Protonation, Tautomerization, and Rotameric Structure of Histidine: A Comprehensive Study by Magic-Angle-Spinning Solid-State NMR

Abstract

Histidine structure and chemistry lie at the heart of many enzyme active sites, ion channels, and metalloproteins. While solid-state NMR spectroscopy has been used to study histidine chemical shifts, the full pH dependence of the complete panel of ¹⁵N, ¹³C, and ¹H chemical shifts and the sensitivity of these chemical shifts to tautomeric structure have not been reported. Here we use magic-angle-spinning solid-state NMR spectroscopy to determine the ¹⁵N, ¹³C, and ¹H chemical shifts of histidine from pH 4.5 to 11. Two-dimensional homonuclear and heteronuclear correlation spectra indicate that these chemical shifts depend sensitively on the protonation state and tautomeric structure. The chemical shifts of the rare π tautomer were observed for the first time, at the most basic pH used. Intra- and intermolecular hydrogen bonding between the imidazole nitrogens and the histidine backbone or water was detected, and N–H bond length measurements indicated the strength of the hydrogen bond. We also demonstrate the accurate measurement of the histidine side-chain torsion angles χ₁ and χ₂ through backbone–side chain ¹³C–¹⁵N distances; the resulting torsion angles were within 4° of the crystal structure values. These results provide a comprehensive set of benchmark values for NMR parameters of histidine over a wide pH range and should facilitate the study of functionally important histidines in proteins.

INTRODUCTION

Histidine is an essential amino acid whose side-chain pK_a (~6) is closest, among all amino acids, to the physiological pH. Thus, small changes in the environmental pH can readily change the histidine charged state. At low pH, both imidazole nitrogens are protonated to give the cationic imidazolium. Near pH 7, two neutral tautomers exist: the Nε2-protonated τ tautomer and the Nδ1-protonated π tautomer. At mildly basic pH, the backbone Nα becomes deprotonated to give an anionic histidine, whose side chain is neutral in either tautomeric state. At even higher pH or when complexed with metal ions, the imidazole can lose another proton to give an imidazolate ion.^1,2 Neutral histidine can serve as a general base and a common coordinating ligand for transition metals, while cationic histidine can serve as a general acid and hydrogen-bond (H-bond) donor. Because of its rich chemistry and pH sensitivity in the physiologically relevant range, histidine is found in the active sites of many proteins and plays key roles in enzyme catalysis,^3,4 proton conduction,^5,6 proton pumps,⁷ photosynthetic complexes,⁸ and metalloproteins.^9,10 In addition to protonation chemistry and metal coordination, the neutral imidazole of histidine can combine tautomerization with ring flips (180° χ₂ angle changes) to interconvert the protonated and unprotonated nitrogens without significantly changing the space occupied by the ring. Thus, histidine side-chain rotamerization is often important for protein function.¹¹

A number of NMR investigations of the chemical structure and dynamics of histidine in proteins have been reported. For example, de Groot and co-workers studied the interactions of histidines in the light-harvesting complex II with bacteriochlorophyll and found that Nε2 was ligated with Mg²⁺ while Nδ1 was protonated and involved in H-bonding.¹² Kay and co-workers¹³ investigated the interconversion of His61 in plastocyanin of Anabana variabilis among three tautomeric and protonated states. Cross and co-workers¹⁴ characterized the protonation state of the histidine responsible for the activation of the influenza A M2 proton channel and found the charged state of the tetrad that coincides with channel opening.

A number of solid-state NMR studies of histidine and imidazole ¹³C and ¹⁵N chemical shifts and bond lengths have also been reported. ¹⁵N isotropic and anisotropic chemical shifts have been used to characterize the acid–base and tautomeric equilibria of histidine.^15,16 The δ₂₂ principal value of the ¹⁵N chemical shielding tensor in the cationic imidazolium was found to depend on the H-bond length.¹⁷ A linear correlation was observed between the imidazole ¹⁵N isotropic chemical shift and the degree of N–H bond stretching due to H-bonding.¹⁸ The imidazole ¹³C chemical shifts of histidine lyophilized from solutions of varying pH were also found to contain information on the pK_a of the parent solution.¹⁹ Quantum chemical calculations showed that the Cγ and Cδ2 chemical shifts were highly correlated and depended on the tautomeric structure.²⁰

Despite these extensive investigations, so far no studies have provided a complete set of ¹⁵N, ¹³C, and ¹H chemical shifts of histidine and its H-bonding properties and rotameric conformations over a wide range of pH values. Moreover, the minor π tautomer has not been observed in small-molecule histidine compounds. Most solid-state NMR studies used site-specifically ¹⁵N-labeled samples with ¹³C in natural abundance, making it difficult to correlate the ¹³C, ¹⁵N, and ¹H chemical shifts. To facilitate structure determination of histidines in protein active sites, we thus carried out a comprehensive study of the NMR structural parameters of histidine as a function of pH using magic-angle-spinning (MAS) NMR techniques. ¹⁵N, ¹³C, and ¹H isotropic chemical shifts and ¹⁵N chemical shift anisotropies were measured from pH 4.5 to 11 on uniformly ¹³C,¹⁵N-labeled histidine and its salts. This pH range allowed us to detect four protonation states of histidine and both the major τ tautomer and the minor π tautomer, the latter being observed for the first time. We also investigated intra- and intermolecular H-bonding through ¹H chemical shifts and N–H bond stretching effects. Finally, we demonstrate that χ₁ and χ₂ torsion angles indicative of the side-chain rotameric conformation can be measured accurately from backbone–side chain ¹³C–¹⁵N distances.

