- Email: [email protected]
13C and 15N spectral editing inside histidine imidazole ring through solid-state NMR spectroscopy
Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
Contents lists available at ScienceDirect
Solid State Nuclear Magnetic Resonance journal homepage: www.elsevier.com/locate/ssnmr
13
C and 15N spectral editing inside histidine imidazole ring through solid-state NMR spectroscopy Shenhui Li a,n,1, Lei Zhou a,1, Yongchao Su b, Bin Han a, Feng Deng a a State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Key Laboratory of Magnetic Resonance in Biological Systems, Wuhan Center for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, The Chinese Academy of Sciences, Wuhan 430071, China b Department of Chemistry, Iowa State University, Ames, IA 50011, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 26 November 2012 Received in revised form 16 April 2013 Available online 15 May 2013
Histidine usually exists in three different forms (including biprotonated species, neutral τ and π tautomers) at physiological pH in biological systems. The different protonation and tautomerization states of histidine can be characteristically determined by 13C and 15N chemical shifts of imidazole ring. In this work, solid-state NMR techniques were developed for spectral editing of 13C and 15N sites in histidine imidazole ring, which provides a benchmark to distinguish the existing forms of histidine. The selections of 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 sites were successfully achieved based on one-bond homo- and hetero-nuclear dipole interactions. Moreover, it was demonstrated that 1H, 13C, and 15 chemical shifts were roughly linearly correlated with the corresponding atomic charge in histidine imidazole ring by theoretical calculations. Accordingly, the 1H, 13C and 15N chemical shifts variation in different protonation and tautomerization states could be ascribed to the atomic charge change due to proton transfer in biological process. Crown Copyright & 2013 Published by Elsevier Inc. All rights reserved.
Keywords: Solid state NMR Spectral editing Histidine DFT calculation
1. Introduction Histidine is an essential cationic residue in proteins because of its structurally flexible imidazole group, e.g. the unprotonated imidazole can function as a weak base, while the protonated form is a weak acid. This aromatic heterocyclic ring is a common coordinating ligand in metalloproteins [1]. In particular, the special resonance structure makes it a common participant in various enzyme-catalyzed reactions and electrical conduction systems [2]. Histidine can also play a key role in stabilizing the folded structures of proteins [3]. In general, it is well recognized that histidine could exist in three different forms: a protonated imidazolium form and two neutral tautomers, namely π and τ at physiological pH. Solid state NMR has been proved to be a very well-established technique to investigate the detailed structure and dynamics property in biological systems [4–8]. So far, a number of reports focus on the study of 13C, 15N chemical shifts tensor and N-H bond length in histidine imidazole ring by means of solid-state NMR and quantum chemical calculations [9–12]. For example, Henry et al. [9] determined the pKa of acid–base properties of histidine by quantitatively measuring the population ratios between different n
Corresponding author. Fax: +86 27 87199291. E-mail address: [email protected] (S. Li). 1 Equal contribution.
forms of histidine during three ionization steps. Oldfield et al. [10] found that the Cγ and Cδ2 NMR chemical shifts in imidazole ring were highly correlated, which enabled good predictions of tautomeric states of histidine. Additionally, McDermott et al. [11] suggested that the 15N chemical shift anisotropy (CSA) especially the δ22 tensor in protonated nitrogen sites could provide valuable information on the hydrogen-bonding geometry of imidazole ring. More recently, the 1H, 13C and 15N chemical shifts, hydrogen bond interaction as well as the side-chain conformations of histidine in different protonation and tautomerization states have been comprehensively investigated by advanced solid-state NMR spectroscopy [12]. In general, the 13C and 15N NMR chemical shifts in histidine imidazole ring can be utilized as a benchmark to distinguish different existing forms of histidine. However, the identification of chemical shifts in biological systems is usually subject to the resolution limitation of solid-state NMR and thus could be ambiguous or even infeasible [13,14]. In this work, we hereby developed one-dimensional (1D) 13C and 15N MAS NMR spectral editing methods to facilitate the chemical shifts assignments for uniformly 13 C, 15N labeled histidine imidazole ring via one-bond dipolar interaction. Meanwhile, the relationship between 1H, 13C and 15N chemical shifts and atomic charge distribution inside histidine imidazole ring was established to clarify the chemical shifts variation mechanism in different protonation and tautomerization states.
0926-2040/$ - see front matter Crown Copyright & 2013 Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ssnmr.2013.05.002
14
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
2. Experimental 2.1. Sample preparation 13
C, 15N-labeled (98%) histidine hydrochloride monohydrate was purchased from Sigma-Aldrich Corporation without further treatment. A total mass of 10 mg dry power was directly packed into a 4 mm rotor for ssNMR measurements. 2.2. Solid-state NMR Solid-state NMR experiments were carried out on a Bruker Avance 500 MHz spectrometer equipped with 4-mm triple-resonance MAS probes under a magic-angle-spinning speed of 8.0 kHz. Typical radiofrequency (RF) field strengths were 12–50 kHz for 13C and 15N, and 50–83 kHz for 1H. 13C and 15N chemical shifts were referenced to the adamantane CH2 signal at 38.5 ppm and the 15N resonance of 15NH4Cl at 35.9 ppm, respectively. The RF field of FSLG [15], LG-CP, TPPM in 1H channel was set to 81, 50 and 62.5– 83 kHz respectively. 1H-13C cross polarization contact time of 800 μs and a recycle delay of 2.5 s were used to collect the NMR signals. The CP contact time during CHHC [16] and 13C–15N SPECIFIC-CP [17,18] period was fixed to 50 μs and 350 μs, respectively. Typically, 16, 32, 512, and 1024 scans were accumulated for the 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 selection experiments, respectively.
