QUATERNARY STRUCTURE OF K_ATP CHANNEL SUR2A NUCLEOTIDE BINDING DOMAINS RESOLVED BY SYNCHROTRON RADIATION X-RAY SCATTERING

Abstract

Heterodimeric nucleotide binding domains NBD1/NBD2 distinguish the ATP-binding cassette protein SUR2A, a recognized regulatory subunit of cardiac ATP-sensitive K⁺ (K_ATP) channels. The tandem function of these core domains ensures metabolism-dependent gating of the Kir6.2 channel pore, yet their structural arrangement has not been resolved. Here, purified monodisperse and interference-free recombinant particles were subjected to synchrotron radiation small-angle X-ray scattering (SAXS) in solution. Intensity function analysis of SAXS profiles resolved NBD1 and NBD2 as octamers. Implemented by ab initio simulated annealing, shape determination prioritized an oblong envelope wrapping NBD1 and NBD2 with respective dimensions of 168 × 80 × 37 ų and 175 × 81 × 37 ų based on symmetry constraints, validated by atomic force microscopy. Docking crystal structure homology models against SAXS data reconstructed the NBD ensemble surrounding an inner cleft suitable for Kir6.2 insertion. Human heart disease-associated mutations introduced in silico verified the criticality of the mapped protein-protein interface. The resolved quaternary structure delineates thereby a macromolecular arrangement of K_ATP channel SUR2A regulatory domains.

1. Introduction

The ATP-binding cassette (ABC) transporters are evolutionarily conserved transmembrane proteins, with canonical family members characterized by ATP hydrolysis-driven substrate translocation that accounts for diverse biological processes ranging from nutrient import to toxin efflux (Dean, 2005; Linton, 2007; Linton and Higgins, 2007; Rees et al., 2009). Atypical ABC proteins have also been identified, and account for substrate translocation-independent functions, such as ion channel gating (Aittoniemi et al., 2009; Burke et al., 2008; Higgins and Linton, 2004; Moreau et al., 2008). A case in point is the sulfonylurea receptor 2A (SUR2A), a distinctive ABC protein that regulates the operation of the ATP-sensitive potassium (K_ATP) channels in cardiac myocytes (Alekseev et al., 2005; Ashcroft, 2006; Inagaki et al., 1996; Inagaki et al., 1995; Nichols, 2006; Wheeler et al., 2008).

Encoded by ABCC9, SUR2A belongs to the ABCC subfamily along the cystic fibrosis transmembrane conductance regulator (CFTR or ABCC7) and the multidrug resistance-related protein (MRP or ABCC1) (Biemans-Oldehinkel et al., 2006; Chutkow et al., 1996; Oram and Vaughan, 2006; Solbach et al., 2006; Yamada and Kurachi, 2005). Through physical association with the potassium channel Kir6.2 pore, SUR2A endows K_ATP channel complexes with a unique metabolic decoding capacity that assures linkage of the cellular energetic state with membrane excitability (Bryan et al., 2006; Dupuis et al., 2008; Lorenz and Terzic, 1999; Zingman et al., 2007). SUR2A harbors essential nucleotide binding domains - NBD1 and NBD2. Each NBD encompasses Walker A (GX₄GKS/T), Walker B (Φ₄DD/E; Φ represents hydrophobic residues) and linker (LSGGQ) signature motifs (Walker et al., 1982). Responsible for adenine nucleotide recognition and processing, the SUR2A NBD1/2 tandem is integral in transduction of metabolic signals to the K_ATP channel pore (Bienengraeber et al., 2000; Karger et al., 2008; Zingman et al., 2001; Zingman et al., 2002). Genetic mutations in SUR2A NBDs alter channel function, and human K_ATP channelopathies have been implicated in cardiac disease susceptibility underscoring the structural integrity of regulatory domains in optimal channel performance (Bienengraeber et al., 2004; Kane et al., 2005; Olson et al., 2007; Reyes et al., 2009; Sattiraju et al., 2008).

To date, over 50 crystal structures of isolated NBDs from both bacterial and eukaryotic ABC transporters have been resolved (Hollenstein et al., 2007; Linton and Higgins, 2007; Moussatova et al., 2008; Rees et al., 2009). Despite the wealth of information pertinent to NBD structures of canonical ABC proteins, little is known regarding atypical ABC counterparts, including the mammalian SUR2A NBDs. In part, the lack of information is due to challenges in protein expression and chaotic orientation in solution that impede crystal formation. Although a molecular model of SUR2A NBD dimers has provided an initial discrete domain map (Park et al., 2008), the actual shape underlying structural arrangement remains uncertain.

