Background

At this Journal Club it was decided to review a historical paper on the pathophysiology underlying autoimmune neuromyotonia. The paper, “Autoantibodies Detected to Expressed K+ Channels Are Implicated in Neuromyotonia”, from Annals of Neurology (1997, 41:238-246), used a novel technique that depended on knowing the gene for the suspected antibody target protein, in this case a potassium channel. The purpose of choosing this paper was partly to highlight how the known range of antibody mediated neurological disease has grown hugely over the subsequent twenty years, and partly to illustrate how positive findings can sometimes be seen in retrospect to have arisen through a degree of serendipity.

Acquired neuromyotonia is now known to be one of a number of neurological conditions that arise through auto-antibodies interfering with voltage gated potassium (KV) channel function. Interference with resting potentials and membrane recovery after action potentials in peripheral nerve results in continual high frequency discharges and continuous muscle activation as cramp, fasciculations and neuromyotonia. Sometimes this can be precipitated by cold, exercise or voluntary muscle activation. Other features included in the spectrum of KV channel auto-immunity include autonomic dysfunction, seizures, psychiatric disturbance and limbic encephalitis. When resulting in a neuropathy  and neuromyotonia, the term Isaac’s syndrome is often used, while a presentation of neuromyotonia with autonomic or CNS involvement is described as Morvan’s syndrome.

Study Design

The techniques used by Hart et al inferred that there would be an affinity of patients’ antibodies for the Kv channel, as it was already known that acquired myotonia results from disturbances of Kv channel function. If there was a known toxin for this channel, as with bungarotoxin for nicotinic acetyl choline receptors, this could be used as a labelled high affinity ligand and form the basis for a radioimmunoassay for detection of circulating antibodies against the channel. Dendrotoxin is a highly specific and high affinity toxin, but binds only a cohort of potassium subunits (Kv 1.1, 1.2 and 1.6).

The first type of assay used in this paper relied upon dendrotoxin; brains containing solubilised Kv channels were treated with radiolabelled dendrotoxin, and then with serum from neuromyotonia patients. An anti-human IgG was used to immunoprecipitate all human antibodies from this solution, which would include any Kv complex-dendrotoxin bound antibodies. When neuromyotonia patient serum was used, the resulting precipitant (which contains any antibody that has bound to its antigen) contained the radiolabel, indicating that the patient antibodies had become coupled to material containing the dendrotoxin and, by inference, become bound to the Kv channel. This result was found in some – but not all – patients (6/12) and reassuringly in none of the control samples (myasthenia gravis, Lambert-Eaton and healthy controls)

Verification that the Kv channel rather than other dendrotoxin-bound material from solubilised brain was provided by demonstrating binding of neuromyotonia patient antibodies to dendrotoxin-bound Kv1 subunits expressed in Xenopus toad oocytes via cRNA after complementary RNA expression. Knowing the gene for Kv1 enabled production of the complementary RNA. Expression of this in the toad oocyte meant that the KV1 protein would now be present in pure form. In this experiment,  4/12 samples were positive of the neuromyotonia cohort (and again, 0/18 controls). This positivity rate was felt to be consistent with the fact that human disease antibodies might be to subunits other than Kv1. The authors offered no information on the correlation between titres on the human brain assay and Xenopus expression system.

The authors then turned to immunohistochemical staining instead of immunoprecipitation. In this assay, antibodies labelled with horseradish peroxidase stain are created that bind to immune complexes. Thus any patient serum anti-Kv channel antibody that has bound to Kv1 channels expressed on the toad oocytes will in turn be bound to by the staining antibody-binding antibodies. The oocytes are then looked at under a microscope. They found positive staining of serum added to different Kv subtypes but not to a number of controls. However, since the oocytes had been fixed, permeabilised and sectioned prior to incubation with patient antibody, one could not confirm that the KV1 channel had actually been expressed on the membrane surface as it would naturally in human neurones. They suggested the technique could be applied to many other putative antigens for pathogenic circulating antibodies, provided the genes for the antigens were known, which is the case for most proteins now.

In another experiment to check for antibody binding to potassium channel subtypes for which dendrotoxin is not a ligand, these Xenopus cells were incubated with sulphur labelled methionine at the time that they were injected with one of three different Kv complementary RNAs, so that when they expressed the channel protein, it would be radiolabelled by the incorporated methionine amino acid, and detectable with autoradiography. The serum of neuromyotonia patients and controls was applied to these preparations and anti-human IgG was used to determine patient antibody-bound Kv material. However, this precipitant did not reveal any labelling with either patient or control serum. The authors suggested that this may be because the antibody binding is conformationally dependent, a feature that somehow did not apply when dendrotoxin had already bound in the other assay. Alternatively, it could also reflect that Kv channels are not really the antigenic target in neuromyotonia – an explanation which has subsequently been confirmed in more recent data.

Historical Context and Journal Club Discussion

Since this paper was published, as mentioned in the background, a more extensive spectrum of disorders associated with potassium channel antibodies has been described, but unfortunately there appears to be no specificity linking disease phenotype to antibodies to particular Kv subunit combinations.

More recently still, the antigenic targets of these antibodies have been clarified to be proteins associated with the potassium channel rather than the channel itself. So the antibodies were not what they were purported by the paper to be after all! It is not surprising therefore that the experiment with directly methionine-labelled subunits yielded negative results. It is not clear why the authors thought the naturally occurring pathological antibodies would bind to a channel better when it had toxin attached to it. But it is also now not clear why the immuno-histochemistry labelling of Kv1 expressing oocytes was positive in some cases, as the actual antigen in most cases was absent. Only in a small minority of neuromyotonia cases have the newer assays demonstrated that the Kv channel proper (and not an associated protein) is the true antigen.

In fact, since 1997, the radioimmunoprecipitation assay by which these antibodies are detected remains largely unchanged from that used before the advance that the paper was supposed to introduce: rodent brain is used as the substrate of Kv channels and still labelled with dendrotoxin. So, since the toxin binds only a small proportion of all Kv channels, there are likely to be many cases of antibodies against other Kv or Kv-associated antigens that are currently undetected by current methods. There is significant scope for improvement in these assays, in terms of range of antigens tested, cross-assay standardisation and, importantly, timescales of test to result.  It was discussed in the journal club how this currently significantly limits appreciation of the potential scope of antibody mediated neurological diseases.

This case was presented and summarised by Dr Sian Alexander, Specialist Registrar in Neurology at Queens Hospital, Romford.