Determination of Histidine Protonation States in Proteins by Fast Magic Angle Spinning NMR.

Histidine residues play important structural and functional roles in proteins, such as serving as metal-binding ligands, mediating enzyme catalysis, and modulating proton channel activity. Many of these activities are modulated by the ionization state of the imidazole ring. Here we present a fast MAS NMR approach for the determination of protonation and tautomeric states of His at frequencies of 40-62 kHz. The experiments combine 1H detection with selective magnetization inversion techniques and transferred echo double resonance (TEDOR)-based filters, in 2D heteronuclear correlation experiments. We illustrate this approach using microcrystalline assemblies of HIV-1 CACTD-SP1 protein.

Introduction

Histidines (His) play important structural and functional roles in proteins such as metal binding (Stryer et al., 1964; Perutz and Mathews, 1966; Adams et al., 1969; Liljas et al., 1972), proton transfer (Hoffee et al., 1967; Blow et al., 1969; Campbell et al., 1974), and stability (Perutz et al., 1969; Lewis et al., 1976; Loewenthal et al., 1992). These functions are often correlated with the ionization state of the histidine sidechain (Figure 1A) (Bachovchin and Roberts, 1978; Kossiakoff and Spencer, 1981; Lewis et al., 1981). While the pKa of the imidazole ring for free histidine is 6.5 (Blomberg et al., 1977), in proteins the pKa values vary widely, from 3 to 9, depending on the interactions with neighboring residues and degree of burial (Zhou et al., 1993; Plesniak et al., 1996). At pH values above the pKa, anionic τ and π tautomers with hydrogens at either Nε2 or Nδ1 are present, while below the pKa the protonated imidazole ring possesses hydrogens at both Nε2 and Nδ1. For a protein at intermediate pH values, it is possible that a fraction of His residues is protonated and the remaining fraction unprotonated (French and Hammes, 1965; Edwards and Sykes, 1980; Hass et al., 2008).

FIGURE 1

(A) Four states of histidine: left to right, charged state, neutral τ tautomer, anionic τ tautomer, and anionic π tautomer. (B) Pulse sequence for the 1H-detected TEDOR-based 15N selective filtered experiment. Tr is the MAS rotor period, τmix is the total TEDOR mixing time. The phase on the individual pulses are: φ1 = 16 × (0) 16 × (2), φ2 = 1, φ3 = 0,φ4 = 0, φ5 = 0213 2031, φ6 = 2, φ7 = 0, φ8 = 02, φ9 = 1133, φ10 = 4 × (0) 4 × (1) 4 × (2) 4 × (3), φ11 = 4 × (1) 4 × (0), φ12 = 4 × (1) 4 × (0) 4 × (1) 4 × (0) 4 × (3) 4 × (2) 4 × (3) 4 × (2), φrec = 3113 0220 1331 2002, where 0 = x, 1 = y, 2 = –x, and 3 = –y. Δ is set to one rotor period during which 1H rf field of ωr amplitude is applied for effective Z-filtering. MISSISSIPPI water suppression sequence is applied during Δ′ time period. (C) Synthetic 1H-detected TEDOR-based 15N selective filtered CH HETCOR spectra showing cross peaks expected for each tautomer. Left to right, soft pulse turned off, soft pulse at 170 ppm, soft pulse at 250 ppm. The filtering patterns for neutral and anionic τ tautomers are identical. (D) 15N (top) and 13C (bottom) CPMAS NMR spectra of crystalline histidine. (E) Aromatic region expansion of 2D NCA spectrum of crystalline histidine. (F) 1D 13C spectra using TEDOR-based 15N selective filtering in the aromatic region. Top to bottom, soft pulse turned off; soft pulse at 250 ppm; soft pulse at 170 ppm. (G) Three complementary 1H-detected TEDOR-based 15N selective filtered CH HETCOR spectra. Left to right, soft pulse turned off, soft pulse at 170 ppm, soft pulse at 250 ppm. The MAS frequency was 60 kHz in all experiments. Signals of charged state are shown in purple, neutral τ tautomer – in magenta, anionic τ tautomer - in grey, and anionic π tautomer – in teal.