Small angle X-ray scattering (SAXS) offers an approach to delineate supramolecular conformations (Hura et al., 2009; Petoukhov and Svergun, 2007; Putnam et al., 2007). Here, SAXS, in tandem with ab initio and rigid body model reconstruction, was applied to decipher the molecular envelope of SUR2A nucleotide binding domains in solution. We report a quaternary structural portrait of NBD1/NBD2 that provides a blueprint of structural constraints within regulatory K_ATP channel domains.

2. Materials and methods

2.1. Purified SUR2A NBD1 and NBD2

Murine cDNA SUR2A (GenBank D86037; kindly provided by Dr. S. Seino) encoding NBD1 (D666-890) and NBD2 (G1301-K1546) were amplified by PCR, and incorporated into a modified pET-15b vector (Novagen) containing the N-terminal (His)₆-tag and TEV protease cleavage site. Plasmids were transformed in the E. coli Rosetta(DE3)pLysS strain (Novagen), and NBD1 as well as NBD2 proteins purified (Park et al., 2008). Pelleted cells were suspended in buffer A (50 mmol/L Tris-HCl, 50 mmol/L NaCl, 1 mmol/L TCEP, 0.5 mmol/L EDTA and 5% glycerol (pH 8.0)), lyzed with a high-pressure microfluidizer Emulsiflex C-5 (Avestin), incubated for 20 min after mixing with 1% Triton X-100, and centrifuged at 20,000g for 30 min. Pellets were resuspended in buffer A containing 1% Triton X-100, homogenized, and centrifuged at 9,000g for 20 min. These steps were repeated three times. Residual Triton X-100 was removed in buffer A followed by repeated centrifugation at 9,000g for 20 min. Washed pellets were resuspended in denaturation buffer (50 mmol/L Tris-HCl, 50 mmol/L NaCl, 5 mmol/L TCEP, 0.5 mmol/L EDTA, and 5% glycerol (pH 8.0)), and incubated with 4.5% N-lauroylsarcosine for 30 min. Solutions were passed through a 0.45 μm filter to exclude aggregates, and dialyzed at least three times against 2 L of 10 mmol/L Tris-HCl (pH 8.0), 0.05 mmol/L EDTA, 0.1 mmol/L TCEP and 0.03% N-lauroylsarcosine, used below the critical micelle concentration (Frankel et al., 1991). Refolded NBDs were subject to gel filtration using a Superdex 200 column (Amersham Biosciences) with 10 mmol/L Tris, 0.05 mmol/L EDTA, 0.1 mmol/L TCEP and 0.03% N-lauroylsarcosine (pH 8.0). The protein concentration was quantified using an extinction coefficient of 34,045 (mol/L)⁻¹cm⁻¹ and 15,845 (mol/L)⁻¹cm⁻¹ for NBD1 and NBD2 respectively, and the purity was determined by densitometry.

2.2. Circular Dichroism Spectroscopy

The far-UV and near-UV circular dichroism (CD) spectra of refolded NBDs were recorded within the 250 to 200 nm and 325 to 250 nm range, respectively, on a JASCO J715 CD spectropolarimeter (Park et al., 2008). The mean residue ellipticity ([θ] in deg cm² dmol⁻¹) was calculated as (mdeg x MRW)/(10lc), where mdeg is the observed ellipticity in millidegrees; MRW, the mean residue weight; l, the optical path in cm; and c, the concentration in mg/ml. Based on UV circular dichroism, the secondary structure of NBDs was analyzed using neural network algorithms (http://geneura.ugr.es/cgi-bin/somcd/index.cgi), while the characteristics defining tertiary structure were evaluated against random coil denatured protein profile (Unneberg et al., 2001).

2.3. ATPase Activity

The ATPase activity of purified NBDs was measured based on inorganic phosphate production detected by colorimetry (Park et al., 2008). Assays were performed in 10 mmol/L Tris, 0.2 mmol/L TCEP, 10 mmol/L MgCl₂, 0.03% N-lauroylsarcosine and 4 mmol/L ATP (pH 8) at 37°C. The produced inorganic phosphate absorption was measured at 850 nm.