carbons from Cγ, Cε1 and Cδ2 sites remain unchanged in all forms. To site-specifically select 1D NMR signals from Nδ1 and Nε2 sites, which is crucial to determine the existing forms of histidine, spectral editing of Cγ and Cδ2 sites in histidine imidazole ring is desirable. Fig. 2 shows the NMR pulse sequences used for Cγ, Cδ2, Nδ1 and Nε2 spectral editing of histidine imidazole ring. The idea of the pulse sequence in Fig. 2a is to select quaternary carbon (Cγ) from tertiary carbons (Cε1 and Cδ2) in histidine imidazole ring. In order to distinguish Cγ from Cε1 and Cδ2 sites, 1H-13C dipole dephasing was performed by using two rotor periods 1H-13C dipole recoupling in REDOR [21,22]. For 13Cγ selection scheme, all of the 13C NMR signals could be irradiated during the first CP. Both the 13Cδ2 and 13Cε1 signals could be removed due to the strong one-bond 1 H-13C dipole interaction while 13Cγ resonance could still survive after 1H-13C REDOR dipole dephasing according to the previous simulations [22,23]. Therefore, the ring Cγ signal could be specifically selected according to the proposed strategy. Fig. 2b shows the pulse sequence for Cδ2 selection. For 13Cδ2 spectral editing, all of the 13C signals could be excited by the first CP. Then, only the directly bonded 13C-13C pair signals (e.g. 13Cγ and 13Cδ2) could survive after SPC-5 [24] double quantum filter as shown in Fig. 2b. To further distinguish the Cδ2 signal from that of Cγ, CHHC [16] with a very short (50 μs) contact time was incorporated to select the carbons which are directly bonded with protons. Therefore,
2.3. Computational methods Three representative tri-peptides, namely Gly-His-Gly (GHG), Ala-His-Ala (AHA) and Leu-His-Leu (LHL), in which histidine was anchored into two neutral residues, were chosen as the computational models. Histidine could exist at any forms including biprotonated, neutral π tautomer and τ tautomer in GHG, AHA and LHL. The geometrical parameters, atomic charge distributions (which were obtained by using a nature population analysis, NPA) [19] in each selected model were calculated at the DFT level by using the Becke's three-parameter hybrid method with the Lee– Yang–Parr correlation functional (B3LYP) and the 6–31Gnn basis set. All the calculations in this work were carried out by using Gaussian09 software package [20].
3. Results and discussion The existing forms of histidine imidazole ring in biological proteins strongly depend on the environmental pH. Fig. 1 shows the schematic structures of three most favorable forms of histidine in different protonated and tautomeric states. As shown in Fig. 1, the biprotonated species, neutral τ tautomer and π tautomer of imidazole ring differ from the protonated and tautomeric states of the Nδ1 and Nε2 sites. Meanwhile, the quaternary and tertiary
Fig. 1. Schematic structures of histidine in different protonation and tautomerization states.
Fig. 2. Pulse sequences for 13C, 15 spectral editing of histidine imidazole ring (a) Cγ selection, (b) Cδ2 selection, (c) Nδ1 selection and (d) Nε2 selection.
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
only 13Cδ2 signal could be selected and both the 13Cε2 and 13Cγ signals would be removed. Since the NMR signals of Cγ and Cδ2 sites could be successfully edited, we can transfer their magnetization to the neighboring Nδ1 and Nε2 sites respectively via onebond polarization transfer. Fig. 2c–d shows the spectral editing pulse sequences for selection of the Nδ1 and Nε2 sites in histidine imidazole ring. 13C–15N SPECIFIC-CP [17,18] with short CP contact time (350 μs) was employed to allow one-bond signal transfer and avoid long-range polarization transfer from Cγ and Cδ2 to Nδ1 and Nε2 sites, respectively. (see Fig. 2c–d) Therefore, all of the 13Cγ, 13 Cδ2, 15Nδ1, and 15Nδ2 signals in uniformly 13C, 15N labeled histidine sample could be selected separately. Uniformly 13C, 15N labeled histidine was used as model compound to demonstrate the feasibility of the proposed 13C and 15N spectral editing techniques in histidine imidazole ring. Fig. 3 shows the Cγ and Cδ2 editing spectra of uniformly 13C, 15N labeled histidine. In the 13C MAS NMR control spectra of histidine, all of the 13C signals could be clearly observed in Fig. 3a and the 13C NMR signals of imidazole ring were located in a region from 110 to 140 ppm. After Cγ spectral editing as shown in Fig. 3b, only the Cγ signal at 128.7 ppm and C′ signal at 173.2 ppm were separately selected. It is noteworthy that the Cδ2 and Cε1 signals were almost completely suppressed by 1H-13C dipole dephasing. In comparison with 1H-13C REDOR, DIPSHIFT [25] cannot largely eliminate the signals from Cε1 and Cδ2 sites under 8 kHz MAS. Fig. 3c shows the Cδ2 selection spectra, in which the major imidazole resonance at 119.4 ppm due to Cδ2 site remains. Besides the CHHC [16] scheme, MAS-J-HMQC technique [26] could be alternatively utilized to distinguish the proton-bonded carbon from quaternary carbon species. As shown in Fig. 3b, the Cα and Cβ signals were removed and C′ signal was still retained in the dipole dephasing experiment, whereas Cα and Cβ signals instead of C′ could survive after Cδ2 spectral editing as displayed in Fig. 3c. As all of the Cα, Cβ and C′ signals distinctly deviated from the imidazole region, the appearance of the Cα, Cβ and C′ signals has negligible influence on the chemical shifts assignment of the signals inside histidine imidazole ring. The selection efficiencies in Cγ and Cδ2 spectral editing experiments are estimated to be around 63% and 13%, respectively. The quaternary carbon signal (13Cγ) would be decreased by less
Fig. 3. 13C NMR spectra of uniformly 13C, 15N-labeled histidine (a) control (b) Cγ selection, and (c) Cδ2 selection acquired under a MAS speed of 8 kHz. 8, 16, and 32 scans were accumulated for (a), (b) and (c) respectively. The first 1H-13C cross polarization (CP) and CHHC contact time were fixed to 800 μs and 50 μs, respectively. A recycle delay of 2.5 s was used to collect the NMR signals. For clarity, the imidazole region was expanded on the right.