Methods to determine His ionization states in proteins are solution NMR (Kilmartin et al., 1973; Markley, 1975; Bachovchin and Roberts, 1978; Perutz et al., 1985; Pelton et al., 1993; Shimba et al., 1998; Hass et al., 2008; Hansen and Kay, 2014) or neutron diffraction (Kossiakoff and Spencer, 1980; Maeda et al., 2004; Kovalevsky et al., 2010), with the latter limited to very large single crystals and requiring a neutron source, both difficult conditions to meet routinely. Therefore, solid-state magic angle spinning (MAS) NMR constitutes a viable alternative (Wei et al., 1999). Similar to solution NMR, the tautomeric state of histidines can be unambiguously determined from a unique combination of 15N sidechain chemical shifts (Munowitz et al., 1982; Wei et al., 1999; Miao et al., 2014) and the corresponding N-H distances can be estimated, allowing for hydrogen bonding studies (Shenderovich et al., 2015). Protonation states for the crystalline histidine amino acid have been determined by MAS NMR for different pH values (Li and Hong, 2011) and crystalline short peptides (Platzer et al., 2014). Using 15N selective filtered, 13C-detected experiments with the inversion pulses at frequencies of the different tautomers (Miao et al., 2014) permits their identification. For proteins containing several histidine residues, the above experiments are challenging due to low sensitivity and spectral overlap. Therefore, only a handful of such studies have been reported to date (Hu et al., 2006; Hu et al., 2010; Miao et al., 2015; Kwon et al., 2019; Maciejko et al., 2019; Vasa et al., 2019; Movellan et al., 2020). In order to increase resolution, the original pulse sequence can be reconfigured as a 2D experiment by introducing a 13C-13C mixing period based on proton-driven spin diffusion (PDSD) (Bloembergen, 1949) and extending the second Z-filter (Miao et al., 2014). 2D and 3D proton-based experiments were also introduced with 1H chemical shifts either recorded in the indirect dimension (Miao et al., 2015) or detected directly (Shenderovich et al., 2015; Vasa et al., 2019; Movellan et al., 2020).

Herein, we present an alternative MAS experiment that uses 1H detected transferred-echo double resonance (TEDOR)-based 15N selectively filtered 2D correlations at fast MAS frequencies of 40–60 kHz. The advantages of the 1H-detected fast-MAS experiments presented here are: i) improved sensitivity due to 1H detection, and ii) improved resolution via the second dimension and selective recoupling of aromatic resonances directly attached to 15N atoms. Microcrystalline assemblies of U-13C,15N- and fractionally deuterated (FD) (Mance et al., 2015) 13C,15N-HIV-1 CACTD-SP1 protein samples, possessing solely a single His residue, His-226, are ideally suited for pulse sequence optimization and therefore were selected for illustrating our current approach. Extension to ultrafast MAS frequencies (up to 110 kHz), should yield even higher sensitivity and resolution for proteins with multiple histidines.

Materials and Methods

Sample Preparation

U-13C,15N-L-histidine was purchased from Cambridge Isotope Laboratories, recrystallized from an aqueous solution at pH 6.0, adjusted by mixing HCl and NaOH. The sample was packed into a 1.3 mm MAS rotor. Microcrystalline assemblies of U-13C,15N- and FD-13C,15N-HIV-1 CACTD-SP1 were prepared in the presence of the assembly cofactor inositol hexakisphosphate (IP6) as described previously (Wagner et al., 2016) except for growing Escherichia coli in M9 medium containing 13C glucose, 15N NH4Cl, isotopically labeled precursors, and (for the deuterated sample) D2O. Proteins were assembled with 1.6 mM IP6 (Sigma-Aldrich), for a final reaction volume of 1 ml at pH 8.0. Assemblies were incubated overnight at 20°C and packed into 3.2 mm (U-13C,15N), 1.9 mm (FD-13C,15N), or 1.3 mm MAS rotors (U-13C,15N).

MAS NMR Spectroscopy

MAS NMR experiments on U-13C,15N-CACTD-SP1 and FD-13C,15N-CACTD-SP1 microcrystalline assemblies were performed on a 20.0 T Bruker AVIII spectrometer outfitted with 3.2 mm E-Free HCN and 1.9 HCN probes, respectively. The MAS frequency was 14 and 40 kHz, respectively, controlled to within ± 10 Hz by a Bruker MAS controller. The actual sample temperature was maintained at 4 ± 1°C throughout the experiments using the Bruker temperature controller.

The Larmor frequencies were 850.4 MHz (1H), 213.9 MHz (13C) and 86.2 MHz (15N). The typical 90° pulse lengths were 2.6–3.0 μs for 1H, 4.3–4.5 μs for 13C, and 4.2–4.7 μs for 15N. The 1H-13C and 1H-15N cross-polarization employed a linear amplitude ramp of 90–110% on 1H, and the center of the ramp was matched to a Hartmann–Hahn condition at the first spinning sideband; contact times of 0.7–1.5 ms and 1.0–1.7 ms were used, respectively. 50 ms CORD (Hou et al., 2013) mixing time was applied to facilitate 13C-13C mixing.

