Applications for measuring scalar and residual dipolar couplings in proteins

Perttu Permi

Abstract

Nuclear magnetic resonance spectroscopic structure determination of proteins has been under rapid development during the last decade. The size limitation impeding structural studies of biological macromolecules in solution has increased from 10 kDa to 30 kDa thanks to exploitation of 15N/13C enrichment. Perdeuteration of non-exchangeable protons has pushed this limit even further, allowing backbone resonance assignment of 40 to 50 kDa proteins. Most recently, transverse relaxation optimized spectroscopy (TROSY) has been demonstrated to lengthen 15N and 1HN spin transverse relaxation times significantly, especially in large perdeuterated proteins, thus extending the size limit beyond 100 kDa systems. However, determination of structurally important nuclear Overhauser enhancements (NOE) suffers from perdeuteration, due to the lower density of proton spins available, eventually leading to imprecise protein structures. Very recently, residual dipolar couplings have been used to supplement NOE information, enabling accurate molecular structures to also be obtained with perdeuterated proteins. This thesis focuses on the measurement of the structurally important 3J-coupling between 1HN and 1H spins, and determination of residual dipolar couplings by utilizing the novel spin-state-selective subspectral editing together with the TROSY methodology. This approach allows precise measurement of a large number of dipolar couplings in larger protonated or perdeuterated proteins.

Dedication

 

