Effective strategy to assign ¹H-¹⁵N heteronuclear correlation NMR signals from lysine side-chain NH₃⁺ groups of proteins at low temperature

Abstract

Recent studies have shown that lysine side-chain NH₃⁺ groups are excellent probes for NMR investigations of dynamics involving hydrogen bonds and ion pairs relevant to protein function. However, due to rapid hydrogen exchange, observation of ¹H-¹⁵N NMR cross peaks from lysine NH₃⁺ groups often requires use of a relatively low temperature, which renders difficulty in resonance assignment. Here we present an effective strategy to assign ¹H and ¹⁵N resonances of NH₃⁺ groups at low temperatures. This strategy involves two new ¹H/¹³C/¹⁵N triple-resonance experiments for lysine side chains. Application to a protein-DNA complex is demonstrated.

Hydrogen bonds and ion pairs involving protein side chains are of fundamental importance in protein functions such as molecular recognition and catalysis. Recently it was demonstrated that lysine (Lys) side-chain NH₃⁺ groups are very useful probes for NMR studies of protein dynamics involving hydrogen bonds and ion pairs (Anderson et al. 2013; Esadze et al. 2011; Zandarashvili et al. 2013; Zandarashvili et al. 2011). Heteronuclear ¹H-¹⁵N cross peaks from Lys NH₃⁺ groups are also useful for identifying molecular interfaces between protein and ligand (Blaum et al. 2010; Poon et al. 2006). Resonance assignment for Lys NH₃⁺ groups typically requires not only standard experiments such as (H)C(CO)NH, H(CCO)NH, and HCCH-TOCSY, which correlate side-chain and backbone resonances, but also Lys-selective NMR experiments (Andre et al. 2007; Iwahara et al. 2007). This assignment process is not trivial because rapid hydrogen exchange renders difficulty in ¹H detection of NH₃⁺ groups. To mitigate this problem, observation of ¹H signals from Lys NH₃⁺ groups often requires the use of relatively low pH and temperature. However, the use of a low temperature can substantially reduce sensitivity of most triple-resonance experiments due to a longer molecular rotational correlation time, and thereby hamper resonance assignment.

Resonance assignment for Lys side-chain NH₃⁺ groups requires long-range correlation spectra because these amino groups are located at the fifth positions (i.e., ζ position) from Cα atoms. Short-range correlations between NH₃⁺ and ¹Hε/¹³Cε resonances are typically insufficient for unambiguous assignment due to poor dispersion of their chemical shifts. While long-range correlation NMR methods such as the 3D (H)CCENH3 (Iwahara et al. 2007) may in principle allow for assignment of Lys NH₃⁺ resonances, this approach can be impractical at low temperatures due to rapid ¹³C transverse relaxation. In fact, it is well known that rapid ¹³C transverse relaxation can diminish the efficiency of ¹³C-¹³C Hartmann-Hahn cross polarization if a molecular rotational correlation time τ_r is relatively long (Fischer et al. 1996).

In this paper, we present two distinct approaches to assign Lys side-chain NH₃⁺ groups at a low temperature. One of the approaches is to obtain long-range ¹⁵N-¹³C correlations by taking advantage of very slow ¹⁵N transverse relaxation of NH₃⁺ groups. Due to bond rotations along the symmetry axis and cross correlation between three ¹⁵N-¹H dipole-dipole interactions, NMR relaxation of the in-phase single-quantum term N_x for NH₃⁺ groups is extremely slow and occurs in a bi-exponential manner (Esadze et al. 2011; Zandarashvili et al. 2013). In our previous studies, initial transverse relaxation rates R_2,ini for NH₃⁺ groups were in the range between 0.9 and 4.5 s⁻¹ for the systems with τ_r = 8.5 and 10.8 ns (Anderson et al. 2013; Esadze et al. 2011); the average R_2,ini for the Lys side-chain NH₃⁺ groups that directly interact with DNA phosphate groups in the HoxD9 homeodomain-DNA complex was 1.7 s⁻¹ (Anderson et al. 2013), which corresponds to a transverse relaxation time of ~580 ms. Therefore, even a time over one hundred milliseconds can readily be used for¹⁵N-¹³C coherence transfer via J-evoluation between N_x and 2N_yC_z terms for a relatively small scalar coupling. This coherence transfer is substantially more efficient than coherence transfer via J-evolution from C_x to 2C_yN_z, because the ¹⁵N transverse relaxation of NH₃⁺ groups is far slower than relevant ¹³C transverse relaxation of CH₂ groups.

