‘Moonlighting’ surface antigens: a paradigm for autoantibody pathogenicity in neurology?

Abstract

This scientific commentary refers to ‘Human autoantibodies to amphiphysin induce defective presynaptic vesicle dynamics and composition’ by Werner et al. (doi:10.1093/awv324).

This scientific commentary refers to ‘Human autoantibodies to amphiphysin induce defective presynaptic vesicle dynamics and composition’ by Werner et al. (doi:10.1093/awv324).

The field of CNS autoantibody-mediated diseases has provided a new source of potentially treatable neurological conditions, with pathophysiology that is of increasing interest to neurologists, immunologists and neuroscientists (Irani et al., 2014). These syndromes can feature seizures, cognitive impairment, movement disorders, dysautonomia, rigidity or startle. Most of the more recently discovered autoantibodies target the extracellular domain of natively-expressed neuronal surface proteins, and have the potential to be pathogenic if they gain access to their antigens in vivo. These syndromes are often treatable and only a minority are associated with malignant tumours (Leypoldt et al., 2015; Varley et al., 2015). By contrast, the more established paraneoplastic antigens, such as Hu, Yo and Ma2, are associated with malignant tumours and a poor response to immunotherapies. These antibodies are directed to intracellular proteins and are not thought to be pathogenic. Concordantly, antibody passive transfer experiments have yielded negative results (Tanaka et al., 1995).

This is one reason why the pathogenesis of the autoantibody-associated disease stiff-person syndrome (SPS) remains enigmatic. The most common antibodies in SPS target the intracellular enzyme glutamic acid decarboxylase (GAD). However, the possible role of GAD antibodies in disease causation has been somewhat superseded by the discovery that GAD antibodies coexist with neuronal surface antibodies with causative potential (Lancaster et al., 2010; Chang et al., 2013). Antibodies against amphiphysin are found in <10% of patients with SPS; these individuals are usually female and often have a breast or small-cell lung carcinoma. Unlike many forms of paraneoplastic disease associated with malignant tumours, a few reports suggest this form of SPS responds to immunotherapies, including plasma exchange, with the response correlating well with amphiphysin-antibody levels (Wessig et al., 2003; Sommer et al., 2005).

The SH3-domain of amphiphysin appears to be the immunodominant region and amphiphysin-antibodies bind well on western blots, suggesting a linearized rather than a native conformationally-dependent epitope. Although there are few histopathology reports on this rare disease, the available data do not lend support to a B-cell or antibody-mediated disorder but have shown a marked predominance of CD8 and CD4 T-cells (Wessig et al., 2003). Therefore, evidence argues both for and against the pathogenicity of amphiphysin-antibodies.

A series of elegant experiments from Geis, Sommer and colleagues in previous publications appeared to clinch the issue. Contrary to conventional wisdom, they showed that amphiphysin-antibodies fulfil Witebsky’s postulates for antibody causality. They showed that an amphiphysin-SH3 domain-specific population of immunoglobulin G (IgG) was internalized by rat spinal motor neuron–interneuron co-cultures. The antigen specificity was confirmed by the absence of this effect in amphiphysin-knockout mice. GABAergic synapses appeared especially vulnerable (Geis et al., 2010). Furthermore, systemic and intrathecal injection of amphiphysin-antibody SPS IgG into experimental rodents produced a phenotype that closely modelled symptoms in patients: mice developed truncal and hindlimb cramps, an exaggerated lordosis, and bursts of EMG activity (Sommer et al., 2005; Geis et al., 2010). Remarkably, the spasms began only 15–30 min after systemic high-titre amphiphysin IgG injections. The rat spinal cords showed human IgG deposits, consistent with the likely site of pathology. In a separate experiment, IgG from one patient with SPS was shown to bind to the amygdala and attenuate exploratory behaviours, an effect akin to the anxiety seen in SPS (Geis et al., 2012). While behavioural and histological analyses were consistent with amphiphysin-antibody pathogenicity, the molecular basis of these observations remained unclear.

