Solid State Nuclear Magnetic Resonance 54 (2013) 13–17

Contents lists available at ScienceDirect

Solid State Nuclear Magnetic Resonance

0926-20http://d

n CorrE-m1 Eq

journal homepage: www.elsevier.com/locate/ssnmr

13C and 15N spectral editing inside histidine imidazole ring throughsolid-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, WuhanCenter for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, The Chinese Academy of Sciences, Wuhan 430071, Chinab Department of Chemistry, Iowa State University, Ames, IA 50011, USA

a r t i c l e i n f o

Article history:Received 26 November 2012Received in revised form16 April 2013Available online 15 May 2013

Keywords:Solid state NMRSpectral editingHistidineDFT calculation

40/$ - see front matter Crown Copyright & 20x.doi.org/10.1016/j.ssnmr.2013.05.002

esponding author. Fax: +86 27 87199291.ail address: lishenhui@wipm.ac.cn (S. Li).ual contribution.

a b s t r a c t

Histidine usually exists in three different forms (including biprotonated species, neutral τ and πtautomers) at physiological pH in biological systems. The different protonation and tautomerizationstates 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 inhistidine imidazole ring, which provides a benchmark to distinguish the existing forms of histidine. Theselections of 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 sites were successfully achieved based on one-bond homo- andhetero-nuclear dipole interactions. Moreover, it was demonstrated that 1H, 13C, and 15 chemical shiftswere roughly linearly correlated with the corresponding atomic charge in histidine imidazole ring bytheoretical calculations. Accordingly, the 1H, 13C and 15N chemical shifts variation in different protonationand tautomerization states could be ascribed to the atomic charge change due to proton transfer inbiological process.

Crown Copyright & 2013 Published by Elsevier Inc. All rights reserved.

1. Introduction

Histidine is an essential cationic residue in proteins because ofits structurally flexible imidazole group, e.g. the unprotonatedimidazole can function as a weak base, while the protonated formis a weak acid. This aromatic heterocyclic ring is a commoncoordinating ligand in metalloproteins [1]. In particular, the specialresonance structure makes it a common participant in variousenzyme-catalyzed reactions and electrical conduction systems [2].Histidine can also play a key role in stabilizing the foldedstructures of proteins [3]. In general, it is well recognized thathistidine could exist in three different forms: a protonatedimidazolium form and two neutral tautomers, namely π and τ atphysiological pH.

Solid state NMR has been proved to be a very well-establishedtechnique to investigate the detailed structure and dynamicsproperty in biological systems [4–8]. So far, a number of reportsfocus on the study of 13C, 15N chemical shifts tensor and N-H bondlength in histidine imidazole ring by means of solid-state NMR andquantum chemical calculations [9–12]. For example, Henry et al.[9] determined the pKa of acid–base properties of histidine byquantitatively measuring the population ratios between different

13 Published by Elsevier Inc. All r

forms of histidine during three ionization steps. Oldfield et al. [10]found that the Cγ and Cδ2 NMR chemical shifts in imidazole ringwere highly correlated, which enabled good predictions of tauto-meric states of histidine. Additionally, McDermott et al. [11]suggested that the 15N chemical shift anisotropy (CSA) especiallythe δ22 tensor in protonated nitrogen sites could provide valuableinformation on the hydrogen-bonding geometry of imidazole ring.More recently, the 1H, 13C and 15N chemical shifts, hydrogen bondinteraction as well as the side-chain conformations of histidine indifferent protonation and tautomerization states have been com-prehensively investigated by advanced solid-state NMR spectro-scopy [12].