METHODS AND MATERIALS

Sample Preparation

¹³C,¹⁵N-Labeled (98%) histidine hydrochloride monohydrate was purchased from Sigma-Aldrich and was recrystallized in aqueous solutions of various pH to obtain histidine samples at pH 4.5, 6.0, 8.5, and 11.0. About 30 mg of the labeled histidine powder was dissolved in 600 μL of solution, the pH of which was adjusted by mixing appropriate volumes of 1 M HCl and NaOH. The solution pH was verified by pH paper to a precision of ±0.5. The four samples were designated as His4.5, His6.0, His8.5, and His11.0. The solutions were slowly dried at ambient temperature in 3–5 days to obtain microcrystalline powders, which were then packed into 4 mm MAS rotors for NMR experiments. For distance experiments to determine the side-chain conformation, it was necessary to remove the effects of intermolecular dipolar couplings. To achieve this we diluted the ¹³C,¹⁵N-labeled histidine to 20% by co-dissolving it with 80% unlabeled histidine hydrochloride monohydrate. Two diluted samples were prepared at pH 4.5 and 8.0.

Solid-State NMR Spectroscopy

Solid-state NMR experiments were carried out on a wide-bore Bruker AVANCE-600 spectrometer (14.1 T) and a DSX-400 spectrometer (Karlsruhe, Germany) on 4-mm triple-resonance MAS probes. Typical radiofrequency field strengths were 35–50 kHz for ¹³C and ¹⁵N and 62–83 kHz for ¹H. ¹³C chemical shifts were referenced to the R-Gly C′ signal at 176.49 ppm on the TMS scale and ¹⁵N chemical shifts were referenced to the ¹⁵N signal of N-acetylvaline at 122 ppm on the liquid ammonia scale. The ¹H chemical shifts were externally referenced to those of N-formyl-U-¹³C,¹⁵N-labeled Met-Leu-Phe-OH .²¹

Three types of 2D correlation experiments were used to determine the ¹³C, ¹⁵N, and ¹H isotropic chemical shifts. 2D ¹³C–¹³C DARR correlation experiments²² were carried out with a 5 ms mixing time under 9 kHz MAS. 2D ¹⁵N–¹³C correlation spectra²³ were measured under 9 kHz MAS using a REDOR²⁴ pulse train of 0.44 ms for ¹³C-¹⁵N coherence transfer. 2D ¹H–¹⁵N and ¹H–¹³C heteronuclear correlation (HETCOR) experiments were carried out with Lee–Goldburg (LG)²⁵ cross-polarization (CP) from ¹H, and the samples were spun at 7.5 kHz. The LG-CP contact time was 800 μs for ¹H–¹⁵N and 300 μs for ¹H–¹³C HETCOR experiments. ¹H homonuclear decoupling during the t₁ dimension was achieved using the FSLG pulse sequence²⁶ with a transverse ¹H field strength of 80 kHz.

To identify histidine–water intermolecular contacts, we carried out a ¹H–¹⁵N 2D HETCOR experiment with a MELODI dipolar filter,^27,28 where two rotor periods of ¹³C and ¹⁵N dipolar dephasing were added before the ¹H evolution period. These ¹⁵N-detected MELODI-HETCOR experiments used a Hartmann–Hahn CP contact time of 3 ms, and the sample was spun at 6859 Hz.

¹⁵N chemical shift anisotropy (CSA) was measured using the 2D SUPER experiment²⁹ under slow spinning speeds of 3.0 and 3.6 kHz. The field strength of the ¹⁵N CSA recoupling pulse was 36.4 and 43.6 kHz to satisfy the ω₁ = 12.12ω_r recoupling condition. A ¹H decoupling field of 80 kHz and a ¹³C decoupling pulse of 3 kHz were applied during ¹⁵N t₁ evolution. The recoupled ¹⁵N CSA powder patterns in the indirect dimension of the 2D spectra gave the three principal values δ_ii, from which the anisotropy parameter δ and the asymmetry parameter η were calculated using the following equations:

$$\delta \equiv {\delta }_{zz}-{\delta }_{\mathrm{iso}}$$ (1)

$$\eta \equiv \frac{{\delta }_{yy}-{\delta }_{xx}}{{\delta }_{zz}-{\delta }_{\mathrm{iso}}},\text{where}\mid {\delta }_{yy}-{\delta }_{\mathrm{iso}}\mid \le \mid {\delta }_{xx}-{\delta }_{\mathrm{iso}}\mid \le \mid {\delta }_{zz}-{\delta }_{\mathrm{iso}}\mid$$ (2)

Another parameter describing the size of the CSA, the span Δσ ≡ δ₁₁ – δ₃₃, was also calculated.

¹⁵N–¹H dipolar couplings were measured using the dipolar-doubled DIPSHIFT experiment³⁰ under 4 kHz MAS. The time domain data were fit to give the apparent dipole coupling strengths and were divided by (2 × 0.54) to account for the dipolar doubling and the FSLG scaling factor to obtain the true coupling strength. The FSLG scaling factor of 0.54 was measured directly by comparing the ¹H chemical shifts of the 1D ¹H spectrum and the indirect dimension of the 2D ¹H–¹³C HETCOR spectrum of methylmalonic acid, where the FSLG decoupling condition was the same as for histidine.