15
than 35% after two rotor periods REDOR dephasing. While the 13C signals of methine (CH) group such as 13Cδ2 and 13Cε1 sites could almost be completely removed due to the strong one-bond 1H–13C dipole interaction on the basis of our previous simulations [23]. The experimental selection efficiency for SPC-5 double quantum filtering was reported to be ca. 30% [27]. The signal transfer efficiency during CHHC period might be estimated to be 50–70%, suggesting that 13Cδ2 selection efficiency could be in the range of 15–21%. In general, our experimental efficiencies for Cγ (63%) and Cδ2 (13%) selections are considerably reasonable. It seems promising and feasible for the application of Cγ and Cδ2 spectral editing techniques to 13C, 15N uniformly labeled biological protein samples. Fig. 4 shows the Nδ1 and Nε2 editing spectra of uniformly 13C, 15 N labeled histidine. In the 15N CP/MAS control spectrum as shown in Fig. 4a, three major signals, Nα (47.6 ppm), Nδ1 (190.0 ppm) and Nε2 (176.3 ppm), could be well resolved. After Nδ1 spectral editing (Fig. 4b), the signal of Nε2 site was completely suppressed with respect to Nδ1. In the Nδ2 spectral editing experiments (Fig. 3c), only the signal of Nε2 survives though its intensity seems relatively weak, confirming the feasibility of the spectral editing technique. In comparison with SPECIFIC-CP [17,18], it is noted that the efficiency for 13C-15 TEDOR [28] is relatively lower for one-bond polarization transfer from 13C to 15N. In 15Nδ1 and 15Nε2 spectral editing experiments, the selection efficiencies are determined to be around 2% and 0.4% respectively. For uniformly 13C, 15N labeled histidine model compound, we use a relatively shorter SPECIFIC-CP contact time (350 μs) instead of 3–8 ms to allow the one-bond 13C–15N transfer and avoid the twobond 13C–15N coherence transfer. Therefore, our selection efficiencies in 15Nδ1 (2%) and 15Nε2 (0.4%) spectral editing methods are quite lower. Even though the transfer efficiencies for 15Nδ1 and 15 Nε2 spectral editing might be evaluated to be around 3–5 times for real biological protein systems, the sensitivity of the 15N signals is still too low to be well resolved for NMR detection. The 13C and 15N chemical shifts in histidine imidazole ring are characteristic to determine different existing forms in proteins. Solid-state NMR has been proved to be a very successful tool to investigate the structure and equilibrium tautomerization states of histidine in biological systems [29,30]. However, the peak overlap due to the broad line width of NMR spectra in solid state [29–31]
Fig. 4. 15N MAS NMR spectra of uniformly 13C, 15N-labeled histidine (a) control, (b) Nδ1 selection, and (c) Nε2 selection. 32, 512, and 1024 scans were accumulated for (a), (b) and (c). 13C-15N SPECIFIC-CP period was set to 350 μs to acquire the NMR spectrum. For clarity, the imidazole region was expanded on the right.
16
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
significantly limits the use of 1D ssNMR techniques to determine the structures of the imidazole ring in proteins. For example, Cross et al. [31] employed two distinct 15N site-specifically labeled histidine residues (H37) to investigate the tautomerization states of histidine in influenza A Virus M2 proton channels. However, it is extremely costly to prepare samples containing 15N sitespecifically labeled protein to conduct NMR experiments. To gain better resolution, Hong et al. [29,30,32,33] utilized twodimensional (2D) 13C-13C DARR and 13C/1H-15N HETCOR ssNMR to investigate the detailed structure, dynamic behavior of uniformly 13C, 15N labeled H37 residues in influenza A Virus M2 proton channel. The results revealed the structure–property relationship for clarifying proton conduction and gating mechanism in influenza M2 proton channels. In addition, de Groot and coworkers studied the interactions of histidines in the lightharvesting complex II with bacteriochlorophyll and found that Νε2 was ligated with Mg2+ while Nδ1 was protonated and involved in H-bonding by using 2D 13C-13C NMR experiments [13]. However, the acquisition of 2D ssNMR spectra was always timeconsuming. Therefore, the structural determination by using ssNMR requires more experimentally efficient techniques. Here, we propose several 1D solid-state NMR techniques for spectral editing of 13C and 15N sites in the histidine imidazole ring. The high efficiency of Cγ and Cδ2 might allow it feasible to select the carbon sites in histidine residues in biological proteins. However, the sensitivity of 15N NMR signals was relatively lower in the Nδ1 and Nε2 spectral editing experiments, which could be ascribed to the lower efficiency of 15N detection as well as the complexity of the pulse sequences (see Fig. 2c–d). In comparison with the 2D solid-state NMR experiments, the 1D spectral editing techniques especially Cγ and Cδ2 selection methods need less spectrometer time, which might be used in combination with REDOR for histidine backbone-side chain distances measurements [12,29]. In order to gain insights into the 1H, 13C and 15N chemical shifts variation of the imidazole ring, theoretical calculations were employed to investigate the relationship between the NMR chemical shifts and the atomic charge distributions. Firstly, three different tri-peptides including GHG, AHA and LHL were chosen as representative computational models. In each computational model, histidine could be present in three different forms: biprotonated species, neutral π and τ tautomers. Fig. S1 shows the optimized geometries of Ala-His-Ala in three different forms. In addition to the optimized structures of different tri-peptides, the NPA atomic charges were directly obtained and summarized in Table S1. To establish the relationship between 1H, 13C, 15N chemical shifts and the NPA atomic charges, we plotted the atomic charges as a function of the 1H, 13C, and 15N chemical shifts [12] in histidine imidazole ring as shown in Fig. 5. It is noteworthy that the chemical shifts of the histidine imidazole ring are roughly linearly correlated with the corresponding atomic charges. The relationships for 1H, 13C, and 15N nuclei can be described as the following equations: Q 1H ¼ 0:024ð 7 0:002Þδ1H þ 0:11ð 7 0:02Þ Q 13C ¼ 0:011ð 7 0:002Þδ13C −1:3ð 7 0:2Þ
R2 ¼ 0:80
R2 ¼ 0:63
Q 15N ¼ 0:00072ð 7 0:00016Þδ15N −0:68ð 7 0:03Þ 1
13
15
R2 ¼ 0:57
ð1Þ ð2Þ ð3Þ
It is demonstrated that H, C, N chemical shifts variation origins from the atomic charge variation in different protonation and tautomerization states, and the 1H, 13C, and 15N chemical shifts can not only function as a benchmark for the histidine existing forms, but also as an indicator to the atomic charge distribution in the histidine imidazole ring. The total atomic charges in protonated and neutral imidazole ring were determined
Fig. 5. The relationships between (a) 1H, (b) 13C, and (c) 15N chemical shifts and the calculated NPA atomic charges inside histidine imidazole ring.