MAS NMR experiments on U-13C,15N-L-histidine and FD-13C,15N-CACTD-SP1 microcrystalline assemblies were performed on a 14.1 T Bruker AVIII spectrometer outfitted with 1.3 mm HCN probe. Larmor frequencies were 599.8 MHz (1H), 150.8 MHz (13C), and 60.7 MHz (15N). The MAS frequency was 60 kHz, controlled to within ± 10 Hz by a Bruker MAS controller. The actual sample temperature was maintained at 40 ± 1°C throughout the experiments using the Bruker temperature controller. The typical 90° pulse lengths were 1.4–1.6 μs for 1H, 2.7–3.0 μs for 13C, and 3.3–3.6 μs for 15N. The 1H-13C and 1H-15N cross-polarization employed a linear amplitude ramp of 90–110% on 1H, center of the ramp was matched to a Hartmann–Hahn condition at the first spinning sideband, with contact times of 1.0–5.0 ms and 1.3–5.0 ms, respectively. Band-selective 15N-13C SPECIFIC-CP contact time was 5.0–6.0 ms. SWFTPPM (Vinod Chandran et al., 2008) decoupling (15 kHz) was used during the TEDOR block and acquisition periods. The selective 15N 180° r-SNOB (Kupce et al., 1995) pulse length in the Z-filtered TEDOR experiments was 500 µs and the bandwidth — 2 kHz; the rf power was 4 kHz. During the Z-filter time period Δ, 60 kHz CW decoupling was applied for τr on 1H channel, while during the time period Δ’, MISSISSIPPI (Zhou and Rienstra, 2008) water suppression was applied. The TEDOR block duration was 1–3 ms.

Data Processing

All MAS NMR data were processed using NMRPipe (Delaglio et al., 1995). The 13C and 15N chemical shifts were referenced with respect to the external standards adamantane (Morcombe and Zilm, 2003) and ammonium chloride (Bertani et al., 2014), respectively. The 2D and 3D data sets were processed by applying 30, 45, 60, and 90° shifted sine bell apodization followed by a Lorentzian-to-Gaussian transformation in both dimensions. Forward linear prediction to twice the number of the original data points was used in the indirect dimension followed by zero filling. The processed spectra were analyzed in NMRFAM-Sparky (Goddard and Kneller, 2004; Lee et al., 2015) and CCPN (Stevens et al., 2011).

Results

Here, we report on a 2D 1H-detected TEDOR-based Z-filtered experiment, which incorporates 15N selective filters for the determination of histidine tautomeric states. The pulse sequence is shown in Figure 1B. The experiment is well suited for fast MAS frequencies of 40 kHz and above. The tautomeric states of His residues are unambiguously determined using a combination of three CH HETCOR experiments comprising: i) 15N selective TEDOR filter, containing 13C resonances of all protonation and tautomeric states present; ii) 15N selective TEDOR filter with a soft pulse at 170 ppm, removing resonances of the protonated state while Cε1 and Cδ2 atoms of π tautomer and Cε1 and Cγ atoms of τ tautomer remain; and iii) 15N selective TEDOR filter with a soft pulse at 250 ppm, retaining all signals of the charged state, Cε1 and Cγ of the π tautomer as well as Cε1 and Cδ2 atoms of the τ tautomer. Cε1 of anionic tautomers is always present in TEDOR filtered spectra, but has reduced peak intensity when 15N selective pulse is applied as C-N dipolar interaction with the non-selectively irradiated nitrogen atom is recoupled. The sequence was first tested on a crystalline L-histidine sample prepared at pH 6.0. The 13C and 15N 1D CPMAS and 2D NCA spectra are shown in Figures 1D, E, respectively. The spectra clearly indicate the presence of two forms of L-histidine, the charged monohydrate and the τ tautomer, in approximately 2:1 ratio. As shown in Figure 1F, conventional 13C-detected TEDOR-based experiments are well suited for the determination of protonation states in this sample. To test the 1H-detected sequences proposed herein, three complementary experiments were performed. As shown in Figure 1G, 15N selective TEDOR-filtered CH HETCOR without or with a soft pulse at 250 ppm (left and right panels, respectively) yield the sidechain signals of both protonation states, while 15N selective TEDOR-filtered CH HETCOR with soft pulse at 170 ppm retains only Cε1 resonance of the τ tautomer (chemical shifts provided in Supplementary Table S1). Water suppression was incorporated into the second Z-filter, allowing to record spectra on hydrated samples.