From this perspective, we have developed the 3D H3NCG experiment (Figure 1), which provides a long-range correlation between ¹H/¹⁵N resonances of the NH₃⁺ groups and ¹³Cγ resonance for Lys side chains. This 3D experiment relies on relatively small scalar coupling constants ³J_NC between the ¹⁵Nζ and ¹³Cγ nuclei. Absolute values of the ³J_NC constants are in a range of 1 – 3 Hz, and depend on Lys χ₄ torsion angles, which can undergo dynamic transitions between gauch and trans conformers (Huang and MacKerell 2013; Zandarashvili et al. 2011). Owing to the slow ¹⁵N transverse relaxation of NH₃⁺ groups during the coherence transfers between the N_x and 2N_yC_z terms, the magnetizations of interest survive a total of 200 ms for the evolution of the small ³J_NC couplings between the ¹⁵Nζ and ¹³Cγ nuclei in the 3D H3NCG experiment. Because ¹⁵N chemical shifts of lysine side-chain NH₃⁺ groups are unique (~33ppm), ¹⁵N r-SNOB pulses (Kupče et al. 1995) selective to ¹⁵Nζ are used to suppress potentially aliased signals from arginine and backbone ¹⁵N nuclei. In this experiment, undesirable coherence transfer between ¹⁵Nζ and ¹³Cε nuclei is avoided by using ¹³C I-BURP2 pulses (Geen and Freeman 1991) that invert ¹³Cγ (~25 ppm) but do not affect ¹³Cε (~42 ppm) nuclei. Although these I-BURP2 pulses also invert ¹³Cδ (~28 ppm), ²J_NCδ coupling constants are undetectably small in lysine side chains (Zandarashvili et al. 2011). Thus, these ¹⁵N and ¹³C shaped pulses permit selective observation of ¹Hζ/¹⁵Nζ/¹³Cγ cross peaks from lysine side chains. Using the 3D H3NCECD (Iwahara et al. 2007) and 3D H3NCG spectra, three resonances of ¹³Cγ, ¹³Cδ, and ¹³Cε nuclei can be obtained for each lysine side-chain NH₃⁺ group. Because the ¹³C-¹⁵N coherence transfer in the H3NCG and H3NCECD experiments is not exposed to rapid ¹³C transverse relaxation, these experiments are more sensitive than the (H)CCENH3 experiment that involves coherence transfer via J-evolution from C_x to 2C_yN_z. It should also be noted that the H3NCG and H3NCECD experiments with ²H decoupling can be readily performed for perdeuterated proteins. Although deuteration should substantially slow ¹³C transverse relaxation and therefore improve ¹³C-¹³C coherence transfer (Yamazaki et al. 1994), the lack of side-chain ¹H nuclei in perdeuterated proteins makes the (H)CCENH3 experiment impractical.

Figure 1.

Pulse sequences for the 3D H3NCG experiment for Lys side-chain NH₃⁺ resonance assignment. Thin and bold bars in black represent hard rectangular 90° and 180° pulses, respectively. Unless indicated otherwise, pulse phases are along x. Carrier positions: ¹H, the position of the water resonance; ¹⁵N, 33 ppm; and ¹³C, 20 ppm. Short-bold bars represent water-selective soft-rectangular ¹H 90° pulses (1.2 ms). The 3-9-19 pulse train-based WATERGATE scheme (Piotto et al. 1992) was used to suppress the water signal. The ¹³Cγ-selective pulses should not affect ¹³Cε nuclei (~42ppm). RF strengths for ¹H and ¹⁵N WALTZ-16 composite pulses (Shaka et al. 1983) were 3.3 kHz and 1.0 kHz, respectively. Shaped pulses: ¹H half-Gaussian 90° pulse (2.1 ms); ¹³C I-BURP2 180° pulse (1.2 ms); and ¹⁵N r-SNOB 180° pulse (1.03 ms). Delays: τ_a = 2.7 ms; δ = 2.6 ms; and T = 50 ms. Phase cycles: ϕ₁ = [2x, 2(−x)], ϕ₂ = [x, −x], ϕ₃ = [4x, 4y], and receiver = [x, −x,−x, x, −x, x, x, −x]. Quadrature detections for indirect ¹³C and ¹⁵N dimensions were achieved using States-TPPI (Marion et al. 1989) for ϕ₁ and ϕ₂, respectively.