2.4. Synchrotron Radiation Small Angle X-ray Scattering

Synchrotron radiation small angle X-ray scattering (SAXS) profiles were collected using the SIBYLS beamline 12.3.1 at the Advance Light Source (ALS) at Lawrence Berkeley National Laboratories (Hura et al., 2009). A wavelength of 1.0332 Å was applied with a sample to detector distance of 1.5 m, which resulted in the scattering vector, q, ranging from 0.01 to 0.32 Å⁻¹. The scattering intensity, I(q), was recorded as a function of the scattering vector, defined as q = 4πsinθ/λ, where 2θ is the angle between the incident and scattered radiation and λ is the wavelength of the incident X-ray beam. Measurements were carried at 10°C. Samples of NBD1 and NBD2 proteins were prepared in the concentration range of 1.5 to 11.4 mg/ml in 10 mmol/L Tris (pH 8), 1 mmol/L TCEP, 0.05 mmol/L EDTA, 0.03% N-lauroylsarcosine and 5% glycerol. Scattering data from buffer (background) versus protein (sample at different concentrations) were collected alternately in the short (6 s), long (60 s), and short (6 s) exposure sequence. Short exposure readouts were compared to ensure that no radiation-induced protein damage had occurred, and data were merged using the PRIMUS program (Svergun and Koch, 2002). In a limited set of experiments, SAXS measurements pertinent to initial reference samples were conducted under similar experimental conditions using the beamline DND-CAT (5IDD) at the Advanced Photon Source (APS), Argonne National Laboratory.

2.5. Data Evaluation and Analysis

The radius of gyration (R_G) was derived by the Guinier approximation using the low q-regions of scattering profiles, I(q) = I(0) exp (-q²R_G²/3) for qR_G < 1.3, which extrapolated scattering intensity to the zero angle, I(0). The radius of gyration (R_G) and the intensity measured at zero angle (I(0)), extracted from the Guinier plot, displayed limited NBD1 or NBD2 concentration-dependence (Supplementary Fig. 1A) indicating lack of interparticle interference and validating monodispersity, prerequisites for SAXS analysis (Petoukhov and Svergun, 2007; Putnam et al., 2007; Svergun and Koch, 2002). In response to escalating NBD1 or NBD2 concentrations, the forward scattering intensity was proven linear (Supplementary Fig. 1B), verifying absence of aggregation and maintenance of NBDs in a refolded state (Putnam et al., 2007; Putnam et al., 2001). The distance distribution function, P(r), representing the probability of finding a point within the observed particle at a distance, r, from a defined point of reference, was calculated using the indirect Fourier transform method implemented in the bioinformatics package GNOM (Svergun, 1992). This analysis provided values of R_G and I(0) based on the entire SAXS data set, and compared favorably to values derived from Guinier plots. To determine maximum protein dimensions (D_max), the P(r) function was computed while constraining the function to zero at r_max, where r_max varied from 130 to 220 Å in 5 Å increments and the r_max yielded a plausible P(r) defined D_max.

2.6. Atomic Force Microscopy

Atomic force microscopy applied in contact mode was performed using the Nanoscope IV PicoForce Multimode AFM equipped with an E-scanner and V shaped silicon nitride cantilevers with a 0.06 N/m spring constant (Veeco Instruments) (Park et al., 2008). NBDs (2–40 μg/ml) were placed on freshly cleaved mica discs, and dried with nitrogen gas. After 30 min, unbound proteins and buffers were washed from the mica surface with water, and NBD-absorbed mica dried again. Data were collected in both height and phase modes. Images (512×512 pixels) were analyzed using the Nanoscope Version 6.13 software.

2.7. Ab initio Reconstruction

The molecular outlines of NBD1 and NBD2 were constructed from experimental SAXS data using the ab initio stimulated annealing programs DAMMIN and GASBOR (Svergun, 1999; Svergun et al., 2001). By minimizing the discrepancy (χ²) between calculated and experimental SAXS data, these programs search a chain-compatible spatial distribution of beads or dummy residues centered on the C_α atoms of the protein:

$${\chi }^2=\frac{1}{N-1}\sum _i{\left[\frac{I{\left(q_i\right)}_{\exp .}-cI{\left(q_i\right)}_{\mathrm{calc}.}}{\sigma \left(q_i\right)}\right]}^2$$

where N is the number of experimental points; c, the scaling factor; I(q_i)_exp. and I(q_i)_calc., the experimental and calculated intensity, respectively; and σ(q_i), the experimental error at the momentum transfer q_i. In order to identify a most probable model, multiple independent runs were required to decrease the risk of over interpretation of underdetermined models (Putnam et al., 2007). Specifically, sixteen independent runs for each NBD were performed using a set of symmetries to generate a series of plausible reconstructions. Reflective of octameric permutations, P1, P2, P4, P8 and P222 were selected to optimize ab initio reconstruction. P1 reflects no symmetry, P2 corresponds to one 2-fold symmetry, P4 to one 4-fold symmetry, P8 to one 8-fold symmetry, and P222 to three perpendicular 2-folds symmetry. Under these symmetry conditions, the χ² values ranged from 0.703 to 0.747 for NBD1, and from 0.633 to 0.701 for NBD2. Using the program package DAMAVER/DAMFILT (Volkov and Svergun, 2003), reconstructed models were aligned, averaged, and scored with a normalized spatial discrepancy (NSD) parameter that characterizes conversions among ab initio structures. Models with a large NSD (> mean NSD + 2 χ variation of NSD) were discarded.