In general, the 13C and 15N NMR chemical shifts in histidineimidazole ring can be utilized as a benchmark to distinguishdifferent existing forms of histidine. However, the identificationof chemical shifts in biological systems is usually subject to theresolution limitation of solid-state NMR and thus could be ambig-uous or even infeasible [13,14]. In this work, we hereby developedone-dimensional (1D) 13C and 15N MAS NMR spectral editingmethods to facilitate the chemical shifts assignments for uniformly13C, 15N labeled histidine imidazole ring via one-bond dipolarinteraction. Meanwhile, the relationship between 1H, 13C and 15Nchemical shifts and atomic charge distribution inside histidineimidazole ring was established to clarify the chemical shiftsvariation mechanism in different protonation and tautomerizationstates.

ights reserved.

S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–1714

2. Experimental

2.1. Sample preparation

13C, 15N-labeled (98%) histidine hydrochloride monohydratewas purchased from Sigma-Aldrich Corporation without furthertreatment. A total mass of 10 mg dry power was directly packedinto a 4 mm rotor for ssNMR measurements.

2.2. Solid-state NMR

Solid-state NMR experiments were carried out on a BrukerAvance 500 MHz spectrometer equipped with 4-mm triple-reso-nance MAS probes under a magic-angle-spinning speed of 8.0 kHz.Typical radiofrequency (RF) field strengths were 12–50 kHz for 13Cand 15N, and 50–83 kHz for 1H. 13C and 15N chemical shifts werereferenced to the adamantane CH2 signal at 38.5 ppm and the 15Nresonance of 15NH4Cl at 35.9 ppm, respectively. The RF field ofFSLG [15], LG-CP, TPPM in 1H channel was set to 81, 50 and 62.5–83 kHz respectively. 1H-13C cross polarization contact time of800 μs and a recycle delay of 2.5 s were used to collect the NMRsignals. The CP contact time during CHHC [16] and 13C–15NSPECIFIC-CP [17,18] period was fixed to 50 μs and 350 μs, respec-tively. Typically, 16, 32, 512, and 1024 scans were accumulated forthe 13Cγ, 13Cδ2, 15Nδ1, and 15Nε2 selection experiments, respectively.

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 wasanchored into two neutral residues, were chosen as the computa-tional models. Histidine could exist at any forms includingbiprotonated, neutral π tautomer and τ tautomer in GHG, AHAand 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 byusing the Becke's three-parameter hybrid method with the Lee–Yang–Parr correlation functional (B3LYP) and the 6–31Gnn basisset. All the calculations in this work were carried out by usingGaussian09 software package [20].

3. Results and discussion

The existing forms of histidine imidazole ring in biologicalproteins strongly depend on the environmental pH. Fig. 1 showsthe schematic structures of three most favorable forms of histidinein different protonated and tautomeric states. As shown in Fig. 1,the biprotonated species, neutral τ tautomer and π tautomer ofimidazole ring differ from the protonated and tautomeric states ofthe Nδ1 and Nε2 sites. Meanwhile, the quaternary and tertiary

Fig. 1. Schematic structures of histidine in different protonation and tautomeriza-tion states.

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 isdesirable.

Fig. 2 shows the NMR pulse sequences used for Cγ, Cδ2, Nδ1 andNε2 spectral editing of histidine imidazole ring. The idea of thepulse sequence in Fig. 2a is to select quaternary carbon (Cγ) fromtertiary carbons (Cε1 and Cδ2) in histidine imidazole ring. In orderto distinguish Cγ from Cε1 and Cδ2 sites, 1H-13C dipole dephasingwas performed by using two rotor periods 1H-13C dipole recou-pling in REDOR [21,22]. For 13Cγ selection scheme, all of the 13CNMR signals could be irradiated during the first CP. Both the 13Cδ2and 13Cε1 signals could be removed due to the strong one-bond1H-13C dipole interaction while 13Cγ resonance could still surviveafter 1H-13C REDOR dipole dephasing according to the previoussimulations [22,23]. Therefore, the ring Cγ signal could be speci-fically selected according to the proposed strategy. Fig. 2b showsthe pulse sequence for Cδ2 selection. For 13Cδ2 spectral editing,all of the 13C signals could be excited by the first CP. Then, only thedirectly bonded 13C-13C pair signals (e.g. 13Cγ and 13Cδ2) couldsurvive 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 selectthe carbons which are directly bonded with protons. Therefore,