to be around 0.87 and 0 respectively, which are generally in consistence with the existing forms of histidine side-chain. The imidazole side-chain of histidine serves as catalytic sites in many biological enzymes, in which the neutral histidine usually accept a proton from serine, threonine, or cysteine to activate it [34–36]. In addition, histidine can quickly shuttle protons [29,30]. Shutting protons by ring flip of histidine side-chain was proposed by Hong et al. to demonstrate the proton conduction mechanism of influenza A M2 channel [29]. The deprotonated nitrogen can adopt a proton to form a positively-charged intermediate, then transfer the proton to the neighboring groups. This progress always accompanies the change of protonation and tautomerization states in the histidine imidazole ring. In addition, it has been widely recognized that polymer containing imidazole rings could be applied as ion conductor materials [2,37,38]. Our results indicate that the atomic charge distribution altered by different
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
protonation states of the imidazole ring induces the 13C, 15N chemical shifts change, forming varied tautomerization states. 4. Conclusion In this work, 1D solid-state NMR methods were developed to specifically select Cγ, Cδ2, Nδ1, and Nε2 signals of imidazole ring in uniformly 13C, 15N labeled histidine sample to determine the tautomeric structure of histidine. The spectral editing techniques successfully facilitate the 13C, 15N chemical shifts assignment of histidine imidazole ring. In addition, it was revealed that the 1H, 13C and 15N chemical shifts were strongly correlated with their corresponding atomic charge in the histidine imidazole ring. The NMR chemical shift variations in different protonation and tautomerization states which are sensitive to the environmental pH were ascribed to the atomic charge change in histidine imidozale ring. Acknowledgments This work was supported by the National Natural Science Foundation of China (21003154, 20933009, 21210005 and 21221064), and the National Basic Research Program of China (2009CB918600). The authors are grateful to Shanghai Supercomputer Center (SSC, China) for the support in computing facilities.
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
[14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ssnmr.2013.05.002.
[28] [29] [30] [31] [32]
References [1] [2] [3]
K.G. Strothkamp, S.J. Lippard, Acc. Chem. Res. 15 (1982) 318–326. B.S. Hickman, M. Mascal, J.J. Titman, I.G. Wood, J. Am. Chem. Soc. 121 (1999) 11486–11490. L.-S.P. Tran, T. Urao, F. Qin, K. Maruyama, T. Kakimoto, K. Shinozaki, K. Yamaguchi-Shinozaki, Proc. Natl. Acad. Sci. USA 104 (2007) 20623–20628.
[33] [34] [35] [36] [37] [38]
17
S.J. Opella, F.M. Marassi, Chem. Rev. 104 (2004) 3587–3606. M. Hong, Y. Zhang, F. Hu, Ann. Rev. Phys. Chem. 63 (2012) 1–24. R. Tycko, Ann. Rev. Phys. Chem. 62 (2011) 279–299. A. McDermott, Ann. Rev. Biophys 38 (2009) 385–403. Y. Su, S. Li, M. Hong, Amino Acids 44 (2013) 821–833. B. Henry, P. Tekely, J.-J. Delpuech, J. Am. Chem. Soc. 124 (2002) 2025–2034. F. Cheng, H. Sun, Y. Zhang, D. Mukkamala, E. Oldfield, J. Am. Chem. Soc. 127 (2005) 12544–12554. Y. Wei, A.C. de Dios, A.E. McDermott, J. Am. Chem. Soc. 121 (1999) 10389–10394. S. Li, M. Hong, J. Am. Chem. Soc. 133 (2011) 1534–1544. J. Alia, C. Matysik, M. Soede-Huijbregts, J. Baldus, J. Raap, P. Lugtenburg, H.J. Gast, A.J. van Gorkom, Hoff, H.J.M. de Groot, J. Am. Chem. Soc. 123 (2001) 4803–4809. M.A.S. Hass, D.F. Hansen, H.E.M. Christensen, J.J. Led, L.E. Kay, J. Am. Chem. Soc. 130 (2008) 8460–8470. M. Lee, W.I. Goldburg, Phys. Rev. 140 (1965) A1261–A1271. A. Lange, S. Luca, M. Baldus, J. Am. Chem. Soc. 124 (2002) 9704–9705. M. Baldus, A.T. Petkova, J. Herzfeld, R.G. Griffin, Mol. Phys. 95 (1998) 1197–1207. A.T. Petkova, M. Baldus, M. Belenky, M. Hong, R.G. Griffin, J. Herzfeld, J. Magn. Reson. 160 (2003) 1–12. A.E. Reed, R.B. Weinstock, F. Weinhold, J. Chem. Phys. 83 (1985) 735–746. M.J. Frisch, et al., Gaussian 09, Revision B.01, Gaussian, Inc., Wallingford, CT, 2010. T. Gullion, J. Schaefer, J. Magn. Reson. 81 (1989) 196–200. S. Li, Y. Su, W. Luo, M. Hong, J. Phys. Chem. B 114 (2010) 4063–4069. S. Li, Y. Su, M. Hong, Solid State Nucl. Magn. Reson. 45–46 (2012) 51–58. M. Hohwy, C.M. Rienstra, C.P. Jaroniec, R.G. Griffin, J. Chem. Phys. 110 (1999) 7983–7992. M.G. Munowitz, R.G. Griffin, G. Bodenhausen, T.H. Huang, J. Am. Chem. Soc. 103 (1981) 2529–2533. A. Lesage, D. Sakellariou, S. Steuernagel, L. Emsley, J. Am. Chem. Soc. 120 (1998) 13194–13201. Y. Su, T. Doherty, A.J. Waring, P. Puchala, M. Hong, Biochemistry 48 (2009) 4587–4595. M. Hong, R.G. Griffin, J. Am. Chem. Soc. 120 (1998) 7113–7114. F. Hu, W. Luo, M. Hong, Science 330 (2010) 505–508. F. Hu, K. Schmidt-Rohr, M. Hong, J. Am. Chem. Soc. 134 (2012) 3703–3713. J. Hu, R. Fu, K. Nishimura, L. Zhang, H.X. Zhou, D.D. Busath, V. Vijayvergiya, T.A. Cross, Proc. Natl. Acad. Sci. USA 103 (2006) 6865–6870. M. Hong, K.J. Fritzsching, J.K. Williams, J. Am. Chem. Soc. 134 (2012) 14753–14755. Y. Su, F. Hu, M. Hong, J. Am. Chem. Soc. 134 (2012) 8693–8702. C. Brenner, Biochemistry 41 (2002) 9003–9014. C. Nishimura, Y. Ohashi, S. Sato, T. Kato, S. Tabata, C. Ueguchi, Plant Cell 16 (2004) 1365–1377. C.T. Walsh, S. Garneau-Tsodikova, G.J. Gatto, Angew. Chem. Int. Ed. 44 (2005) 7342–7372. S. Scheiner, M.Y. Yi, J. Phys. Chem. 100 (1996) 9235–9241. R. Graf, Solid State Nucl. Magn. Reson. 40 (2011) 127–133.