Using a 750-MHz spectrometer, we recorded 3D H3NCG and 3D H3NCECD spectra for the Egr-1 – DNA complex at 10 °C. This complex contains the Egr-1 zinc-finger protein (90 amino acids) and 12-bp DNA with Egr-1’s target sequence. Our group uses the complex to study dynamics in DNA scanning and recognition by Egr-1 (Esadze et al. 2014; Zandarashvili et al. 2012). The Egr-1 zinc-finger protein and 12-bp DNA were purified as previously described (Esadze and Iwahara 2014; Takayama et al. 2010). A 280-μl solution of 0.8 mM complex in a buffer of 20 mM potassium succinate (pH 5.8), 2 mM ZnCl₂, and 20 mM KCl was sealed into an inner tube of co-axial NMR tube. To avoid deuterated species of NH₃⁺ groups, D₂O for NMR lock was separately sealed in a thin outer layer of the co-axial NMR tube (Iwahara et al. 2007). The relatively low temperature, 10 °C, was required to observed ¹H-¹⁵N HISQC (Iwahara et al. 2007) cross peaks from all 6 Lys side-chain NH₃⁺ groups of the Egr-1 – DNA complex under these solution conditions. 3D (H)CCENH3 spectrum (Iwahara et al. 2007) recorded at this temperature did not show signals from any Lys residues other than Lys89, which is located at the disordered C-terminal tail (data not shown). Because the (H)CCENH3 experiment involves the ¹³C-¹⁵N coherence transfer schemes via J-evolution from C_x to 2C_yN_z terms and the ¹³C-¹³C cross polarization scheme, rapid ¹³C transverse relaxation can severely reduce sensitivity at low temperatures. In contrast, the 3D H3NCECD and 3D H3NCG spectra clearly show signals from all Lys side-chain NH₃⁺ groups (Figure 2). As described above, higher sensitivity of these experiments can be attributed largely to efficient ¹³C-¹⁵N coherence transfer via J-evolution from N_x to 2N_yC_z terms.

Figure 2.

Strips of the 3D H3NCG and H3CECD spectra for all Lys side-chain NH₃⁺ groups of the Egr-1 – DNA complex at 10 °C. Negative contours are shown in green. The ¹H and ¹⁵N spectral widths for were 18.0 ppm and 4.7 ppm, respectively, in both experiments. The ¹³C spectral widths were 20.7 ppm in H3NCG and 26.0 ppm in H3NCECD. In each experiment, 32 scans were accumulated per FID, and numbers of complex points for ¹H, ¹³C, and ¹⁵N dimensions were 810, 32, and 32, respectively. The total experimental time was 60 hours each. These spectra were recorded at the ¹H frequency of 750 MHz using a Bruker Avance III spectrometer equipped with a non-cryogenic TXI probe.

Our second approach to assign lysine side-chain NH₃⁺ resonances at low temperatures is to observe ¹⁵Nζ resonances through ¹H detection of lysine side-chain CH₂ groups at higher temperatures. In this approach, we use a new 2D heteronuclear long-range correlation spectrum, (H2C)N(CC)H-TOCSY, which shows cross peaks of lysine side-chain ¹⁵Nζ and CH₂ proton resonances. Figure 3 illustrates the pulse sequence for this experiment. Owing to the detection of carbon-attached ¹H nuclei, this experiment is not hindered by rapid hydrogen exchange of NH₃⁺ groups, and therefore can be performed using a high temperature at which ¹³C-¹³C Hartmann-Hahn cross polarization can occur efficiently. As described previously (Iwahara et al. 2007; Andre et al. 2007), ¹⁵N transverse magnetization is maintained to be in-phase with respect to ¹H via ¹H-WALTZ-16 decoupling so that scalar relaxation due to hydrogen exchange is avoided. Although this experiment can be implemented as a 3D experiment with a ¹³C dimension, we used the 2D version to achieve a higher resolution in the ¹⁵N dimension as well as a higher sensitivity. Figure 4 shows the 2D (H2C)N(CC)H-TOCSY spectrum recorded for the Egr-1 – DNA complex. While rapid hydrogen exchange makes it impossible to detect ¹H signals from the Lys NH₃⁺ groups of the Egr-1 – DNA complex at 35 °C, the (H2C)N(CC)H-TOCSY spectrum for the same sample at this temperature exhibited many ¹H-¹⁵N cross peaks from Lys side chains (Figure 4C). In this spectrum, there was significant variation in signal intensity, presumably because individual Lys side chains exhibit different degrees of relaxation loss, depending on fast internal motions or slow conformational exchange (Foster et al. 1997). Because multiple side-chain ¹H resonances are observed at high resolution along the ¹H direct-detection dimension with excellent separation by sharp ¹⁵Nζ resonances, the (H2C)N(CC)H-TOCSY spectrum greatly facilitates ¹⁵Nζ resonances assignment for lysine NH₃⁺ groups. The temperature dependence of the H2(C)N spectra (Andre et al. 2007) allows us to track the change of ¹⁵Nζ resonances as a function of temperature (Figure 4B). These data together with the 3D H3NCG and H3NCECD spectra allowed us to unambiguously assign ¹H-¹⁵N HISQC signals from the Lys NH₃⁺ groups of the Egr-1 – DNA complex at 10 °C (Figure 4A).