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 15

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 magneti-zation to the neighboring Nδ1 and Nε2 sites respectively via one-bond polarization transfer. Fig. 2c–d shows the spectral editingpulse sequences for selection of the Nδ1 and Nε2 sites in histidineimidazole ring. 13C–15N SPECIFIC-CP [17,18] with short CP contacttime (350 μs) was employed to allow one-bond signal transfer andavoid long-range polarization transfer from Cγ and Cδ2 to Nδ1 andNε2 sites, respectively. (see Fig. 2c–d) Therefore, all of the 13Cγ,13Cδ2, 15Nδ1, and 15Nδ2 signals in uniformly 13C, 15N labeledhistidine sample could be selected separately.

Uniformly 13C, 15N labeled histidine was used as model com-pound to demonstrate the feasibility of the proposed 13C and 15Nspectral editing techniques in histidine imidazole ring. Fig. 3shows the Cγ and Cδ2 editing spectra of uniformly 13C, 15N labeledhistidine. In the 13C MAS NMR control spectra of histidine, all ofthe 13C signals could be clearly observed in Fig. 3a and the 13CNMR signals of imidazole ring were located in a region from 110 to140 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 separatelyselected. It is noteworthy that the Cδ2 and Cε1 signals were almostcompletely suppressed by 1H-13C dipole dephasing. In comparisonwith 1H-13C REDOR, DIPSHIFT [25] cannot largely eliminate thesignals from Cε1 and Cδ2 sites under 8 kHz MAS. Fig. 3c shows theCδ2 selection spectra, in which the major imidazole resonance at119.4 ppm due to Cδ2 site remains. Besides the CHHC [16] scheme,MAS-J-HMQC technique [26] could be alternatively utilized todistinguish the proton-bonded carbon from quaternary carbonspecies. As shown in Fig. 3b, the Cα and Cβ signals were removedand C′ signal was still retained in the dipole dephasing experi-ment, whereas Cα and Cβ signals instead of C′ could survive afterCδ2 spectral editing as displayed in Fig. 3c. As all of the Cα, Cβ and C′signals distinctly deviated from the imidazole region, the appear-ance of the Cα, Cβ and C′ signals has negligible influence on thechemical shifts assignment of the signals inside histidine imida-zole ring. The selection efficiencies in Cγ and Cδ2 spectral editingexperiments 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) Cγselection, and (c) Cδ2 selection acquired under a MAS speed of 8 kHz. 8, 16, and 32scans were accumulated for (a), (b) and (c) respectively. The first 1H-13C crosspolarization (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. Forclarity, the imidazole region was expanded on the right.

than 35% after two rotor periods REDOR dephasing. While the 13Csignals of methine (CH) group such as 13Cδ2 and 13Cε1 sites couldalmost be completely removed due to the strong one-bond 1H–13Cdipole interaction on the basis of our previous simulations [23].The experimental selection efficiency for SPC-5 double quantumfiltering was reported to be ca. 30% [27]. The signal transferefficiency during CHHC period might be estimated to be 50–70%,suggesting that 13Cδ2 selection efficiency could be in the range of15–21%. In general, our experimental efficiencies for Cγ (63%) andCδ2 (13%) selections are considerably reasonable. It seems promis-ing and feasible for the application of Cγ and Cδ2 spectral editingtechniques to 13C, 15N uniformly labeled biological proteinsamples.

