Structural and dynamic study of the tetramerization region of non-erythroid alpha-spectrin: a frayed helix revealed by site-directed spin labeling electron paramagnetic resonance

Abstract

The N-terminal region of alpha-spectrin is responsible for its association with beta-spectrin in a heterodimer, forming functional tetramers. Non-erythroid alpha-spectrin (alphaII-spectrin) has a significantly higher association affinity for beta-spectrin than the homologous erythroid alpha-spectrin (alphaI-spectrin). We have previously determined the solution structure of the N-terminal region of alphaI-spectrin by NMR methods, but currently no structural information is available for alphaII-spectrin. We have used cysteine scanning, spin labeling electron paramagnetic resonance (EPR), and isothermal titration calorimetry (ITC) methods to study the tetramerization region of alphaII-spectrin. EPR data clearly show that, in alphaII-spectrin, the first nine N-terminal residues were unstructured, followed by an irregular helix (helix C'), frayed at the N-terminal end, but rigid at the C-terminal end, which merges into the putative triple-helical structural domain. The region corresponding to the important unstructured junction region linking helix C' to the first structural domain in alphaI-spectrin was clearly structured. On the basis of the published model for aligning helices A', B', and C', important interactions among residues in helix C' of alphaI- and alphaII-spectrin and helices A' and B' of betaI- and betaII-spectrin are identified, suggesting similar coiled coil helical bundling for spectrin I and II in forming tetramers. The differences in affinity are likely due to the differences in the conformation of the junction regions. Equilibrium dissociation constants of spin-labeled alphaII and betaI complexes from ITC measurements indicate that residues 15, 19, 37, and 40 are functionally important residues in alphaII-spectrin. Interestingly, all four corresponding homologous residues in alphaI-spectrin (residues 24, 28, 46, and 49) have been reported to be clinically significant residues involved in hematological diseases.

Figure 1

NMR solution structure of the first 156 residues of erythroid αI-spectrin (ref. 17) (A) and the proposed structure for the first 149 residues of αII-spectrin (ref. 30) (B). The NMR structure identifies the first helix (Helix C′) consisting of residues 21 – 45 and the first structural domain of triple helical bundle (Helices A₁, B₁ and C₁) consisting of residues 53 – 154. The junction region between Helix C′ and Helix A₁ is unstructured and consists of residues 46 – 52. Through sequence homology, the corresponding residues for αII are also labeled, with residues 37 – 43 corresponding to the unstructured junction region in αI.

Figure 2

Circular dichcroism (CD) spectra for all αII proteins (wild type, cysteine-less and 39 αIIΔR1). The αII cysteine-less protein was shown as a thick black line; all other spectra were shown in gray. Spectra were recorded at 20 °C with sample ~ 10 μM in PBS7.4. Raw ellipticity was normalized by protein concentration to give molar ellipticity. Helical content was calculated from molar ellipticity at 222 nm using 36, 000 deg cm² dmol⁻¹ as 100 %. The helical content of each αIIΔR1 is shown in the inset.

Figure 3

Representative ITC titration curves and analyzed data for αIIΔR1 samples with typical affinities for association with βI, such as that of L9R1 (at a concentration of 85 μM; with βI at 5.5μM in PBS7.4) with a K_d value of 28 nM (A). The titration curve and analyzed data of L40R1 (at 126 μM; with βI at 7.8 μM) represent those with relatively low association affinities with βI, with a K_d value of 730 nM at 25 °C (B). The K_d values for αIIΔR1 proteins scanning from position 9 to position 47 (C) show that the affinity in association with βI for most of the proteins are similar to each other and to that of the WT parent protein (12 nM), except for I15R1 (237 nM), R19R1 (1.14 μM), R37R1 (177 nM) and L40R1 (730 nM). The values for the proteins with R1 at positions 29, 32, 39, 43 and 44 were slightly higher than that of WT, ranging from 50 – 80 nM. The K_d values of αIIΔR1 normalized by the K_d value of WT, K_d/K_d (WT), (D) shows that replacing the native residues with R1 caused the largest perturbation at position 19, followed by position 40 and then positions 15 and 37.

Figure 4

EPR spectra of αIIΔR1 family scanning residues 9 – 47 (thin solid line) at 20 °C, with background signals removed. All samples were in PBS7.4 with 30 % (w/w) sucrose. The protein concentrations were generally about 100 μM, and the spin label to protein ratios were about 0.8. A total of 64 scans were acquired for each spectrum. The corresponding simulated spectra (thick dotted lines) were also plotted. See Figure 5 for fitting and simulation details.

Figure 5

Simulated spectra of L9R1 (Component I) (top spectrum in A) and of Component III (bottom spectrum in A). L9R1 spectrum (29 %) and Component III (11 %) were subtracted from the experimental spectrum of Y26R1 to give the Y26R1 difference spectrum, which was used to simulate Component II of Y26R1 (middle spectrum in A). The sum of these three components, in their proper proportion (B), gave the fitted spectrum (dash line) of Y26R1 (C), which matched the experimental spectrum (thin solid line) well. (D) The sum of two simulated spectra generated by the simulation program directly, with a τ_c = 1.2 ns for one component (36 %) and a τ_c = 6.3 ns for another component (dash line) (64 %). This fit was not used since the spectrum did not fit the experimental spectrum well (thin solid line), but is shown to indicate the uniqueness of the three-component fit. Components I and III were the same for all αIIΔR1 spectra, whereas Component II varies as a function of residue position. Component II for each protein was obtained and used for fitting to give best fit of experimental spectra in Fig 2. The parameters used in simulation were as follow: gib0 = 0.02, N = 0.80, rbar = 8.08 and c₂₀ = 0.06 for L9R1 (Component I); gib0 = 3.79, N = 0.58, rbar = 6.50 and c₂₀ = 5.50 for Component III, and gib0 = 0.02, N = 0.80, rbar = 7.69 and c₂₀ = 0.01 for Component II of Y26R1.

Figure 6

Relative amounts (%) of Component I (○) (top), Component II (■) (middle) and Component III (Δ) (bottom) were used to give best fit spectra for αIIΔR1 proteins. Component I and Component III for all αIIΔR1 proteins were the same as those shown in Fig 5A. Component II spectrum for each αIIΔR1 protein was individualized. See text for details. The amounts of Component III were generally less than 20 % for all proteins, except for E41R1, Y46R1 and R47R1. The amounts of Component I exhibited a general “decreasing” trend, whereas the amounts of Component II exhibited a general “increasing” trend, from positions 9 – 47. There were no Components I for positions 39 – 46. The uncertainties were 9 % for Component I, 8 % for Component II and 4 % for Component III.

Figure 7

Rotational correlation time (τ_c) used to simulate Component II spectrum of each αIIΔR1 protein, except that at position 9, which had no Component II. The values exhibited a periodicity in labeled residue positions in αII. A nonlinear least square sine wave fit with a periodicity of 3.5 (y = sin (x−A)/3.5 * 2π, where A is the “phase”), using all data points (lower dotted line, sine wave 1), matched the data at positions 10 – 18 and 38 – 45. However, for the data of positions 19 – 37, the lower dotted line did not fit well. A separately fitted sine wave, using just the data set of positions 19 – 37 gave the upper dotted line (sine wave 2). The two dotted lines were shifted by 102°, or the periodicity of positions 20 – 37 was shifted by one residue toward the N-terminal end (one residue deletion) when compared with the lower dotted line. Also shown (top x-axis) are the sequence of residues 9 – 47 and the corresponding heptad assignment.