One man’s constant is another man’s variable

 
--A. J. Perlis 
Table of Contents
Acknowledgments
Abbreviations
List of original articles
1. Introduction
2. Aims of the study
3. Problems associated with measurement of coupling constants in proteins
4. Methods for determination of scalar and dipolar couplings in proteins
4.1. J-resolved experiments
4.2. Double- and zero-quantum experiments
4.3. E.COSY experiments
4.4. J-correlation experiments
4.5. Spin-state-selective filtering
5. Applications
5.1. Methods for measuring 3JHNHα coupling constants in 15N and 15N/13C-labeled protein samples
5.1.1. J-resolved experiments for measuring 3JHNHα
5.1.2. DQ/ZQ experiments for measuring 3JHNHα
5.1.3. Methods utilizing the E.COSY principle
5.1.4. Quantitative J-correlation
5.2. Methods for measuring scalar and dipolar couplings in dilute liquid crystal medium
5.2.1. Pulse sequences for determination of 1JNC’ and 2JHNC’ couplings
5.2.2. Determination of 1JC’Cα coupling
5.2.3. Access to 1JNCα, 2JHNCα, 2JNCα, and 3JHNCα
5.2.4. Insight to side-chains, 1JCαCβ
6. Conclusions
References
List of Tables
1. Transverse relaxation times for different spins in the 29 kDa, uniformly 15N/13C and 15N/13C/2H labeled, HCA II protein at 600 1H frequency. (Farmer & Venters 1999).
2. Transverse relaxation times for the TROSY and anti-TROSY components of backbone 15N in 8.6 kDa 15N/13C and 15N/13C/2H labeled ubiquitin, and in 30 kDa 15N/13C and 15N/13C/2H labeled EIN at 600 and 800 MHz 1H frequency (Kontaxis et al. 2000).
List of Figures
1. Protein resonance line widths as a function of isotropic molecular rotational correlation time τc. (a) Plots are shown for (--) 1H spins, (…) 1H spins directly bonded to 13C, and (- - -) 1H spins directly bonded to 15N spin. (b) Line widths for 13C and 15N spins. Plots are shown for (---) 13C, (- - -) 13C1Hz, (- . - . - . -) 15N, and (…) 15N1Hz coherences (Cavanagh et al. 1994).
2. A schematic representation of double/zero-quantum coherence spectroscopy for the determination of coupling constants.
3. A schematic representation of the E.COSY principle for the measurement of coupling constants.
4. A schematic representation of spin-state-selective subspectral editing for the measurement of coupling constants.
5. Different spin-state-selective filter elements for the subspectral editing. (A) Original S3E filter element (Meissner et al. 1997a). (B) Long α/β-half-filter element (Andersson et al. 1998). (C) Long α/β-half-filter element for averaging CSA and DD cross-correlation rates for doublet components (Andersson et al. 1998). (D) Long α/β-half-filter element with PFG-z-filtration and CSA/DD cross-correlation averaging (IV). (E) S3CT filter element (Meissner et al. 1997b).
6. Intensity (%) of the minor component with respect to the principal component as a function of 1JNH, 1JC"Cα, and 1JNC" couplings. Plots for the filters matched on 1JNH, 1JC"Cα, and 1JNC" were calculated using 94, 55, and 15 Hz as nominal values, respectively.
7. Pulse sequences of the (A) CT-HMQC-HA and (B) IM-HSQC experiments for the determination of 3JHNHα coupling constants in 15N/(13C) labeled proteins. Narrow and wide bars denote 90º and 180º pulses, respectively. The delays employed are: Δ = 1/(4JNH); 4T+4Δ = 2τ = J-modulation delay. (A) Phase cycling: φ1 = x; φ2 = 16(x), 16(y), 16(-x), 16(-y); φ3 = x, y, -x, -y; φ4 = 4(x), 4(y), 4(-x), 4(-y); φrec. = 2(2(2(x, -x), 2(-x, x)), 2(2(-x, x), 2(x, -x))). φ1 is incremented in the usual States-TPPI manner for quadrature detection in F1 (Marion et al. 1989). (B) Phase cycling: φ1 = 2(x), 2(y), 2(-x), 2(-y); φ2 = x, -x; φ3 = x; φ4 = 8(x), 8(-x); φrec. = 2(x, 2(-x) x), 2(-x, 2(x) −x). φ2 is incremented in the usual States-TPPI manner for quadrature detection in F1. Two spectra for each experiment are recorded in an interleaved manner, with and without alpha proton decoupling during the J-modulation delay, 2τ. Semi-selective alpha proton decoupling can be achieved either by applying a selective decoupling field, e.g., G3 pulse cascade (Emsley & Bodenhausen 1990), or two selective inversion pulses to the alpha proton region. 15N is decoupled during acquisition by WALTZ-16 decoupling field (Shaka et al. 1983). Efficient water suppression can be obtained using the WET scheme (Smallcombe et al. 1995).
8. A selected region of the (A) J-modulated CT-HMQC-HA and the (B) corresponding IM-HSQC spectrum of 1.1 mM U-15N/13C HB-GAM in 95%/5% H2O/D2O, pH 4.7, 30°C, recorded on the Varian INOVA 500 spectrometer. Experimental parameters IM-HSQC (CT-HMQC-HA): t1, max = 141.1 (29.4) ms, t2 = 128 (128) ms, number of transients = 32 (32). Data were zero-filled to 2K (4K) in F1 (F2) dimensions in both experiments and apodized using a cosine bell weighting function in both dimensions.