Contents lists available at ScienceDirect
Solid State Nuclear Magnetic Resonance journal homepage: www.elsevier.com/locate/ssnmr
13
C and 15N spectral editing inside histidine imidazole ring through solid-state NMR spectroscopy Shenhui Li a,n,1, Lei Zhou a,1, Yongchao Su b, Bin Han a, Feng Deng a a State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Key Laboratory of Magnetic Resonance in Biological Systems, Wuhan Center for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, The Chinese Academy of Sciences, Wuhan 430071, China b Department of Chemistry, Iowa State University, Ames, IA 50011, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 26 November 2012 Received in revised form 16 April 2013 Available online 15 May 2013
Histidine usually exists in three different forms (including biprotonated species, neutral τ and π tautomers) at physiological pH in biological systems. The different protonation and tautomerization states of histidine can be characteristically determined by 13C and 15N chemical shifts of imidazole ring. In this work, solid-state NMR techniques were developed for spectral editing of 13C and 15N sites in histidine imidazole ring, which provides a benchmark to distinguish the existing forms of histidine. The selections of 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 sites were successfully achieved based on one-bond homo- and hetero-nuclear dipole interactions. Moreover, it was demonstrated that 1H, 13C, and 15 chemical shifts were roughly linearly correlated with the corresponding atomic charge in histidine imidazole ring by theoretical calculations. Accordingly, the 1H, 13C and 15N chemical shifts variation in different protonation and tautomerization states could be ascribed to the atomic charge change due to proton transfer in biological process. Crown Copyright & 2013 Published by Elsevier Inc. All rights reserved.
Keywords: Solid state NMR Spectral editing Histidine DFT calculation
1. Introduction Histidine is an essential cationic residue in proteins because of its structurally flexible imidazole group, e.g. the unprotonated imidazole can function as a weak base, while the protonated form is a weak acid. This aromatic heterocyclic ring is a common coordinating ligand in metalloproteins [1]. In particular, the special resonance structure makes it a common participant in various enzyme-catalyzed reactions and electrical conduction systems [2]. Histidine can also play a key role in stabilizing the folded structures of proteins [3]. In general, it is well recognized that histidine could exist in three different forms: a protonated imidazolium form and two neutral tautomers, namely π and τ at physiological pH. Solid state NMR has been proved to be a very well-established technique to investigate the detailed structure and dynamics property in biological systems [4–8]. So far, a number of reports focus on the study of 13C, 15N chemical shifts tensor and N-H bond length in histidine imidazole ring by means of solid-state NMR and quantum chemical calculations [9–12]. For example, Henry et al. [9] determined the pKa of acid–base properties of histidine by quantitatively measuring the population ratios between different n
Corresponding author. Fax: +86 27 87199291. E-mail address: [email protected] (S. Li). 1 Equal contribution.
forms of histidine during three ionization steps. Oldfield et al. [10] found that the Cγ and Cδ2 NMR chemical shifts in imidazole ring were highly correlated, which enabled good predictions of tautomeric states of histidine. Additionally, McDermott et al. [11] suggested that the 15N chemical shift anisotropy (CSA) especially the δ22 tensor in protonated nitrogen sites could provide valuable information on the hydrogen-bonding geometry of imidazole ring. More recently, the 1H, 13C and 15N chemical shifts, hydrogen bond interaction as well as the side-chain conformations of histidine in different protonation and tautomerization states have been comprehensively investigated by advanced solid-state NMR spectroscopy [12]. In general, the 13C and 15N NMR chemical shifts in histidine imidazole ring can be utilized as a benchmark to distinguish different existing forms of histidine. However, the identification of chemical shifts in biological systems is usually subject to the resolution limitation of solid-state NMR and thus could be ambiguous or even infeasible [13,14]. In this work, we hereby developed one-dimensional (1D) 13C and 15N MAS NMR spectral editing methods to facilitate the chemical shifts assignments for uniformly 13 C, 15N labeled histidine imidazole ring via one-bond dipolar interaction. Meanwhile, the relationship between 1H, 13C and 15N chemical shifts and atomic charge distribution inside histidine imidazole ring was established to clarify the chemical shifts variation mechanism in different protonation and tautomerization states.
0926-2040/$ - see front matter Crown Copyright & 2013 Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ssnmr.2013.05.002
14
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
2. Experimental 2.1. Sample preparation 13
C, 15N-labeled (98%) histidine hydrochloride monohydrate was purchased from Sigma-Aldrich Corporation without further treatment. A total mass of 10 mg dry power was directly packed into a 4 mm rotor for ssNMR measurements. 2.2. Solid-state NMR Solid-state NMR experiments were carried out on a Bruker Avance 500 MHz spectrometer equipped with 4-mm triple-resonance MAS probes under a magic-angle-spinning speed of 8.0 kHz. Typical radiofrequency (RF) field strengths were 12–50 kHz for 13C and 15N, and 50–83 kHz for 1H. 13C and 15N chemical shifts were referenced to the adamantane CH2 signal at 38.5 ppm and the 15N resonance of 15NH4Cl at 35.9 ppm, respectively. The RF field of FSLG [15], LG-CP, TPPM in 1H channel was set to 81, 50 and 62.5– 83 kHz respectively. 1H-13C cross polarization contact time of 800 μs and a recycle delay of 2.5 s were used to collect the NMR signals. The CP contact time during CHHC [16] and 13C–15N SPECIFIC-CP [17,18] period was fixed to 50 μs and 350 μs, respectively. Typically, 16, 32, 512, and 1024 scans were accumulated for the 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 selection experiments, respectively.
carbons from Cγ, Cε1 and Cδ2 sites remain unchanged in all forms. To site-specifically select 1D NMR signals from Nδ1 and Nε2 sites, which is crucial to determine the existing forms of histidine, spectral editing of Cγ and Cδ2 sites in histidine imidazole ring is desirable. Fig. 2 shows the NMR pulse sequences used for Cγ, Cδ2, Nδ1 and Nε2 spectral editing of histidine imidazole ring. The idea of the pulse sequence in Fig. 2a is to select quaternary carbon (Cγ) from tertiary carbons (Cε1 and Cδ2) in histidine imidazole ring. In order to distinguish Cγ from Cε1 and Cδ2 sites, 1H-13C dipole dephasing was performed by using two rotor periods 1H-13C dipole recoupling in REDOR [21,22]. For 13Cγ selection scheme, all of the 13C NMR signals could be irradiated during the first CP. Both the 13Cδ2 and 13Cε1 signals could be removed due to the strong one-bond 1 H-13C dipole interaction while 13Cγ resonance could still survive after 1H-13C REDOR dipole dephasing according to the previous simulations [22,23]. Therefore, the ring Cγ signal could be specifically selected according to the proposed strategy. Fig. 2b shows the pulse sequence for Cδ2 selection. For 13Cδ2 spectral editing, all of the 13C signals could be excited by the first CP. Then, only the directly bonded 13C-13C pair signals (e.g. 13Cγ and 13Cδ2) could survive after SPC-5 [24] double quantum filter as shown in Fig. 2b. To further distinguish the Cδ2 signal from that of Cγ, CHHC [16] with a very short (50 μs) contact time was incorporated to select the carbons which are directly bonded with protons. Therefore,
2.3. Computational methods Three representative tri-peptides, namely Gly-His-Gly (GHG), Ala-His-Ala (AHA) and Leu-His-Leu (LHL), in which histidine was anchored into two neutral residues, were chosen as the computational models. Histidine could exist at any forms including biprotonated, neutral π tautomer and τ tautomer in GHG, AHA and LHL. The geometrical parameters, atomic charge distributions (which were obtained by using a nature population analysis, NPA) [19] in each selected model were calculated at the DFT level by using the Becke's three-parameter hybrid method with the Lee– Yang–Parr correlation functional (B3LYP) and the 6–31Gnn basis set. All the calculations in this work were carried out by using Gaussian09 software package [20].
3. Results and discussion The existing forms of histidine imidazole ring in biological proteins strongly depend on the environmental pH. Fig. 1 shows the schematic structures of three most favorable forms of histidine in different protonated and tautomeric states. As shown in Fig. 1, the biprotonated species, neutral τ tautomer and π tautomer of imidazole ring differ from the protonated and tautomeric states of the Nδ1 and Nε2 sites. Meanwhile, the quaternary and tertiary
Fig. 1. Schematic structures of histidine in different protonation and tautomerization states.
Fig. 2. Pulse sequences for 13C, 15 spectral editing of histidine imidazole ring (a) Cγ selection, (b) Cδ2 selection, (c) Nδ1 selection and (d) Nε2 selection.
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
only 13Cδ2 signal could be selected and both the 13Cε2 and 13Cγ signals would be removed. Since the NMR signals of Cγ and Cδ2 sites could be successfully edited, we can transfer their magnetization to the neighboring Nδ1 and Nε2 sites respectively via onebond polarization transfer. Fig. 2c–d shows the spectral editing pulse sequences for selection of the Nδ1 and Nε2 sites in histidine imidazole ring. 13C–15N SPECIFIC-CP [17,18] with short CP contact time (350 μs) was employed to allow one-bond signal transfer and avoid long-range polarization transfer from Cγ and Cδ2 to Nδ1 and Nε2 sites, respectively. (see Fig. 2c–d) Therefore, all of the 13Cγ, 13 Cδ2, 15Nδ1, and 15Nδ2 signals in uniformly 13C, 15N labeled histidine sample could be selected separately. Uniformly 13C, 15N labeled histidine was used as model compound to demonstrate the feasibility of the proposed 13C and 15N spectral editing techniques in histidine imidazole ring. Fig. 3 shows the Cγ and Cδ2 editing spectra of uniformly 13C, 15N labeled histidine. In the 13C MAS NMR control spectra of histidine, all of the 13C signals could be clearly observed in Fig. 3a and the 13C NMR signals of imidazole ring were located in a region from 110 to 140 ppm. After Cγ spectral editing as shown in Fig. 3b, only the Cγ signal at 128.7 ppm and C′ signal at 173.2 ppm were separately selected. It is noteworthy that the Cδ2 and Cε1 signals were almost completely suppressed by 1H-13C dipole dephasing. In comparison with 1H-13C REDOR, DIPSHIFT [25] cannot largely eliminate the signals from Cε1 and Cδ2 sites under 8 kHz MAS. Fig. 3c shows the Cδ2 selection spectra, in which the major imidazole resonance at 119.4 ppm due to Cδ2 site remains. Besides the CHHC [16] scheme, MAS-J-HMQC technique [26] could be alternatively utilized to distinguish the proton-bonded carbon from quaternary carbon species. As shown in Fig. 3b, the Cα and Cβ signals were removed and C′ signal was still retained in the dipole dephasing experiment, whereas Cα and Cβ signals instead of C′ could survive after Cδ2 spectral editing as displayed in Fig. 3c. As all of the Cα, Cβ and C′ signals distinctly deviated from the imidazole region, the appearance of the Cα, Cβ and C′ signals has negligible influence on the chemical shifts assignment of the signals inside histidine imidazole ring. The selection efficiencies in Cγ and Cδ2 spectral editing experiments are estimated to be around 63% and 13%, respectively. The quaternary carbon signal (13Cγ) would be decreased by less
Fig. 3. 13C NMR spectra of uniformly 13C, 15N-labeled histidine (a) control (b) Cγ selection, and (c) Cδ2 selection acquired under a MAS speed of 8 kHz. 8, 16, and 32 scans were accumulated for (a), (b) and (c) respectively. The first 1H-13C cross polarization (CP) and CHHC contact time were fixed to 800 μs and 50 μs, respectively. A recycle delay of 2.5 s was used to collect the NMR signals. For clarity, the imidazole region was expanded on the right.
