ObjectivesAutoantibodies against the extracellular domains of the voltage-gated potassium channel (VGKC) complex proteins, leucine-rich glioma-inactivated 1 (LGI1) and contactin-associated protein-2 (CASPR2), are found in patients with limbic encephalitis, faciobrachial dystonic seizures, Morvan's syndrome and neuromyotonia. However, in routine testing, VGKC complex antibodies without LGI1 or CASPR2 reactivities (double-negative) are more common than LGI1 or CASPR2 specificities. Therefore, the target(s) and clinical associations of double-negative antibodies need to be determined.MethodsSera (n=1131) from several clinically defined cohorts were tested for IgG radioimmunoprecipitation of radioiodinated α-dendrotoxin (125I-αDTX)-labelled VGKC complexes from mammalian brain extracts. Positive samples were systematically tested for live hippocampal neuron reactivity, IgG precipitation of 125I-αDTX and 125I-αDTX-labelled Kv1 subunits, and by cell-based assays which expressed Kv1 subunits, LGI1 and CASPR2.ResultsVGKC complex antibodies were found in 162 of 1131 (14%) sera. 90 of these (56%) had antibodies targeting the extracellular domains of LGI1 or CASPR2. Of the remaining 72 double-negative sera, 10 (14%) immunoprecipitated 125I-αDTX itself, and 27 (38%) bound to solubilised co-expressed Kv1.1/1.2/1.6 subunits and/or Kv1.2 subunits alone, at levels proportionate to VGKC complex antibody levels (r=0.57, p=0.0017). The sera with LGI1 and CASPR2 antibodies immunoprecipitated neither preparation. None of the 27 Kv1-precipitating samples bound live hippocampal neurons or Kv1 extracellular domains, but 16 (59%) bound to permeabilised Kv1-expressing human embryonic kidney 293T cells. These intracellular Kv1 antibodies mainly associated with non-immune disease aetiologies, poor longitudinal clinical–serological correlations and a limited immunotherapy response.ConclusionsDouble-negative VGKC complex antibodies are often directed against cytosolic epitopes of Kv1 subunits and occasionally against non-mammalian αDTX. These antibodies should no longer be classified as neuronal-surface antibodies. They consequently lack pathogenic potential and do not in themselves support the use of immunotherapies.