Published on Web 05/17/2007 A Solid State 13C NMR, Crystallographic, and Quantum Chemical Investigation of Phenylalanine and Tyrosine Residues in Dipeptides and Proteins Dushyant Mukkamala,† Yong Zhang,‡ and Eric Oldfield*,†,‡ Contribution from the Center for Biophysics and Computational Biology, 607 South Mathews AVenue, UniVersity of Illinois at Urbana-Champaign, Urbana, Illinois 61801, and Department of Chemistry, UniVersity of Illinois at Urbana-Champaign, 600 South Mathews AVenue, Urbana, Illinois 61801 Received February 20, 2007; E-mail: [email protected] Abstract: We report the results of a solid-state NMR and quantum chemical investigation of the 13Cγ NMR chemical shifts in phenylalanine and tyrosine in dipeptides and proteins. Accurate computation of the experimental shifts is shown to require a good description of local electrostatic field effects, and we find the best results (R2 ) 0.94, rmsd ) 1.6 ppm, range ) 17.1 ppm, N ) 14) by using a self-consistent reaction field continuum model. There are no obvious correlations with φ, ψ, ø1,orø2 torsion angles, unlike the results seen with other amino acids. There is, however, a linear relation between computed Cγ atomic charges and shifts for the 14 peptide as well as 18 protein residues investigated. This result is similar to the correlation reported in the 1960s between π-electron density and 13C shifts for classical 4n + 2(n ) 0, 1, 2) π-electron aromatic species, such as cyclopentadienide and the tropylium cation, and in fact, we found that the shielding/atomic charge correlation seen in the peptides and proteins is virtually identical to that seen with a broad range of aromatic carbocations/carbanions. These results suggest the dominance of an electrostatic field polarization model in which increasing π electron density results in an increase in γ C atomic charge and increased shielding (of σ11 and σ22, perpendicular to the π orbital) in Phe and Tyr, as well as in the other aromatic species. These results are of general interest since they demonstrate the importance of electrostatic field effects on Phe and Tyr Cγ chemical shifts in peptides and proteins and imply that inclusion of these effects will be necessary in order to interpret the shifts of other aromatic species, such as drug molecules, bound to proteins. Introduction a [5-19F] Trp labeled protein would clearly be too long-range 19 Folding a protein into its native conformation causes a large to affect F shifts directly. The results of a number of range of chemical shift nonequivalence, of obvious importance investigations have shown that most chemical shifts in proteins from the perspective of NMR structure determination. For 1H, (including metal shifts and paramagnetic or hyperfine shifts) 6,7 chemical shifts due to folding are typically about 2 ppm; for can now be computed with good accuracy, but a remaining γ 13C, chemical shift ranges due to folding are typically ∼10 ppm; problem has been that of C of the aromatic amino acids, γ for 19F (in labeled proteins), ∼20 ppm, and for 15N, up to about phenylalanine, and tyrosine. The shift ranges for C in Phe and 8 ∼ - 35 ppm. For 1H, quite good correlations have been found Tyr were found in early studies to be quite small ( 3 4 ppm), 8,9 ∼ - between experiment and prediction using, e.g., ring-current shifts but the shifts of Trp residues were much larger ( 6 7 ppm), and empirical descriptions of electrostatics, but for the heavier and this enhanced shift range facilitates the overall precision nuclei, 13C, 15N, and 19F, quantum chemical methods have to (versus experiment) of quantum chemical shielding predictions, ∼ - γ 2 be used.1 In previous work, we found that backbone (13CR) shifts which are typically 1 2 ppm for Trp C . Here, we report the first 13C NMR chemical shift predictions for Cγ in Phe and were dominated by electronic effects due to φ, ψ differences, R and in sidechains, ø /ø or other (gamma-gauche) effects Tyr residues, both in dipeptides and in proteins. Unlike C (and 1 2 â contribute to shielding. Similar conformational contributions C ), there are no obvious correlations with either the backbone were found to dominate 15N shifts,1-3 but for 19F in aromatic torsion angles or ø1/ø2. Rather, our results suggest that electro- amino acids,1,4,5 purely electrostatic field effects dominated static field effects make an important contribution to shielding shielding, since backbone or side chain torsional effects for say (4) Pearson, J. G.; Montez, B.; Le, H.; Oldfield, E.; Chien, E. Y.; Sligar, S. G. Biochemistry 1997, 36, 3590. † Center for Biophysics and Computational Biology. (5) Pearson, J. G.; Oldfield, E.; Lee, F. S.; Warshel, A. J. Am. Chem. Soc. ‡ Department of Chemistry. 1993, 115, 6851. (1) de Dios, A. C.; Pearson, J. G.; Oldfield, E. Science 1993, 260, 1491. (6) Mao, J.; Zhang, Y.; Oldfield, E. J. Am. Chem. Soc. 2002, 124, 13911. (2) Sun, H.; Oldfield, E. J. Am. Chem. Soc. 2004, 126, 4726. (7) Zhang, Y.; Sun, H.; Oldfield, E. J. Am. Chem. Soc. 2005, 127, 3652. (3) Szabo, C. M.; Sanders, L. K.; Le, H. C.; Chien, E. Y.; Oldfield, E. FEBS (8) Oldfield, E.; Norton, R. S.; Allerhand, A. J. Biol. Chem. 1975, 250, 6381. Lett. 2000, 482, 25. (9) Oldfield, E.; Allerhand, A. J. Biol. Chem. 1975, 250, 6403. 