lines and enlightening for clinicians, no appreciable differences were extracted by comparing NMDAR1-AB+ and NMDAR1-AB− encephalitis cases, except for typical NMDAR antagonistic (ketamine-like) symptoms. CSF analysis, EEG monitoring, MRI scanning, accompanying tumor frequency and outcomes were not different [47]. These findings would support that ‘anti-NMDA encephalitis’ is not a separate condition, but marks any type of encephalitis in which the highly prevalent NMDAR1-AB happen to be present, are boosted, and dramatically shape the clinical picture. Dependent on the circumstances, namely absence or presence of an underlying brain inflammation, NMDAR1-AB may contribute to chronic processes (dementia, psychosis, epilepsy of unknown cause, personality changes) or fulminant courses (‘anti-NMDAR encephalitis’), provided access to the brain (BBB dysfunction or intrathecal synthesis). The prodromal phase of many encephalitides is ‘infectious-like’ [22–24,47], and infections cannot easily be excluded by routine analyses. Later symptoms may be ‘ketamine-like’ due to the presence of NMDAR1-AB among many other brain antigen-directed AB [45,48]. The underlying condition finally determines the highly variable clinical and pathological picture, namely further symptoms, diagnostic readouts (EEG, MRI, CSF/blood parameters), treatment response, outcome and autopsy findings. It may be problematic if the search for causes stops after detection of NMDAR1-AB (IgG). We note, however, that in most cases, ‘polypragmatic’ treatment anyhow includes antibiotics, antiviral compounds and corticosteroids/immunosuppressants on top of a wide range of supporting measures.

The question whether NMDAR1-AB can induce inflammation in the brain has recently been experimentally addressed [49^▪▪]. Active immunization of mice against four peptides of the extracellular NMDAR1 domain (including NTD-G7; N368/G369) leads to high circulating levels of specific AB. After 4 weeks, the endogenously formed NMDAR1-AB (IgG) induce psychosis-like symptoms upon MK-801 challenge in ApoE^−/− mice, characterized by open BBB, but not in ApoE^+/+ littermates, which are indistinguishable from ovalbumin controls. Most importantly, NMDAR1-AB do not induce any sign of inflammation in the brain. Immunohistochemical staining for microglial activation markers and T-lymphocytes yields comparable results, irrespective of immunization. Thus, NMDAR1-AB (here IgG) shape behavioral phenotypes upon access to the brain but do not cause encephalitis [49^▪▪]. This result may put recent reports into perspective that analyzed plasma cell clones from CSF of encephalitis patients and purport that ‘human CSF monoclonal NMDAR1-AB are sufficient for encephalitis pathogenesis’ [48] or claim to have proven their pathogenic relevance by passive transfer of disease symptoms from man to mouse, based on a single behavior test, novel object recognition, but no histology [45].

INCREASING SEROPREVALENCE OF N-METHYL-d-ASPARTATE RECEPTOR 1-AB WITH AGE: WHAT MAY BE THE (PATHO) PHYSIOLOGICAL ROLE AND WHICH ARE THE INDUCERS?

The increasing seroprevalence of NMDAR1-AB with age is particularly interesting, as it is in contrast to most studies in humans and rodents that report natural AB of the IgM class to decrease or lose their effectiveness with age (reviewed in [1]). At least for NMDAR1-AB, this is clearly not the case, as all classes (IgM, IgA, IgG) rise with age [27,28,49^▪▪]. The reason for this discrepancy is still obscure, but induction of NMDAR1-AB by for example influenza infection [27,28] or by chronic life stress [49^▪▪], stimulants with accumulating relevance upon aging, may partly explain this finding.

Remarkably, not only humans, but all other mammals investigated, dogs, cats, mice, rats and monkeys, show high seroprevalence of NMDAR1-AB, with Ig class distribution comparable with humans [49^▪▪]. We note that NMDAR1 has >99% sequence homology across mammals. Functionality of NMDAR1-AB was proven for all species by NMDAR endocytosis in human IPSC-derived cortical neurons. However, the age dependence was lost in monkeys (rhesus macaques and baboons) in which already young animals, captured/transported or born in captivity, displayed high seroprevalence. Hypothesizing that this finding may be related to chronic life stress, for example caused by an unfamiliar or unfriendly environment, we screened human migrants and likewise found that the age-dependence had disappeared. In both migrants and monkeys, IgA was the predominant class accounting for the early AB rise [49^▪▪]. Mechanisms behind the selective increase in IgA still need clarification. Nevertheless, it fits well to early work showing distinct serum IgA response to stress [50].

In the context of natural as well as pathological NMDAR1-AB, the emerging role of NMDAR in the immune system itself is intriguing: NMDAR antagonists dampen B cell [51] as well as T-cell functions [52]. NMDAR is rapidly upregulated upon CD4+ T-cell activation in humans and neurotransmitter/agonist signaling via NMDAR leads to decreased T helper type 1-like and enhanced T helper type 2-like immune balance, affecting proliferation, cytokine production and cell survival [53^▪▪]. Even though ultimate proof of NMDAR1 functionality on blood cells is still pending, it is intriguing to speculate that NMDAR1-AB as humoral (auto)immunity players could feedback-modulate the immune system.

Many other questions in the field of humoral autoimmunity in general and of NMDAR1-AB in particular remain open: The old question [5], raised again in a recent review [54], of how brain constituents get antigens, is pivotal also for NMDAR1-AB. Central nervous system constituents are naturally not antigenic but may belong to the normal autoimmune repertoire of mammals and/or be converted to be immunogens via release of brain debris into circulation, formation of complexes with other substances or alteration of configuration [55], posttranslational protein modification, for example citrullination [56,57], or exposure in ‘free solution form’ to AB as a result of cellular injury or death [54,58]. These and other mechanisms, for example molecular mimicry, may account for the predisposition observed upon oncological conditions (teratoma; [22]), infections (e.g. influenza A and B), genetic constellation (SNP-rs524991, with an adjacent gene being NMDAR biology-related) or chronic life stress [27,28,49^▪▪]. Exposure of the brain to the immune system via leaky BBB alone, however, may not suffice to induce NMDAR1-AB formation, as APOE4 carrier status or ApoE knockout alone, does not seem to predispose to NMDAR1-AB formation [28,35,49^▪▪].

CONCLUSION

All naturally occurring NMDAR1-AB, irrespective of immunoglobulin isotype or epitope, are potentially syndrome-relevant upon access to or production within the brain. The intriguing fact that NMDAR1-AB age-dependently increase to an extremely high seroprevalence across mammals may indicate additional not yet understood roles in natural/physiological autoimmunity.

Acknowledgements

None.

Financial support and sponsorship

The current work was supported by the Max Planck Society, the DFG (CNMPB), EXTRABRAIN EU-FP7, and the Niedersachsen-Research Network on Neuroinfectiology (N-RENNT).

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest