5.2. Methods for measuring scalar and dipolar couplings in dilute liquid crystal medium

5.2.1. Pulse sequences for determination of ¹J_NC’ and ²J_H^N_C’ couplings

The simplest way of determining ¹J_NC’ and ²J_H^NC’ coupling constants is to record a two-dimensional ¹³C’-coupled ¹⁵N-HSQC spectrum (Delaglio et al. 1991). Because the backbone ¹⁵N nuclei are coupled to three different carbon spins, i.e. ¹³C’(i-1), ¹³C^α(i-1), and ¹³C^α(i), it is necessary to decouple the ¹³C^α spins to measure the ¹J_NC’ splitting adequately. Thus, ¹J_NC’ is measured from the indirectly detected dimension, whereas ²J_H^NC’ can be measured from the orthogonal, i.e. proton, dimension in a familiar E.COSY fashion provided that the ¹³C’ spin-state is left untouched between the t₁ evolution period and acquisition. When this pulse sequence is combined with the spin-state-selective IPAP-HSQC (Ottiger et al. 1998), three different coupling constants can be measured from the same experiment (Wang et al. 1998). This pulse sequence has good sensitivity and the appearance of a ¹⁵N, ¹H correlation map, but it creates doublets in the spectrum and thus increases spectral overlap. However, this can be avoided by inserting a spin-state-selective filter, sensitive to the ¹³C’-spin-state, prior to the t₁ evolution period [II, IV]. Figure 10 shows the gradient-selected, sensitivity-enhanced versions of the HN(α/β-NC’-J) and HN(α/β-NC’-J)-TROSY pulse schemes, which enable simultaneous measurement of ¹J_NC’, ²J_H^NC’, and ¹J_NH couplings from a two-dimensional ¹⁵N, ¹H correlation spectrum superficially resembling ¹⁵N-HSQC or TROSY. In this triple-spin-state-selective experiment, the magnetization is initially transferred from the amide proton to its directly bound nitrogen using the familiar INEPT step. Subsequently, with the aid of a spin-state-selective filter element, either antiphase or in-phase ¹⁵N magnetization with respect to the ¹³C’ spin is created. During the following t₁ period, the ¹⁵N chemical shift is recorded with simultaneous evolution of ¹⁵N-¹³C’ coupling. Ultimately, by inserting a generalized TROSY scheme (Andersson et al. 1998; Weigelt 1998) prior to the acquisition period, the most slowly relaxing ¹⁵N and ¹H^N multiplet components can be selected, thus providing a ¹⁵N, ¹H correlation spectrum with high sensitivity and resolution.

This approach provides obvious advantages: 1) The spectral overlap diminishes. 2) As we separate the α- and β-spin-states of the ¹J_NC’ doublet into different subspectra, i.e. remove the overlap, the measured coupling is not an underestimate of the true coupling. The above-mentioned statements are valid only if an adequate subspectral editing is obtained, i.e. corresponding α- and β-spin-states are well separated into the two subspectra. As can be seen in Figure 6, good suppression of the undesired multiplet component is achieved for ¹J_NC’ couplings in the range between 11.5 and 18.5 Hz. Within this range, the principle component is at least 30 times larger than the undesired minor component. We can thus obtain a very good filtering in isotropic phase, where the variation of the ¹J_NC’ couplings is negligible with respect to the filtering capability. In the anisotropic phase, especially when strongly orienting Pf1 phages (Hansen et al. 1998) are exploited, insufficient subspectral editing is expected for some residues. However, this can easily be avoided by scaling the corresponding in- or antiphase spectra before the reconstruction of subspectra. Figure 11 shows overlaid subspectra recorded using the pulse sequence shown in Figure 10B from the uniformly ¹⁵N, ¹³C labeled human cardiac troponin C (cTnC), which has a molecular weight of 18 kDa (161 residues). Dipolar contribution to ¹J_H^NC’ can clearly be seen in the F₂-dimension by observing the direction of the slope connecting the E.COSY multiplets.