10.1021/ja071227y CCC: $37.00 © 2007 American Chemical Society J. AM. CHEM. SOC. 2007, 129, 7385-7392 9 7385 ARTICLES Mukkamala et al. and that use of a “solvent” (self-consistent reaction field, SCRF) model enables the best accord with experiment. Remarkably, we also find that the computed shifts are highly correlated with Cγ atomic charges, reminiscent of the correlations reported many years ago10-14 between 13C shifts and π-electron densities in a variety of other classical aromatic 4n + 2 π-electron systems. Experimental Details NMR Spectroscopy: 13C NMR spectra were obtained by using the cross-polarization “magic-angle” sample-spinning technique15,16 with either full proton decoupling17 or interrupted decoupling18 (using a dipolar dephasing time of 140 µs for phenylalanine and tyrosine residues), for selection of the nonprotonated aromatic carbons (Cγ and Cú). The 1H and 13C90° pulse widths used were 2.6 µs, and the spectra in all cases were collected using a recycle time of 5 s. The chemical shifts were referenced to external glycine, setting the Gly CR carbon to 43.6 ppm downfield from tetramethylsilane (TMS). All the experiments were performed on a Varian (Palo Alto, CA) Infinityplus 600 MHz (1H) NMR spectrometer at a 10 kHz spinning speed using a 3.2 mm Varian/Chemagnetics HXY probe. Fully 1H decoupled experiments were carried out at two different spinning speeds (8 kHz and 10 kHz), in order to determine the isotropic chemical shifts. The compounds for the NMR experiments were obtained from Bachem (King of Prussia, PA). X-ray Crystallography. The structures of the 11 compounds (1- 11) investigated are shown in Supporting Information Figure S1. To verify their actual structures, we obtained powder diffraction data for each system and compared the results with calculated powder spectra for crystal structures obtained from the Cambridge Crystallographic Data Center (CCDC). The powder spectra for each sample were calculated by using DIFFRACplus TOPAS.19 For six dipeptides (1, 2, Figure 1. Structures and models: (a) Crystal structure of compound 6; 4, 7-9), we found a very good match between the calculated and (b) crystal structure of compound 11; (c) N-formyl-phenylalanine-amide model; (d) N-terminus charged model; (e) C-terminus charged model; (f) experimental powder diffraction data, so these structures were used monomer from crystal structure of Phe-Tyr (11); and (g) monomer from directly in the calculations. For three compounds (3, 5, 10), we obtained crystal structure of Gly-Phe (6) with surrounding hydrogen-bond partners. a good match between the computed and experimental diffraction results after refining the unit-cell parameters using the Pawley fitting method,20 approach as that used in studying tryptophan chemical shifts,2 with the plus 19 as incorporated in DIFFRAC TOPAS. The experimental and torsion angles (φ, ψ, ø1, and ø2, as appropriate) set to the values seen calculated powder patterns for two compounds (6, 11) did not match in the X-ray structures. After this, we performed calculations on charged those reported, so we recrystallized them and determined their structures models of the N-formyl-phenylalanine-amide structures, converting the + - by X-ray crystallography. Compound 6 was crystallized from water, terminal N-formyl group to NH3 or the amide to CO2 , for N-terminal while 11 was crystallized from 5% DMSO-water, and the structures and C-terminal models, respectively (Figure 1d and e). The next set of so obtained are shown in Figure 1a,b. Full crystallographic details are calculations used the monomers found in the X-ray crystal structures, given in Table S1 in the Supporting Information. as shown for example in Figure 1f. In the fourth set of calculations, Computational Aspects: We performed eight sets of calculations we used hydrogen-bond partner cluster models in which the effects of in order to compute the 13C NMR shieldings for the phenylalanine and neighboring residues were included by incorporating acetate or me- tyrosine dipeptides (1-11) shown in Figure S1. In six cases, we used thylammonium ions, or formamide or methanol molecules, to represent the Hartree-Fock method as incorporated in Gaussian 9821 and the lattice partners, e.g., Figure 1g. And, in the case of tyrosine, since Gaussian 03,22 using the locally dense basis set scheme (6-311++G some structures had what appeared to be bad (∼0.6 Å) OH bond lengths, (2d,2p)/6-31G*) with the denser basis set on the atoms of interest (and we performed geometry optimization of the Cú hydroxyl group (using their nearest neighbors), combined with the gauge-including atomic the HF method and a locally dense basis set scheme, 6-311++G orbitals (GIAO) method.23 The first set of calculations utilized the (2d,2p)/6-31G*).