Excitation of Selected Proton Signals in NMR of Isotopically Labeled Macromolecules Philippe Pelupessy,* Elisabetta Chiarparin,* and Geoffrey Bodenhausen* , † ,1 *Section de Chimie, Universite ´ de Lausanne, BCH, 1015 Lausanne, Switzerland; and †De ´partement de Chimie, associe ´ au CNRS, Ecole Normale Supe ´rieure, 24 rue Lhomond, 75231 Paris Cedex 05, France Received September 15, 1998; revised December 21, 1998 In isotopically labeled macromolecules, it is possible to excite the signal of a selected proton by shuttling magnetization back and forth between the chosen proton and a heteronucleus such as 13 C or 15 N, using two-way doubly selective heteronuclear cross-polar- ization. Selective excitation of a chosen proton can be followed by homonuclear coherence transfer to identify side-chain resonances of the corresponding amino acid in proteins. The resulting one- dimensional experiments yield information that can usually only be obtained from three-dimensional HSQC–TOCSY spectra. The method also provides efficient suppression of solvent signals with- out affecting resonances close to the solvent peak. © 1999 Academic Press Key Words: selective excitation; cross-polarization in liquids; labeled macromolecules; selective TOCSY. Much has been written about the virtues of selective exci- tation in magnetic resonance (1). Crowded spectra can be unraveled in exquisite detail (2–4). Band-selective versions of multidimensional spectra can provide excellent resolution (5, 6). Cumbersome three- or four-dimensional spectra can be reduced to two- or one-dimensional spectra (1, 7). Doubly selective homonuclear coherence transfer allows one to mea- sure relaxation rates of individual protons (8, 9). Synchronous nutation (10) and doubly selective inversion (11) make it possible to measure Overhauser effects while suppressing spin diffusion (12). All of these selective methods normally rely on a variety of amplitude- and phase-modulated pulses (13, 14). Unfortunately, these ideas are not truly suitable for macro- molecules with severely overlapping proton spectra. The more crowded the spectra, the longer the selective pulses need to be and the greater the toll taken by transverse relaxation and evolution under homonuclear couplings during the pulses. Typically, a selective Gaussian 270° pulse must be as long as 30 ms to excite a multiplet of 30 Hz width without phase distortion, leading to prohibitive losses in signal amplitude in macromolecules. For this reason, one is usually compelled to resort to cumbersome three- and four- dimensional methods (15). In this Communication, we propose a way to extend the benefits of selective experiments to isotopically labeled macro- molecules. Selective heteronuclear cross-polarization using two weak radio-frequency fields has been shown to be very effective in transfering magnetization from a selected proton to a scalar-coupled heteronucleus (16). Two-way coherence transfer I 3 S 3 I can be carried out within an interval on the order of 2  = 2( 1 J IS ) -1 (e.g., 2 = 20 ms if 1 J IS = 100 Hz). Selective spin-locking effectively decouples homonuclear J - couplings, which helps to minimize signal losses. The principle is illustrated in Fig. 1. The I x magnetization of a nucleus that one wishes to excite selectively (usually 1 H) is transformed by cross-polarization into S x magnetization of an S nucleus ( 13 C or 15 N). Provided that the RF fields applied to both I and S spins are weak (typically about 1 J IS /2), the Hartmann–Hahn condition (17) need not be fulfilled accurately and inhomogeneous RF fields do not significantly affect the transfer efficiency (16). Both intervals  1 =  2 are set approx- imately to ( 1 J IS ) -1 . The carrier frequencies of the I and S channels must be set to the chemical shifts of selected I and S resonances, which can be taken from a heteronuclear correla- tion HSQC spectrum such as that shown in Fig. 2 (in this case, ubiquitin, with 76 amino acids, 8.5 kDa,  c  4 ns at 300 K, and a typical amide proton T 2  40 ms). In analogy to techniques employing broadband cross-polarization (18), gra- dients and RF pulses are used to destroy residual transverse and longitudinal proton magnetization (in particular solvent mag- netization), while the 15 N magnetization of interest is stored along the z axis. Note that the suppression of the solvent signal is excellent without resorting to selective proton pulses and gradient labeling methods. Figure 3c shows the amide proton of glutamine Q62 in ubiquitin detected immediately after the second cross-polariza- tion step (i.e., by skipping the TOCSY sequence in Fig. 1). The success of the method depends on the selectivity of the RF fields during two-way doubly selective cross-polarization, and on the fact that the relevant cross peak must be resolved in the HSQC spectrum (a condition which must also be fulfilled for the success of 3D experiments). In Fig. 2, the Q62 signal appears fairly close to the D21 signal (shifted by 0.3 and 0.8 ppm in the 1 H and 15 N dimensions, which at 300 MHz amounts to 100 and 25 Hz, respectively) (19, 20). In Fig. 2, a simulated 1 To whom correspondence should be addressed. Fax: +33 1 44 32 33 97. E-mail: Geoffrey.Bodenhausen@ens.fr. Journal of Magnetic Resonance 138, 178 –181 (1999) Article ID jmre.1999.1715, available online at http://www.idealibrary.com on 178 1090-7807/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.