It is also possible to measure ¹J_NC’ and ²J_H^NC’ couplings from a HNCO-type spectrum by utilizing a short filter element matched to the ¹J_HN coupling. Ottiger (et al. 1998) have devised a pulse sequence in which the ²J_H^NC’ coupling can be measured from a spin-state-selective two-dimensional H(N)CO spectrum. In their approach, ²J_H^NC’ is measured from the ¹³C’-dimension. Advantages of measuring ²J_H^NC’ from the ¹³C’ rather than from the proton dimension are more favorable relaxation properties and the absence of large proton-proton dipolar contributions in the anisotropic phase if protonated samples are used. Unfortunately, the ¹³C’ chemical shift range is rather small (~10 ppm), and owing to its large chemical shift anisotropy, relaxation properties are not optimal at the highest magnetic fields. Kay and co-workers have used TROSY-based accordion spectroscopy (Bodenhausen & Ernst 1981) to measure ¹J_NC’ and ²J_H^NC’ from 3D-HNCO spectrum (Yang et al. 1999). In this case, ¹J_NC’ and ²J_H^NC’ are determined from the ¹⁵N- and ¹H-dimensions from the slowly relaxing cross-peak. However, as the ¹⁵N-¹³C’ coupling is not refocused, the couplings are measured from an antiphase splitting, and errors similar to those found in the phase-sensitive COSY are likely to occur in unfavorable cases. In addition, spectral crowding increases since the number of peaks is doubled.

Figure 12 illustrates two-dimensional H(α/β-NC’-J)CO and corresponding 3D-HNCO(α/β-NC’-J) pulse schemes (II; Permi et al. 2000) for determination of ¹J_NC’ coupling either from spin-state-selective ¹³C’-¹H^N or ¹³C’-¹⁵N-¹H^N correlation spectrum, respectively.

In both schemes, the magnetization is transferred from ¹H^N to ¹³C’ via directly bound ¹⁵N, and overall sensitivity is improved by employing proton decoupling during the ¹⁵N-¹³C’ INEPT steps (Figure 12). Subsequently, proton single-quantum coherence is excited and edited with respect to ¹⁵N. Editing is based on the large and uniform ¹J_HN coupling, providing an excellent subspectral editing with respect to the J-mismatch (Figure 6). Thus, at the beginning of the t₁ period, magnetization is in the form of 4H^N_zN_zC’_y in one experiment, whereas it is in the form of 2H^N_zC’_y in the other. After labeling the ¹³C’ chemical shift and ¹J_NC’ coupling frequencies during the t₁ evolution period, the magnetization is transferred back to the amide proton. In the 3D-HNCO(α/β-NC’-J) scheme (Permi et al. 2000; Figure 12B), the ¹⁵N chemical shift is detected during ¹³C’-¹⁵N back-transfer in the usual constant-time manner, utilizing the gradient-selected sensitivity-enhancement scheme. Eventually, after a post-acquisitional addition and subtraction of the corresponding in- and antiphase data sets, the cross-peaks appear at ω_C"(i-1) + πJ_C"N, ω_HN(i) and ω_C"(i-1) - πJ_C"N, ω_HN(i) in 2D, and ω_C"(i-1) + πJ_C"N, ω_N(i), ω_HN(i) and ω_C"(i-1) - πJ_C"N, ω_N(i), ω_HN(i) in 3D data sets, respectively. The separation of cross-peak placements in the F₁-dimension between the two subspectra yields the ¹J_NC" couplings.

The improvement over the pulse sequence presented by Kay and co-workers is the spin-state-selective filtering utilized in the ¹³C’-dimension. Hence, overlapping ¹⁵N-¹³C’ doublet components are separated into two subspectra. Although transverse relaxation of ¹³C’ is faster than that of ¹⁵N in larger proteins, α/β-filtering establishes the use of a shorter acquisition time in the ¹³C’-domain. It is also beneficial to scale ¹J_NC’ coupling up with respect to the ¹³C’ chemical shift in order to reduce the experimental time (at the cost of transverse relaxation, of course). Therefore, the number of t₁ increments can be reduced in the 3D version of the experiment. This also enables a more precise measurement of variations in ¹J_NC’, and reduces random measurement errors since the measured coupling is divided by 1+κ. For large, perdeuterated proteins, the measurement of ¹J_NC" is better carried out using the 3D HNCO(α/β-NC’-J)-TROSY scheme (Permi et al. 2000).