Intramolecular ¹³C–¹⁵N distances were measured on 20% diluted His4.5 and His8.0 samples to determine the rotameric structure. These distances were measured using the frequency-selective (FS) REDOR experiment³¹ with ¹³C detection and ¹⁵N dephasing. For the ¹⁵N inversion pulses every half a rotor period, composite 90° 180° 90° pulses were used to reduce the effects of flip angle errors and enhance the distance accuracy.³² During the REDOR mixing period, 75–83 kHz of ¹H TPPM decoupling was applied, and the ¹⁵N rf field strength was 36 kHz. A pair of soft Gaussian ¹³C and ¹⁵N inversion pulses were applied in the middle of the REDOR mixing period to selectively invert the peaks of interest. The Gaussian pulse lengths were rotor-synchronized to be 2 ms for ¹³C and 4 ms for ¹⁵N. The long ¹⁵N soft pulse was necessary to selectively invert the Nδ1 and Nε2 peaks in the pH 4.5 sample, since the two peaks differ by only 14 ppm. For each REDOR mixing time (t_m), a control spectrum (S₀) with the central ¹⁵N Gaussian pulse off and a dephasing spectrum (S) with the ¹⁵N Gaussian pulse on were measured. The intensity ΔS/S₀ ≡ (S₀ – S)/S₀ as a function of t_m gives the ¹³C–¹⁵N dipolar coupling. The FS-REDOR experiments were carried out under 7 kHz MAS. In simulating the REDOR data, we corrected for the presence of natural-abundance ¹³C and ¹⁵N spins according to

$$\left[1-{\left(\frac{S}{S_0}\right)}_{\mathrm{obs}}\right]=93.8\%\times 1-{\left(\frac{S}{S_0}\right)}_{\text{true},100\%}$$ (3)

REDOR dephasing curves were simulated using the SIMPSON program.³³ The ¹⁵N CSA was not included for protonated nitrogens, since they are sufficiently small to cause no detectable changes in the REDOR dephasing curves. For the Nδ1 of the neutral τ tautomer, which exhibited the largest ¹⁵N CSA among all sites (δ = 210 ppm, Table 2), the ¹⁵N CSA was included in the simulation. The Nδ1 chemical shift tensor orientation was taken from ref 34, with Euler angles of 153°, 77°, and 101°, to describe the Cα–Nδ1 vector orientation in the Nδ1 chemical shift tensor frame.

Table 2. ¹⁵N CSA Principal Values, Anisotropy Parameters (δ), Asymmetry Parameters (η), and Spans (Δσ) in Histidine; Hydrogen-Bond Distances Are Also Included.

state	site	δiso (ppm)	δ33 (ppm)	δ22 (ppm)	δ11 (ppm)	δ (ppm)	η	Δσ (ppm)	H-bond (Å)
cationic	Nδ1–H	190	71	219	280	−119	0.51	209	2.63a
	Nε2–H	176	63	195	270	−113	0.66	207	2.81a
neutral τ tautomer	Nδ1	249	39	319	389	−210	0.33	350	2.76b
	Nε2–H	171	73	184	256	−98	0.73	183	2.76b
anionic τ tautomer	Nδ1	253	100	288	371	−153	0.54	271	–
	Nε2–H	167	71	175	255	−96	0.83	184	–
anionic π tautomer	Nδ1–H	172	69	189	258	−103	0.67	189	–
	Nε2	248	101	271	362	−147	0.62	261	–

^aThe cationic histidine hydrogen-bonding distances are based on the crystal structure of histidine hydrochloride monohydrate C₆H₁₂ClN₃O₃, measured on the pH 4.5 sample.

^bThe distances for the neutral τ tautomer are R_NN, based on the crystal structure of histidine C₆H₉N₃O₂, measured on the pH 8.5 sample.

RESULTS AND DISCUSSION

¹³C and ¹⁵N Isotropic Chemical Shifts

To determine the protonation and tautomeric structure of histidine, we measured the ¹³C and ¹⁵N chemical shifts of histidine. Figure 1 shows 1D ¹³C and ¹⁵N CP-MAS spectra of histidine from pH 4.5 to 11.0. In this paper, we use green to designate the low-pH biprotonated and cationic histidine, red for the neutral τ tautomer, blue for the backbone anionic τ tautomer, and purple for the backbone anionic π tautomer. Figure 1 shows that both backbone and side-chain chemical shifts vary with the pH of the solution from which the samples were prepared. For the pH 4.5 sample, the ¹⁵N isotropic shifts of both Nδ1 and Nε2 lie between 170 and 190 ppm, consistent with a biprotonated imidazolium, and no signal at the unprotonated ¹⁵N chemical shift of 250 ppm was observed. For all other pH states, the unprotonated~15N signal was present. The Nα chemical shift reflects the charged state of the amino acid backbone. At pH 4.5 and 6.0, only an NH₃⁺ peak at ~45 ppm was detected, while at pH 8.5 and 11, a 100 ppm ¹⁵N NH₂ peak, characteristic of an anionic backbone, was also present. At pH 11, this NH₂ peak was the dominant Nα signal.

Figure 1.

¹³C (left) and ¹⁵N (right) CP-MAS spectra of histidine at pH 4.5 (a,e), 6.0 (b,f), 8.5 (c,g), and 11.0 (d,f). Peak assignments were obtained from 2D correlation spectra shown in Figure 2 and are color-coded as shown in the box. Assignment in black indicates a mixture of two different tautomeric or charged states of histidine.