Fig. 4 shows the Nδ1 and Nε2 editing spectra of uniformly 13C,15N labeled histidine. In the 15N CP/MAS control spectrum asshown in Fig. 4a, three major signals, Nα (47.6 ppm), Nδ1

(190.0 ppm) and Nε2 (176.3 ppm), could be well resolved. AfterNδ1 spectral editing (Fig. 4b), the signal of Nε2 site was completelysuppressed with respect to Nδ1. In the Nδ2 spectral editingexperiments (Fig. 3c), only the signal of Nε2 survives though itsintensity seems relatively weak, confirming the feasibility of thespectral editing technique. In comparison with SPECIFIC-CP[17,18], it is noted that the efficiency for 13C-15 TEDOR [28] isrelatively lower for one-bond polarization transfer from 13C to 15N.In 15Nδ1 and 15Nε2 spectral editing experiments, the selectionefficiencies are determined to be around 2% and 0.4% respectively.For uniformly 13C, 15N labeled histidine model compound, we usea relatively shorter SPECIFIC-CP contact time (350 μs) instead of3–8 ms to allow the one-bond 13C–15N transfer and avoid the two-bond 13C–15N coherence transfer. Therefore, our selection efficien-cies in 15Nδ1 (2%) and 15Nε2 (0.4%) spectral editing methods arequite lower. Even though the transfer efficiencies for 15Nδ1 and15Nε2 spectral editing might be evaluated to be around 3–5 timesfor real biological protein systems, the sensitivity of the 15N signalsis still too low to be well resolved for NMR detection.

The 13C and 15N chemical shifts in histidine imidazole ring arecharacteristic to determine different existing forms in proteins.Solid-state NMR has been proved to be a very successful tool toinvestigate the structure and equilibrium tautomerization states ofhistidine in biological systems [29,30]. However, the peak overlapdue 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 accumulatedfor (a), (b) and (c). 13C-15N SPECIFIC-CP period was set to 350 μs to acquire the NMRspectrum. For clarity, the imidazole region was expanded on the right.

Fig. 5. The relationships between (a) 1H, (b) 13C, and (c) 15N chemical shifts and thecalculated NPA atomic charges inside histidine imidazole ring.

S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–1716

significantly limits the use of 1D ssNMR techniques to determinethe structures of the imidazole ring in proteins. For example, Crosset al. [31] employed two distinct 15N site-specifically labeledhistidine residues (H37) to investigate the tautomerization statesof histidine in influenza A Virus M2 proton channels. However, it isextremely costly to prepare samples containing 15N site-specifically labeled protein to conduct NMR experiments. To gainbetter resolution, Hong et al. [29,30,32,33] utilized two-dimensional (2D) 13C-13C DARR and 13C/1H-15N HETCOR ssNMRto investigate the detailed structure, dynamic behavior of uni-formly 13C, 15N labeled H37 residues in influenza A Virus M2proton channel. The results revealed the structure–property rela-tionship for clarifying proton conduction and gating mechanism ininfluenza M2 proton channels. In addition, de Groot and co-workers studied the interactions of histidines in the light-harvesting complex II with bacteriochlorophyll and found thatΝε2 was ligated with Mg2+ while Nδ1 was protonated and involvedin H-bonding by using 2D 13C-13C NMR experiments [13]. How-ever, the acquisition of 2D ssNMR spectra was always time-consuming. Therefore, the structural determination by usingssNMR requires more experimentally efficient techniques.

Here, we propose several 1D solid-state NMR techniques forspectral editing of 13C and 15N sites in the histidine imidazole ring.The high efficiency of Cγ and Cδ2 might allow it feasible to selectthe carbon sites in histidine residues in biological proteins.However, the sensitivity of 15N NMR signals was relatively lowerin the Nδ1 and Nε2 spectral editing experiments, which could beascribed to the lower efficiency of 15N detection as well as thecomplexity of the pulse sequences (see Fig. 2c–d). In comparisonwith the 2D solid-state NMR experiments, the 1D spectral editingtechniques especially Cγ and Cδ2 selection methods need lessspectrometer time, which might be used in combination withREDOR for histidine backbone-side chain distances measurements[12,29].

