Different Forms of Disorder in NMDA-Sensitive Glutamate Receptor Cytoplasmic Domains Are Associated with Differences in Condensate Formation

Abstract

The N-methyl-D-aspartate (NMDA)-sensitive glutamate receptor (NMDAR) helps assemble downstream signaling pathways through protein interactions within the postsynaptic density (PSD), which are mediated by its intracellular C-terminal domain (CTD). The most abundant NMDAR subunits in the brain are GluN2A and GluN2B, which are associated with a developmental switch in NMDAR composition. Previously, we used single molecule fluorescence resonance energy transfer (smFRET) to show that the GluN2B CTD contained an intrinsically disordered region with slow, hop-like conformational dynamics. The CTD from GluN2B also undergoes liquid-liquid phase separation (LLPS) with synaptic proteins. Here, we extend these observations to the GluN2A CTD. Sequence analysis showed that both subunits contain a form of intrinsic disorder classified as weak polyampholytes. However, only GluN2B contained matched patterning of arginine and aromatic residues, which are linked to LLPS. To examine the conformational distribution, we used discrete molecular dynamics (DMD), which revealed that GluN2A favors extended disordered states containing secondary structures while GluN2B favors disordered globular states. In contrast to GluN2B, smFRET measurements found that GluN2A lacked slow conformational dynamics. Thus, simulation and experiments found differences in the form of disorder. To understand how this affects protein interactions, we compared the ability of these two NMDAR isoforms to undergo LLPS. We found that GluN2B readily formed condensates with PSD-95 and SynGAP, while GluN2A failed to support LLPS and instead showed a propensity for colloidal aggregation. That GluN2A fails to support this same condensate formation suggests a developmental switch in LLPS propensity.

Keywords: discrete molecular dynamics; glutamate receptor; intrinsically disordered protein; liquid-liquid phase separation; single molecule fluorescence.

Figure 1

Prediction and classification of intrinsic disorder in the cytoplasmic domains of the GluN2A and GluN2B. (A) Cartoon schematic of domain organization in the intracellular C-terminal domain (CTD) of the GluN2A and GluN2B subunits of the NMDA receptor. The CTD is connected to the M4 helix within the transmembrane domain [3]. The essential palmitoylation sites (yellow) mediate attachment to the membrane [17]. The subdomains demarked by palmitoylation are termed CTD1 (purple) and CTD2 (brown). (B) The disorder propensity from PONDR is plotted for the CTDs of GluN2A and GluN2B with regions predicted to be order-prone (PONDR scores < 0.5) highlighted in red. (C) Classification of Intrinsically Disordered Ensemble Regions (CIDER) analysis [19] of the CTD subdomains. Colored regions indicate conformational classes of IDPs showing the boundaries for positive polyelectrolytes (red), negative polyelectrolytes (blue), strong polyampholytes (dark green) intermediate polyampholytes (mint green) and weak polyampholytes (pea green, lower left) [20]. Circles representing the CTD subdomains are placed based on their classification by CIDER analysis. The CTD1 of GluN2A (yellow circle) and CTD2 of GluN2A (magenta circle) are classified as intermediate polyampholytes. The CTD1 of GluN2B (purple circle) is also classified as an intermediate polyampholyte. However, CTD2 of GluN2B (pink circle) is classified as a weak polyampholyte. (D) The separation between residues within CTD2A and CTD2B are represented by a boxplot with the Gaussian distribution of its recurrence. Shown are the distributions for arginine (R), tyrosine (Y), phenylalanine (F), total aromatics (F + Y), lysine (K), aspartate (D), glutamate (E) and histidine (H). The mean frequency, of all the above-mentioned residues within each isoform, is highlighted with a red line. The standard deviation for the boxplot indicated by black bars.

Figure 2

Discrete molecular dynamics simulation of the CTD2 subdomains from GluN2A and GluN2B. (A) Distribution of the radius of gyration derived from simulations for GluN2A and (B) GluN2B. GluN2A favors more extended states. (C) The calculated per residue probabilities of different secondary structures based on occupancy observed during simulations for GluN2A and (D) GluN2B. Shown are the probability of an individual residue adopting α-helical (light blue), β-sheet (red), random coil (black), and turn (dark blue) conformations. Random coil was the dominant secondary structure for both CTD2s, although GluN2A showed more stable secondary structural elements. (E) The pairwise residue-contact frequency maps show the intramolecular interactions observed in simulations of the CTD2 from GluN2A and (F) GluN2B. The associated color scale gives the probability of contact between two residues. GluN2A showed stable short-range interactions involved in stabilizing the local, ordered secondary structures. GluN2B showed more long-range contacts.

Figure 3

Single molecule FRET measurements of the CTD2 subdomains from GluN2A and GluN2B. Representative single molecule intensity time traces for the CTD2 subdomains. (A) Representative GluN2A molecules in low and high FRET states. Emission of donor (orange) and acceptor (blue) fluorophores show stable intensity in GluN2A but vary between molecules within the population. (B) Representative GluN2B molecules showing slow timescale, anticorrelated changes in intensity, which is the predominant state as observed previously [15,18,40]. (C) Population histogram of raw FRET efficiency (proximity ratio) accumulated from each frame captured before photobleaching for the CTD2 subdomain of GluN2A and (D) GluN2B. Shown are the experimental data (red circles) along with the global fit (black line). The number of individual states from global fitting (grey lines) differed. GluN2A adopted three states while GluN2B was well fit with a two state model containing wider peaks (Table 2). The number of molecules analyzed is indicated in each panel.

Figure 4

Condensate formation by the CTD2 subdomains from GluN2A and GluN2B. (A) Measurement of turbidity at 550 nm for the binary and ternary protein mixtures indicated beneath the panel. Samples contained 20 µM of each protein including full-length PSD-95, the CC-PBM fusion from synGAP, and the CTD2A domain from GluN2A or the CTD2B domain from GluN2B. CTD2B shows maximal turbidity while the same concentration of CTD2A remains clear. (B) Analysis of protein composition in the condensed phase isolated by sedimentation. Samples were resolved using SDS-PAGE. Left, the individual proteins were run separately followed by the low molecular weight markers (LMW). Right, soluble (S) and pellet (P) fractions from sedimentation of ternary mixtures containing a 1:1:1 ratio of PSD-95, CC-PBM, and CTD2 at 20 µM for CTD2A (left) and CTD2B (right). The molecular weights are indicated to the left of the gel (in kDa). The identity of each protein band is indicated to the right of the gel (C,D). Representative images from differential interference contrast (DIC) microscopy of the same ternary protein mixtures used for sedimentation analysis. (C) CTD2A does not form droplets, although some scattering is observed at high contrast. (D) CTD2B forms droplets with a range of different sizes. The scale bars are 100 µm.