The spectral line widths of the histidines were narrow between pH 4.5 and 8.5 but significantly broadened at pH 11. Below pH 9, the line widths were 1.3 ± 0.4 ppm for ¹³C and 1.9 ± 0.2 ppm for ¹⁵N, but at pH 11 the line widths increased to 5.5 ± 1.1 ppm for ¹³C and 9.3 ± 2.2 ppm for ¹⁵N (Figure 1d,h). The broader line widths indicate that the crystal packing was disrupted due to the coexistence of the τ and π tautomers. This is interesting because the rare π tautomer had not been detected in previous NMR studies of small histidine compounds at less basic pH conditions,¹⁹ suggesting that the Nδ1-protonated π tautomer may be stabilized by inter-molecular interactions with other imidazole rings.

Assignment of the ¹³C and ¹⁵N signals in Figure 1 was made using 2D ¹³C–¹³C and ¹⁵N–¹³C correlation experiments (Figure 2). A short mixing time of 5 ms was used for the ¹³C–¹³C DARR experiment to confine cross peaks mostly to one- and two-bond correlations. Figure 2a,b shows the full 2D ¹³C–¹³C and ¹⁵N–¹³C correlation spectra of His6.0, and Figure 2c,d shows the spectra of His11. A comparison of the aromatic side-chain region of the 2D ¹³C–¹³C spectra for all four pH’s is given in Figure 2e, and Figure 2f shows the 2D ¹⁵N–¹³C HETCOR spectra of His4.5 and His8.5.

Figure 2.

Representative 2D ¹³C–¹³C (a,c,e) and ¹⁵N–¹³C (b,d,f) correlation spectra of histidine from pH 4.5 to 11.0: (a,b) His6.0; (c,d) His11.0; (e) imidazole regions of the 2D ¹³C–¹³C correlation spectra of histidine; (f) imidazole side-chain regions of the 2D ¹⁵N–¹³C correlation spectra of His4.5 and His8.5.

While only cationic histidine was present at pH 4.5, both cationic (green) and neutral τ tautomer (red) of histidine were observed as the pH increased to 6. The latter was evidenced by the Nδ1 peak at 249.4 ppm (Figure 2b). The ¹³C chemical shifts nicely distinguish the cationic and neutral histidines. For example, the Cγ–Cδ2 cross peak resonated at (128.7, 119.4) ppm in the cationic state but shifted to (137.7, 113.6) ppm in the neutral τ tautomer. The backbone Cα, C′, and Nα chemical shifts also differed between the cationic histidine and the neutral τ tautomer. At pH 6, the intensity ratio of the cationic to the neutral histidine is about 54:46 (Figure 1b,f), consistent with the side-chain pK_a¹⁶ of about 6 in aqueous solution (Scheme 1).

Scheme 1.

At pH 8.5, τ tautomers with a neutral backbone (red) and an anionic backbone (blue) coexisted at a ratio of about 3:1 (Figure 1c,g), in good agreement with the pK_a of 9.1 for the backbone amino group (Scheme 1). The absence of the π tautomer indicates that the τ tautomer is much more stable at pH 8.5, possibly due to intramolecular H-bonding. As the pH increased to 11, the neutral τ tautomer disappeared, and a second anionic histidine with a π tautomer (purple) was observed (Figure 2c,d). The pronounced line broadening of the spectra was directly related to the coexistence of the π and τ tautomers in the dry sample, whose poor packing caused inhomogeneous local environments. When the sample was well hydrated at the same pH, we found that the anionic π tautomer transformed to the τ tautomer with concomitant line narrowing (Figure S1, Supporting Information). Thus, the anionic π tautomer is metastable and only found in the absence of water, suggesting that water–histidine H-bonding stabilizes the τ tautomer, which in turn implies that increased percentages of the π tautomer in proteins must result from other stabilizing interactions, such as H-bonding with neighboring residues or coordination by metal ions.^20,35 The ¹³C and ¹⁵N isotropic chemical shifts of all four histidines are summarized in Table 1.

Table 1. ¹H, ¹³C, and ¹⁵N Isotropic Chemical Shifts (ppm) of Histidine in Different Protonation and Tautomeric States.

site		cationic	neutral τ tautomer	anionic τ tautomer	anionic π tautomer
13C	C′	173.2	175.6	183.4	183.2
	Cα	54.1	57.0	58.0	59.4
	Cβ	26.0	27.0	32.6	28.4
	Cγ	128.7	137.7	136.8	129.2
	Cε1	136.3	135.3	135.6	135.8
	Cδ2	119.4	113.6	113.0	125.4
15N	Nα	47.6	41.5	96.3	96.3
	Nδ1	190.0	249.4	252.7	171.8
	Nε2	176.3	171.1	167.3	248.2
1H	HN	8.6	9.0	7.4	6.2
	Hα	3.5	4.3	5.2	4.0
	Hβ	3.3	2.7	2.7	2.7
	Hδ1	16.8	NA	NA	12.7
	Hε2	12.6	13.7	12.2	NA
	Hδ2	8.0	4.9	5.3	6.4
	Hε1	9.3	6.1	5.6	7.2