In order to gain insights into the 1H, 13C and 15N chemical shiftsvariation of the imidazole ring, theoretical calculations wereemployed to investigate the relationship between the NMR che-mical shifts and the atomic charge distributions. Firstly, threedifferent tri-peptides including GHG, AHA and LHL were chosen asrepresentative computational models. In each computationalmodel, histidine could be present in three different forms:biprotonated species, neutral π and τ tautomers. Fig. S1 showsthe 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 inTable S1. To establish the relationship between 1H, 13C, 15Nchemical shifts and the NPA atomic charges, we plotted the atomiccharges as a function of the 1H, 13C, and 15N chemical shifts [12] inhistidine imidazole ring as shown in Fig. 5. It is noteworthy thatthe chemical shifts of the histidine imidazole ring are roughlylinearly correlated with the corresponding atomic charges. Therelationships for 1H, 13C, and 15N nuclei can be described as thefollowing equations:

Q1H ¼ 0:024ð70:002Þδ1H þ 0:11ð70:02Þ R2 ¼ 0:80 ð1Þ

Q13C ¼ 0:011ð70:002Þδ13C−1:3ð70:2Þ R2 ¼ 0:63 ð2Þ

Q15N ¼ 0:00072ð70:00016Þδ15N−0:68ð70:03Þ R2 ¼ 0:57 ð3ÞIt is demonstrated that 1H, 13C, 15N chemical shifts variation

origins from the atomic charge variation in different protonationand tautomerization states, and the 1H, 13C, and 15N chemicalshifts can not only function as a benchmark for the histidineexisting forms, but also as an indicator to the atomic chargedistribution in the histidine imidazole ring. The total atomiccharges in protonated and neutral imidazole ring were determined

to be around 0.87 and 0 respectively, which are generally inconsistence with the existing forms of histidine side-chain.

The imidazole side-chain of histidine serves as catalytic sites inmany biological enzymes, in which the neutral histidine usuallyaccept 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 proposedby Hong et al. to demonstrate the proton conduction mechanismof influenza A M2 channel [29]. The deprotonated nitrogen canadopt a proton to form a positively-charged intermediate, thentransfer the proton to the neighboring groups. This progressalways accompanies the change of protonation and tautomeriza-tion states in the histidine imidazole ring. In addition, it has beenwidely recognized that polymer containing imidazole rings couldbe applied as ion conductor materials [2,37,38]. Our resultsindicate that the atomic charge distribution altered by different

S. Li et al. / Solid State Nuclear Magnetic Resonance 54 (2013) 13–17 17

protonation states of the imidazole ring induces the 13C, 15Nchemical shifts change, forming varied tautomerization states.

4. Conclusion

In this work, 1D solid-state NMR methods were developed tospecifically select Cγ, Cδ2, Nδ1, and Nε2 signals of imidazole ring inuniformly 13C, 15N labeled histidine sample to determine thetautomeric structure of histidine. The spectral editing techniquessuccessfully facilitate the 13C, 15N chemical shifts assignment ofhistidine imidazole ring. In addition, it was revealed that the 1H, 13Cand 15N chemical shifts were strongly correlated with their corre-sponding atomic charge in the histidine imidazole ring. The NMRchemical shift variations in different protonation and tautomeriza-tion states which are sensitive to the environmental pH wereascribed to the atomic charge change in histidine imidozale ring.

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China (21003154, 20933009, 21210005 and21221064), and the National Basic Research Program of China(2009CB918600). The authors are grateful to Shanghai Super-computer Center (SSC, China) for the support in computingfacilities.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ssnmr.2013.05.002.

References

[1] K.G. Strothkamp, S.J. Lippard, Acc. Chem. Res. 15 (1982) 318–326.[2] B.S. Hickman, M. Mascal, J.J. Titman, I.G. Wood, J. Am. Chem. Soc. 121 (1999)

11486–11490.[3] 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.

