Hydrogen-bond detection, configuration assignment and rotamer correction of side-chain amides in large proteins by NMR spectroscopy through protium/deuterium isotope effects

Abstract

The configuration and hydrogen-bonding network of side-chain amides in a 35 kDa protein were determined by measuring differential and trans-hydrogen-bond H/D isotope effects by using the isotopomer-selective (IS)-TROSY technique, which leads to a reliable recognition and correction of erroneous rotamers that are frequently found in protein structures. First, the differential two-bond isotope effects on carbonyl (13)C' shifts, which are defined as Delta(2)Delta(13)C'(ND) = (2)Delta(13)C'(ND(E))-(2)Delta(13)C'(ND(Z)), provide a reliable means for the configuration assignment for side-chain amides, because environmental effects (hydrogen bonds and charges, etc.) are greatly attenuated over the two bonds that separate the carbon and hydrogen atoms, and the isotope effects fall into a narrow range of positive values. Second and more importantly, the significant variations in the differential one-bond isotope effects on (15)N chemical shifts, which are defined as Delta(1)Delta(15)N(D) = (1)Delta(15)N(D(E))-(1)Delta(15)N(D(Z)) can be correlated with hydrogen-bonding interactions, particularly those involving charged acceptors. The differential one-bond isotope effects are additive, with major contributions from intrinsic differential conjugative interactions between the E and Z configurations, H-bonding interactions, and charge effects. Furthermore, the pattern of trans-H-bond H/D isotope effects can be mapped onto more complicated hydrogen-bonding networks that involve bifurcated hydrogen-bonds. Third, the correlations between Delta(1)Delta(15)N(D) and hydrogen-bonding interactions afford an effective means for the correction of erroneous rotamer assignments of side-chain amides. Rotamer correction by differential isotope effects is not only robust, but also simple and can be applied to large proteins.

Figure 1

Regions of A) 2D ¹⁵N–¹H IS-TROSY and B) 2D IS-TROSY-H(N)CO spectra of yCD showing Asn/Gln side-chain amide resonance correlations. On top of A) is a schematic diagram that illustrates how the differential H/D isotope effects on one-bond ¹⁵N chemical shifts, Δ¹Δ¹⁵N(D) = ¹Δ¹⁵N(D^E) − ¹Δ¹⁵N(D^Z), are measured in the 2D ¹⁵N–¹H IS-TROSY spectrum. Both spectra A) and B) were recorded on a Bruker AVANCE 900 MHz NMR spectrometer equipped with a TCI cryoprobe at 25 °C with a u-²H/¹³C/¹⁵N labeled protein sample in the 50% H₂O/50% D₂O solvent. The differential H/D isotope effects on one-bond ¹⁵N chemical shifts, Δ¹Δ¹⁵N(D) = ¹Δ¹⁵N(D^E) − ¹Δ¹⁵N(D^Z) (modified by scalar couplings), and on two-bond ¹³C′ chemical shifts, Δ²Δ¹³C′ (ND) = ²Δ¹³C′ (ND^E) − ²Δ¹³C′ (ND^Z), where E stands for the trans and Z the cis side-chain amide hydrogen atoms, can be accurately measured from the chemical shift difference in the indirect dimension for the same carboxyamide moiety. Paired side-chain NH{D} resonances of each Asn/Gln residue is linked by two dash lines across the center of each resonance, and the chemical shift difference is indicated with paired arrows. Stereospecific distinction between the E and Z protons can be achieved based on the difference in isotope effects (refer to the text for details). Spectrum of A) was recorded with 8 scans and a 2 s delay time, t_1max(¹⁵N) = 93 ms and t_2max(¹H^N) = 285 ms, resulting in the experimental time of 2.4 h and spectrum of B) with 64 scans and a 1.8 s delay time, t_1max(¹³C) = 110 ms and t_2max(¹H^N) = 285 ms, resulting in the experimental time of 29.8 h Before Fourier transformation, the raw data were zero filled resulting in a 0.5 Hz digital resolution in the indirect dimension for spectrum A) and 1.0 Hz for B).

Figure 2

Regions of A) the 2D ¹⁵N–¹H IS-TROSY spectrum and B) the 2D projection of 3D ¹⁵N–¹H IS-TROSY-dispersed NOESY spectrum of yCD with u-²H/¹³C/¹⁵N labeling in 50% H₂O/50% D₂O. The spectrum was recorded on a Bruker AVANCE 900 MHz NMR spectrometer. Experimental and data processing conditions are the same as in Figure 1. Illustrated in the inset in panel A) is suggested hydrogen-bonding network of Asn113, which is derived from the 1.14 Ǻ crystal structure of yCD (accession code: 1P6O) after flipping the side-chain oxygen and nitrogen positions.

Figure 3

Regions of the 2D ¹⁵N–¹H IS-TROSY spectra of u-²H/¹³C/¹⁵N labeled yCD in A) 50% H₂O/50% D₂O and B) 95% H₂O/5% D₂O. The spectra were recorded on a Bruker AVANCE 900 MHz NMR spectrometer. Experimental and data processing conditions are the same as in Figure 1. The doublets of Gln55-NH^E{H^Z} resonance are caused by across H-bond H/D isotope effects mediated via the bifurcated H-bond of His50-H^ε/D^ε as illustrated in the top diagram.

Figure 4

2D IS-TROSY-H(N)CO spectrum of u-²H/¹³C/¹⁵N labeled yCD in 50% H₂O/50% D₂O. The spectrum was recorded on a Bruker AVANCE 900 MHz NMR spectrometer. The doublet Asn/Gln side-chain amide cross-peaks, enclosed with dashed and enlarged with solid squares, are caused by trans-H-bond isotope effects ^2hΔ¹H and ^4hΔ¹³C′. The spectrum was recorded with 128 scans and a 1.8 s delay time, t_1max(¹³C) = 71 ms and t_2max(¹H^N) = 285 ms, resulting in the experimental time of 41 h. Before Fourier transformation, the raw data were zero filled resulting in a 1.0 Hz digital resolution in the indirect dimension.

Figure 5

Evolution of the C^α RMSDs of residues 15–158 of subunit 1 (solid line) and subunit 2 (dashed line) from those of the 1.14 Ǻ-resolution crystal structure during the entire MD simulation period.

Figure 6

Variations of the torsion angles for the side-chain amides in the last 4-ns MD simulation. The torsion angles measured from the 1.14 Ǻ-resolution crystal structure are indicated by solid squares. The subunit identities are labeled at the top and residue identities on the left side.