5.2.2. Determination of ¹J_C’Cα coupling

Relatively large dipolar contributions, i.e. 5-6 Hz, can be expected for one-bond scalar coupling between ¹³C’(i-1) and ¹³C^α(i-1) in the protein backbone. The scalar coupling varies between 51-55 Hz, but to the best of our knowledge, no dependence on backbone local conformation has been reported. The most obvious way of measuring this coupling is to record a ¹³C-HSQC spectrum, in which the coupling between ¹³C^α and ¹³C’ is allowed to evolve during the t₁ evolution period. This evolution period is usually implemented as a constant-time to decouple the relatively large couplings between aliphatic ¹³C^α and ¹³C^β. This is not very suitable for larger protonated protein samples due to the rapid relaxation of the ¹³C^α spins and the constant-time nature of the experiment. Clearly, neither is the method applicable to perdeuterated samples. Alternatively, starting from the ¹H^N magnetization, one could record a two-dimensional ¹³C’-¹H correlation spectrum, in which the ¹J_C’Cα coupling evolves concomitantly with the ¹³C’ chemical shift during t₁. However, as the dispersion of resonances in the ¹³C’-dimension is rather limited, the resulting 2D-spectrum may have severely overlapping signals.

As already mentioned, a ¹⁵N, ¹H correlation map usually gives best results when considering minimum overlap in a two-dimensional spectrum. Thus, it is advantageous to record the ¹⁵N, ¹H correlation spectrum, in which one is able to measure the ¹J_C’Cα coupling from cross-peak displacements in the F₁-dimension. Several slightly different approaches can be used, one of which is illustrated in Figure 13. In a simple HN(α/β-COCA-J) experiment (III; Figure 13A), the ¹H^N magnetization is first transferred to its preceding ¹³C’ spin. At this point, the half-filter sensitive to the ¹³C^α spin-state is inserted into the pulse sequence to create either N_zC’_y or N_zC’_xC^α_z coherence for the in-phase and antiphase experiment, respectively. Subsequently, a mixed DQ/ZQ coherence between ¹⁵N and the preceding ¹³C’ is created by applying a 90° pulse for ¹⁵N (time point a in Figure 13A). During the following t₁ evolution period, the ¹⁵N chemical shift evolves concomitantly with the ¹J_C’Cα coupling. Ultimately, the magnetization is transferred back to ¹⁵N single-quantum coherence and is brought back to the amide proton. Eventually, after addition and subtraction of the in-phase and antiphase data sets and their corresponding quadrature counterparts, correlations at ω_N(i) + πJ_C"Cα, ω_HN(i) and ω_N(i) - πJ_C"Cα, ω_HN(i) appear. It is then obvious that ¹J_C’Cα can be measured from the cross-peak displacement in the F₁-dimension between the two subspectra.

An alternative approach is to utilize the double-semi-constant-time scheme (DSCT) to evolve part of the ¹⁵N chemical shift during the ¹⁵N-¹³C’ out- and ¹³C’-¹⁵N back-transfer steps. By combining this approach with the sensitivity-enhanced TROSY scheme, more sensitive spectra for larger proteins can be obtained (IV). Alternatively, one could first record the coupling between ¹³C’ and ¹³C^α under ¹³C’ single-quantum coherence and monitor the ¹⁵N chemical shift under ¹⁵N single quantum coherence during the ¹³C’-¹⁵N back-transfer step, by employing the semi-constant-time TROSY scheme in order to obtain sufficient resolution in the ¹⁵N dimension. Additional sensitivity improvement is obtained by concatenating the first ¹⁵N, ¹H spin-state-selective filter element in the TROSY scheme with the ¹³C’-¹⁵N back-INEPT step (Salzmann et al. 1999; Permi et al. 2000; Figure 13B). Regrettably, use of both out- and back- ¹⁵N-¹³C’ INEPT steps for ¹⁵N shift evolution results in concomitant downscaling of ¹J_C’Cα and increases the error in the measured coupling. However, improvement in overall sensitivity may compete with the downscaling of apparent splitting and thus, no clear distinction between the accuracy of the two experiments can be made. One interesting aspect is the function of the last 90° pulse on ¹³C’ following the ¹³C’-¹⁵N back-INEPT step. This pulse purges the undesired dispersive magnetization components arising from the J-mismatch of delay 2T_a, used to refocus ¹J_NC’ (III, IV). In most cases, due to relaxation, the delay 2T_a is set shorter than 1/(2J_NC’), and as a result, the dispersive antiphase term also contributes to the detected signal. Therefore, it is essential to use a purge pulse on ¹³C’ to yield an absorptive line shape. It may also be advantageous to apply selective decoupling of either the ¹³C^α and/or ¹³C’ spins during acquisition (Yang & Kay 1999). This is because two- and three-bond dipolar contributions between ¹H^N(i) and ¹³C^α(i/i-1)/¹³C’(i) spins can be quite large in the anisotropic medium, and decoupling of these spins may improve the sensitivity of the experiments, even with larger proteins.