Amphiphysin is known to regulate clathrin-coated vesicle-associated endocytosis, a major mechanism by which synaptic vesicles are recycled after depolarization-mediated fusion. In this issue of Brain, Werner and co-workers report the results of detailed experiments examining the effects of amphiphysin-antibodies on the composition of presynaptic vesicular pools and the implications for neurotransmission (Werner et al., 2016).

Glossary

Amphiphysin: Amphiphysin belongs to the BAR (Bin-Amphiphysin-Rvsp) family of proteins and is associated with the cytoplasmic surface of endocytotic vesicles. Amphiphysin is involved in clathrin-mediated endocytosis via its interaction with cytoskeletal proteins such as dynamin.

Witebsky’s postulates for antibody causality: Ernest Witebsky proposed criteria to determine whether a disease could be termed autoimmune. These include the detection of the antibodies in cases, recognition of the autoantigen, and the demonstration that an experimental animal develops similar changes to affected humans. More recently, the criteria have been slightly modified by others to include circumstantial clinical clues (such as tight correlations between antibody levels and clinical features).

Ultrastructural analyses revealed that in vivo supramaximal sciatic nerve stimulation caused increases in synaptic density, vesicle pool size, and clathrin-coated vesicle numbers in spinal cord boutons. This may serve to sustain responses to prolonged high-frequency stimulation. Chronic intrathecal application of amphiphysin-specific IgG reversed these effects in rats, which also developed clinical features of SPS. The most marked reduction in vesicles was observed in the presynapses with the highest GABA density. However, the effect was dependent on which vesicle population was under scrutiny; there was a decrease in the resting pool (characterized by synaptobrevin 7) and an increase in the readily releasable pool (expressing synaptobrevin 2). This suggests both depletion of resting vesicles, and impaired endocytosis of other vesicles at the plasma membrane after fusion. In addition, there were alterations in the distribution of endophilin, a protein known to interact with amphiphysin.

Taken together, these findings provide a cogent potential molecular mechanism for the pathogenesis of amphiphysin-antibody mediated SPS. Amphiphysin antibodies appear to disrupt vesicular recycling leading to slower endocytosis, more rapid synaptic exhaustion and a failure of exocytotic synaptic transmission. This mechanism has a predilection for GABAergic interneurons, partly because these neurons often show a high turnover of vesicles. In turn, this may lead to increased motor unit firing with consequent stiffness and spasms, and may account for the recognized therapeutic benzodiazepine response.

This well-designed series of experiments shows several potential limitations. First, the number of patient samples was very small. This calls into question the generalizability of these findings to other patients with amphiphysin antibodies. Second, some patients with amphiphysin antibodies have been reported to have ataxia, dysautonomia and cerebellar features, findings not accounted for by these observations (Pittock et al., 2005). Finally, the presence of amphiphysin antibodies in up to 2% of healthy and disease controls suggests the antibody alone may be insufficient for symptom causation in many adults (Dahm et al., 2014).

Nevertheless, in the samples studied it is intriguing to consider how and where the precise antibody–antigen interaction might take place. It may be that presynaptic boutons have a relatively indiscriminate mechanism of antibody uptake, and any antibodies that survive an intracellular environment can then access their targets. Alternatively, amphiphysin may gain access to the extracellular environment upon fusion of vesicles with the presynaptic membrane. This may give the autoantibodies a brief window of opportunity in which to interact specifically with their transiently-exposed antigenic target. Indeed, the site of antibody–antigen interaction would benefit from more detailed visualization in future experiments and could be a site for therapeutic interventions. While the concept of a transiently-exposed antigen is less likely to translate to the RNA-binding proteins Hu, Yo and Ri, it may be a mechanism by which GAD-antibodies mediate a similar antigen-specific effect (Chang et al., 2013; Hansen et al., 2013). Indeed, ‘transient moonlighting’ of such antigens to the surface of the synaptic cleft may be a paradigm that allows other intracellular antigens to be targeted by autoantibodies.