15
than 35% after two rotor periods REDOR dephasing. While the 13C signals of methine (CH) group such as 13Cδ2 and 13Cε1 sites could almost be completely removed due to the strong one-bond 1H–13C dipole interaction on the basis of our previous simulations [23]. The experimental selection efficiency for SPC-5 double quantum filtering was reported to be ca. 30% [27]. The signal transfer efficiency during CHHC period might be estimated to be 50–70%, suggesting that 13Cδ2 selection efficiency could be in the range of 15–21%. In general, our experimental efficiencies for Cγ (63%) and Cδ2 (13%) selections are considerably reasonable. It seems promising and feasible for the application of Cγ and Cδ2 spectral editing techniques to 13C, 15N uniformly labeled biological protein samples. Fig. 4 shows the Nδ1 and Nε2 editing spectra of uniformly 13C, 15 N labeled histidine. In the 15N CP/MAS control spectrum as shown in Fig. 4a, three major signals, Nα (47.6 ppm), Nδ1 (190.0 ppm) and Nε2 (176.3 ppm), could be well resolved. After Nδ1 spectral editing (Fig. 4b), the signal of Nε2 site was completely suppressed with respect to Nδ1. In the Nδ2 spectral editing experiments (Fig. 3c), only the signal of Nε2 survives though its intensity seems relatively weak, confirming the feasibility of the spectral editing technique. In comparison with SPECIFIC-CP [17,18], it is noted that the efficiency for 13C-15 TEDOR [28] is relatively lower for one-bond polarization transfer from 13C to 15N. In 15Nδ1 and 15Nε2 spectral editing experiments, the selection efficiencies are determined to be around 2% and 0.4% respectively. For uniformly 13C, 15N labeled histidine model compound, we use a relatively shorter SPECIFIC-CP contact time (350 μs) instead of 3–8 ms to allow the one-bond 13C–15N transfer and avoid the twobond 13C–15N coherence transfer. Therefore, our selection efficiencies in 15Nδ1 (2%) and 15Nε2 (0.4%) spectral editing methods are quite lower. Even though the transfer efficiencies for 15Nδ1 and 15 Nε2 spectral editing might be evaluated to be around 3–5 times for real biological protein systems, the sensitivity of the 15N signals is still too low to be well resolved for NMR detection. The 13C and 15N chemical shifts in histidine imidazole ring are characteristic to determine different existing forms in proteins. Solid-state NMR has been proved to be a very successful tool to investigate the structure and equilibrium tautomerization states of histidine in biological systems [29,30]. However, the peak overlap due to the broad line width of NMR spectra in solid state [29–31]
Fig. 4. 15N MAS NMR spectra of uniformly 13C, 15N-labeled histidine (a) control, (b) Nδ1 selection, and (c) Nε2 selection. 32, 512, and 1024 scans were accumulated for (a), (b) and (c). 13C-15N SPECIFIC-CP period was set to 350 μs to acquire the NMR spectrum. For clarity, the imidazole region was expanded on the right.
16
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
significantly limits the use of 1D ssNMR techniques to determine the structures of the imidazole ring in proteins. For example, Cross et al. [31] employed two distinct 15N site-specifically labeled histidine residues (H37) to investigate the tautomerization states of histidine in influenza A Virus M2 proton channels. However, it is extremely costly to prepare samples containing 15N sitespecifically labeled protein to conduct NMR experiments. To gain better resolution, Hong et al. [29,30,32,33] utilized twodimensional (2D) 13C-13C DARR and 13C/1H-15N HETCOR ssNMR to investigate the detailed structure, dynamic behavior of uniformly 13C, 15N labeled H37 residues in influenza A Virus M2 proton channel. The results revealed the structure–property relationship for clarifying proton conduction and gating mechanism in influenza M2 proton channels. In addition, de Groot and coworkers studied the interactions of histidines in the lightharvesting complex II with bacteriochlorophyll and found that Νε2 was ligated with Mg2+ while Nδ1 was protonated and involved in H-bonding by using 2D 13C-13C NMR experiments [13]. However, the acquisition of 2D ssNMR spectra was always timeconsuming. Therefore, the structural determination by using ssNMR requires more experimentally efficient techniques. Here, we propose several 1D solid-state NMR techniques for spectral editing of 13C and 15N sites in the histidine imidazole ring. The high efficiency of Cγ and Cδ2 might allow it feasible to select the carbon sites in histidine residues in biological proteins. However, the sensitivity of 15N NMR signals was relatively lower in the Nδ1 and Nε2 spectral editing experiments, which could be ascribed to the lower efficiency of 15N detection as well as the complexity of the pulse sequences (see Fig. 2c–d). In comparison with the 2D solid-state NMR experiments, the 1D spectral editing techniques especially Cγ and Cδ2 selection methods need less spectrometer time, which might be used in combination with REDOR for histidine backbone-side chain distances measurements [12,29]. In order to gain insights into the 1H, 13C and 15N chemical shifts variation of the imidazole ring, theoretical calculations were employed to investigate the relationship between the NMR chemical shifts and the atomic charge distributions. Firstly, three different tri-peptides including GHG, AHA and LHL were chosen as representative computational models. In each computational model, histidine could be present in three different forms: biprotonated species, neutral π and τ tautomers. Fig. S1 shows the optimized geometries of Ala-His-Ala in three different forms. In addition to the optimized structures of different tri-peptides, the NPA atomic charges were directly obtained and summarized in Table S1. To establish the relationship between 1H, 13C, 15N chemical shifts and the NPA atomic charges, we plotted the atomic charges as a function of the 1H, 13C, and 15N chemical shifts [12] in histidine imidazole ring as shown in Fig. 5. It is noteworthy that the chemical shifts of the histidine imidazole ring are roughly linearly correlated with the corresponding atomic charges. The relationships for 1H, 13C, and 15N nuclei can be described as the following equations: Q 1H ¼ 0:024ð 7 0:002Þδ1H þ 0:11ð 7 0:02Þ Q 13C ¼ 0:011ð 7 0:002Þδ13C −1:3ð 7 0:2Þ
R2 ¼ 0:80
R2 ¼ 0:63
Q 15N ¼ 0:00072ð 7 0:00016Þδ15N −0:68ð 7 0:03Þ 1
13
15
R2 ¼ 0:57
ð1Þ ð2Þ ð3Þ
It is demonstrated that H, C, N chemical shifts variation origins from the atomic charge variation in different protonation and tautomerization states, and the 1H, 13C, and 15N chemical shifts can not only function as a benchmark for the histidine existing forms, but also as an indicator to the atomic charge distribution in the histidine imidazole ring. The total atomic charges in protonated and neutral imidazole ring were determined
Fig. 5. The relationships between (a) 1H, (b) 13C, and (c) 15N chemical shifts and the calculated NPA atomic charges inside histidine imidazole ring.