When considering large or highly alpha-helical proteins where resonance overlap is likely to occur more frequently, the need for a three-dimensional experiment is inevitable. The most obvious way to achieve sufficient resonance dispersion is then to record a HNCO-type experiment where ¹³C’ is allowed to couple to its preceding ¹³C^α during t₁. Kay and co-workers have successfully applied this approach with a TROSY implementation for two large, perdeuterated proteins (Yang et al. 1999). However, analogously to their HNCO-TROSY scheme for the measurement of ¹J_NC’ and ²J_H^NC’ couplings, the ¹³C^α coupled HNCO-TROSY experiment creates doublets in the spectrum, which may degrade a number of adequately separated cross-peaks. Thus, to achieve a minimum resonance overlap, it is advantageous to record an α/β-filtered HNCO-experiment (III; Permi et al. 2000; Figure 13C). Analogously to the 3D-HNCO(α/β-NC’-J) scheme, post-acquisitional addition and subtraction of the in- and antiphase data sets result in two spectra with correlations at ω_C"(i-1) + πJ_C"Cα, ω_N(i), ω_HN(i) and ω_C"(i-1) - πJ_C"Cα, ω_N(i), ω_HN(i), respectively. This ensures a minimum resonance overlap and allows direct extraction of ¹³C"-¹³C^α coupling constants by subtracting cross-peak frequencies in the F₁-dimension in the two subspectra. Since the ¹³C"-¹³C^α coupling is large (~53 Hz) and the spin-state-selective filtering is utilized, t_1,max can be sufficiently short, i.e. 20 ms. Without spin-state separation, a limited acquisition time for the ¹J_C"Cα coupling would result in ¹³C"-¹³C^α doublets, which are not resolved to the baseline. Consequently, the separation of doublet components would underestimate the true coupling values.

For larger proteins, measurement of ¹J_C"Cα is best carried out by using the transverse-relaxation-optimized HNCO(α/β-C’C^α-J)-TROSY scheme (Permi et al. 2000). The TROSY version is essentially similar to the non-TROSY version, excluding a few modifications. In this case, the time period for ¹⁵N frequency labeling is implemented in a manner similar to the semi-constant-time (SCT) TROSY evolution (IV). The ¹³C’-¹⁵N back-INEPT step and the first spin-state-selective filter element of the generalized TROSY scheme are concatenated to obtain an optimum sensitivity (Permi et al. 2000). In addition, for the selection of the most slowly relaxing ¹⁵N-¹H multiplet component, it is possible to take advantage of the gradient- and sensitivity-enhanced TROSY implementation (Weigelt 1998). This enables minimal phase cycling needed for the coherence selection, and also provides for excellent water suppression.

The corresponding cross-peaks in the HNCO(α/β-C’C^α-J)-TROSY experiment, after addition and subtraction, appear at ω_C’(i-1) + πJ_C’Cα, ω_N(i) - πJ_NH, ω_HN(i) + πJ_NH and ω_C’(i-1) – πJ_C’Cα, ω_N(i) - πJ_NH, ω_HN(i) + πJ_NH, respectively. Therefore, ¹J_C’Cα couplings can be measured analogously to the HNCO(α/β-C’C^α-J) experiment, but from the most slowly relaxing ¹⁵N-¹H multiplet component. A three-dimensional HNCO(α/β-C’C^α-J)-TROSY spectrum recorded from the 30.4 kDa (286 amino acid residues), uniformly ¹⁵N/¹³C and 80% ²H-labeled protein, E2, is shown in Figure 14. The in- and antiphase data sets were recorded in an interleaved manner over 18 hours at a 600 MHz ¹H frequency.