2.8. Model Building

As no crystallographic structure of SUR2A NBD1 or NBD2 currently exists, homology models of NBDs were built. To this end, NBDs sequences from SUR2A were aligned using ClustalX with hemolysin B (HylB) that shares 29% sequence identity. Three-dimensional models of homodimer NBD1, NBD2, or heterodimer NBD1/NBD2 were generated using the homology modeling program MODELLER 6 based on the 2.5 Å crystal structure of the HylB-NBD dimer template (PBD: 1XEF) (Park et al., 2008). Purified recombinant NBDs had 23 N-terminal pre-residues randomly oriented in solution, accounting for 20 conformations resolved by NMR (PDB: 2HFV). Distinct conformations were added to the N-terminus of the NBD homology models. Rigid body modeling of the NBD quaternary structure was performed in the q range of 0.01–0.22 Å⁻¹. The scattering amplitudes of homology models were calculated using the CRYSOL program, which predict a theoretical SAXS pattern with given atomic coordinates (Svergun et al., 1995). The quaternary structure was generated by GLOBSYMM, a brute-force modeling program (Petoukhov and Svergun, 2005), using exhaustive searches to screen for the best fit produced by rigid body movements and rotations of the crystallographic NBD dimer under P222 symmetry without steric clashes. Primary contacts between subunits within the resolved quaternary ensemble were delineated by randomly oriented N-terminal pre-residues, deleted from final models for clarity.

3. Results

3.1. Secondary and Tertiary Structures from Purified SUR2A NBDs

The nucleotide binding domains, NBD1 and NBD2, of the K_ATP channel ABCC9-encoded SUR2A were individually expressed in E. coli, refolded and purified by gel filtration chromatography. Recombinant proteins, with purity >92%, migrated at corresponding molecular weights on SDS-PAGE (Fig. 1A), and formed oligomers detected on size-exclusion chromatography (Fig. 1B). Purified NBDs displayed a distinctive secondary structure in far-UV circular dichroism spectroscopy reflecting peptide backbone transitions with profiles composed of 66% α-helix, 9% β-sheet, 7% turn and 18% random coil for NBD1, and 40% α-helix, 20% β-sheet, 11% turn and 29% random coil for NBD2 (Fig. 1C). In near-UV circular dichroism spectroscopy, where tertiary conformation is attributed to aromatic amino acids or disulfide bond interactions, purified NBD1 and NBD2 demonstrated a well-folded structure (Fig. 1D) with detectable ATPase activity (Fig. 1D inset), indicating intrinsic structural and functional integrity.

Figure 1.

Structural and functional integrity of refolded SUR2A NBD1/NBD2 proteins. Recombinant nucleotide binding domains, NBD1 and NBD2, were purified, refolded, and extracted from size exclusion chromatography. (A) Loaded at 7 μg/protein, Coomassie blue-stained NBD1 and NBD2 migrated at SDS-PAGE following expected molecular weights of 27.8 kDa and 29.8 kDa, respectively. (B) Superdex 200 gel filtration column elution of NBD1 and NBD2 demonstrated oligomeric profiles away from the monomeric 28 kDa protein. The far-UV (C) and near-UV (D) circular dichroism spectra revealed distinctive secondary and tertiary structures of NBD1 and NBD2. Inset: ATPase activity of NBD1 (0.51±0.37 nmol/mg/min, n=4) and NBD2 (1.62±0.41 nmol/mg/min, n=4) demonstrated functionality of refolded protein.

