JOURNAL OF MAGNETIC RESONANCE 67,565-569 (1986)

I

Sensitivity-Enhanced Two-Dimensional Heteronuclear Shift Correlation NMR Spectroscopy

AW BAX AND SANKARAN sUBRAMANIAN* I

bboretoty of Chemical Physics2 National Institute of Arthrilh, Diabetes and Digestive and Kidney Disease& 1Vatlod Inctiiutes of Health, Eeches& M Q r y h d 20892

Reoeived January 3, 1986

Several years ago, it was demonstrated convincingly (1-3) that 'H-detected chemid- shift correlation via heternnuclear multiplequantum coherence (HMQC) (3-5) offers a substantial sensitivity advantage over the conventional shiftcorrelation e x r d r n m t (6-9), in which the low-y nucleus is detected directly during the data acquisition period. The 'H-detmed HMQC experiment, however, has not yet gained the pepularity one might expect on the basis of its intrinsic advantages. Major problems with the HMQC experiment are the dynamic range problem that is introduced by the presence of large signals from protons that are not ooupkd to the low- nucleus and the required suppression of these intense signals in a difference experiment. In principle, both problems can be soIv,d by presaturating the proton signals and transferring the NOEdenhanced low? signal to the protons (IO), but the sensitivity advantage of the 'H-detected experiments i s parhlly lost in this process.

Here, we describe a simple method that al1eviate-s the dynamic range problem and that hcilitazes the suppression of signals from protons that are not mupled to the low- y nucleus. The idea is to saturate all protons not directly attached to the low-y nucleus, leaving the protons coupled to the low-y nucleus unaffected or slightly intended by the homonuclear NOE effect (for molecules in the fast motion limit). The 2D pulse sequence is sketched in Fig. 1. In the following discussion, the low? nucleus is assumed to be "C. All protons not coupled to I3C are inverted by the bilinear (BIRD) pulse (12, 12): 90,"('H)-I/~ZJ)-lSO~('H, '3C)-1//2Jb90!('H), where J is the one-bond 13C-1H scalar coupling constant. Pcotons coupled to 13C are not affected by t h i s bilinear pulse, whereas the magnetization of all other protons is inverted. At the time, T, when the inverted magnetization changes from negative to positive (i.e., when protons not coupled to 13C are nearly saturated), the first 90" pulse of the HMQC experiment is applied. In practice, the Ti's of the various protons in the mdecule will vary, which at fmt sight makes it impossible to select a single T value for whch all protons not coupled to I3C are near saturation. However, by keeping the deIay time, T, between experiments short, one can 1argeIy circumvent th& problem. This will be briefly explained below.

We define the delay period, T, as the time between the start of data aCqUiSitiQn in one scan and the end of the preparation period of the next scan. The bilinear pulse is

I

* On leave from Indian Institute of Technology, Madras, India.

0022-2364186 $3.00 Copy+t 8 1986 by Acxdemic Pm!q Inc All: rights of repduction in any form &.

565

566 COMMUNICATIONS

'+-BIRD+!

FIG. 1. Pulse scheme of the heteronucbr multiplequantum experiment with presatUratiOn of protons not coupId to 'v. The phase cycling is given in Table I. The BIRD puke unit inverts protons that are not ooupled to IT. The delay time, A, is set to 1/(2J). The use of an 0lTsc-I compens;tted 180' ("0 pulse, S0~lSO;Wl: (20) or 90~1802J70,O (21), i s recommended to avoid partial saturation of protons that are coupkd to 13C nuclei that have a large offset h m the I3C carrier. The first 90; (I3C) puke m e s to eliminate mal1 artifacts from the 2D spectrum that otherwise a~ise from longitudinal I3C magnetization present at the end of the preparation period. Broadband 13G decoupling is used during 'H data acquisition.

applied at a time, r = T/2.7, before the end of this delay perid. The time Tis chosen to be about 1.3 times the TI value of the fastst relaxing proton in the molecule. At the start of the data acquisition for the first scan, longitudinal magnetization may be assumed to be cIose to zero. The buildup of longitudinal magnetization during the delay period i s depicted in Fig. 2. The bitinear pulse inverts the magnetization, and just afkr this inversion, the magnetization of the fastest relaxing proton is the most negative. At a time T/2.7 later, the Iongiturllnal magnetizations of all protons not coupId to I3C are close to zero. At this time the HMQC experiment is begun. In practice, we "fme tune" the duration of the delay time, 7, by choosing the value that minimizes the signal that is recorded in a single s m preceded by two dummy scans. It may appear that for the slowly relaxing protons in the molecuIe such a relatively short delay period between exwrhents is far from the optimum delay of 1 .3Tj (13).

