Chemical shift assignments of calmodulin bound to a cytosolic domain of GluN2A (residues 1004–1024) from the NMDA receptor

Abstract

N-methyl-D-aspartate receptors (NMDARs) consist of glycine-binding GluN1 and glutamate-binding GluN2 subunits that form tetrameric ion channels. NMDARs in the neuronal post-synaptic membrane are important for controlling neuroplasticity and synaptic transmission in the brain. Calmodulin (CaM) binds to the cytosolic C0 domains of both GluN1 (residues 841–865) and GluN2 (residues 1004–1024) that may play a role in the Ca²⁺-dependent desensitization of NMDAR channels. Mutations that disrupt Ca²⁺-dependent desensitization of NMDARs are linked to Alzheimer’s disease, depression, stroke, epilepsy, and schizophrenia. NMR chemical shift assignments are reported here for Ca²⁺-saturated CaM bound to the GluN2A C0 domain of NMDAR (BMRB no. 51821).

Biological context

NMDA receptors at the post-synaptic membrane in the brain bind to the neurotransmitter, glutamate (and/or co-agonist, glycine) and serve as ligand-gated Na⁺/Ca²⁺ channels (Traynelis et al. 2010). NMDARs are tetrameric channels comprised of two protein subunits (GluN1a-b and GluN2A-D) that assemble as a 2:2 complex, (GluN1)₂:(GluN2)₂. NMDAR channels open upon GluN2 binding to glutamate and/or GluN1 binding to glycine (Benveniste and Mayer 1991; Clements and Westbrook 1991). The ligand-gated opening of NMDAR channels leads to a neuronal Ca²⁺ influx, which causes the intracellular Ca²⁺ concentration to increase into the micromolar range (Wadel et al. 2007) and promote many different Ca²⁺-dependent processes (Kunz et al. 2013; Puri 2020). Prolonged opening of NMDAR channels causes an elevated intracellular Ca²⁺ concentration that can be cytotoxic (Peng et al. 1998). High cytosolic Ca²⁺ levels desensitize the opening of NMDAR channels by a process known as Ca²⁺-dependent inactivation (CDI) (Iacobucci and Popescu 2017, 2019, 2020). The Ca²⁺-induced desensitization of NMDAR channels requires CaM binding to the cytosolic C0 domains in both GluN1 (Iacobucci and Popescu 2017, 2019, 2020; Zhang and Majerus 1998) and GluN2A (Bajaj et al. 2014). CaM binding to NMDARs is hypothesized to promote conformational changes in the cytosolic C0 domains that lead to channel desensitization (Iacobucci and Popescu 2020; Krupp et al. 1999; Wang et al. 2008).

Atomic resolution structures of NMDARs (Chou et al. 2020; Jalali-Yazdi et al. 2018; Karakas Furukawa 2014; Lee et al. 2014; Regan et al. 2018) reveal subunit interactions between the extracellular ligand-binding domain and transmembrane channel domain. The structure of the NMDAR C-terminal cytosolic domain (that mediates CDI) is not visible in any of the published structures and is believed to be dynamically disordered or otherwise undetected. The cytosolic region of GluN2A (residues 840–1464) contains a helical C0 domain (residues 1004–1024) that binds to CaM (Bajaj et al. 2014). We recently reported NMR assignments of CaM bound to the GluN1 C0 domain (Bej and Ames 2023). We report here NMR chemical shift assignments of Ca²⁺-saturated CaM bound to the C0 domain of GluN2A. These assignments will be important for elucidating the structure of CaM bound to GluN2A within the context of the full-length channel, which may provide insights into the mechanism of CDI.

Methods and experiments

Sample preparation of CaM and GluN2A C0

Recombinant human CaM was expressed and purified as described previously (Bej and Ames 2022a). A peptide fragment of the GluN2A C0 domain of the NMDA receptor (residues 1004–1024) was purchased from GenScript. A 2.5-fold excess of the peptide was added to CaM and the complex was concentrated to 0.4 mM.