¹⁵N Chemical Shift Anisotropies

Chemical shift anisotropy gives more complete information than isotropic shifts on the local electronic environment and on H-bonding.^17,20,34,36 Since the ¹³C and ¹⁵N isotropic chemical shifts already vary systematically with the protonation and tautomeric structure of the imidazole, the anisotropic chemical shifts are expected to show even larger variations. We measured the ¹⁵N CSA of Nδ1 and Nε2 as a function of pH using the 2D SUPER experiment,²⁹ where the CSA line shapes were recoupled in the indirect dimension and separated according to their isotropic shifts in the direct dimension. While some of these ¹⁵N CSAs were reported before,^15-17 the CSAs for the minor π tautomers were not known. Figure S2 (Supporting Information) shows the recoupled CSA patterns of Nδ1 and Nε2 at the four pH values. The three principal values, δ₁₁, δ₂₂, and δ₃₃, defined from the most downfield (left) to the most upfield chemical shifts, were directly read from the two edges and the maximum of the powder patterns. It can be seen that the protonated nitrogens exhibit smaller CSAs than unprotonated nitrogens. The span Δσ ranges from 180 to 210 ppm (Table 2), in good agreement with literature values.^15,17 The protonated nitrogens in the cationic imidazolium have the upper bound of ~210 ppm, whereas those in the neutral imidazoles adopt lower-bound values of 180–190 ppm. These spans are moderately larger than those of the backbone amides (~150 ppm).³⁷ The asymmetry parameter η differs more significantly between the imidazole and amide nitrogens: η is ~0.5 for imidazole NH groups but only 0.2 (i.e., nearly uniaxial) for backbone amides,³⁸ the latter due to the dominating influence of the carbonyl group on its electronic environment. Table 2 also shows that the middle principal value, δ₂₂, ranges from 175 to 219 ppm for the protonated imidazole nitrogens. This principal value was known to be sensitive to H-bond formation:¹⁷ δ₂₂ shifts downfield by ~50 ppm as R_NO decreases from 3.0 to 2.5 Å. We found Nδ1–H in the cationic histidine to exhibit the most downfield δ₂₂ value (219 ppm), suggesting that it was involved in the strongest H-bond among all imidazole nitrogens.

Unprotonated imidazole nitrogens have much larger CSA spans of 260–350 ppm.^16,34 The CSA tensor orientation is known to differ between the unprotonated and protonated imidazole nitrogens: the direction of the most deshielded element, δ₁₁, is tangential to the ring for unprotonated nitrogens but radial to the ring in protonated nitrogens.³⁴ Density functional theory calculations suggested that the most downfield principal axis was sensitive to intermolecular H-bonding.³⁴ Table 2 shows that the unprotonated nitrogen in the neutral histidine has a significantly larger span (~350 ppm) than the unprotonated nitrogens (~270 ppm) in either tautomer of the anionic histidine. Below we examine the origin of this CSA difference by detecting intra- and intermolecular H-bonding through ¹H chemical shifts and N–H bond lengths.

¹H Chemical Shifts and Hydrogen Bonding

¹H chemical shifts provide a sensitive indicator of the chemical structure and H-bonding of imidazoles. The ¹H isotropic chemical shift is well known to increase (move downfield) with increasing H-bond strength.^39,40 We measured the ¹H chemical shifts using 2D ¹H–¹³C and ¹H–¹⁵N HETCOR experiments. LG-CP was used to transfer the ¹H polarization to ¹³C or ¹⁵N in a site-specific fashion, and strong¹H homonuclear decoupling was applied during t₁ to ensure site resolution and to prevent ¹H spin diffusion.²⁶ Figure 3 shows the HETCOR spectra of His6.0 and His8.5, where narrow line widths of 0.8 ± 0.3 ppm were observed in the ¹H dimension. At pH 6, where the cationic imidazolium (green) coexists with the neutral τ tautomer (red), the carbon-bonded Hδ2 and Hε1 resonate ~3 ppm downfield in the cationic histidine compared to those in the neutral τ tautomer (Figure 3a), which can be attributed to the delocalized positive charge creating a more deshielded environment for the protons. In the ¹⁵N-detected HETCOR spectrum (Figure 3b), the unprotonated Nδ1 exhibits cross peaks both with Hε1 two bonds away and with the backbone NH₃. The amino ¹H chemical shift is 0.5 ppm more downfield in the neutral τ tautomer (9.1 ppm) than in the cationic imidazolium (8.6 ppm), supporting the existence of a NH₃ ⋯ Nδ1 H-bond.

Figure 3.

¹H chemical shifts of histidine from 2D ¹H–¹³C and ¹H–¹⁵N HETCOR spectra: (a,b) His6.0; (c,d) His8.5; (a,c) ¹H–¹³C HETCOR spectra; (b,d) ¹H–¹⁵N HETCOR spectra. At both pH values, a mixture of two states was observed. Note the large downfield ¹H chemical shifts of Nδ1 in cationic histidine.

At pH 8.5, where both neutral (red) and anionic (blue) τ tautomers exist, most aliphatic and aromatic ¹H’s exhibit similar chemical shifts between the two states (Figure 3c,d). The main exceptions are the backbone NH₃ and side-chain Hε2 protons, which show lower chemical shifts in the anionic than in the neutral histidine. In addition, a cross peak between the unprotonated Nδ1 and backbone H^N was detected for the neutral τ tautomer, suggesting side chain–backbone H-bonding in the neutral histidine but not in the anionic histidine.

In general, unprotonated imidazole nitrogens can serve as H-bond acceptors, while the protonated nitrogens can act as H-bond donors. For the latter, the H-bond acceptors can be either backbone carbonyl or water molecules. To determine whether H-bonds indeed exist between water and Nδ1–H or Nε2–H, we measured ¹⁵N-detected MELODI-HETCOR spectra. This experiment eliminates the signals of immobile ¹H spins directly bonded to a ¹³C or ¹⁵N spin by ¹³C and ¹⁵N dipole dephasing, thus ensuring that only water protons or dynamic protons can give rise to cross peaks in the 2D spectra. Figure 4 shows ¹⁵N-detected MELODI-HETCOR spectra of His6.0 without (a) and with (b) dipolar dephasing pulses. The control spectrum exhibited the expected cross peaks between ¹⁵N and aliphatic, amino, and water protons, while in the ¹³C and ¹⁵N-dephased spectrum, the signals of the aliphatic and aromatic protons were completely removed, leaving only water and mobile NH₃ signals. Interestingly, the spectrum shows that only the cationic histidine has cross peaks with water, while the neutral τ tautomer does not. These results agree with the crystal structures, which showed four water molecules in the unit cell of cationic histidine but no water molecules in neutral histidine (Figure S3, Supporting Information). The spectra confirm that the protonated and neutral molecules at pH 6 are not in molecular contact but pack in separate microcrystalline environments.