5.2.3. Access to ¹J_NCα, ²J_H^NCα, ²J_NCα, and ³J_H^NCα

Determination of ¹J_NCα and ²J_NCα from proteins is a difficult task because both intra- and interresidual couplings between the amide nitrogen and ¹³C^α are comparable, and rather small in magnitude. The intraresidual coupling varies between 7 and 12 Hz, whereas interresidual ²J_NCα shows a discrepancy between 4 and 9 Hz (Bystrov 1976). If a ¹⁵N-HSQC spectrum without ¹³C^α decoupling during the t₁ evolution period (Delaglio et al. 1991) is recorded, the cross-peaks split into a doublet of doublets in the ¹⁵N-dimension. This is due to the coupling of ¹⁵N(i) to both the ¹³C^α(i) and ¹³C^α(i-1) spins. Analogously to the case of ¹J_NC’, it is necessary to decouple the ¹³C’ spins from ¹⁵N during t₁ in order to maintain the simplified multiplet structure. In addition, if the spin-state of ¹³C^α is not perturbed, the ¹⁵N, ¹H cross-peaks show a tilted E.COSY pattern because both the intra- and interresidual ¹³C^αs act as passive spins during the t₁ and t₂ periods. In practice, a triplet-like cross-peak is found due to the overlapping α- and β-states of the corresponding doublet of doublets. Even in small proteins, it is difficult to resolve the two center lines of the multiplet. Recording the corresponding TROSY spectrum, with the most slowly relaxing multiplet component, provides narrower line widths, but adequate separation of the multiplet components is still difficult in most cases, as can be seen in Figure 15, illustrating the ¹⁵N, ¹H cross-peak from Lys63 in ubiquitin (IV).

One possibility is to make use of accordion-style J-multiplication, as illustrated in Figure 16. The pulse sequence is a simple modification of the {¹³C^α}-¹⁵N-TROSY. In this case, the ¹⁵N chemical shift is recorded during the t₁ period, whereas the coupling between ¹³C^α and ¹⁵N evolves simultaneously for κ*t₁, where κ is the multiplication coefficient. This approach necessitates, however, that the ¹⁵N spin relaxes at a favorable rate due to the long period of time during which the ¹⁵N spin is in the transverse plane.

It is obvious that the multiplet pattern could be simplified if one of the ¹³C^α spins coupled to the ¹⁵N spin could be selectively decoupled. This approach is not suitable for proteins due to numerous ¹³C^α spins, therefore, an alternative method is required to obtain a simplified multiplet pattern. We have used the HN(α/β-NC^α-J)-TROSY experiment (IV; Figure 17), allowing exploitation of spin-state-selective filtering sensitive to the ¹³C^α(i-1) spin-state. The pulse sequence in Figure 17 has been modified similar to the one presented in Figure 13B to obtain higher sensitivity.

Initially, magnetization is transferred from the ¹H^N spin to ¹³C’, as in the TROSY-type HNCO scheme. The subsequent spin-state-selective filter element is employed to edit the ¹³C^α(i-1) spin-state in order to create 4H^N_zN_zC’_z coherence in the in-phase experiment and 8H^N_zN_zC’_zC^α_z coherence in the corresponding antiphase experiment (time point a). During the following ¹³C’-¹⁵N back-INEPT step, which is implemented as a semi-constant time, the ¹⁵N chemical shift evolves simultaneously with couplings to ¹³C^α(i) and ¹³C^α(i-1) on the most slowly relaxing ¹⁵N-¹H multiplet component. Post-acquisitional addition and subtraction yields two subspectra, in which the centers of visible doublets are separated by ²J_NCα. The individual doublet components in either subspectrum are separated by ¹J_NCα. Figure 18 clearly illustrates this spectral simplification, showing previously overlapping doublet components overlaid with thick and thin contours. Two additional couplings, namely ³J_H^NCα and ²J_H^NCα, are readily available from this HN(α/β-NC^α-J) experiment in the ¹H dimension. A familiar E.COSY pattern emerges since the ¹³C^α spin acts as a common passive spin to both the ¹⁵N and ¹H^N spins during the t₁ and acquisition periods, respectively. Thus, ³J_H^NCα can be determined from cross-peak displacement in the F₁-dimension between two subspectra, whereas ²J_H^NCα can be measured from the tilt between doublet components in the F₂-dimension within each subspectrum. Analogously to the former HN(α/β-C’C^α-J) schemes, to preserve the purely absorptive line shape in F₁ and F₂, an additional 90° pulse on ¹³C’ is applied prior to the acquisition period.