NMR spectroscopy

NMR samples were prepared by dissolving CaM/GluN2A C0 complex in 20 mM Tris-d₁₁ (pH 7.0) and 1 mM CaCl₂ containing 8% or 100% (v/v) D₂O in which 0.3 mL was transferred to a Shigemi NMR tube (Shigemi Inc.). NMR experiments on isotopically labeled CaM bound to the unlabeled GluN2A C0 peptide were performed at 308 K on a Bruker Avance III 800 MHz spectrometer equipped with a four-channel interface and triple resonance cryogenic (TCI) probe. A series of NMR experiments (HNCACB, HN(CO) CACB, HNCO, HBHA(CO)NH, HBHANH, C(CO)NH, and H(CCO)NH) were recorded and analyzed to assign the backbone resonances. HCCCONH-TOCSY, ¹³C-edited NOESY-HSQC, HBCBCGCDHD, and HBCBCGCDCEHE were recorded and analyzed to assign the side chain resonances. NMRPipe (Delaglio et al. 1995) was used to process the NMR spectra and Sparky (Lee et al. 2015) was used to analyze NMR spectra and catalog the chemical shift assignments.

Extent of assignments and data deposition

Two-dimensional ¹⁵N-¹H HSQC NMR spectra of ¹⁵N-labeled CaM bound to unlabeled GluN2A C0 peptide illustrate representative backbone resonance assignments (Fig. 1A–B). Side chain aliphatic resonance assignments are illustrated in the constant-time ¹³C-¹H HSQC spectrum (Fig. 1C). The resonance assignments were determined by analyzing 3D triple resonance NMR spectra of ¹³C/¹⁵N-labeled CaM bound to unlabeled GluN2A C0 peptide. The stereo-specific assignment of side chain methyl resonances from Val and Leu was obtained as described by (Neri et al. 1989). The resolved NMR peaks with uniform intensities indicate the CaM/GluN2A C0 complex forms a stable and folded structure. The CaM amide resonances assigned to G26, G62, G99 and G135 exhibited downfield-shifted chemical shifts that indicate Ca²⁺ is bound to the four EF-hands (Fig. 1A). Amide resonances assigned to I28, I101 and N138 exhibited noteworthy downfield shifts, perhaps because these residues are located in β-sheet regions that are stabilized by strong main chain hydrogen bonds to these amide groups. Spectral assignments were obtained for more than 93% of the main chain ¹³C resonances (¹³Cα, ¹³C β, and ¹³CO), 96% of non-proline backbone amide resonances (¹HN, ¹⁵N), and 89% of side chain resonances (Fig. 1C). The unassigned residues (A2, D3, E121, V122, and D134) had weak HSQC peaks that prevented their assignment. The chemical shift assignments (¹H, ¹⁵N, ¹³C) for CaM bound to GluN2A C0 peptide have been deposited in the BioMagResBank (http://www.bmrb.wisc.edu) under accession number 51821.

Fig. 1.

NMR spectra of isotopically labeled CaM bound to unlabeled GluN2A C0 peptide. A ¹⁵N-¹H HSQC spectrum of ¹⁵N-labeled CaM bound to unlabeled GluN2A C0 illustrates backbone resonance assignments indicated by the labeled peaks. The black-dashed box encompasses the central crowded region. Side chain amide resonances were not assigned and are not labeled (upper right). B Expanded view of assigned resonances in the spectrally crowded region. C Constant-time ¹³C-¹H HSQC spectrum of ¹³C-labeled CaM bound to unlabeled GluN2A C0 peptide illustrates side chain methyl assignments. The blue peaks represent methionine methyl resonances that have negative intensity

Chemical shift index (Wishart et al. 1992) and secondary structure prediction using TALOS + (Shen et al. 2009) were used to calculate the secondary structure of CaM bound to the GluN2A C0 peptide (Fig. 2). The residues located in regions of regular secondary structure (cyan and magenta in Fig. 2A) exhibit RCI order parameter values greater than 0.9 (Fig. 2B), suggesting these residues have a rigid and stable backbone conformation. The NMR-derived secondary structure of CaM bound to GluN2A C0 peptide is similar to that reported for free CaM (Bej and Ames 2022b), and is shown as cylinders and triangles in Fig. 2A. The CaM secondary structure is composed of eight α-helices and four β-strands that form the four EF-hands (EF1: residues 7–39, EF2: residues 45–76, EF3: residues 83–112 and EF4: residues 119–144) as seen in the CaM crystal structure (Babu et al. 1988). EF1 and EF2 interact to form the CaM N-lobe, while EF3 and EF4 are connected to form the C-lobe shown schematically in Fig. 2A.

Fig. 2.