3.2. Synchrotron Radiation X-Ray Scattering Profiles of NBD1 and NBD2

Synchrotron radiated NBD1 and NBD2 molecules displayed an intensity (I) versus scattering vector (q) profile (Fig. 2A) with linear dependence, solved by the Guinier function, i.e., log(I(q)) versus (q)², that demonstrated monodispersity (Fig. 2A inset), distinguishable from aggregated samples indicated by non-linear dependence (Putnam et al., 2007). Synchrotron radiation-based quality control established a profile of recombinant NBD1 and NBD2 proteins amenable for structural deconvolution (Supplementary Fig. 1A and 1B). The distance distribution function P(r), which represents the probable distribution of interatomic distances, was implemented through indirect Fourier transformation (Svergun, 1992), and displayed a radius of gyration (R_G) and a maximum molecular dimension (D_max) of 55.3±0.2 Å and 170±8 Å for NBD1 (n=5), and 59.3±0.3 Å and 180±12 Å for NBD2 (n=5), respectively (Fig. 2B). The size of NBDs was independently validated by atomic force microscopy at nanoscale resolution (n=8; Fig. 2C). NBD1 and NBD2 particles corresponded to eight assembled monomers extrapolated from the normalized SAXS intensity function (q=0, I(0)/conc) (Mylonas and Svergun, 2007; Putnam et al., 2007) through comparison with protein standards (Supplementary Fig. 1C). Thus, SAXS resolved interference-free NBD1 and NBD2 proteins as octameric architectures with maximum dimensions of 170 and 180 Å, respectively.

Figure 2.

Molecular outlines from interference-free NBD1 and NBD2. (A) Superimposed experimental SAXS curves from refolded NBD1 and NBD2. Inset: Guinier plots reflect homogeneous and monodisperse NBD1 and NBD2 in solution. (B) Distance distribution function, (P(r)), computed from experimental data and normalized to the maximum value of unity, demonstrated a maximum molecular dimension (D_max) at 170 and 180 Å for NBD1 and NBD2, respectively. (C) Atomic force microscopy in contact mode mapped the topography of NBD2 molecules (186±18 Å, n=8).

3.3. Ab initio Reconstruction of NBD Envelope

The macromolecular fit for NBD1 or NBD2 proteins was constructed by imposing symmetry constraints using ab initio simulated annealing (Svergun, 1999). To attain optimal symmetry, five different conditions (P1, P2, P4, P8 and P222) reflective of octameric permutations were assessed (Putnam et al., 2007). Initial computation, launched with a chain-compatible spatial distribution of beads (Volkov and Svergun, 2003), built a spectrum of macromolecular shapes. Sixteen independent runs per each symmetry condition were aligned, averaged, and scored with the normalized spatial discrepancy (NSD) index to prioritize the best constraint while minimizing discrepancies (χ²) between calculated and experimental SAXS outputs (Fig. 3A). Average χ² values from NBD1 ab initio reconstructions were 0.7340±0.0009 for P1, 0.7370±0.0006 for P2, 0.7337±0.0011 for P4, 0.7238±0.0027 for P8 and 0.7340±0.0007 for P222 (n=16; Fig. 3A). In addition, mean values of normalized spatial discrepancy (NSD) of NBD1 were 0.575 for P1, 1.195 for P2, 0.766 for P4, 1.090 for P8, and 0.687 for P222 (n=16; Fig. 3A). Although simulations produced under P8 symmetry exhibited the lowest χ² value, the NSD under P8 demonstrated a wide variation similar to P2. Envelopes constructed under P4 symmetry or without symmetry restraint (P1) provided less variation than P8 or P2 models. Imposing the highest symmetry (P222) produced the most convergent model with the lowest discrepancy between calculated and experimental data (Fig. 3A). Indeed, residual curves imposed by P222 symmetry obtained from the fit to the desmeared and smoothed data from GNOM showed lower variation compared to the P8 symmetry constrain (Supplementary Fig. 2). The P222-based ab initio NBD1 model disclosed an oblong shape with composite dimensions of 168 × 80 × 37 ų (Fig. 3B and Fig. 3C). Similarly, ab initio modeling of NBD2 under different symmetry conditions was evaluated. Average χ² values from NBD2 reconstructions were 0.6767±0.0013 for P1, 0.6768±0.0011 for P2, 0.6841±0.0017 for P4, 0.6640±0.0046 for P8 and 0.6794±0.0022 for P222 (n=16; Fig. 3D). The mean values of NSD were 0.621 for P1, 1.059 for P2, 0.678 for P4, 1.497 for P8, and 0.789 for P222 (n=16; Fig. 3D). An imposed P222 symmetry retained an optimized χ² agreement and NSD values, revealing an oblong shape with averaged dimensions of 175 × 81 × 37 ų (Fig. 3E and Fig. 3F). Reconstruction was independently verified through a progressive spatial distribution run of dummy residues centered on the C_α atoms of the protein (Supplementary Fig. 3). Thus, ab initio reconstruction divulged oblong envelopes holding NBD1 and NBD2 proteins.

Figure 3.