to be around 0.87 and 0 respectively, which are generally in consistence with the existing forms of histidine side-chain. The imidazole side-chain of histidine serves as catalytic sites in many biological enzymes, in which the neutral histidine usually accept a proton from serine, threonine, or cysteine to activate it [34–36]. In addition, histidine can quickly shuttle protons [29,30]. Shutting protons by ring flip of histidine side-chain was proposed by Hong et al. to demonstrate the proton conduction mechanism of influenza A M2 channel [29]. The deprotonated nitrogen can adopt a proton to form a positively-charged intermediate, then transfer the proton to the neighboring groups. This progress always accompanies the change of protonation and tautomerization states in the histidine imidazole ring. In addition, it has been widely recognized that polymer containing imidazole rings could be applied as ion conductor materials [2,37,38]. Our results indicate that the atomic charge distribution altered by different
S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17
protonation states of the imidazole ring induces the 13C, 15N chemical shifts change, forming varied tautomerization states. 4. Conclusion In this work, 1D solid-state NMR methods were developed to specifically select Cγ, Cδ2, Nδ1, and Nε2 signals of imidazole ring in uniformly 13C, 15N labeled histidine sample to determine the tautomeric structure of histidine. The spectral editing techniques successfully facilitate the 13C, 15N chemical shifts assignment of histidine imidazole ring. In addition, it was revealed that the 1H, 13C and 15N chemical shifts were strongly correlated with their corresponding atomic charge in the histidine imidazole ring. The NMR chemical shift variations in different protonation and tautomerization states which are sensitive to the environmental pH were ascribed to the atomic charge change in histidine imidozale ring. Acknowledgments This work was supported by the National Natural Science Foundation of China (21003154, 20933009, 21210005 and 21221064), and the National Basic Research Program of China (2009CB918600). The authors are grateful to Shanghai Supercomputer Center (SSC, China) for the support in computing facilities.
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
[14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ssnmr.2013.05.002.
[28] [29] [30] [31] [32]
References [1] [2] [3]
K.G. Strothkamp, S.J. Lippard, Acc. Chem. Res. 15 (1982) 318–326. B.S. Hickman, M. Mascal, J.J. Titman, I.G. Wood, J. Am. Chem. Soc. 121 (1999) 11486–11490. L.-S.P. Tran, T. Urao, F. Qin, K. Maruyama, T. Kakimoto, K. Shinozaki, K. Yamaguchi-Shinozaki, Proc. Natl. Acad. Sci. USA 104 (2007) 20623–20628.
[33] [34] [35] [36] [37] [38]
17
S.J. Opella, F.M. Marassi, Chem. Rev. 104 (2004) 3587–3606. M. Hong, Y. Zhang, F. Hu, Ann. Rev. Phys. Chem. 63 (2012) 1–24. R. Tycko, Ann. Rev. Phys. Chem. 62 (2011) 279–299. A. McDermott, Ann. Rev. Biophys 38 (2009) 385–403. Y. Su, S. Li, M. Hong, Amino Acids 44 (2013) 821–833. B. Henry, P. Tekely, J.-J. Delpuech, J. Am. Chem. Soc. 124 (2002) 2025–2034. F. Cheng, H. Sun, Y. Zhang, D. Mukkamala, E. Oldfield, J. Am. Chem. Soc. 127 (2005) 12544–12554. Y. Wei, A.C. de Dios, A.E. McDermott, J. Am. Chem. Soc. 121 (1999) 10389–10394. S. Li, M. Hong, J. Am. Chem. Soc. 133 (2011) 1534–1544. J. Alia, C. Matysik, M. Soede-Huijbregts, J. Baldus, J. Raap, P. Lugtenburg, H.J. Gast, A.J. van Gorkom, Hoff, H.J.M. de Groot, J. Am. Chem. Soc. 123 (2001) 4803–4809. M.A.S. Hass, D.F. Hansen, H.E.M. Christensen, J.J. Led, L.E. Kay, J. Am. Chem. Soc. 130 (2008) 8460–8470. M. Lee, W.I. Goldburg, Phys. Rev. 140 (1965) A1261–A1271. A. Lange, S. Luca, M. Baldus, J. Am. Chem. Soc. 124 (2002) 9704–9705. M. Baldus, A.T. Petkova, J. Herzfeld, R.G. Griffin, Mol. Phys. 95 (1998) 1197–1207. A.T. Petkova, M. Baldus, M. Belenky, M. Hong, R.G. Griffin, J. Herzfeld, J. Magn. Reson. 160 (2003) 1–12. A.E. Reed, R.B. Weinstock, F. Weinhold, J. Chem. Phys. 83 (1985) 735–746. M.J. Frisch, et al., Gaussian 09, Revision B.01, Gaussian, Inc., Wallingford, CT, 2010. T. Gullion, J. Schaefer, J. Magn. Reson. 81 (1989) 196–200. S. Li, Y. Su, W. Luo, M. Hong, J. Phys. Chem. B 114 (2010) 4063–4069. S. Li, Y. Su, M. Hong, Solid State Nucl. Magn. Reson. 45–46 (2012) 51–58. M. Hohwy, C.M. Rienstra, C.P. Jaroniec, R.G. Griffin, J. Chem. Phys. 110 (1999) 7983–7992. M.G. Munowitz, R.G. Griffin, G. Bodenhausen, T.H. Huang, J. Am. Chem. Soc. 103 (1981) 2529–2533. A. Lesage, D. Sakellariou, S. Steuernagel, L. Emsley, J. Am. Chem. Soc. 120 (1998) 13194–13201. Y. Su, T. Doherty, A.J. Waring, P. Puchala, M. Hong, Biochemistry 48 (2009) 4587–4595. M. Hong, R.G. Griffin, J. Am. Chem. Soc. 120 (1998) 7113–7114. F. Hu, W. Luo, M. Hong, Science 330 (2010) 505–508. F. Hu, K. Schmidt-Rohr, M. Hong, J. Am. Chem. Soc. 134 (2012) 3703–3713. J. Hu, R. Fu, K. Nishimura, L. Zhang, H.X. Zhou, D.D. Busath, V. Vijayvergiya, T.A. Cross, Proc. Natl. Acad. Sci. USA 103 (2006) 6865–6870. M. Hong, K.J. Fritzsching, J.K. Williams, J. Am. Chem. Soc. 134 (2012) 14753–14755. Y. Su, F. Hu, M. Hong, J. Am. Chem. Soc. 134 (2012) 8693–8702. C. Brenner, Biochemistry 41 (2002) 9003–9014. C. Nishimura, Y. Ohashi, S. Sato, T. Kato, S. Tabata, C. Ueguchi, Plant Cell 16 (2004) 1365–1377. C.T. Walsh, S. Garneau-Tsodikova, G.J. Gatto, Angew. Chem. Int. Ed. 44 (2005) 7342–7372. S. Scheiner, M.Y. Yi, J. Phys. Chem. 100 (1996) 9235–9241. R. Graf, Solid State Nucl. Magn. Reson. 40 (2011) 127–133.