Secondary structure and RCI order parameters (S²) of CaM bound to GluN2A C0 peptide calculated on the basis of the assigned backbone chemical shifts. A TALOS + probability of each residue forming an α-helix or β-strand (cyan for helix and magenta for strand) and B RCI S² order parameter for each residue calculated using TALOS + server (Shen et al. 2009). The schematic diagram of secondary structure (cylinder for helix and triangle for strand) was determined from the crystal structure of CaM structure (PDB ID: 2VAY (Halling et al. 2009))

The binding of the GluN2A C0 peptide to CaM causes chemical shift perturbations (CSPs) for hydrophobic residues in the CaM C-lobe (I86, A89, F93, M125, A129, F142, M145) (Fig. 3A). The large CSP values suggest that the GluN2A C0 peptide contacts the CaM C-lobe like that seen in the NMR structure of CaM bound to the C-terminal helix (called CaM2) in the retinal cyclic nucleotide gated channel β-subunit (CNGB1) (Bej and Ames 2022c). Surprisingly, the GluN2A C0 peptide has very little effect on the chemical shifts assigned to the CaM N-lobe, and the CSP values of residues in the CaM N-lobe are below experimental error (Fig. 3A–B). The small CSP values suggest that the CaM N-lobe does not interact with the GluN2A C0 peptide, in contrast to the CaM N-lobe interaction with the GluN1 C0 peptide (Bej and Ames 2023). The amino acid sequence of GluN2A C0 (Fig. 3C, left panel) reveals that the peptide likely forms an amphipathic helix (Fig. 3C, right panel) in which hydrophobic residues (W1010, V1014, I1017 highlighted red in Fig. 3C) on one side of the helix may contact the CaM C-lobe, while charged or polar residues (R1007, K1011, D1015 shown blue in Fig. 3C) on the opposite side of the helix might prevent its binding to the hydrophobic target binding site in the CaM N-lobe. Similar amphipathic target helices (CaM1, CaM2 and creatine kinase or CK in Fig. 3C) were seen in the structures of CaM bound to helical sites in CNGB1 (CaM1 and CaM2) (Bej and Ames 2022c) and CK (Sprenger et al. 2021). Therefore, we hypothesize that the CaM C-lobe binds to the hydrophobic side of the amphipathic helix in GluN2A (highlighted red in Fig. 3C), while the N-lobe remains unbound to the opposite polar surface of the amphipathic helix (highlighted blue in Fig. 3C).

Fig. 3.

Chemical shift perturbation (CSP) for CaM caused by binding of the GluN2A C0 peptide. A Backbone amide CSP was calculated as: $CSP=\sqrt{{\left(\Delta H^N\right)}^2+{(0.14\times \Delta N)}^2}$ (Williamson 2013). ΔH^N and ΔN are the difference in ¹H^N and ¹⁵N chemical shifts, respectively between bound and free CaM (Bej & Ames 2022b). B Side-chain methyl CSP was calculated as: $CSP=\sqrt{{(\Delta H)}^2+{(0.14\times \Delta C)}^2}$. ΔH and ΔC are the difference in the ¹H and ¹³C methyl chemical shifts, respectively between bound and free CaM (Bej & Ames 2022b). Color coded CSP values are mapped on the CaM crystal structure (PDB ID: 2VAY). Residues with largest CSP values are highlighted by spheres and marked with residue labels. Residues that lack side chain methyl groups or are otherwise unassigned are colored gray in panel B. C Amino acid sequence of GluN2 C0 aligned with other amphipathic helical CaM targets (CaM1, CaM2 and CK) is shown in the left panel. The helical structure of GluN2A C0 is shown in the right panel. Hydrophobic amino acids (W1010, V1014, I1017) on the same side of the peptide helix are highlighted red. Polar and charged amino acids (R1007, K1011, D1015) on the opposite side of the amphipathic helix are highlighted blue. CaM1 and CaM2 are amphipathic helical CaM binding sites found in CNGB1 (Bej & Ames 2022c) and CK is an amphipathic helix from creatine kinase (Sprenger et al. 2021)

Future studies are needed to determine the NMR structure of CaM bound to the GluN2A C0 peptide to test our hypothesis and confirm whether the CaM C-lobe is bound to the hydrophobic residues in the amphipathic target helix as seen for CaM bound to CaM2 in CNGB1 (Bej and Ames 2022c). The NMR assignments of CaM bound to GluN2A C0 peptide reported here will be important for determining the structure of CaM bound to GluN2A in future studies.

Data availability

The assignments have been deposited to the BMRB under the accession code: 51821.