Small angle X-ray scattering reconstruction of NBD1 and NBD2. Ab initio models of NBDs were determined from experimental SAXS data under different point-group symmetries using DAMMIN. The fits for NBD1 (A) and NBD2 (E) under different symmetry conditions are displayed as a function of normalized spatial discrepancy (NSD), discrepancy between the calculated and experimental SAXS curves (χ²), and standard errors (SE) of χ². Sixteen independent runs were performed using P1, P2, P4, P8 and P222 symmetry. Ensemble average and respective fits for NBD1 (B and C) and NBD2 (E and F) based on P222 symmetry.

3.4. Quaternary NBDs Structures Resolved by Rigid Body Modeling

By docking crystal structure homology models (Putnam et al., 2007) against experimentally defined SAXS data, the quaternary structures of NBD1 (Fig. 4A) and NBD2 (Fig. 4B) were mapped onto SAXS envelopes. Based on P222 symmetry constraints, NBDs were resolved as tetrameric dimers that fulfilled the SAXS deduced octameric arrangement. Starting from a stochastically positioned crystallographic homodimer under P222 symmetry, atomic coordinates for NBD1 or NBD2 were computed using the CRYSOL platform (Svergun et al., 1995) to predict X-ray scattering amplitude patterns. Iterations with the modeling program GLOBSYMM (Petoukhov and Svergun, 2005) revealed a homodimeric quaternary structure for NBD1 (Fig. 4A) and NBD2 (Fig. 4B). From independent program runs with different Fibonacci grid orders, the best reconstructed model fitted experimental data with χ²=2.69 for NBD1 and χ²=4.09 for NBD2. It should be noted that a homology model used to fit the low resolution SAXS profile, as well as randomly oriented N-terminal pre-residues, could impact fitting (Putnam et al., 2007). The reconstructed models under P222 symmetry had a significantly better correlation with SAXS data than other symmetry models (e.g., for P4 symmetry see Supplementary Fig. 4). As the root mean square deviation between homodimer NBD1 or NBD2 and heterodimer NBD1/NBD2 was 0.54 Å or 0.58 Å, respectively, i.e., values <1 Å, the quaternary structure of the NBD1/NBD2 heterodimer was derived from homomeric NBD1 and NBD2 SAXS profiles. Although this approach is abstract, in the predicted model of heterodimeric NBD1/NBD2 quaternary structure (χ²=2.86), NBD2 was consistently oriented at the outside of the octameric ring while NBD1 was positioned at the inside of the octameric ring facing a central opening with a diameter of ~35 Å (Fig. 4C). Such opening could incorporate a macromolecular structure with dimensions equivalent to the Kir6.2 channel pore (Mikhailov et al., 2005) reconstructing a K_ATP channel complex.

Figure 4.

Rigid body models of SUR2A NBD1(A), NBD2 (B) and NBD1/NBD2 (C) under P222 symmetry superimposed on ab initio envelopes. χ² values are 2.69 for NBD1, 4.09 for NBD2, and 2.86 for NBD1/NBD2. The inside and outside orientations of the octameric ring are denoted in yellow and blue for NBD1, and in green and purple for NBD2, respectively. In the NBD1/NBD2 structure, NBD1 is in yellow and NBD2 in purple. Scattering profiles (purple) from quaternary structure models of NBD1, NBD2 and NBD1/NBD2 were overlaid on experimental scattering curves (black).

3.5. Structural Consequences of Pathogenic SUR2A NBD1/NBD2 Mutations

Susceptibility to human heart disease has been linked to genetic defects in ABCC9-encoded SUR2A (Bienengraeber et al., 2004; Minoretti et al., 2006; Olson et al., 2007). This includes the NBD1 V734I variant (Fig. 5A), associated with myocardial infarction, and the NBD2 A1513T and T1547I mutations implicated in dilated cardiomyopathy and stress-induced atrial fibrillation, respectively (Fig. 5B). Here, the resolved structures based on low resolution SAXS data guided a targeted evaluation of human heart disease-associated mutations. The A1513 mutation, replaced with a bulky nucleophilic T residue, mapped in the NBD2 monomer to the C-terminal β-strand in close proximity with the Walker A motif, and disrupted the tertiary protein structure leading to disarray of this function-critical region (Fig. 5B). While the monomeric structure could not decisively address the structural consequences of either V734I or T1547I mutations due to their remote positions from critical functional motifs (Fig. 5A and 5B), resolved heterodimer and quaternary structures were able to trace anticipated structural implications. Specifically, the NBD2 T1547I, whereby a nucleophilic residue is substituted with a somewhat larger hydrophobic residue found at the tail of the isoform-specific SUR2A C-terminal α-helix (Fig. 5B), compromised solvent accessibility due to proximity from the NBD1/NBD2 heterodimer interface (Fig. 5C). Though the NBD1 V734 mutation swapped with a larger I residue was charted at a loop region distant to either Walker A or Walker B motifs (Fig. 5A), the resolved quaternary structure indicated that the afflicted residue was at the juncture between regulatory NBD1/NBD2 domains and the groove for channel protein insertion (Fig. 5D). Pathogenic mutations are thereby demonstrated to cluster at locales critical for protein-protein interaction underlying the structural integrity of the K_ATP channel complex.