HIV-1 CACTD-SP1 (Figure 2A) contains a single His residue, His-226. The outstanding high spectral resolution in the microcrystalline FD-13C,15N-CACTD-SP1 sample allows for the determination of histidine protonation and tautomeric states even in the 13C-detected mode (Figure 2B). The Cε1 and Cγ resonances are present in 1D experiments, while the Cδ2 resonance is absent in the 15N selective TEDOR-filtered 13C CPMAS experiment with the soft pulse at 170 ppm since its magnetization does not build up during the TEDOR block due to the very weak dipolar coupling to Nδ1 (chemical shifts provided in Supplementary Table S2). The 2D 13C-13C CORD spectrum clearly shows a single set of resonances, indicating the presence of only one histidine species (Figure 2C), although the protonation and tautomeric state cannot be determined without additional experiments. The three complementary 1H-detected TEDOR-based 15N selective CH HETCOR spectra (Figure 2D) also indicate the presence of a single species, which is unambiguously assigned as τ tautomer. These 1H-detected 2D spectra contain no resonances of aromatic residues other than His (shown in black in the CH HETCOR spectrum) and Trp (these are weak or absent in the spectra of the deuterated sample), as only carbons attached to nitrogens are selected, making assignment of histidine resonances straightforward.

FIGURE 2

(A) A hexameric unit of HIV-1 CACTD-SP1 in the microcrystalline assembly (PDB 5I4T) shown as side view (left) and top view (right). (B) 1D 13C MAS NMR spectra of FD-13C,15N-CACTD-SP1 with TEDOR-based 15N selective filtering in the aromatic region. Top to bottom: CPMAS spectrum; TEDOR-based 15N selectively filtered spectra with soft pulse turned off, soft pulse at 170 ppm, and soft pulse at 250 ppm. (C) 2D CORD spectrum of FD-13C,15N-CACTD-SP1 (MAS frequency 14 kHz). (D) Aromatic regions of 1H-detected TEDOR-based 15N selective filtered CH HETCOR spectra in FD-13C,15N-CACTD-SP1: TEDOR filter and soft pulse turned off (top left), soft pulse turned off (top right), soft pulse at 170 ppm (bottom left), soft pulse at 250 ppm (bottom right). The MAS frequency was 40 kHz in all experiments, unless indicated otherwise. Signals of τ tautomer are shown in magenta.

In contrast to the FD-13C,15N-CACTD-SP1, the His-226 protonation state in U-13C,15N-CACTD-SP1 assemblies cannot be easily determined using the 1D 13C-detected version of TEDOR-based 15N selective filtered experiments due to low resolution and spectral overlap (Figure 3A). In contrast, the 2D 1H-detected TEDOR-based 15N selective filtered spectra (Figure 3B) suggest the presence of a small fraction of π tautomer along with the predominant τ tautomer in this sample (chemical shifts provided in Supplementary Table S2).

FIGURE 3

(A) 1D 15N CPMAS and 13C MAS NMR spectra of U-13C,15N-CACTD-SP1 with TEDOR-based 15N selective filtering in the aromatic region (MAS frequency 14 kHz). Top to bottom: 15N CPMAS spectrum; 13C CPMAS spectrum; TEDOR-based 15N selectively filtered spectra with soft pulse turned off, soft pulse at 170 ppm, soft pulse at 250 ppm, and a reference (S 0) experiment. (B) Aromatic regions of 1H-detected TEDOR-based 15N selective filtered CH HETCOR spectra in U-13C,15N-CACTD-SP1: TEDOR filter and soft pulse turned off (top left), soft pulse turned off (top right), soft pulse at 170 ppm (bottom left), soft pulse at 250 ppm (bottom right). The first contour was set at 5× the noise rmsd. (C) 1D 1H slices of 1H-detected TEDOR-based 15N selective filtered CH HETCOR spectra in U-13C,15N-CACTD-SP1, extracted at 13C shifts shown as gray dashed lines in panel (B). Left to right: TEDOR filter and soft pulse turned off, soft pulse turned off, soft pulse at 170 ppm, soft pulse at 250 ppm. The MAS frequency was 60 kHz. Signals of τ tautomer and π tautomer are shown in magenta and teal, respectively.

In addition to the His signals, the indole ring signals of the Trp184 residue are also present in the 1H-detected TEDOR-based experiments when the soft pulse is either turned off or centered at 250 ppm. This is expected due to the nitrogen atom Nε1 in the indole ring, which allows for magnetization build up on adjacent carbon atoms (Cδ1 and Cε2) during TEDOR transfer. Tryptophan sidechain resonances appear much stronger in non-deuterated protein assemblies compared to the FD-13C,15N -CACTD-SP1 and can be distinguished from those corresponding to the histidine based on chemical shift.

Conclusion

We demonstrated that 1H-detected 2D Z-filtered TEDOR experiments incorporating 15N selective filters permit unambiguous assignment of histidine protonation and tautomeric states in microcrystalline proteins and protein assemblies. This approach combines all the advantages of fast MAS and proton detection. Extending the experiments to MAS frequencies of 110 kHz and above can further improve the quality of data sets and allow unambiguous assignment of His protonation and tautomeric states in larger proteins and protein assemblies.