9. The estimated maximal dipolar contributions to various scalar couplings between different nuclei in the protein main-chain when the maximal contribution to 1JNH is 25 Hz in a dilute liquid crystal.
10. Pulse schemes of the (A) HN(α/β-NC"-J) and (B) HN(α/β-NC"-J)-TROSY experiments for the measurement of 1JNC" and 2JHNC" couplings from the 15N, 1H correlation spectrum. Narrow and wide bars denote 90º and 180º pulses, respectively, whereas selective 90º pulses for water are denoted by half-ellipses. 13C" 180˚ pulses are applied with a strength of δ/√3, where δ is the frequency between centers of the 13C" and 13Cα regions. Aliphatic carbons are selectively decoupled during t1 with the semi-selective SEDUCE-1 decoupling scheme (McCoy & Mueller 1992). Alternatively, an 180º pulse selective for α-carbons can be used. If the gradient-selected, sensitivity-enhanced HSQC is used, 15N is decoupled during acquisition using the WALTZ-16 decoupling field (Shaka et al. 1983). The delays employed are: Δ = 1/(4JNH); Ta = 1/(4JNC’); δ = gradient duration + recovery delay. (A) Phase cycling for the in-phase experiment: φ1 = x, -x; φ2 = x; φ3 = 2(y), 2(-y); φ4 = x; φrec. = x, -x. For the antiphase experiment, the phases of the φ1 and φ2 pulses are incremented by 90º. Frequency discrimination in F1 is obtained using the PEP sensitivity-enhanced gradient selection (Kay et al. 1992; Schleucher et al. 1993) by inverting the sign of the Gs gradient pulse together with the inversion of φ4. (B) Phase cycling for the in-phase experiment: φ1 = x, -x; φ2 = x; φ3 = y; φ4 = x; φ5 = 2(y), 2(-y); φrec. = x, -x. For the in-phase experiment, φ5 is incremented by 90º. For the axial peak suppression, φ1, φ4, and φ5 are incremented in the usual States-TPPI manner (Marion et al. 1989). Quadrature detection and TROSY selection in F1 is obtained by collecting two data sets, (I): φ2 = y; φ3 = x, (II): φ2 = -y; φ3 = -x, with simultaneous change in the gradient polarity (Weigelt 1998).
11. Expansion of the HN(α/β-NC’-J) spectrum recorded from U-(15N, 13C) 18 kDa cTnC (0.5 mM) in a dilute liquid crystal at 40ºC, t1,max (t2) = 71 (128) ms, 16 transients. The upfield and downfield multiplet components are shown overlaid. 1(J+D)NC’ and 2(J+D)HNC’ couplings can be measured along the 15N- and 1H-dimensions, respectively. The spectrum was recorded on the Varian Unity 600 NMR spectrometer. The data were zero-filled to 2048x2048 points prior to Fourier transformation, and phase-shifted squared sine-bell window functions were applied in both dimensions.
12. Pulse sequences of the (A) H(α/β-NC"-J)CO and (B) HNCO(α/β-NC"-J) experiments for measuring 1JNC" couplings from the two (three)-dimensional 13C", (15N), 1H correlation spectrum. Narrow and wide bars denote 90º and 180º pulses, respectively, whereas selective 90º pulses for water are denoted by half-ellipses. Aliphatic carbons are selectively decoupled during t1 with the semi-selective SEDUCE-1 decoupling scheme (McCoy & Mueller 1992). Alternatively, a 180º pulse selective for α-carbons can be used. The WALTZ-16 sequence (Shaka et al. 1983) was used to decouple 1H during heteronuclear coherence transfer and 15N during acquisition. 13C 90˚ (180˚) pulses were applied with a strength of δ/√15 (δ/√3), where δ is the frequency between centers of the 13C" and 13Cα regions. All 13C" pulses were applied on-resonance and 13Cα pulses off-resonance with phase modulation by δ. The vertical arrow indicates the position of the off-resonance compensation pulse. The delays employed are: Δ = τ = 1/(4JNH); Ta = 1/(4JNC’); Tb = 1/(4JC"Cα); 0 ≤ κ ≤ Ta’/t1,max; λ ≥ 0; δ = gradient duration + recovery delay. (A) Phase cycling for the in-phase experiment: φ1 = x, -x; φ2 = 2(y), 2(-y); φ3 = 4(x), 4(-x); φrec. = x, 2(-x), x. For the antiphase experiment, the phase of φ2 is incremented by 90º. Frequency discrimination in F1 is achieved by incrementing φ1 according to the States-TPPI protocol (Marion et al. 1989). (B) Phase cycling for the in-phase experiment: : φ1 = y; φ2 = x, -x; φ3 = x; φ4 = 2(y), 2(-y); φ5 = 4(x), 4(-x); φrec. = x, 2(-x), x. For the antiphase experiment, phase of the φ4 pulse is incremented by 90º. Frequency discrimination in F1 is achieved by incrementing φ2 according to the States-TPPI protocol. Frequency discrimination in F2 is obtained using the PEP sensitivity-enhanced gradient selection (Kay et al. 1992). The echo and anti-echo signals are collected separately by inverting the sign of the Gs gradient pulse together with the inversion of φ3. In addition to echo/anti-echo selection, φ1 and φrec. are inverted according to the States-TPPI protocol for axial peak suppression. A 90º pulse on the carbonyl carbon after the t2 period serves as a purge pulse for the undesired dispersive magnetization component arising from the 1/(2JNC") mismatch (III-IV).
