Diagnostic and Pathogenic Significance of Glutamate Receptor Autoantibodies FREE

More than 100 autoantibodies that recognize known antigens have been documented in human diseases, and many of these autoantibody-associated diseases involve the nervous system. Some autoantibodies (eg, those against nicotinic acetylcholine receptors in myasthenia gravis) have been established as the proximate cause of neurologic deficits. Others (eg, anti-HuD in paraneoplastic encephalomyelitis), although not known to be pathogenic, are helpful in differential diagnosis.¹

In 1994, autoantibodies against the ionotropic glutamate receptor protein GluR3 were reported in 3 of 4 children with Rasmussen encephalitis.² Plasma exchange resulted in transient improvement in seizure control and cognition in 1 of the 3 autoantibody-positive children. Since that initial article, plasma and cerebrospinal fluid autoantibodies that recognize GluR3 and other glutamate receptor proteins have been described in focal epilepsies, systemic lupus erythematosus (SLE), central nervous system (CNS) ischemia, and paraneoplastic encephalopathies. We discuss the diagnostic utility and pathophysiologic significance of these autoantibodies.

Glutamate receptors transduce excitatory signals from glutamatergic presynaptic terminals to postsynaptic neurons. Glutamate receptors are also expressed by nonneuronal cells, including neuroglia and T lymphocytes, where, as in neurons, they serve to convey glutamate signals across the plasma membrane. Glutamate receptors are classified into 2 broad groups based on their structures and modes of operation: ionotropic glutamate receptors are heterotetrameric or homotetrameric channels that are opened by glutamate, thus causing sodium influx and plasma membrane depolarization, and metabotropic glutamate receptors are plasma membrane homodimers that modulate enzyme and channel functions and gene transcription via second messenger–dependent mechanisms. The composition and some properties of these receptors are outlined in Table 1.

Table 2summarizes the neurologic disorders in which GluR autoantibodies have been reported. Since the initial publication by Rogers et al,² the case for an association between Rasmussen encephalitis and GluR3 autoantibodies has been weakened by the failure to detect GluR3 antibodies in many patients who meet the clinical and pathologic criteria for diagnosis of this disorder and by the demonstration of GluR3 autoantibodies in patients with noninflammatory focal epilepsies.^{3 - 7} Serum and cerebrospinal fluid NR2B autoantibodies have been reported in patients with Rasmussen encephalitis and were found in other forms of chronic epilepsia partialis continua and in nonherpetic acute limbic encephalitis but not in patients with the Lennox-Gastaut syndrome or infantile spasms (West syndrome). In patients in whom serial autoantibody assays were available, IgM antibodies appeared after the onset of seizures and later became undetectable.^{8 - 9} NR2A/NR2B autoantibodies are detectable in more than a third of patients with SLE.^{10 - 11} Whether titers of these autoantibodies correlate with abnormalities in cognition and other neuropsychiatric complications of SLE remains controversial.^{11 - 16}

Elevated titers of IgG autoantibodies against an NR2A/NR2B peptide have been reported in patients with acute ischemic infarction or transient ischemic attack. These autoantibodies were not present in patients with intracerebral hemorrhage or hypertension without neurologic deficits. There was a strong correlation between antibody titer and severity of neurologic deficits in the ischemic infarction group.¹⁷ The same laboratory has subsequently reported that an elevated preoperative titer of NR2 autoantibodies was highly predictive of poor neurologic outcome after cardiac surgery in high-risk surgical patients.¹⁸ These results suggest that N-methyl-D-aspartate receptor autoantibodies are biomarkers for CNS ischemia, but this requires confirmation by other laboratories. Furthermore, the 2-fold increase in IgG autoantibody titers within 12 hours after admission to the intensive stroke unit that these investigators reported in patients with ischemic infarction¹⁷ seems unusually rapid for even a memory B-cell antibody response.

