GluN1 splice variant control of GluN1/GluN2D NMDA receptors

Abstract

NMDA receptors are ionotropic glutamate receptors that mediate a slow, Ca2+-permeable component of excitatory synaptic transmission in the central nervous system. Recombinant GluN1-1a/GluN2D receptors are characterized by low channel open probability and prolonged deactivation time course following the removal of agonist. Here, we show that the deactivation time course, agonist potency, and single channel properties of GluN2D-containing NMDA receptors are modulated by alternative RNA splicing of GluN1. Our results demonstrate that GluN1 exon 5, which encodes a 21-amino-acid insert in the amino-terminal domain, is a key determinant of GluN1/GluN2D receptor function. GluN1-1b/GluN2D receptors, which contain the residues encoded by exon 5, deactivate with a dual exponential time course described by a τFAST of 410 ms and a τSLOW of 1100 ms. This time course is 3-fold more rapid than that for exon 5-lacking GluN1-1a/GluN2D, which deactivates with a τFAST of 1100 ms and a τSLOW of 3400 ms. Exon 5-containing NMDA receptors also have a two-fold higher open probability (0.037) than exon 5-lacking receptors (0.017). Furthermore, inclusion of exon 5-encoded residues within the GluN1-1b subunit decreases the potency for the endogenous agonist l-glutamate. Evaluation of receptor kinetics for NMDA receptors containing mutated GluN1-1b subunits and wild-type GluN2D identified residue Lys211 in GluN1-1b as a key determinant of exon 5 control of the deactivation time course and glutamate potency. Evaluation of a kinetic model of GluN1/GluN2D gating suggests that residues encoded by exon 5 influence several rate-limiting steps. These data demonstrate that the GluN1 subunit is a key determinant of the kinetic and pharmacological properties of GluN2D-containing NMDA receptors.

Figure 1. Splice variants that include GluN1 exon 5 decrease the potency ofl-glutamate

A, a model of GluN1/GluN2D based on the GluA2 crystal structure (Sobolevsky et al. 2009) is shown (Acker et al. 2011). The GluN1 subunits are in yellow, and the GluN2D subunits are in purple. The region within the GluN1-1b amino-terminal domain where exon 5-encoded residues would be present is shown as dark grey. The intracellular carboxyl-terminal domain is omitted from the model. B, a linear representation of the GluN1 polypeptide chain is shown for 8 alternative splice variants. The GluN1 subunits are composed of the amino-terminal domain (ATD), S1 and S2 domains which form the ligand-binding domain, three transmembrane helices (M1, M3, and M4) and one reentrant loop (M2) that comprise the ion channel pore, and an intracellular carboxyl-terminal domain (CTD). Exon 5 (red) encodes a 21-amino-acid region within the ATD. Exon 21 (grey) encodes a 37-amino-acid segment of the carboxyl-terminal tail, while exon 22 encodes a 38-amino-acid segment of the CTD (dark grey). Deletion of exon 22 creates a shift in the open reading frame that results in the production of an alternate exon 22′ (black), encoding a 22-amino-acid region of the CTD. C, two-electrode voltage-clamp recordings from Xenopus oocytes were used to determine the EC₅₀ values (see Table 1) of glycine for each GluN1 splice variant in the presence of 100 μm l-glutamate (pH 7.4, V_hold = −40 mV). D, the EC₅₀ value for l-glutamate was determined for each GluN1 splice variant in the presence of 30 μm glycine (pH 7.4). GluN1 splice variants that contained exon 5 showed significantly higher (i.e. less potent) glutamate EC₅₀ (P < 0.05, one-way ANOVA with Tukey's post hoc test, Table 1).

Figure 2. Residues encoded by GluN1 exon 5 accelerate the deactivation time course of GluN1/GluN2D NMDA receptors

A and B, representative voltage-clamp recordings from HEK 293 cells expressing GluN1/GluN2D show the mean deactivation time courses following rapid removal of l-glutamate (1 mm, 2 s application, grey bar); 0.05 mm glycine was present in all solutions. GluN1-1a/GluN2D receptors deactivate slowly upon removal of glutamate with a dual exponential time course. The average fitted τ_FAST was 1100 ± 200 ms and τ_SLOW was 3400 ± 370 ms (n = 5). By contrast, GluN1-1b/GluN2D receptors deactivate more rapidly, with τ_FAST of 410 ± 26 ms and τ_SLOW of 1100 ± 74 ms (n = 10). C and D, the fitted individual exponential components (dashed lines) are superimposed on the expanded deactivation time course for GluN1-1a (C) and GluN1-1b (D).

Figure 3. Lys211 of the GluN1 subunit ATD influences GluN2D L-glutamate potency and deactivation time course

A, a sequence alignment of GluN1-1a and GluN1-1b shows the highly charged 21-amino-acid insert encoded by exon 5 in the GluN1 ATD. Lys211 is in bold font. B, GluN1-1b mutants in which one or more of the charged residues encoded by exon 5 has been mutated were evaluated for changes in l-glutamate EC₅₀ using two-electrode voltage-clamp recordings of Xenopus oocytes. C–E, whole cell voltage-clamp recordings of transfected HEK 293 cells were conducted by rapidly applying 1 mm l-glutamate to cells for 2 s; 0.05 mm glycine was present in all solutions. The average responses from representative cells show that GluN1-1b/GluN2D has a rapid deactivation (C), whereas GluN1-1b (K211A) (D) or GluN1-1b (K211L) (E) expressed with GluN2D have slower deactivation time courses that are similar to that of GluN1-1a/GluN2D. A dual-exponential fit of a representative wild type GluN1-1a/GluN2D recording is superimposed in grey in C, D and E.

