NMR analysis of the conformational properties of the trapped on-pathway folding intermediate of the bacterial immunity protein Im7

Abstract

Previous work shows that the transiently populated, on-pathway intermediate in Im7 folding contains three of the four native alpha-helices docked around a core stabilised by native and non-native interactions. To determine the structure and dynamic properties of this species in more detail, we have used protein engineering to trap the intermediate at equilibrium and analysed the resulting proteins using NMR spectroscopy and small angle X-ray scattering. Four variants were created. In L53AI54A, two hydrophobic residues within helix III are truncated, preventing helix III from docking stably onto the developing hydrophobic core. In two other variants, the six residues encompassing the native helix III were replaced with three (H3G3) or six (H3G6) glycine residues. In the fourth variant, YY, two native tyrosine residues (Tyr55 and Tyr56) were re-introduced into H3G6 to examine their role in determining the properties of the intermediate ensemble. All four variants show variable peak intensities and broad peak widths, consistent with these proteins being conformationally dynamic. Chemical shift analyses demonstrated that L53AI54A and YY contain native-like secondary structure in helices I and IV, while helix II is partly formed and helix III is absent. Lack of NOEs and rapid NH exchange for L53AI54A, combined with detailed analysis of the backbone dynamics, indicated that the hydrophobic core of this variant is not uniquely structured, but fluctuates on the NMR timescale. The results demonstrate that though much of the native-like secondary structure of Im7 is present in the variants, their hydrophobic cores remain relatively fluid. The comparison of H3G3/H3G6 and L53AI54A/YY suggests that Tyr55 and/or Tyr56 interact with the three-helix core, leading other residues in this region of the protein to dock with the core as folding progresses. In this respect, the three-helix bundle acts as a template for formation of helix III and the creation of the native fold.

Figure 1

(a) Cartoon of the structure of Im7 (pdb:1ayi) constructed with Molscript. The side-chains of important helix III residues, Leu53, Ile54, Tyr55 and Tyr56, are shown. (b) Schematic diagram of the folding mechanism of Im7^⁎. The four helices of native Im7 are coloured differently. The scheme highlights that this small, single-domain protein folds via a three-helical intermediate and rate-limiting transition state, that differ from each other in the packing of the helices. Intermediate formation occurs on the sub-milliseconds timescale. Data presented in this article indicate that the C-terminal region of helix II is not uniquely formed in the intermediate ensemble, but remains conformationally dynamic at this stage of folding.

Figure 2

Effect of sodium sulphate on ¹H-¹⁵N HSQC spectra of wild-type Im7^⁎ and its variants. The absence (left panels) and presence (right panels) of 400 mM sodium sulphate is shown for (a) and (f) wild-type Im7^⁎, (b) and (g) L53AI54A, (c) and (h) YY, (d) and (i) H3G3 and (e) and (j) H3G6. All spectra were measured at 500 MHz, 298 K on samples containing ∼1 mM protein concentration in 50 mM sodium phosphate buffer (pH 7.0), 90% H₂O/10% ²H₂O.

Figure 3

(a) X-ray scattering profile of Im7 in solution. Experimental data points are shown in red. The smooth curve represents the theoretical scattering profile based on the crystal structure of Im7 (pdb:1ayi). The inset depicts the Kratky plot of native Im7 compared with the protein in the presence of 0, 3, 6 or 8 M urea derived from the solution X-ray scattering data recorded at 10 °C. (b) Solution X-ray scattering profiles of Im7^⁎ and the L53AI54A variant, measured at 10 °C in the presence of 400 mM sodium sulphate. The inset shows the experimental distance distribution functions p(r) for both proteins.

Figure 4

TALOS-derived secondary structure of wild-type Im7^⁎, L53AI54A and YY aligned with that of the X-ray structure of wild-type Im7 (pdb:1ayi). Cylinders represent helices and connecting lines represent random coil, as determined by N, C^α, C^β and C′ chemical shifts in TALOS. Breaks in the secondary structure along the sequence denote residues for which the C^α atoms are unassigned and/or the amide resonances could not be detected due to exchange-broadening.

Figure 5

Two-dimensional ¹H-¹H projections of 3D ¹H-¹H-¹⁵N NOESY-HSQC data. (a) 500 MHz spectrum of wild-type Im7^⁎ and (b) 750 MHz spectrum of L53AI54A. Both spectra were acquired with a mixing time of 100 ms at 298 K. Protein concentration was ∼1 mM in 50 mM sodium or potassium phosphate buffer (pH 7.0), 400 mM sodium sulphate, 90% H₂O/10% ²H₂O and a trace amount of sodium azide. Contour levels are comparable.