13. Pulse sequences of the (A) HN(α/β-C"Cα-J), (B) HN(α/β-C"Cα-J)-TROSY, and (C) HNCO(α/β-C"Cα-J) experiments for the measurement of 1JC"Cα couplings from the two (three)-dimensional (13C"), 15N, 1H correlation spectra. Narrow and wide bars denote 90˚ and 180˚ pulses, respectively, whereas selective 90˚ pulses for water are denoted by half-ellipses. 13C 90˚ (180˚) pulses are applied with a strength of δ/√15 (δ/√3), where δ is the frequency between centers of the 13C" and 13Cα regions. All 13C" pulses are applied on-resonance and 13Cα pulses off-resonance with phase modulation by δ. The vertical arrow indicates the position of the off-resonance compensation pulses. The WALTZ-16 sequence (Shaka et al. 1983) is used to decouple 1H during heteronuclear transfer and 15N during acquisition in non-TROSY experiments. The delays employed are: Δ = 1/(4JNH); Ta = 1/(4JNC’); Tb = 1/(4JC"Cα). (A) Phase cycling scheme for the in-phase experiment is φ1 = x, -x; φ2 = 2(x), 2(-x); φ3 = x; φ4 = x; φ5 = 4(x), 4(-x); φrec. = x, 2(-x), x. For the antiphase experiment, φ2 and φ3 are incremented by 90˚. Frequency discrimination in F1 is obtained using the PEP sensitivity-enhanced gradient selection (Kay et al. 1992). The echo and anti-echo signals are collected separately by inverting the sign of Gs gradient pulse together with inversion of φ4. (B) Phase cycling for the in-phase spectrum: φ1 = y; φ2 = x; φ3 = y; φ4 = x, -x; φ5= 4(x), 4(-x); φ6 = 2(x), 2(-x); φrec. = x, 2(-x), x. ). For the antiphase experiment, φ1 and φ6 are incremented by 90˚. Delays as in (A) except for Ta = 1/(4JNC’) + Tb/2 – Δ/2 - κ*t1/4; T’a = 1/(4JNC’) - Tb/2 - Δ/2 + κ*t1/4. 0 ≤ κ ≤ Ta/t1,max; δ = gradient duration + recovery delay. Frequency discrimination in F1 is obtained using the sensitivity-enhanced TROSY scheme with gradient selection (Weigelt 1998). The echo and anti-echo signals are collected separately by inverting the sign of Gs gradient pulse together with inversion of φ2 and φ3. (C) The phase cycling scheme for cos(πJC"Cαt1) cos(ωC"t1) modulated data is φ1 = x; φ2 = x, -x; φ3 = 2(x), 2(-x); φ4 = 4(x), 4(-x); φ5 = x; φrec. = x, 2(-x), x. For sin(πJC"Cαt1) sin(ωC"t1) modulated data, φ3 is incremented by 90˚. Frequency discrimination in F1 is achieved by incrementing φ2 according to the States-TPPI protocol (Marion et al. 1989). Frequency discrimination in F2 is obtained using the PEP sensitivity-enhanced gradient selection. The echo and anti-echo signals are collected separately by inverting the sign of Gs gradient pulse together with inversion of φ5.
14. A representative 2D-plane from the HNCO(α/β-C"Cα-J)-TROSY spectrum recorded from the 30.4 kDa protein E2. The corresponding upfield and downfield 1JC"Cα multiplet components are shown overlaid. A 1D trace is taken at the 13C’ chemical shift of 176.0 ppm. The spectrum was recorded on the Varian Unity INOVA 600 NMR spectrometer using 4 transients per FID from 1.0 mM U-(15N, 13C) and 80% 2H-labeled E2, 95%/5% H2O/D2O, 40 ºC, t1,max, t2,max, (t3) = 17, 18, (64) ms. Resolution in the F1-domain was doubled using forward linear prediction. Data were zero-filled to 128x512x512 data matrices and apodized with shifted squared sine-bell functions in all dimensions.
15. Expansion of the Lys63 15N, 1H cross-peak recorded using the {13Cα}-15N-TROSY experiment. The spectrum was recorded on the Varian Unity 600 NMR spectrometer from 1.0 mM U-(15N, 13C) ubiquitin, 90/10% H2O/D2O, 25ºC, t1,max (t2) = 222 (128) ms. Data were zero-filled to 4kx4k data matrices and apodized with shifted squared sine-bell functions in both dimensions. The data were processed using a squared cosine bell weighting functions in both dimensions.
16. The J-multiplied {13Cα}-15N-TROSY experiment for the measurement of 1JNCα, 2JNCα, 2JHNCα, and 3JHNCα couplings in 15N, 13C, (2H)-labeled proteins. Narrow and wide bars denote 90˚ and 180˚ pulses, respectively, whereas selective 90˚ pulses for water are denoted by half-ellipses. 13C" 180˚ pulses were applied with a strength of δ/√3, where δ is the frequency between centers of the 13C" and 13Cα regions. Aliphatic carbons are selectively decoupled during t1 with the semi-selective SEDUCE-1 decoupling scheme (McCoy & Mueller 1992). Alternatively, a 180˚ pulse selective for α-carbons can be used. The delays employed are: Δ = 1/(4JNH); δ = gradient duration + recovery delay; κ ≥ 0. Phase cycling: φ1 = x, -x; φ2 = x; φ3 = y; φrec. = x, -x. Quadrature detection and TROSY selection in F1 is obtained by collecting two data sets, (I): φ2 = x; φ3 = y, (II): φ2 = -x; φ3 = -y, with simultaneous change in the gradient polarity (Weigelt 1998).