TABLE I

mtase Cycling Used in the Scheme of fie I

Scan 0 # ACq.'

I x X X

2 X Y X

3 X --x --x 4 X -Y -X 5 -X X X

6 --x Y x 7 -x -X --x 8 -X -Y -X

'Data in odd- and even-numbered scans are stored in separate memory locations.

3us not 1% are ate t at

en At be he nd 1st lot In lat IS.

:ly 3).

.

coMMuNIcATroNs 567

However, the presence of a I3C nucleus adjacent to the protons of interest shortens the Tl of these protons significantly. The dipolar interaction between the proton and its directly coupled 13C nudeus is of the same order of magnitude as the dipolar coupling between geminal methylene protons. Therefore, dl protons directly coupled to s3C relax relatively efficiently, and a short delay period between experiments d o e not adversely affect the sensitivity.

As an example, we have applied the technique to a sample of 12 mg of the octapeptide angiotensin-Il ( M W 1044, dsolved in 0.35 ml D20, in a 5 mm sample tube, pH 3.5. The sample is not spun. Experiments are recorded on a Nicolet NT-500 spectrometer, using a Cryornagnet Systems probe (14), which has a broadband (15N to 3'P) &coupling coil outside the 'H observe coil. Eight scans preceded by two dummy scans are recorded for every t l value. A 1 s delay time ktween experiments (including the 125 ms t 2

acquisition period) and a 335 ms 7 period are used. The experiment is repeated for I72 t l increments of 120 ps , resulting in a total measuring time of 29 min. Data for odd- and even-numbed scans are stored separately and the data are processed in the standard fashion (15-17) to yield a 2D phase-sensitive spectrum. Phase modulation effects due to homonuclear J coupling are very small for the short dura~ms of the evolution period used in this experiment and consequently, the 2D spectrum can be p h a d , to a good approximation, to the purely absorptive mode. The spectrum ob- tained h s way is shown in fig+ 3, along with the regular 'H spYmrn and the projdon of the 2D spectrum onto the I3C axis. The 3C hquency scale is relative to hypothetical internal TSP. This referen&ng is obtained in a way similar to t h e method previously used by five ef a[. (18) for 15N. We measure the absolute frequency of the residual

0.1

0.2

- MO

- 0.2

-0.4

RG. 2. Time evolution of longitudinal magnetization during the delay m e between experiments for three spins, nut coupld to I3C with longitudinal retaxation times of (A) 0.5, (B) 0.7, and (C) 1.0 S. Tis the time bctween the start of data acquisition of one scan and the end of the preparation period of the next scan. The BIRD pulse, which inverts magnetization of protons not coupled to 'v is applied a t h e T before the end of the preparation p o d , such that at the end of the preparation period the magn&zations Of all t h m spins are close to zero. Recommended values are T = 1.3Tl of the rastest relaxing protun and T = 0.35T.

568 COMMUNICATIONS

- Pro -vai

c I l e Phe

Tyr -I

- H i s -Asp

Arg

Pro

5 9

k I=- - >

I l e * F FIG. 3. Absorption-made heteronuclear cbemical-sbift-com3ation spectrum of 12 mg angotensin-11, rw

corded in 29 min. The spectrum results from a 172 X 2048 data matrix. 30 and 8 Hz Gaussian line broadening have lwen used in the ti and tz dimensions, respectively. Some &XU artifacts at the *3C carrier frequency (37.3 ppm), and an impurity at 38.0i2.96 ppm are also visible. The 13C carrier artifacts disapm if four inswad of two dummy s c a ~ s are recorded for every ti value. The projection of the 2D spectnun onto the I3C (FJ axis and the regular high-resolution 'H spectrum are shown dong the sides ofthe 2D spectrum.