Figure 4.

2D ¹H–¹⁵N MELODI-HETCOR spectra of His6.0 to identify intermolecular water–histidine hydrogen bonding. (a) Control spectrum without dipolar filter. (b) Spectrum with two rotor periods of ¹³C and ¹⁵N dipolar dephasing. Only water protons and mobile protons remain in (b).

N–H Bond Lengths and Hydrogen Bond Formation

¹⁵N–¹H bond lengths provide an independent probe of the presence of H-bonds in histidine. Hydrogen bonding stretches the N–H bond from 1.05 Å⁴¹ and thus reduces the N–H dipolar coupling from the rigid-limit covalent-bond value of 10 kHz.¹⁸ Figure 5 shows the N–H DIPSHIFT results of Nδ1 and Nε2 in all four histidines. Among the protonated nitrogens, Nδ1 of the cationic histidine exhibited the longest N–H bond of 1.09 Å, while Nε2 in the same sample exhibited a modestly increased bond length of 1.06 Å. In comparison, the imidazole nitrogens in the neutral and anionic histidines showed unstretched bond lengths of 1.05 Å. The prominent Nδ1–H bond stretching in cationic histidine is in excellent agreement with its large downfield ¹⁵N δ₂₂ principal value of 219 ppm (Table 2), its significantly downfield Hδ1 isotropic shift of 16.8 ppm (Figure 3b), and the presence of a strong Nδ1–water cross peak of the sample in the 2D MELODI-HETCOR spectrum (Figure 4b). Indeed, the crystal structure of histidine at pH 4.5 showed a short R_NO of 2.63 Å (Table 2), indicating a strong H-bond. In comparison, Nε2 in the same cationic histidine displayed a less robust panel of H-bonding effects: the ¹⁵N δ₂₂ principal value (195 ppm) and the Hε2 chemical shift (12.6 ppm) are not as far downfield, and the N–H bond stretching is modest (1.06 Å). Consistently, the crystal structure indicates a 0.2 Å longer R_NO distance of 2.81 Å for Nε2 (Table 2).

Figure 5.

Nδ1 and Nε2 ¹⁵N–¹H dipolar couplings of histidine as a function of pH and tautomeric structure. The dipolar dephasing curves are extracted from the t₁ dimension of 2D DIPSHIFT spectra. (a) Cationic histidine at pH 4.5. (b) Neutral τ tautomer at pH 6. (c) Anionic τ tautomer at pH 8.5. (d) Anionic π tautomer at pH 11. The coupling strengths and N–H distances are indicated.

The different N–H bond lengths of protonated nitrogens between the cationic and neutral histidines can be understood on the basis of the different proton affinities of these histidines. According to a recent DFT calculation,⁴² the proton affinity of Nδ1 and Nε2 ranged from −250 to −230 kcal/mol in cationic histidine but from −340 to −360 kcal/mol in neutral histidine. Thus, the protons in the cationic imidazolium are more easily removed than protons in the neutral imidazole.

For unprotonated nitrogens, the N–H dipolar couplings are much weaker, as expected. Since there are several proximal protons contributing to the observed couplings, the nearest-neighbor distance to a proton determined from the dipolar couplings should be systematically smaller than the true nearest-neighbor distance. Between pH 6 and 11, the strongest dipolar coupling was found for Nδ1 (3.5 kHz) in the neutral τ tautomer at pH 6, corresponding to an effective N–H distance of 1.51 Å (Table 3). This distance suggests a strong H-bond, possibly with the backbone amino group, because of the clear H^N–Nδ1 cross peak in the 2D ¹H–¹⁵N HETCOR spectra at pH 6 and 8.5 (Figure 3b,d). In comparison, the unprotonated Nδ1 in the anionic τ tautomer (pH 8.5) showed a significantly weaker dipolar coupling of 2.3 kHz, consistent with the lack of a backbone NH₂ cross peak with Nδ1 in the HETCOR spectrum (Figure 3d).

Table 3. N–H Bond Length (R_NH) in Different Protonated and Tautomeric States of Histidine Determined from N–H Dipolar Couplings (ω_NH).

sample	state	site	δN (ppm)	ωNH (kHz)a	RNH (Å)
His4.5	cationic	Nδ1–H	190.0	9.3 ± 0.1	1.09 ± 0.01
		Nε2–H	176.3	10.2 ± 0.1	1.06 ± 0.01
His6.5	neutral τ	Nδ1	249.4	3.5 ± 0.1	1.51 ± 0.06
	tautomer	Nε2–H	171.1	10.4 ± 0.2	1.05 ± 0.02
His8.5	anionic τ	Nδ1	252.7	2.3 ± 0.4	1.75 ± 0.20
	tautomer	Nε2–H	167.3	10.5 ± 0.1	1.05 ± 0.01
His11.0	anionic π	Nδ1–H	171.8	10.6 ± 0.3	1.05 ± 0.02
	tautomer	Nε2	248.2	2.2 ± 0.3	1.77 ± 0.25

^aA FSLG scaling factor of 0.540 was measured from model compound experiments and used in fitting the N–H dipolar couplings.