Figure 6

Comparison of backbone ¹⁵N relaxation parameters determined for wild-type Im7^⁎ (left panels) and L53AI54A (right panels) at 600 MHz, 298 K. (a) and (b) T₁ relaxation times; (c) and (d) T₂ relaxation times; (e) and (f) {¹H}-¹⁵N heteronuclear NOE; (g) and (h) T₁/T₂ ratio. The average errors in the relaxation parameters for wild-type Im7^⁎ and L53AI54A, respectively, are 1.6% and 1.2% (T₁), 1.3% and 3.2% (T₂), 14.1% and 32.8% (NOE), 2.0% and 3.4% (T₁/T₂ ratio). Error bars are displayed although in some cases the error bar is smaller than the size of the symbols used. Residues in Im7^⁎ for which no data are shown results from very weak signal intensity (residue 54), severe resonance overlap (residues 19, 33, 39, 40, 63, 71, 85), undetectable resonances (residues 1, 2, 44), or residues that are proline (residues 48, 57, 65, 82). Likewise, in L53AI54A no data results from very weak signal intensity (residues 6, 8, 30, 51, 52, 53, 55, 56, 59, 60, 66, 68, 79), severe resonance overlap (residues 5, 20, 58, 61, 69, 74, 84), undetectable resonances (residues 1, 2, 7, 27, 28, 32, 39–49, 54), or residues that are proline (residues 48, 57, 65, 82). The secondary structure of wild-type Im7^⁎ and L53AI54A as predicted by TALOS is depicted above the left and right columns of the data, respectively. Protein concentration was ∼1 mM in 50 mM sodium/potassium phosphate buffer (pH 7.0), 400 mM sodium sulphate, 90% H₂O/10% ²H₂O and a trace amount of sodium azide.

Figure 7

Reduced spectral density functions, J(0) ((a) and (b)), J(ω_N) ((c) and (d)), and J(0.87 ω_H) ((e) and (f)) for wild-type Im7^⁎ (left panels) and L53AI54A (right panels) derived from R₁, R₂ and NOE relaxation data at 600 MHz, 298 . The secondary structure of wild-type Im7^⁎ and L53AI54A as predicted using TALOS is depicted above each column of data for reference. The average errors for wild-type Im7^⁎ and L53AI54A, respectively, are 1.4% and 3.3% (J(0)), 0.2% and 0.3% (J(ω_N)), 33.0% and 64.4% (J(0.87 ω_H). Error bars are displayed although in some cases the error bar is smaller than the size of the symbols used. Residues in Im7^⁎ for which no data are shown results from very weak signal intensity (residue 54), severe resonance overlap (residues 19, 33, 39, 40, 63, 71, 85), undetectable resonances (residues 1, 2, 44), or residues that are proline (residues 48, 57, 65, 82). Likewise, in L53AI54A no data results from very weak signal intensity (residues 6, 8, 30, 51, 52, 53, 55, 56, 59, 60, 66, 68, 79), severe resonance overlap (residues 5, 20, 58, 61, 69, 74, 84), undetectable resonances (residues 1, 2, 7, 27, 28, 32, 39–49, 54), or residues that are proline (residues 48, 57, 65, 82).

Figure 8

Plots of J(0) against (a) J(ω_N) and (b) J(0.87 ω_H) for residues of Im7^⁎ (left panels) and L53AI54A (right panels), respectively. Residues located in helices are colour-coded as follows: helix I (red), helix II (green), helix III (orange), helix IV (blue). Black symbols depict non-helical residues. Continuous lines represent the line of best fit through the data and were calculated using a linear least-squares fit. ▴ symbols indicate residues in Im7^⁎ that experience conformational exchange and were not included in the linear least-squares fit of the Im7^⁎ data. The slopes for the line of best fits in (a) are 0.0073 ± 0.007 (Im7^⁎) and 0.0014 ± 0.004 (L53AI54A), with correlation coefficient, R, values of 0.1207 and 0.0487, respectively; and in (b) are −0.0036 ± (7.16 × 10⁻⁴) (Im7^⁎) and −0.0033 ± (5.69 × 10⁻⁴) (L53AI54A), with correlation coefficient, R, values of −0.5329 and −0.6797, respectively.