Autoantibodies against the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptor (AMPAR) subunit proteins GluR1 and GluR4, the kainate receptor subunit protein GluR5, the N-methyl-D-aspartate receptor NR1/NR2 heteromers, and the metabotropic receptor subunit protein mGluR1 have been well characterized in paraneoplastic syndromes.^{19 - 22} Antibodies that bound to NR1/NR2B heteromers but not to individual N-methyl-D-aspartate receptor subunit proteins were found in the serum samples and cerebrospinal fluid of 12 women with teratomas and paraneoplastic encephalitis but not in patients with other paraneoplastic and nonparaneoplastic forms of encephalitis.²²

Whereas CNS-specific antigens are normally shielded from the systemic immune system, infarction and other pathologic processes in which cellular necrosis occurs could result in the release of antigenic glutamate receptor peptides from the CNS for processing by the peripheral immune system. In Rasmussen encephalitis and other inflammatory disorders, release of glutamate receptor antigens to the periphery might not be required. Instead, antigen presentation and autoantibody production could be initiated by immune cells already resident in the CNS. Peripheral tissues may also provide glutamate receptor peptides to antigen-presenting cells. For example, T lymphocytes express GluR3-containing AMPARs, and their GluR3 undergoes proteolytic cleavage during T-lymphocyte activation.²³ In the paraneoplastic encephalopathy associated with ovarian teratomas, glutamate receptors expressed by neurons in the teratomas may serve as the source of N-methyl-D-aspartate receptor heteromers for autoantibody production.²² Finally, glutamate receptor autoantibody production might be a result of molecular mimicry. This phenomenon has been well documented in mice that develop autoantibodies against multiple AMPAR subunit proteins when infected with murine leukemia virus.²⁴ A well-documented example in rabbits and mice that may occur in humans with SLE is the production of autoantibodies that recognize double-stranded DNA epitopes and NR2A/NR2B after sensitization to an NR2 peptide.^{10 ,14 ,25 - 26}

Are glutamate receptor autoantibodies pathogenic? Animal studies suggest that they can be. Rabbits immunized with an NR2B or GluR3 peptide develop seizures, and rabbit GluR3 autoantibodies can activate AMPARs expressed by transfected oocytes and cultured neuronal AMPARs and kill the cultured neurons by means of a complement-dependent mechanism.^{2 ,27 - 29} Mouse anti-GluR3 peptide antibodies activate AMPARs and induce excitotoxicity in cultured neurons,³⁰ and mice immunized with either NR2/NR3 or GluR3 peptides demonstrate CNS neuronal loss in vivo.^{26 ,30} One study³¹ argues that glutamate receptor antibodies can also be neuroprotective; rats expressing high-titer NR1 autoantibodies were refractory to kainate-induced seizures and were partially protected against forebrain ischemic infarction. This observation requires confirmation by other laboratories.

With respect to the pathogenicity of human glutamate receptor autoantibodies, more than a decade has passed since the initial report of a therapeutic response to plasma exchange in Rasmussen syndrome, but support for the pathogenicity of glutamate receptor autoantibodies in this disorder is still equivocal. Many patients with Rasmussen encephalitis do not have detectable GluR3 autoantibodies, and a consensus has not yet been reached as to whether progressive CNS damage in this disorder is mediated by autoantibodies, pathogenic T lymphocytes, or both.³²

There is preliminary evidence of the neuropathogenicity of glutamate receptor autoantibodies in SLE, sporadic olivopontocerebellar atrophy, and paraneoplastic encephalopathies. Stereotaxic injection of autoantibodies prepared from the brain specimen of a patient with SLE and severe encephalopathy killed mouse hippocampal neurons.¹⁴ Encephalopathic patients with NR2A/NR2B autoantibodies showed diffusion-weighted imaging abnormalities in the amygdala that were not present in encephalopathic patients who do not express this class of autoantibodies.¹³ GluR2 autoantibodies from a patient with sporadic olivopontocerebellar atrophy activated AMPARs in cultured mouse neurons,³³ and GluR5 autoantibodies from a group of patients with paraneoplastic encephalopathy activated kainate receptor–mediated currents,¹⁹ thus suggesting that these autoantibodies are capable of inducing neuronal excitotoxicity. Perhaps the strongest evidence of a pathophysiologic role for glutamate receptor autoantibodies was obtained in 2 patients who, while in remission from Hodgkin disease, developed severe cerebellar ataxia and expressed mGluR1 autoantibodies that blocked glutamate-induced inositol phosphate formation. These autoantibodies, when administered intrathecally to mice, elicited reversible ataxia.^{20 - 21}

Correspondence:David Pleasure, MD, University of California, Davis, School of Medicine, c/o Shriners Hospital Northern California, 2425 Stockton Blvd, Sacramento, CA 95817 (david.pleasure@ucdmc.ucdavis.edu).

Accepted for Publication:August 8, 2007.

Financial Disclosure:None reported.

Funding/Support:This study was supported by grant RO1 NS25044 from the National Institutes of Health.