17. The pulse scheme of the HN(α/β-NCα-J)-TROSY experiment for determination of 1JNCα, 2JNCα, 2JHNCα, and 3JHNCα couplings in 15N/13C/(2H)-labeled protein samples. The delays employed are: Δ = 1/(4JNH); Ta = 1/(4JNC"); T’a = 1/(4JNC") - Δ; Tb = 1/(4JC"Cα); 0 ≤ κ ≤ T’a/t1,max. Phase cycling for the in-phase spectrum: φ1 = x, -x; φ2 = x; φ3 = y; φ4 = 2(x), 2(-x); φ5 = 4(x), 4(-x); φrec. = x, 2(-x), x; for the antiphase spectrum, φ4 is incremented by 90˚. The arrow indicates the position of the Bloch-Siegert compensation pulse in the antiphase filter. Frequency discrimination in F1 is obtained using the sensitivity-enhanced TROSY scheme with gradient selection. The echo and anti-echo signals are collected separately by inverting the sign of Gs gradient pulse together with inversion of φ2 and φ3. Resolution in the 15N-dimension is improved by implementing an evolution period for the 15N chemical shift and the 15N-13Cα couplings in a semi-constant time manner. Narrow and wide bars denote 90˚ and 180˚ pulses, respectively, whereas selective 90˚ pulses for water are denoted by half-ellipses. 13C 90˚ (180˚) pulses are applied with a strength of δ/√15 (δ/√3), where δ is the frequency between centers of 13C" and 13Cα regions. All 13C" pulses are applied on-resonance and 13Cα pulses off-resonance with phase modulation by δ.
18. A selected F2-F3 plane from the 3D-HNCO(α/β-NCα-J)-TROSY experiment recorded from the 30.4 kDa, uniformly 15N/13C and 80% 2H-labeled protein, E2. The spectrum was recorded on the Varian Unity INOVA 600 NMR spectrometer using 8 transients per FID from 1.0 mM U-(15N, 13C) and 80% 2H-labeled E2, 95%/5% H2O/D2O, 40 ºC, t1,max, t2,max, (t3) = 12, 35, (56) ms. Resolution in the F1-domain was doubled using forward linear prediction. Data were zero-filled to 128x1kx1k data matrices and apodized with shifted squared sine-bell functions in all dimensions. The upfield and downfield 1JNCα multiplet components are shown overlaid. The visible splittings for the 1JNCα and 2JNCα couplings were multiplied by a factor of 4.
19. The pulse sequence of the HNCO(CαCβ-J)-TROSY experiment for the measurement of 1JCαCβ couplings in 15N/13C/(2H)-labeled proteins. The delays employed are: Δ = 1/(4JNH); Ta = 1/(4JNC"); T’a = 1/(4JNC") - Δ; Tb = 1/(4JC"Cα); 0 ≤ κ ≤ Ta/t2,max; λ ≥ 0; 0 ≤ µ ≤ Tb/t1,max. Phase cycle: φ1 = y; φ2 = x; φ3 = y; φ4 = 2(x), 2(-x); φ5 = x, -x; φ6 = 4(x), 4(-x); φrec. = x, 2(-x), x. The arrow indicates the position of the Bloch-Siegert compensation pulse in the antiphase filter. Frequency discrimination in F1 is obtained using the sensitivity-enhanced TROSY scheme with gradient selection. The echo and anti-echo signals are collected separately by inverting the sign of Gs gradient pulse together with inversion of φ2 and φ3. The resolution in the 15N-dimension is improved by implementing an evolution period for the 15N chemical shift and 15N-13Cα couplings in a semi-constant time manner. Narrow and wide bars denote 90˚ and 180˚ pulses, respectively, whereas selective 90˚ pulses for water are denoted by half-ellipses. 13C 90˚ (180˚) pulses are applied with a strength of δ/√15 (δ/√3), where δ is the frequency between centers of the 13C’ and 13Cα regions. All 13C pulses are applied on-resonance and 13Cα pulses off-resonance with phase modulation by δ.
20. A representative portion of the HNCO(CαCβ-J)-TROSY spectrum of ubiquitin. Cross-peaks are shown for K6, I13, L67, and V70 residues at the 15N cross-section of I13. The cross-peaks are split by apparent 2*1JCαCβ in the F1-dimension (λ = 2). The spectrum was recorded on the Varian Unity INOVA 500 NMR spectrometer using 24 transients per FID from 1.0 mM U-(15N, 13C) ubiquitin, 90%/10% H2O/D2O, 30ºC, t1,max, t2,max, (t3) = 37.6, 18.8, (64) ms. The data were post-processed to a 1024x128x1024 matrix prior to Fourier transformation, and phase-shifted squared sine-bell window functions were applied in all dimensions.
  Next
  Acknowledgments

Homepage of this publication

Library Units | Collections | Databases | Library News | Library Services | Electronic Collection | Links elsewhere | Alphabetical Index

© 2000 Oulu University Library