H D o present in the sample, which at 22°C resonates at 4.80 ppm relative to TSP. By multiplying the 'R frequency of the HDO resonance by 0.25 144954 (the "C/'H f r e quency ratio in TSP) one obtains the carbon frequency that corresponds to 4.8 ppm downfield from internal TSP. In all cases used we 6nd this method of referencing accurate to at least 0.1 ppm. For TMS, the I3C/lH frequency ratio is 0.25 145002. All I3C chemical shifts agree fairly well (k 1.5 ppm) with t y p i d values reported for linear peptides (19). This suggests that this type of heteronuclear chemical-shift correlation may be of use for assignment of 'H spectca of linear peptides. The signal-tenoise ratio on cross sections taken parallel to the Fl axis through the

spectrum of Fig. 3 ranges from 8:1 for the IIe Cr methylene resonances that have a

0 5 % :

COMMUNICATIONS 569

broad multiplet structure in the F2 dimension to 15: I for the Ca resonances and 40: 1 for the methy1 groups. For relatively small. molecules (<2000 daltons) we find in practice that N pg of sample of molecular weight, N, generalty yields heternnuclear shift-correlation spectra with an acceptable sensitivity in less than I2 hours. The main advantage of the modification proposed here is that on larger quantities (but stiU small relative to conventional I3C NMR quantities) it becomes possible to record 2D shift- cornlation spectra that are essentialIy free oft , noise in a very short period of time. We find the indir- detected '% method attractive for routine use in our laboratory since no probe changes are necessary and the time needed for shimming on our 5 mm probe is much less than usually needed for our broadband 10 mm probe. The addition of a bilinear pulse as descriM in this communication is not recommended for the study of proteins and other macromolecules because during the delay time, 7, t h e negative NOE effect decreases the intensity of the protons coupled to 13C.

ACKNOWLEDGMENTS

We thank Rolf Tschudin for continuous technical suppwt and Dr. Laura Lemer for userut suggestions during the prepmiion of the manuscript Dr. Jam= A. Ferretti kindly provided the sample of angio- tensin-n.

REFERENCES

1. A. BAX, R. H. G R I ~ Y , AND B. L. HAWKINS, J Am. C h m . Soc. 105,7ISS (1983). 2. D. H. LIVE, D. G. DAW, W. C AGOSTA, AND D. COWURN, J. Am. Chem. Soc. 106,6104 (1984). 3 A. BAX, R. H. G m , ANDB. L. HAWKINS, J. Mc?@. Reson 55,301 (1983). 4. L. MVUER,J. Am. chew Soc 101,4481 (1979). 5 M. R BENDALL, D. T. PEW, AND D. M. DODDRELL, X Mugia Reson. 52,81 (1983). 6. A. A. ~ U D S L E Y , L. MULLER, ANDR. R. ERNST, 1 M a p . R a m . 28,463 (1977). 7. G. BODENHAUSEN AND R FREEMAN, J. M g n . Reson. 28,47 I ( 1.977). 8. A. BAX AND G. MORRIS, J Mugn. Reson. 4&50 1 ( 198 1). 9. A. BAX ANDs. K. S A W J. khgp1. Reson 60, 170 11984).

10. D. NEUHAUS, J. KEELER, AND R. FREEMAN, J. M q n . Resox 61,553 (1985). I 1 J . R. GARBOW, D. P. W ~ E K A M P , AND A. PINE, Chern P h p . Lest. 93,504 (1982). 12. A. BAX, J. M q n . Resun. 52,330 (1983). 13. J. S. WAWGH, J. Mol. Specirosc. 35,298 (1970). 14. Cryornagnet Systems Inc., 4101 Cashard Ave. No. 103, Indianawlii, Ind. 46203. 15. L. MULLER ANDR. R. ERNST, Mol. Phys. 38,963 (1979). Ib. D. J . STATES, R. A. HABERKORN, APID D. J. RUBEN, J. h'qn Reson. 48,286 (1982). 17. A. B~x,BdI.Magn. Reson 7, 167 (1985). 18. I). H. LrVe D. G. DAVE, W. C. A m A , AND D. COWBURN, J. Am. Chem. SQC 106, 1939 (1984). 19. 0. W. HAWARTI AND D. M. J. L i t t n , Prugr. NMR Spectrosc 12, 1 (1978). 20. R. FREEMAN, s. P. KEMPSEtL, A M 3 M. H. LEVITT, I Magn. Reson. 38,453 (1980). 21. A. J . SHAKA, J. KEELER, T. FRENKIEL, AND R. FR!E!ZMAN, J. Magn. Reson. 52, 335 (3983).