Figure 5.

Structure of NBDs incorporating pertinent pathogenic mutations found in human disease. (A) Homology structural model of NBD1 (A), NBD2 (B), NBD1/NBD2 (C), and SAXS resolved quaternary structure of NBD1/NBD2 (D) with mutated residues clustered at protein-protein interaction locales. NBD1 in yellow; NBD2 in purple; Walker A (W_A), Walker B (W_B), and linker motifs in blue; pathogenic human mutations in green.

4. Discussion

The atypical ABC protein SUR2A serves as a regulatory subunit of heteromultimeric ATP-sensitive K⁺ (K_ATP) channels in heart muscle, in particular within the ventricle (Du et al., 2006; Flagg et al., 2008; Inagaki et al., 1996; Li et al., 2000; Nichols, 2006; Shi et al., 2005). Coupling of the pore-forming Kir6.2 with cellular energetic signaling systems relies on cooperative interaction of the SUR2A nucleotide binding domains, NBD1 and NBD2, endowing high-fidelity metabolic sensing properties to the channel complex (Abraham et al., 2002; Alekseev et al., 2005; Zingman et al., 2001; Zingman et al., 2002). While the function of NBD1/NBD2 has been linked to the homeostatic cardioprotective role afforded by K_ATP channels (Arrell et al., 2009; Hodgson et al., 2003; Kane et al., 2005; Nichols, 2006; Zingman et al., 2007; Zlatkovic et al., 2009), here synchrotron radiation X-ray scattering was used to resolve the quaternary architecture of these unique nucleotide binding domains. The delineated macromolecular arrangement outlines structural constraints for inter-molecular communication within regulatory domains of the K_ATP channel.

Small angle X-ray scattering (SAXS) provided an approach to reveal nanometer-scale structural information and domain organization without the requirement for a crystalline sample. SAXS is particularly valuable to study biomacromolecules within a native solution environment (Hura et al., 2009; Petoukhov and Svergun, 2007; Putnam et al., 2007), and although successful in resolving voltage gated channel domains (Pioletti et al., 2006) or typical ATPases (Davies et al., 2005), atypical ABC proteins - such as the K_ATP channel subunit SUR2A - have not been characterized in this manner. Subjected here to synchrotron radiation in solution, the refolded and purified SUR2A NBD1 and NBD2 proteins fulfilled quality control criteria necessary for structural deconvolution (Putnam et al., 2007). Obtained SAXS profiles mapped an octameric appearance for both NBD1 and NBD2, with the resolved shape and size verified by atomic force microscopy-based nanoimaging.

The presented quaternary ensemble profile based on low resolution SAXS data provides direct structural evidence, at the level of nucleotide binding domain molecules, for the proposed K_ATP channel subunit stoichiometry previously limited to anticipated patterns of SUR/Kir6.2 biogenesis and assembly (Clement et al., 1997; Inagaki et al., 1997; Shyng and Nichols, 1997). Captured on electron micrographs, NBD oligomers correspond to a set of eight to nine monomers (de Wet et al., 2007) consistent with the present SAXS readout. Ab initio computation here further prioritized candidate oblong envelopes holding NBD1/NBD2 paired octameric ensembles. Structure modeling into constraints of the collected SAXS envelopes provided a refined arrangement that delineated preferential alignment for NBD1 and NBD2 facing toward the inside and outside of the octameric ring, respectively. This organization provides a structural substrate for the functional asymmetry reported for SUR2A NBD1 versus NBD2, unique among ABC proteins (Alekseev et al., 2005; Park et al., 2008). Indeed, cooperative NBD1/NBD2 heterodomain interaction, imposed by architectural subunit constraints, ensures intrinsic catalytic activity with vitality for optimal K_ATP channel gating (Park et al., 2008; Zingman et al., 2002).

