Abstract
Chemical Corrector Treatment Ameliorates Increased Seizure Susceptibility in a Mouse Model of Familial Epilepsy
Yokoi N, Fukata Y, Kase D, Miyazaki T, Jaegle M, Ohkawa T, Takahashi N, Iwanari H, Mochizuki Y, Hamakubo T, Imoto K, Meijer D, Watanabe M, Fukata M. Nat Med 2015;21:19–26.
Epilepsy is one of the most common and intractable brain disorders. Mutations in the human gene LGI1, encoding a neuronal secreted protein, cause autosomal dominant lateral temporal lobe epilepsy (ADLTE). However, the pathogenic mechanisms of LGI1 mutations remain unclear. We classified 22 reported LGI1 missense mutations as either secretion defective or secretion competent, and we generated and analyzed two mouse models of ADLTE encoding mutant proteins representative of the two groups. The secretion-defective LGI1(E383A) protein was recognized by the ER quality-control machinery and prematurely degraded, whereas the secretable LGI1(S473L) protein abnormally dimerized and was selectively defective in binding to one of its receptors, ADAM22. Both mutations caused a loss of function, compromising intracellular trafficking or ligand activity of LGI1 and converging on reduced synaptic LGI1-ADAM22 interaction. A chemical corrector, 4-phenylbutyrate (4PBA), restored LGI1(E383A) folding and binding to ADAM22 and ameliorated the increased seizure susceptibility of the LGI1(E383A) model mice. This study establishes LGI1-related epilepsy as a conformational disease and suggests new therapeutic options for human epilepsy.
Commentary
Most of the attention in “personalized medicine” genetics focuses on problems that result from genetic mutations in nucleic acids, which ultimately affect three-dimensional protein folding. However, editing nucleic acid abnormalities in humans can be challenging. The ability to “correct” the abnormalities that result from protein misfolding without changing nucleic acid sequences has been demonstrated for some genetic diseases but not epilepsy (1).
LGI1 (Leucine-rich, glioma inactivated 1a) is a leucine-rich protein that is expressed in brain and plays a role in regulating postnatal glutamatergic synapse development. Mutations in the gene coding for this protein (LGI1) result in autosomal dominant lateral temporal lobe epilepsy (ADLTE), which is characterized by an onset in the second or third decades of life, focal-onset seizures, and in most cases, auditory auras (2–4). The clinical course is variable but some people experience refractory seizures. LGI1 has a number of binding partners that affect excitatory neurotransmission but it is a secreted protein, rather than a channel (like most known causes of genetic epilepsy). Therefore, a deeper understanding of its pathological mechanisms remains elusive (5).
Because LGI1 is expressed at the cell membrane, Yokoi et al. (6) classified mutations from a patient database based on the ability of the abnormal protein to be secreted in cultured cells. They ultimately used one mutant that was “secretion-competent” (S473L) and one that was “secretion-defective” (E383A) to explore the cell biology and epilepsy-related implications of LGI1-associated pathophysiology. The authors expressed the mutant genes (in mice lacking native LGI1), showing that both mutations led to decreased survival and increased spontaneous seizure frequencies compared wild-type mice. When the mutant proteins were expressed in mice that had some native LGI1, increased seizure susceptibility suggested that mutations in LGI1 are caused by loss of function, not dominant-negative effects. As expected, expression of the secretion-defective mutant was lower than the secretion-competent mutant in the brain, but elevated mRNA levels of the secretion-defective mutant gene suggested that its protein was unstable. Immunohistochemical analysis showed that the secretion-defective mutant protein was mislocalized to neurons only in the hippocampus, while the normal protein was localized to both neurons and the neuropil. The secretion-competent protein had an expression pattern that was intermediate between the other two phenotypes.
Not surprisingly, the secretion-defective product bound to proteins involved in the endoplasmic reticulum (ER) quality-control pathway, including a folding sensor (UGT1) and an ER chaperone protein (BiP). Binding of the ER-associated proteins to the secretion-defective mutant product was greater than the secretion-competent product. The secretion-competent product also had a post-translational processing (N-glycosylation) pattern similar to that of wild-type protein, in contrast to the secretion-defective product. Together, the data suggest that the secretion-competent mutant passes quality-control checkpoints more consistently than the secretion-defective mutant, which likely is retained in the ER for degradation.
LGI1 has two native receptors belonging to the ADAM (
Identification of ADLTE as a protein misfolding-associated disease makes it a candidate for treatment with chemical correctors (compounds known to facilitate bypass around the ER quality-control process). The authors screened a number of such chemical correctors and found two that increased LGI1 secretion in cultured cells: sodium phenylbutyrate (4PBA, which is used clinically in combination with other agents for patients with urea cycle defects such as ornithine transcarbamylase deficiency (7)) and suberoylanilide hydroxamic acid (SAHA, used clinically for its antitumor effects (8)). The secretion-defective mutant LGI1 was also able to bind to ADAM22 after 4PBA treatment, raising the possibility that treatment with a chemical corrector might improve the underlying epilepsy seen in ADLTE.
To test this hypothesis, heterozygous mice expressing some native LGI1 and the secretion-defective mutant LGI1 were treated with 4PBA for 10 days and showed lower seizure scores after pentylenetetrazol (PTZ) injection; this effect was dependent on the dose of 4PBA, and there was evidence that both secretion of the mutant product and its binding to ADAM22 were increased. The secretion-competent mutant showed no improvement, which also supports the hypothesis that 4PBA was acting via a secretion-dependent mechanism and not a nonspecific effect on the two mutants. 4PBA did not have an effect in wild-type mice subjected to PTZ, suggesting that it does not have intrinsic acute antiseizure properties in response to this convulsant agent. 4PBA treatment in mice completely lacking native LGI1 but expressing the secretion-defective mutant led to slightly longer survival than those treated with vehicle. The number of mice exhibiting behavioral seizures was also decreased. Treatment with 4PBA also partially corrected the amount of mutant protein, ADAM22, and ADAM23 expressed in the neuropil in these mice. The authors conclude by noting the potential for 4PBA use in patients with missense mutations in ion channels (and by logical extension, other proteins), noting that only low levels of secreted protein may be highly beneficial.
Interestingly, both 4PBA and SAHA are also histone deacetylase (HDAC) inhibitors (8, 9), and while the effect on protein folding is likely unrelated, it would still be interesting to know if there is any contribution to treatment as a result of HDAC inhibition. It is also unclear if treatment with small molecules than enhance protein function would provide additional efficacy once the protein is correctly targeted to the cell membrane. The seizure-related outcomes in this study were unclear, as seizure susceptibility, one indicator of neuronal excitability, may not necessarily equate to epilepsy (results of EEG monitoring for spontaneous seizures were not reported, and video recording sessions were brief).
Yokoi et al. have made a substantial contribution to our understanding of ADLTE as a loss-of-function disease involving protein misfolding. Although not all mutations were tested, these data provide a framework for understanding the pathology of other types of epilepsy due to protein misfolding and its direct consequences (i.e., failure to bind to other structural proteins or form multimers). Treatment with chemical correctors was also shown as proof-of-concept in vivo, reinforcing the notion that correction of nucleotide sequences may not be necessary to ameliorate symptoms and signs of genetic disease.