4690 J. Am. Chem. Soc. 1999, 121, 4690-4695 How Tetrahedral Are Methyl Groups in Proteins? A Liquid Crystal NMR Study Marcel Ottiger and Ad Bax* Contribution from the Laboratory of Chemical Physics, National Institute of Diabetes and DigestiVe and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892-0520 ReceiVed December 29, 1998. ReVised Manuscript ReceiVed March 30, 1999 Abstract: A small degree of protein alignment with an external magnetic field can be obtained in a dilute aqueous liquid crystalline solution of dimyristoylphosphatidylcholine (DMPC) and dihexanoylphosphatidyl- choline (DHPC). It is demonstrated that residual one-bond 13C-13C and 13C-1H dipolar couplings of methyl groups in weakly aligned human ubiquitin can be measured with high accuracy. Experimentally, the ratio between 13C-1H and 13C-13C dipolar couplings is found to be -3.17 ( 0.03. Assuming a static conformation of the methyl group, rapidly spinning about its 3-fold symmetry axis, this ratio corresponds to an average C-C-H bond angle of 110.9 ( 1°, which is larger than the ideal tetrahedral value of 109.5°. Data indicate that the geometry of the various methyl groups is quite uniform, but that small (e1°) deviations between the C-C vector and the axis connecting the methyl carbon to the geometric center of the three methyl protons may occur. The largest outlier is found for Ala46, which has a positive φ backbone angle, causing its methyl group to be within van der Waals contact of the preceding carbonyl oxygen. Introduction The issue of methyl group geometry recently has regained new interest9-11 because 13C methyl group NMR relaxation Methyl groups commonly are assumed to adopt an ideal parameters present a potentially powerful probe for the study tetrahedral geometry, with H-C-H and C-C-H bond angles - of side-chain mobility in proteins.9,10,12 17 Relaxation of such of 109.5°. However, small deviations from such idealized 13C signals is dominated by the dipolar interaction with the tetrahedral geometry previously were found on the basis of NMR methyl protons, but fast rotation about the methyl group’s C dipolar coupling measurements of small molecules dissolved 3 1-4 symmetry axis reduces the effective dipolar couplings by in nematic liquid crystals, and single-crystal neutron diffrac- 2 P2(cos â), with P2(x) ) (3x - 1)/2, and â being the C-C-H tion studies of alanine and valine.5,6 Neutron diffraction for these angle. Similarly, when studying methyl group dynamics using two amino acids yielded C-C-H angles of 110.0° and 111.9°, deuterium NMR, after averaging over the 3-fold rotation the respectively. Small molecule liquid crystal NMR cannot deter- effective quadrupole coupling scales with P2(cos â). In the mine the angle directly, unless an assumption about bond length ) ° 1 vicinity of â 109.5 , P2(cos â) is a particularly steep function is made. However, the ratio of the one-bond DCH and two- 2 of â, and accurate knowledge of â is therefore essential for bond DHH dipolar couplings is strongly correlated with this 1 making quantitative interpretation of such relaxation measure- angle. This ratio ranges from 0.68 in acetaldehyde to 0.88 in - 3 ments. Substantial differences, on the order of 30 40% between CH I. This is in qualitative agreement with results from gas- R R 3 the order parameters observed for Ala C -H and Câ-Hâ in phase rotational spectroscopy, which yields the ideal C-C-H 3 the protein staphylococcal nuclease,9 for example, potentially angle of 109.5° for the methyl group of acetaldehyde (but a very short C-H bond length of 1.073 ( 0.002 Å)7 and 113.0° (9) Nicholson, L. K.; Kay, L. E.; Torchia, D. A. In NMR Spectroscopy for CH I(r ) 1.085 Å).8 Thus, these results indicate that and its Application to Biomedical Research; Sarkar, S. K., Ed.; Elsevier: 3 CH New York, 1996; pp 241-279. substantial variation in methyl group geometry can indeed occur, (10) Wand, A. J.; Urbauer, J. L.; McEvoy, R. P.; Bieber, R. J. depending on the substituent. Biochemistry 1996, 35, 6116-6125. (11) Chatfield, D. C.; Szabo, A.; Brooks, B. R. J. Am. Chem. Soc. 1998, * Corresponding author. 120, 5301-5311. (1) Emsley, J. W.; Lindon, J. C.; Tabony, J. J. Chem. Soc., Faraday (12) Jones, W. C.; Rothgeb, T. M.; Gurd, F. R. N. J. Biol. Chem. 1976, Trans. 2 1975, 71, 586-595. 251, 7452-7460. Richarz, R.; Nagayama, K.; Wu¨thrich, K. Biochemistry (2) Khetrapal, C. L.; Saupe, A. Mol. Cryst. Liq. Cryst. 1973, 19, 367- 1980, 19, 5189-5196. 