[4] S.J. Opella, F.M. Marassi, Chem. Rev. 104 (2004) 3587–3606.[5] M. Hong, Y. Zhang, F. Hu, Ann. Rev. Phys. Chem. 63 (2012) 1–24.[6] R. Tycko, Ann. Rev. Phys. Chem. 62 (2011) 279–299.[7] A. McDermott, Ann. Rev. Biophys 38 (2009) 385–403.[8] Y. Su, S. Li, M. Hong, Amino Acids 44 (2013) 821–833.[9] B. Henry, P. Tekely, J.-J. Delpuech, J. Am. Chem. Soc. 124 (2002) 2025–2034.[10] F. Cheng, H. Sun, Y. Zhang, D. Mukkamala, E. Oldfield, J. Am. Chem. Soc. 127

(2005) 12544–12554.[11] Y. Wei, A.C. de Dios, A.E. McDermott, J. Am. Chem. Soc. 121 (1999)

10389–10394.[12] S. Li, M. Hong, J. Am. Chem. Soc. 133 (2011) 1534–1544.[13] 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.

[14] 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.

[15] M. Lee, W.I. Goldburg, Phys. Rev. 140 (1965) A1261–A1271.[16] A. Lange, S. Luca, M. Baldus, J. Am. Chem. Soc. 124 (2002) 9704–9705.[17] M. Baldus, A.T. Petkova, J. Herzfeld, R.G. Griffin, Mol. Phys. 95 (1998)

1197–1207.[18] A.T. Petkova, M. Baldus, M. Belenky, M. Hong, R.G. Griffin, J. Herzfeld, J. Magn.

Reson. 160 (2003) 1–12.[19] A.E. Reed, R.B. Weinstock, F. Weinhold, J. Chem. Phys. 83 (1985) 735–746.[20] M.J. Frisch, et al., Gaussian 09, Revision B.01, Gaussian, Inc., Wallingford,

CT, 2010.[21] T. Gullion, J. Schaefer, J. Magn. Reson. 81 (1989) 196–200.[22] S. Li, Y. Su, W. Luo, M. Hong, J. Phys. Chem. B 114 (2010) 4063–4069.[23] S. Li, Y. Su, M. Hong, Solid State Nucl. Magn. Reson. 45–46 (2012) 51–58.[24] M. Hohwy, C.M. Rienstra, C.P. Jaroniec, R.G. Griffin, J. Chem. Phys. 110 (1999)

7983–7992.[25] M.G. Munowitz, R.G. Griffin, G. Bodenhausen, T.H. Huang, J. Am. Chem. Soc.

103 (1981) 2529–2533.[26] A. Lesage, D. Sakellariou, S. Steuernagel, L. Emsley, J. Am. Chem. Soc. 120

(1998) 13194–13201.[27] Y. Su, T. Doherty, A.J. Waring, P. Puchala, M. Hong, Biochemistry 48 (2009)

4587–4595.[28] M. Hong, R.G. Griffin, J. Am. Chem. Soc. 120 (1998) 7113–7114.[29] F. Hu, W. Luo, M. Hong, Science 330 (2010) 505–508.[30] F. Hu, K. Schmidt-Rohr, M. Hong, J. Am. Chem. Soc. 134 (2012) 3703–3713.[31] 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.[32] M. Hong, K.J. Fritzsching, J.K. Williams, J. Am. Chem. Soc. 134 (2012)

14753–14755.[33] Y. Su, F. Hu, M. Hong, J. Am. Chem. Soc. 134 (2012) 8693–8702.[34] C. Brenner, Biochemistry 41 (2002) 9003–9014.[35] C. Nishimura, Y. Ohashi, S. Sato, T. Kato, S. Tabata, C. Ueguchi, Plant Cell 16

(2004) 1365–1377.[36] C.T. Walsh, S. Garneau-Tsodikova, G.J. Gatto, Angew. Chem. Int. Ed. 44 (2005)

7342–7372.[37] S. Scheiner, M.Y. Yi, J. Phys. Chem. 100 (1996) 9235–9241.[38] R. Graf, Solid State Nucl. Magn. Reson. 40 (2011) 127–133.