The HN(α/β-NC^α-J)-TROSY scheme can be modified to obtain the three-dimensional HNCO(α/β-NC^α-J)-TROSY experiment (Permi et al. 2000), which provides improved resolution necessary for highly α-helical proteins. In this case, the ¹³C’ chemical shift is recorded during the t₁ period. The α/β-filter is inserted after the t₁ evolution period and for the in-phase (antiphase) experiment, the 4H^N_zN_zC’_z (8H^N_zN_zC’_zC^α_z) coherence is created before the t₂ evolution period, which is again implemented in a semi-constant-time TROSY manner. However, due to the relatively small size of the ¹J_NCα coupling, an additional spin-echo period must be inserted for the ¹⁵N-¹³C^α coupling evolution to reduce the number of increments needed in the F₂-dimension. Hence, the ¹J_NCα and ²J_NCα couplings scale up by 1+λ (λ ≥ 0) with respect to the ¹⁵N chemical shift. After addition and subtraction of the in- and antiphase data sets and their corresponding quadrature counterparts, cross-peaks are found at ω_C’(i-1), ω_N(i) - πJ_NH + (1+λ)π²J_NCα ± (1+λ)π¹J_NCα, ω_HN(i) + πJ_NH + π³J_H^NCα ± π²J_H^NCα and ω_C’(i-1), ω_N(i) - πJ_NH - (1+λ)π²J_NCα ± (1+λ)π¹J_NCα, ω_HN(i) + πJ_NH - π³J_H^NCα ± π²J_H^NCα, for the two- and three-dimensional subspectra, respectively.

5.2.4. Insight to side-chains, ¹J_Cα_Cβ

Up to this point, we have obtained a wealth of data from which the orientation of several internuclear vectors in the protein backbone can be derived. It is obvious that it is crucial to gain information on the orientation of side-chains, as well. A logical approach is to measure the one-bond couplings between the ¹³C^α and ¹³C^β spins. This is readily available from a ¹³C-HSQC spectrum. However, several reasons limit the applicability of a ¹³C-HSQC method to larger proteins. First, the high transverse relaxation rates of the ¹³C^α and ¹H^α spins dramatically reduce the intensity and resolution of the ¹³C-HSQC spectrum. Second, the ¹H^α spins resonate close to the water signal, which hampers the interpretation of data. Third, if perdeuterated samples are used, it is clear that this method cannot be used. We can alleviate these problems with a HNCO -based TROSY-type experiment, as shown in Figure 19 (Permi et al. 2000).

Initially, magnetization is transferred from the ¹H^N spin to the preceding ¹³C^α spin. Depending on the sample, we can use two different schemes to record ¹J_Cα_Cβ. Consider first a protonated sample. During the following t₁ evolution period, ¹J_Cα_Cβ evolution is allowed to take place, while the large ¹J_Cα_Hα and the chemical shift of ¹³C^α are refocused by inserting a 180° (¹³C^α/C^β) pulse in the middle of λt₁. Instead of ¹³C^α, we record the ¹³C’ chemical shift during the following ¹J_C’Cα refocusing delay. This period is implemented in a semi-constant time manner in order to obtain sufficient resolution in the F₁-dimension. Eventually, the ¹⁵N chemical shift is incremented during the t₂ evolution period by using the semi-constant-time TROSY scheme (IV; Permi et al. 2000). It should be noted that, by using this approach, the 180° (¹H) pulse during the t₁ time for ¹J_Cα_Hα refocusing can be omitted. As this pulse would interchange the fast and slow relaxing ¹⁵N-¹H multiplet components, the presented scheme is preferred. On the other hand, the ¹³C^α chemical shift can be recorded without mixing the ¹⁵N-¹H spin-states if we apply during t₁ a semi-selective ¹H^α decoupling or, alternatively, a spin-locking. Finally, the ¹³C’, ¹⁵N, ¹H^N correlation map results, where the multiplets in the F₁-dimension are resolved by λ* ¹J_Cα_Cβ, as illustrated in Figure 20 (Permi et al. 2000).

For perdeuterated samples, a scheme in which the ¹³C^α chemical shift is recorded simultaneously with ¹J_Cα_Cβ, is preferred. In this case, it is also possible to use J-scaling by inserting an additional spin-echo period for the ¹J_Cα_Cβ evolution. As the relaxation rate of the ¹³C^α spin is rather long in perdeuterated samples, this implementation improves the accuracy in data analysis, thanks to the properly resolved coupling. Additionally, since the measured coupling is divided by 1+λ, more precise values with respect to random measurement error can be obtained. Subsequently, magnetization is transferred back to ¹⁵N via the ¹³C’ spin, and the ¹⁵N chemical shift is recorded during the SCT TROSY period.