381 (13) Nicholson, L. K.; Kay, L. E.; Baldisseri, D. M.; Arango, J.; Young, (3) Wooton, J. B.; Savitsky, G. B.; Jacobus, J.; Beyerlein, A. L. J. Chem. P. E.; Bax, A. ; Torchia, D. A. 1992, 31, 5253-5263. Phys. 1977, 66, 4226-4230. Wooton, J. B.; Savitsky, G. B.; Jacobus, J.; (14) Muhandiram, D. R.; Yamazaki, T.; Sykes, B. D.; Kay, L. E. J. Am. Beyerlein, A. L.; Emsley, J. W. J. Chem. Phys. 1979, 70, 438-442. Chem. Soc. 1995, 117, 11536-11544. Yang, D. W.; Mok, Y. K.; Forman- (4) Emsley, J. W.; Longeri, M.; Veracini, C. A.; Catalone, D.; Pedulli, Kay, J. D.; Farrow, N. A.; Kay, L. E. J. Mol. Biol. 1997, 272, 790-804. G. F. J. Chem. Soc., Perkin Trans. 2 1982, 1289-1296. Yang, D. W.; Mittermaier, A.; Mok, Y. K.; Kay, L. E. J. Mol. Biol. 1998, (5) Lehnmann, M. S.; Koetzle, T. F.; Hamilton, W. C. J. Am. Chem. 276, 939-954. Soc. 1972, 94, 2657-2660. (15) LeMaster, D. M.; Kushlan, D. M. J. Am. Chem. Soc. 1996, 118, (6) Koetzle, T. F.; Golic, L.; Lehnmann, M. S.; Verbist, J. J.; Hamilton, 9255-9264. W. C. J. Chem. Phys. 1974, 60, 4690-4696. (16) Straus, S. K.; Bremi, T.; Ernst, R. R. J. Biomol. NMR 1997, 10, (7) Duncan, J. L. J. Mol. Struct. 1970, 6, 447-459. 119-128. (8) Iijima, T.; Kimura, M. Bull. Chem. Soc. Jpn. 1969, 42, 2159-2164. (17) Lee, A. L.; Urbauer, J. L.; Wand, A. J. J. Biomol. NMR 1997, 9, Iijima, T.; Tsuchiva, S. J. Mol. Spectrosc. 1972, 44,88-107. 437-440. 10.1021/ja984484z This article not subject to U.S. Copyright. Published 1999 by the American Chemical Society Published on Web 05/04/1999 Methyl Group Geometry in Proteins J. Am. Chem. Soc., Vol. 121, No. 19, 1999 4691 could be explained by increasing the C-C-H angle by only and cetyltrimethylammonium bromide (CTAB).28 Final sample condi- 3° from its ideal value. However, ab initio calculations on tions were: 0.7 mM protein, 5% (w/w) DMPC/DHPC/CTAB in a molar individual amino acids or molecular dynamics calculations on ratio of 30:10:1 in water, 10 mM phosphate buffer, pH 6.6, 7% D2O. - - NMR spectra were recorded on a Bruker DMX750 spectrometer the intact protein yield rather uniform C C H angles of ca. 1 110.5°,11 suggesting that other factors must play a role too. operating at a H resonance frequency of 750 MHz equipped with a triple resonance, three-axis pulsed-field-gradient probehead. Data sets Recently, it has become possible to measure dipolar couplings in the aligned state were recorded at 35 °C; isotropic spectra were in macromolecules aligned with the magnetic field in a dilute recorded at 22 °C. Spectra were processed using the NMRPipe software lyotropic liquid crystalline phase of phospholipid particles,18 package.29 19 1 known as bicelles. Alternatively, such alignment now can also Residual DCH dipolar couplings were derived from the difference be obtained in media containing filamentous phages,20 purple of the modulation frequencies in the aligned and the isotropic states of membrane fragments,21 or a dilute lamellar phase of nonlipid 3D J-modulated constant-time [13C,1H] HSQC spectra30 which were molecules.22 In such media, it is possible to accurately measure recorded as data matrices of 224* × 16 × 1024* points (n* denotes n 13 1 one- and two-bond dipolar interactions in isotopically enriched complex points), with acquisition times of 28 (t1, C), 28 (t2, JCH), and 57 ms (t , 1H). The modulation frequencies were obtained by time- proteins.18,20-22 We have shown that comparison of the one bond 3 domain fitting in the t constant-time dimension, as described previ- 13 -1 13 -15 15 -1 N 13 -13 2 C H, C N, N H , and C C dipolar interactions ously.30 in the protein ubiquitin yields information on the ratios of the 1 13 13 DCC values were derived from the difference in one-bond C- C 23 effective bond lengths, corrected for rapid angular fluctuations. J splittings in the aligned and the isotropic states measured for the The ratio for the rCN and rCC bond lengths was found to be in methyl group resonances in the 2D [13C,1H] HSQC spectra. These excellent agreement with results from a high-resolution crystal spectra were recorded as data matrices of 460* × 1024* complex points, 24 13 1 structure database, but the effective rCH and rNH bond lengths with acquisition times of 92 (t1, C), and 57.0 ms (t2, H). Acquired were about 3% larger than their equilibrium values, when using data were apodized with a squared sine bell in the directly and a sine r and r as a reference. This latter result is not unexpected bell in the indirectly detected dimension, both shifted by 72° and CN CC ° because it is known that r and r undergo larger angular truncated at 176 .