Figure 4. GluN1/GluN2D single channel properties

A, a voltage-clamp recording of an excised outside-out patch from a transfected HEK 293 cell is shown with 1 active GluN1-1a/GluN2D channel in response to 1 mm l-glutamate and 0.05 mm glycine (V_hold–80 mV). The boxed region is shown below on an expanded time scale. B, open duration histogram for this representative patch can be fitted by two exponential components, whereas the shut time histogram (C) was fitted with seven exponential components. D and E, GluN1-1a channels opened to two conductance levels that had asymmetrical transitions between sublevels. We detected 2035 direct transitions between the two conductance states, with 1310 transitions from the higher to lower conductance level (66 ± 2.0%) and 725 transitions from the lower to higher conductance level (34 ± 2.0%). F, a representative recording of one active GluN1-1b/GluN2D channel in an outside-out patch from transfected HEK 293 cells in response to 1 mm l-glutamate and 0.05 mm glycine (V_hold−80 mV). G, two exponential components were required to describe the open duration histogram for GluN1-1b/GluN2D channels. H, the shut duration histogram was fitted with six components. I and J, GluN1-1b/GluN2D channels have two conductance levels, with direct transitions between these levels being asymmetric. In a total of 1733 direct transitions between the two conductance states, 1066 transitions were from the high conductance level to the subconductance level (63 ± 2.2%), while 667 transitions (37 ± 2.2%) were observed for transitions from the lower to higher conductance level. For panels B, C, G and H, the y-axes of the histograms are plotted on a square root scale.

Figure 5. Kinetic model of GluN1/GluN2D function

A, Scheme 1 contains 2 similar arms, a fast-gating arm (upper arm) and a slow-gating arm (lower arm). Each arm is composed of 3 shut states and 2 open states. B, the probability density function predicted from maximum interval likelihood (MIL) fitting of Scheme 1 with binding rates b+ and b– held constant to data from one representative GluN1-1a/GluN2D outside-out single channel recording is superimposed on the shut duration histogram (left panel; Table 5). The predicted responses are shown following simultaneous least squares fitting of Scheme 1 to macroscopic recordings of GluN1-1a/GluN2D activated by 1 mm l-glutamate for 10 s (HL), 1 mm l-glutamate for 10 ms (HS), and 0.0025 mm l-glutamate for 10 s (LL) with gating rates held constant (right panel; Table 5; see Methods). Macroscopic response waveforms predicted from the fitted rate constants are superimposed as light grey lines. C, single channel recordings were analysed to determine the mean open time when an opening precedes a specified shut range. Time constants describing the closed duration histogram for individual patches were used to determine critical closed times, as described in Jackson et al. (1983). The average critical closed times from 10 patches containing one active GluN1-1a/GluN2D receptor were 0.10, 0.74, 43, 190 and 1800 ms; for some patches the critical closed times related to τ₂ in Table 4 were not well determined, and thus the shut range for this component was combined with the next component. Conditional mean open durations for 10 single channel recordings in 1 mm l-glutamate and 0.05 mm glycine at pH 8.0 were determined for openings preceding (▪) the shut duration within a specified range. The conditional mean open times are plotted against the centre of the shut ranges, which were 0.031–0.10, 0.10–0.74, 0.74–43, 43–190 and 190–1800 ms. The dashed line is the mean open time from 10 GluN1-1a/GluN2D patches. The open circles (○) are the values predicted from Scheme 1. *Significantly different from the two briefest shut ranges (P < 0.05; ANOVA, Tukey's post hoc test).

Figure 6. Exon 5 preferentially increases occupancy of the open state in the fast-gating arm

A, the block representation of Scheme 1 illustrates the fast- and slow-gating arms of the model, as well as the binding, gating, and open states. B, the occupancy of each state is given at steady state in response to a maximally active concentration of glutamate; occupancy is given to two significant figures in increments of 0.0001. The open probabilities of the fast- and slow-gating arms are calculated. Occupancy is colour-coded as indicated on the right to visually illustrate how exon 5 is predicted by Scheme 1 to alter receptor function.

Figure 7. Effects of exon 5 on NMDA receptor kinetics

A, the MIL fit of a representative GluN1-1b/GluN2D excised outside-out single channel recording with Scheme 1 is given (left panel; Table 5); b+ and b− were fixed. The result of least squares fitting of Scheme 1 (right panel; Table 5) to macroscopic recordings of GluN1-1b/GluN2D activated by 1 mm l-glutamate for 10 s (HL), 1 mm l-glutamate for 10 ms (HS), and 0.0025 mm l-glutamate for 10 s (LL) is given. Gating rates were held constant during fitting. Waveforms predicted from the fitted rate constants are superimposed as white lines. B, rates are colour-coded based on the change between fitted GluN1-1b compared to GluN1-1a. Dark red indicates a 4-fold increase, light red indicates 1.5-fold increase, grey indicates no change, light blue indicates a 1.5-fold decrease, medium blue indicates a 2-5-fold decrease, and dark blue indicates 8-fold decrease for exon 5.