χ₁ and χ₂ Torsion Angles from Backbone–Side Chain Distances

The side-chain conformation of histidines in proteins has important implications for protein function. We now demonstrate that it is possible to measure the side-chain χ₁ and χ₂ angles accurately. A number of methods have been introduced to determine the side-chain rotameric structure of amino acids: for example, direct dipolar correlation techniques such as HCCH are useful for β-branched amino acids,⁴³ and methyl ¹³C chemical shifts of doubly methylated amino acid residues (Val, Leu, and Ile) are sensitive to the side-chain conformation.⁴⁴ Here we chose to measure backbone–side chain ¹³C–¹⁵N distances, using the frequency-selective REDOR technique,^31,45 to quantify the χ₁ and χ₂ angles.

The Cα-to-imidazole ¹⁵N distances depend on the χ₂ angle, and the Nα-to-side-chain carbon distances depend on both χ₁ and χ₂ angles (Figure 6), according to the following equations:

$$d_{\mathrm{C}\alpha -\mathrm{N}\delta 1}^2={(d_{\mathrm{C}\gamma \mathrm{N}\delta 1}\sin {\chi }_2\sin 120°)}^2+{(-d_{\mathrm{C}\alpha \mathrm{C}\beta }\sin 109°+d_{\mathrm{C}\gamma \mathrm{N}\delta 1}\cos {\chi }_2\sin 120°)}^2+{(-d_{\mathrm{C}\alpha \mathrm{C}\beta }\cos 109°+d_{\mathrm{C}\beta \mathrm{C}\gamma }-d_{\mathrm{C}\gamma \mathrm{N}\delta 1}\cos 120°)}^2$$ (4)

$$d_{\mathrm{C}\alpha -\mathrm{N}\epsilon 2}^2={\left[d_{\mathrm{C}\gamma \mathrm{N}\epsilon 2}\sin \right({\chi }_2-180°\left)\sin {\theta }_1\right]}^2+{[-d_{\mathrm{C}\alpha \mathrm{C}\beta }\sin 109°+d_{\mathrm{C}\gamma \mathrm{N}\epsilon 2}\cos ({\chi }_2-180°\left)\sin {\theta }_1\right]}^2+{(-d_{\mathrm{C}\alpha \mathrm{C}\beta }\cos 109°+d_{\mathrm{C}\beta \mathrm{C}\gamma }-d_{\mathrm{C}\gamma \mathrm{N}\epsilon 2}\cos {\theta }_1)}^2$$ (5)

$$d_{\mathrm{C}\delta 2-\mathrm{N}\alpha }^2={[d_{\mathrm{N}\alpha \mathrm{C}\alpha }\sin {\chi }_1\sin 109°-d_{\mathrm{C}\gamma \mathrm{C}\delta 2}\sin ({\chi }_2-180°)\sin 120°]}^2+{[d_{\mathrm{C}\alpha \mathrm{C}\beta }-d_{\mathrm{N}\alpha \mathrm{C}\alpha }\cos 109°-(d_{\mathrm{C}\beta \mathrm{C}\gamma }-d_{\mathrm{C}\gamma \mathrm{C}\delta 2}\cos 120°)\cos 109°+d_{\mathrm{C}\gamma \mathrm{C}\delta 2}\sin 120°\cos ({\chi }_2-180°)\sin 109°)]}^2+{[-d_{\mathrm{N}\alpha \mathrm{C}\alpha }\sin 109°\cos {\chi }_1+d_{\mathrm{C}\beta \mathrm{C}\gamma }\sin 109°-d_{\mathrm{C}\gamma \mathrm{C}\delta 2}\cos 109°\cos ({\chi }_2-180°)\sin 120°]}^2$$ (6)

$$d_{\mathrm{C}\epsilon 1-\mathrm{N}\alpha }^2={[d_{\mathrm{N}\alpha \mathrm{C}\alpha }\sin {\chi }_1\sin 109°+d_{\mathrm{C}\gamma \mathrm{C}\epsilon 1}\sin {\chi }_2\sin {\theta }_2]}^2+{[d_{\mathrm{C}\alpha \mathrm{C}\beta }-d_{\mathrm{N}\alpha \mathrm{C}\alpha }\cos 109°-(d_{\mathrm{C}\beta \mathrm{C}\gamma }-d_{\mathrm{C}\gamma \mathrm{C}\epsilon 1}\cos {\theta }_2)\cos 109°+d_{\mathrm{C}\gamma \mathrm{C}\epsilon 1}\sin {\theta }_2\cos {\chi }_2\sin 109°)]}^2+{[-d_{\mathrm{N}\alpha \mathrm{C}\alpha }\sin 109°\cos {\chi }_1+d_{\mathrm{C}\beta \mathrm{C}\gamma }\sin 109°-d_{\mathrm{C}\gamma \mathrm{C}\epsilon 1}\cos 109°\cos {\chi }_2\sin {\theta }_2]}^2$$ (7)

Figure 6.

Intramolecular ¹³C–¹⁵N distances between the side chain and backbone of histidine to determine (χ₁, χ₂) angles: (a–d) cationic histidine at pH 4.5; (e–h) neutral τ tautomer at pH 8.0; (a,e) Cα–Nδ1 REDOR data; (b,f) Cα–Nε2 REDOR data; (c,g) Cδ2–Nα REDOR data; (d,h) Cε1–Nα REDOR data. Left: Schematic representation of the χ₂-dependent Cα–Nδ1 and Cα-Nε2 distances and the (χ₁, χ₂)-dependent Cδ2–Nα and Cε1–Nα distances.