Rigid body modeling of dimer NBDs against SAXS profiles unmasked an inner cleft, amenable for insertion of the Kir6.2 pore. Single-particle electron microscopy measurements have previously suggested that NBDs point towards the central channel pore with a significant tilt away from transmembrane domains (Mikhailov et al., 2005). Such structural implication is in line with independent electron micrography of annular NBDs formations surrounding a void delineated by channel subunit dimensions (de Wet et al., 2007) equivalent to those resolved herein. The putative plane of SUR2A NBDs and channel pore interaction remains at present speculative, and could range along the height of the cytoplasmic Kir6.2 region. Notwithstanding, the spatial arrangement of SUR2A NBD1/NBD2, mapped by SAXS rigid body refinement, indicated that one NBD, namely NBD1, is primarily facing the Kir6.2 channel pore consistent with non-equivalent roles of nucleotide binding domains (Mikhailov et al., 2002). Although speculative, it is noteworthy that based on other ABC proteins a cross-over in the arrangement of SUR NBDs and transmembrane domains (TMDs) (Procko et al., 2009; Rees et al., 2009) could bring a second set of transmembrane domains (TM12-TM17) into proximity to the transmembrane domain of Kir6.2. In fact, about 60 residues located between transmembrane domain 17 and NBD2 have been proposed to physically interact with Kir6.2 (Chan et al., 2003; Dupuis et al., 2008; Rainbow et al., 2004).

Such privileged relationship, in turn, provides a structural rationale for dysregulated channel subunit interaction in the setting of human disease-causing genetic mutations. A case in point is the recently reported variant V734I, implicated in increased susceptibility towards myocardial infarction (Minoretti et al., 2006). The resolved functional quaternary structure of NBDs revealed propinquity between V734 and the delineated inner cleft, a site potentially strategic in channel subunit interaction. More broadly, pathogenic mutations were found to cluster at locales critical for the protein-protein interphase (Ashcroft, 2005; Campbell et al., 2003; Masia and Nichols, 2008; Nichols, 2006) underlying an inherent vulnerability of the K_ATP channel complex structure susceptible to genetic variance. Thus, collectively these findings provide an initial structural rationale underlying dysregulation of channel subunit interaction in the setting of human heart disease.

In summary, synchrotron radiation small angle X-ray scattering resolved for the first time the macromolecular arrangement of nucleotide binding domains of ABCC9-encoded SUR2A. The delineated quaternary structure, based on low resolution SAXS data, revealed architectural constraints within the charted molecular envelope wrapping NBD1/NBD2 octamers, with dimensions validated by atomic force microscopy. The asymmetric domain alignment within the rigid body ensemble unmasked a structural substrate vulnerable to disease-provoked distortion. Deconvoluted NBD1 and NBD2 heterodomains establish, thereby, a structural map of critical K_ATP channel regulatory domains potentially useful in anticipating the consequences imposed by genetic variations among individuals.

Supplementary Material

01

Supplementary Figure 1. Small angle X-ray scattering (SAXS) demonstrated interference-free NBD1 and NBD2. (A) Lack of fluctuation as a function of increasing concentration for radius of gyration (R_G) and intensity measured at zero angle (I(0)/concentration) demonstrated lack of aggregation and interparticle interference. (B) The forward scattering intensity (I(0)) was linear in response to NBDs concentrations ruling out aggregation and disassembly of refolded NBDs. (C) Based on linear regression analysis, from I(0)/concentration versus molecular weights of 4 standard proteins, NBD1 or NBD2 were determined to be octamers.

02

Supplementary Figure 2. (A) Small angle X-ray scattering reconstruction of NBD1 under P8 symmetry using DAMMIN. (B) The residual curves (I_exp-I_model) of NBD1 observed for P8 and P222 symmetry. (C) The alternative residual curves (I_GNOM-I_model) of NBD1 obtained from the fit to the desmeared and smoothed data from GNOM revealed less variation under P222 than P8 symmetry.

03

Supplementary Figure 3. Small angle X-ray scattering reconstruction of NBD1 and NBD2 under P222 symmetry using GASBOR. Ab initio models were determined through a progressive spatial distribution run of dummy residues centered on the C_α atoms of the protein, and the ensemble average of the most probable models are shown.

04

Supplementary Figure 4. Rigid body models of SUR2A NBD1 (A) and NBD2 (B) under P4 symmetry were superimposed on ab initio models with χ² of 4.9 and 6.4, respectively. The inside and outside orientations of the octameric ring are denoted in yellow and blue for NBD1, and in green and purple for NBD2, respectively. Scattering profiles (purple) from quaternary structure models of NBD1 and NBD2 were overlaid on experimental scattering curves (black).