Figure 3.

The pulse sequence of the 2D (H2C)N(CC)H-TOCSY experiment. This experiment provides correlation between Lys side-chain ¹⁵Nζ and ¹H resonances, and corresponds to a hybrid of the H2CN (Andre et al. 2007) and HCCH-TOCSY (Kay et al. 1993) experiments for H₂O samples. ¹³C shaped pulses are Lys ¹³Cε-selective r-SNOB pulses (980 us). Length of the ¹³C- spin lock (SL) was 1 ms. ¹³C carrier position was 43 ppm. The rf strength of the DIPSI-3 scheme (the total length, 16.9 ms) was 9.6 kHz and that of hard ¹³C pulses was 20.8 kHz. ¹⁵N shaped pulses are Lys ¹⁵Nζ-selective r-SNOB pulses (1.0 ms). ¹⁵N carrier position was 32 ppm. Delays: τ₁ = 1.6 ms; τ₂ = 1.1 ms; δ = 1.7 ms; and T_c = 20 ms. Phase cycles: ϕ₁ = [x, −x]; ϕ₂ = [2x, 2(−x)], ϕ₃ = [4x, 4y]; ϕ₄ = [4x, 4(−x)]; and receiver = [x, −x, −x, x, −y, y, y, −y]. Quadrature detections for indirect ¹⁵N dimensions was achieved using States-TPPI for ϕ₁.

Figure 4.

Use of 2D (H2C)N(CCH)-TOCSY spectrum and temperature dependence of 2D H2(C)N spectra for assignment of Lys NH₃⁺ groups. The spectra were recorded for 0.8 mM ¹³C/¹⁵N-labeled Egr-1 – DNA complex at pH 5.8. (A) The lysine NH₃⁺-selective HISQC (Iwahara et al. 2007) spectrum recorded at 10 °C. (B) Lysine side-chain-specific H2(C)N (Andre et al. 2007) spectra recorded at 10 °C (black), 18 °C (blue), 27 °C (magenta), and 35 °C (red). (C) (H2C)N(CC)H-TOCSY spectrum recorded at 35 °C. The spectra widths and numbers of complex points: 13.4 ppm and 400 points for ¹H; and 2.2 ppm and 50 points for ¹⁵N. The mixing time of the ¹³C DIPSI-3 scheme (Shaka et al. 1988) was 16.9 ms. 1024 scans were accumulated per FID. The total titme to record the 2D (H2C)N(CC)H-TOCSY spectrum was 56 hours. All spectra shown in this figure were recorded with a Bruker Avance III spectrometer equipped with a QCI cryogenic probe operated at the ¹H frequency of 600 MHz.

In conclusion, we have presented an effective strategy for resonance assignment of lysine NH₃⁺ groups that can be observed only at relatively low temperatures. Our new approaches can help expand applicability of the NMR methods for characterizing Lys side-chain NH₃⁺ groups (Anderson et al. 2013; Andre et al. 2007; Esadze et al. 2011; Iwahara et al. 2007; Segawa et al. 2008; Tomlinson et al. 2009; Williamson et al. 2013; Zandarashvili et al. 2013; Zandarashvili et al. 2011). As a consequence, dynamics of hydrogen bonds and ion pairs involving Lys side chains may become more assessable for a larger number of proteins and protein complexes.