In eqs 4–7, the bond lengths and covalently fixed two-bond distances (d’s) were set to crystallographic values for histidine hydrochloride monohydrate (HISTCM12, Cambridge Structure Database). The bond angles were 109° for ∠NαCαCβ and ∠CαCβCγ and 120 for ∠CβCγNδ1, ∠CβCγCδ2, and ∠CβCγCε1. The angles ∠CβCγNε2 (θ₁) and ∠CβCγCε1 (θ₂) were also fixed by the covalent geometry to be 166.7 and 157.2°, respectively.

Figure 6 shows the ¹³C-¹⁵N REDOR ΔS/S₀ curves of 20% diluted histidine at pH 4.5 and 8.0. Significant differences were observed between the two samples for the Cα–Nδ1 and Cε1–Nα couplings, indicating that the χ₁ and χ₂ angles differ between the cationic and neutral histidines. The resulting intramolecular distances (Table 4) agree well with the crystal structure, within ±0.2 Å, for most sites. The largest deviation was observed for the Cα–Nδ1 distance in the neutral τ tautomer: the NMR distance was longer by 0.3 Å than the crystal structure value. This may be partly due to the large ¹⁵N CSA of this unprotonated site, even though SIMPSON simulations were carried out to include the CSA effect.

Table 4. ¹³C–¹⁵N Intramolecular Distances and Side-Chain (χ₁,χ₂) Angles in Cationic and Neutral τ Tautomer of Histidine.

		intramolecular distance (Å)
state	method	Cα-Nδ1	Cα-Nε2	Cα2-Nα	Cε1-Nα	χ 1	χ 2
cationic	SSNMR	3.84 ± 0.20	4.46 ± 0.20	4.10 ± 0.20	4.70 ± 0.20	75	−120
						−75	120
	X-raya	3.64	4.47	3.95	4.77	72.0	−121.1
neutral τ tautomer	SSNMR	3.32 ± 0.20	4.55 ± 0.20	4.10 ± 0.20	4.10 ± 0.20	−55	60
						55	−60
	X-rayb	3.06	4.57	4.24	4.00	−58.3	56.1

^aThese distances were extracted from the crystal structure of histidine hydrochloride monohydrate C₆H₁₂ClN₃O₃, measured on the pH 4.5 sample.

^bThese distances were extracted from the crystal structure of histidine C₆H₉N₃O₂, measured on the pH 8.5 sample.

The backbone–side chain ¹³C–¹⁵N distances were converted to (χ₁, χ₂) angles according to eqs 4–7. Figure 7 shows contour plots of distances as a function of (χ₁, χ₂) angles. The overlap between the Cα–Nδ1 and Cα–Nε2 distance contours constrains the χ₂ angle, while the overlap between the Cδ2–Nα and Cε1–Nα distances constrains both χ₁ and χ₂ angles. The availability of multiple distances reduced the degeneracy of the dihedral angles to two. Cationic histidine yielded (χ₁, χ₂) angles of (−75°, +120°) or (+75°, −120°), the second set of values being within 3 of the crystal structure values obtained on the same compound. For the neutral τ tautomer, the best-fit (χ₁, χ₂) angles were (−55°, +60°) or (+55°, −60°), the first set of values agreeing with the crystal structure values to ±4°. Thus, the backbone–side chain ¹³C–¹⁵N distances can be measured accurately to determine the side-chain rotameric conformation, and the experiments are applicable to proteins to determine functionally important rotameric structures of histidines.¹¹

Figure 7.

Contour plots of intramolecular ¹³C–¹⁵N distances as a function of (χ₁, χ₂) torsion angles. Only contour lines matching the measured distance values are shown. (a,b) Measured Cα–Nδ1 (red) and Cα–Nε2 (black) distances for cationic imidazolium at pH 4.5 (a) and neutral τ tautomer at pH 8.5 (b). (c,d) Measured Cδ2–Nα (red) and Cε1–Nα (black) distances in cationic histidine (c) and neutral τ tautomer (d). Best-fit torsion angles were read from the positions where the contours overlap.

CONCLUSIONS

The protonation state, tautomeric structure, hydrogen bonding, and rotameric structures of histidines were comprehensively investigated in a wide pH range using MAS solid-state NMR techniques. Two-dimensional correlation experiments resulted in a complete set of ¹H, ¹³C, and ¹⁵N isotropic chemical shifts for four states of histidine: the cationic histidine, the neutral τ tautomer, and the anionic τ and π tautomers. The ¹⁵N and ¹³C chemical shifts are sensitive to both the protonation state and the tautomeric structure, while ¹⁵N and ¹H chemical shifts are sensitive to hydrogen bonding of the imidazole ring. Multiple lines of evidence, including heteronuclear correlation spectra, N–H bond length, and ¹⁵N CSA, consistently indicate strong H-bonds between the protonated Nδ1 and water in the cationic but not the neutral histidine. Hydrogen bonding was also observed between backbone NH₃ and unprotonated Nδ1 in the neutral τ tautomer, with a measured R_N⋯H distance of 1.5 Å. The side-chain dihedral angels χ₁ and χ₂ can be accurately measured, to within 4 of the crystal structure value, through backbone–side chain ¹³C–¹⁵N distances. These results extend our knowledge of the influence of histidine chemical structure and three-dimensional structure on NMR parameters, and provide a large panel of benchmark values to facilitate the study of the high-resolution structure, dynamics, and pH-dependent chemistry of histidines in proteins.