Autoimmunity due to molecular mimicry as a cause of neurological disease

Abstract

One hypothesis that couples infection with autoimmune disease is molecular mimicry. Molecular mimicry is characterized by an immune response to an environmental agent that cross-reacts with a host antigen, resulting in disease. This hypothesis has been implicated in the pathogenesis of diabetes, lupus and multiple sclerosis (MS). There is limited direct evidence linking causative agents with pathogenic immune reactions in these diseases. Our study establishes a clear link between viral infection, autoimmunity and neurological disease in humans. As a model for molecular mimicry, we studied patients with human T-lymphotropic virus type 1 (HTLV-1)-associated myelopathy/tropical spastic paraparesis (HAM/TSP), a disease that can be indistinguishable from MS (refs. 5,6,7). HAM/TSP patients develop antibodies to neurons. We hypothesized these antibodies would identify a central nervous system (CNS) autoantigen. Immunoglobulin G isolated from HAM/TSP patients identified heterogeneous nuclear ribonuclear protein-A1 (hnRNP-A1) as the autoantigen. Antibodies to hnRNP-A1 cross-reacted with HTLV-1-tax, the immune response to which is associated with HAM/TSP (refs. 5,9). Immunoglobulin G specifically stained human Betz cells, whose axons are preferentially damaged. Infusion of autoantibodies in brain sections inhibited neuronal firing, indicative of their pathogenic nature. These data demonstrate the importance of molecular mimicry between an infecting agent and hnRNP-A1 in autoimmune disease of the CNS.

Fig. 1

Specificity of HAM/TSP IgG for CNS neurons and isolation of the neuronal antigen. a, Immunohistochemistry of human tissues. Biotinylated HAM/TSP IgG stained CNS neurons (cortex, brown reaction product) (scale bar, 100 µm). There was no staining of glia (arrowhead), dorsal root ganglion or systemic organs. A tax mAb mimicked HAM/TSP IgG staining indicated by the red stain. Neurofilament (gray) confirmed localization of tax mAb to neurons. b, Western blots of proteins derived from CNS neurons. HAM/TSP IgG showed an intense signal at 33 kD (lane 1). There was no reactivity using IgG isolated from HTLV-1 seronegative (lane 2) or HTLV-1 seropositive, neurologically asymptomatic individuals (lane 3). The tax mAb recognized the 33-kD neuronal antigen (lane 4). c, Purification and western blots of the neuronal protein following high salt extraction and centrifugation. Molecular weight markers (lane 1), precipitate (lane 2) and supernatant (lane 3). HAM/TSP IgG reacted with the supernatant (lane 3), but not the precipitate (lane 2). d, 2-D gel electrophoresis and western blot of the neuronal protein purified from the supernatant. Coomassie stain of gel (left). Western blot of the neuronal extract (right), HAM/TSP IgG reacted at 33–38 kD, pI = 9.3 (arrow).

Fig. 2

Immunoreactivity of HAM/TSP IgG with the neuronal extract and hnRNP-A1. a, Left panel, Coomassie staining; right panel, HAM/TSP IgG immunoreactivity by western blot. Neuronal extract (lane 1), molecular weight markers (lane 2) and rhnRNP-A1 (lane 3). rhnRNP-A1 is identified as a 38-kD band that is recognized by HAM/TSP IgG (right panel, third lane) and corresponds to an immunoreactive band in the neuronal extract (right panel, lane 1). b, Reactivity by western blot following 2-D gel electrophoresis of neuronal extract is adsorbed by pre-incubation with rhnRNP-A1. Immunoreactivity of HAM/TSP IgG with neuronal proteins (top panel) was adsorbed by pre-incubation with rhnRNP-A1 (bottom panel, arrows). There was immunoreactivity to other neuronal proteins that did not adsorb with rhnRNP-A1 (arrowheads). M, molecular weight markers; N, neurons; A1, rhnRNP A1. c, Pre-incubation of HAM/TSP IgG with 0, 0.1 and 1.0 µg rhnRNP-A1 followed by western blot of the neuronal extract demonstrated concentration-dependent adsorption of immunoreactivity. d, IgG purified from the CSF and CNS of a patient with HAM/TSP reacted with neurons and rhnRNP-A1. e, The tax mAb reacted with neurons and rhnRNP-A1. f, Screening of patient sera demonstrated 10/10 HAM/TSP patients (numbers 2, 3, 5–7, 9, 10 and 12–14) and 0/10 HTLV-1 seronegative controls (1, 4, 8, 11 and 15–20) reacted with rhnRNP-A1. *, patient 13 was used as a positive control for the assay that tested patients 15–20.

Fig 3

Target specificity and biologic activity of HAM/TSP IgG. a–f, CNS neurons were isolated from the precentral gyrus (a–c) and posterior parietal-occipital cortex (d–f). a and d, Trypan blue staining confirmed the presence of large pyramidal neurons and Betz cells (a) compared with typical cortical neurons (d). All cells stained with neurofilament (d, inset). b and e, Coomassie staining and western blots of neuronal proteins. There was an intense HAM/TSP IgG signal from pyramidal neurons derived from the pre-central gyrus that included Betz cells (b), compared with parietal-occipital cortex (e) . Coomassie staining shows comparable protein loading. c and f, Immunohistochemistry of the pre-central gyrus of human motor cortex and of the parietal-occipital cortex. c, A Betz cell is shown surrounded by pyramidal neurons. Staining is present throughout the cytoplasm and extends into neuronal processes (arrow). f, Staining is less intense in parietal-occipital cortex. g–i, Patch clamp recording of neurons in rat brain following extracellular infusion of antibodies (normal CSF - red). g, Left, Physiological concentrations of HAM/TSP IgG (5, 10, 15 and 20 µg/ml, respectively) decreased neuronal firing without affecting calcium (data not shown). HAM/TSP IgG localized to the cytoplasm of neurons (inset). Right, Experiments were replicated using the monospecific antibody to hnRNP-A1 (5 and 10 µg/ml, respectively). h, Application of normal IgG (10 µg/ml; green) showed no slowing in neuronal firing. i, The tax mAb decreased (10 µg/ml; green), then completely inhibited neuronal firing (20 µg/ml; blue).