GRIN and Bear the Diverse Functional Effects of Rare NMDA Receptor Variants

Commentary

Mechanistic Insight into NMDA Receptor Dysregulation by Rare Variants in the GluN2A and GluN2B Agonist Binding Domains.

Swanger SA, Chen W, Wells G, Burger PB, Tankovic A, Bhattacharya S, Strong KL, Hu C, Kusumoto H, Zhang J, Adams DR, Millichap JJ, Petrovski S, Traynelis SF, Yuan H. Am J Hum Genet 2016;99:1261–1280.

Epilepsy and intellectual disability are associated with rare variants in the GluN2A and GluN2B (encoded by GRIN2A and GRIN2B) subunits of the N-methyl-D-aspartate receptor (NMDAR), a ligand-gated ion channel with essential roles in brain development and function. By assessing genetic variation across GluN2 domains, we determined that the agonist binding domain, transmembrane domain, and the linker regions between these domains were particularly intolerant to functional variation. Notably, the agonist binding domain of GluN2B exhibited significantly more variation intolerance than that of GluN2A. To understand the ramifications of missense variation in the agonist binding domain, we investigated the mechanisms by which 25 rare variants in the GluN2A and GluN2B agonist binding domains dysregulated NMDAR activity. When introduced into recombinant human NMDARs, these rare variants identified in individuals with neurologic disease had complex, and sometimes opposing, consequences on agonist binding, channel gating, receptor biogenesis, and forward trafficking. Our approach combined quantitative assessments of these effects to estimate the overall impact on synaptic and non-synaptic NMDAR function. Interestingly, similar neurologic diseases were associated with both gain- and loss-of-function variants in the same gene. Most rare variants in GluN2A were associated with epilepsy, whereas GluN2B variants were associated with intellectual disability with or without seizures. Finally, discerning the mechanisms underlying NMDAR dysregulation by these rare variants allowed investigations of pharmacologic strategies to correct NMDAR function.

In recent years, significant advances in human genome sequencing and computational analysis of sequencing data have yielded a discovery boom for genetic variants that cause epilepsy (1–3). The identification of these causative genes holds great promise for improving our understanding of epilepsy and how to treat it (4), but there are currently significant barriers between the identification of variants and the application of this knowledge, including creating suitable animal models and model systems in which to interrogate the altered functions of these genes, their protein products, and the networks in which they operate.

From a genetic standpoint, this discovery boom has also taught us that there are many ways to make a brain epileptic. Although the standards for calling a gene or a particular variant causative can vary, there are well over 100 genes that have been called such, and within each gene, there can be hundreds of genetic variants that are causative, each of which can affect the expression of the gene or function of the protein in unique ways. This staggering number of potential variants raises a fundamental question when it comes to both mechanistic understanding and treatment design: Do we need to consider each gene variant as a separate case, or do their effects converge at common pathological mechanisms? If the former is true, then there could be a long road ahead, especially if we need to create animal models for each variant.

In a recent article, Swanger et al. approached these issues by focusing on genetic variants found in GRIN2A and GRIN2B, both of which are members of the GRIN family of ligand-gated ion channels that code for the NMDA receptor subunits Glu-N2A and GluN2B, respectively. These variants were missense mutations, meaning that a single nucleotide was changed that switched one amino acid in the protein, and were considered to be pathogenic. The majority of the variants were associated with epilepsy, especially the GRIN2A variants, but some of the GRIN2B variants were associated with intellectual disability.

NMDA receptors are well-characterized proteins (5). Extensive structure-function experiments, in combination with crystal structures of the ligand-binding domains and well-defined electrophysiological assays, have led to a good understanding of how different parts of these proteins bind glutamate, flux ions, and assemble into functional tetramers (6). Using this available breadth of knowledge, the authors first looked at which functional domains of the proteins were most vulnerable to missense mutations. Although missense mutations can be disease causing, another insight from human genetics is that they, and other genetic variants, are also common in healthy people. One way to determine whether a particular variant is pathogenic is to compare the frequency of the variant in healthy people to those with neurologic disease (7). Amino acids or regions of a protein that show higher variability in control populations are less likely to harbor disease-causing variants, whereas those with lower variability are more likely (8). Comparing the distribution of missense variants in GRIN2A and GRIN2B in the healthy control group versus that in the neurologic disease group showed that the regions of the proteins that bind glutamate, the transmembrane domains, and the links between the two had very few missense mutations in the control group but were overrepresented in the disease group, suggesting that changes to amino acids in these regions are particularly likely to cause functional, pathogenic changes.

Next, the authors cloned human GRIN2A and GRIN2B genes harboring 25 disease-associated variants within these vulnerable regions and expressed them in oocytes or HEK cells to assess their functional impact. They put them through a battery of tests designed to measure fundamental aspects of ligand-gated ion channel biology, such as the EC₅₀ for glutamate and glycine, the time course of the measured current, the probability of channel opening, and the number of receptors that make it to the cell surface.

Many of the variants were in the glutamate-binding region of the protein, so the authors first measured whether the potency of glutamate was altered. Of the 22 variants that had a measurable response to glutamate, 20 caused a significant shift in the glutamate EC₅₀; 11 variants had a higher EC₅₀ (up to 1000-fold), and 9 had a lower EC₅₀ (up to 4.5-fold). In agreement with the changes in EC₅₀ values, the rate of decay of the receptor currents in response to brief applications of glutamate (deactivation) were also affected, as these two measurements are highly correlated. Surface expression was also reduced by all the variants that reduced glutamate potency, and even by some of the variants that increased potency.

Because the effects of the variants on individual receptor properties were variable and often opposing, the authors next attempted to quantify the magnitude of the overall impact on receptor function by estimating the degree of the increase or decrease in charge transfer through the receptors. According to this estimation, about a quarter of the variants enhanced receptor function (up to 3.7-fold), whereas the majority decreased receptor function (down to 5.1 × 10⁻⁵-fold). Gain-of-function and loss-of-function variants were found in both GRIN2A and GRIN2B, further demonstrating that the variants resist simple classification schemes. Despite this complexity, the authors did show that the gain-of-function variants were sensitive to the NMDA receptor antagonist memantine, whereas the loss-of-function variants showed enhanced charge transfer by three different positive allosteric modulators, suggesting that these or similar drugs may be beneficial to patients with GRIN variants.

By performing a detailed and thorough analysis of a relatively large number of variants in two NMDA receptor-encoding genes, this work advances our knowledge of the nature and functional consequences of GRIN variants and gives a glimpse of what attempts at personalized therapies based on identified genetic variants might look like. Based on this work, one could imagine that an epilepsy patient with an identified NMDA receptor mutation could have this mutant receptor cloned and expressed, parameters measured, and then screened against a library of compounds to identify those that normalize the overall impact on receptor function and treat the neurologic symptoms. However, for many other epilepsy-causing mutations in less well-characterized genes, this sort of personalized therapy procedure may not yet be feasible. At the same time, this work also serves as a warning that the task of finding common mechanisms may be daunting. Even though these variants were associated with similar neurologic symptoms and caused single amino-acid changes in the same domain of a protein, their functional effects were very different. This suggests that higher-level cellular, circuit, and developmental mechanisms, such as differential expression in excitatory versus inhibitory neurons, are crucial to dictating the disease phenotype, and that each variant may lead the brain down a unique path to epilepsy. To better understand these complex mechanisms, we will need additional animal models and complementary model systems, potentially one for each variant, to effectively design precision therapies and achieve a more complete understanding of the underlying mechanisms.