Single-Carbon Resonances of Hen Egg-White Lysozymet

he development of a probe for sample tubes of 20-mm outside diameter has increased the sensitivity of natural abundance 3C Fourier transform nuclear magnetic resonance to the point that single-carbon resonances of proteins can be studied. Numerous narrow single-carbon resonances are observed in the aromatic region of the I3C spectrum of native hen egg-white lysozyme. Theoretical and experimental evidence is presented to show that these narrow resonances are those of the 28 nonprotonated aromatic carbons. The 59 protonated aromatic carbons give rise to a N uclear magnetic resonance (nmr) is suitable for detailed studies of biopolymers in solution because it can, in principle, yield an individual signal for every atom with a nonzero nuclear spin. Proton nmr, which has been widely t Contribution No. 2168 from the Department of Chemistry, Indiana University, Bloomington, Indiana 47401. ReceiGed Nocember 6, 1972. This research was supported by the National Science Foundation (Grant GP17966), the donors of the Petroleum Research Fund administered by the American Chemical Society (Grant 4559-AC5), and by Eli Lilly and Co. One of us (E. 0.) thanks the European Molecular Biology Organization and the Gilbert Foyle Trust of Great Britain for partial support. background of broad peaks. Partial assignments for the nonprotonated carbon resonances are presented. Significant chemical-shift variations occur upon folding of the protein into its native conformation. For example, the y carbons of the six tryptophan residues resonate at 81.4, 82.1, 83.2, 83.8, and 85.2 ppm upfield from CS2. The peak at 85.2 ppm is a two-carbon resonance. Upon denaturation with guanidinium chloride, all six carbons resonate at about 83.8 ppm. chemical shifts and assignments for aqueous tryptophan at pH 4 are also presented. used for studies of proteins (Roberts and Jardetzky, 1970), is characterized by a small range of chemical shifts, and spectral splittings arising from homonuclear scalar coupling. As a result, proton spectra of proteins are relatively unresolved envelopes. Only a few resonances, such as those of H'1 of histidine residues, fall outside the main spectral region and thus can be resolved into single-proton resonances. It is known (Levy and Nelson, 1972) that the range of chemical shifts is much greater than that of protons. Moreover, because of the low natural abundance (1.1%) of 13C, splittings from homonuclear coupling are normally not observed. Splittings caused by heteronuclear coupling to B I O C H E M I S T R Y , V O L . 1 2 , N O . 7, 1 9 7 3 1335 A L L E R H A N D , C H I L D E R S , A N D O L D F I E L D protons (or, less frequently, other nuclei) can be eliminated by strong irradiation at the resonance frequencies of the protons (or other nuclei). Thus, 13C nmr spectra of large molecules are more resolved and simpler to interpret than their proton spectra. We have already shown (Allerhand et a/., 1970) that existing Fourier transform (FT) nmr equipment provides enough sensitivity for recording proton-decoupled naturalabundance 13C nmr spectra of proteins. The potential of 13C nmr for detailed studies of biopolymers in solution depends on the magnitude of 13C chemical-shift nonequivalence induced by the folding of the molecule into its native conformation. Several natural-abundance 13C FT nmr spectra of proteins have been reported, some recorded on “homebuilt” equipment operating at about 15 MHz (Allerhand et al., 1970, 1971a; Glushko et a/., 1972), others recorded on commercial instruments operating above 20 MHz (Chien and Brandts, 1971 ; Conti and Paci, 1971 ; Moon and Richards, 1972). In all cases, however, the signal-to-noise ratio was not sufficient for observing single-carbon resonances. It is pertinent to our discussion that the signal-to-noise ratio in a proton-decoupled nmr spectrum depends on the nuclear Overhauser effect (NOE) (Kuhlmann and Grant, 1968; Kuhlmann et a/., 1970), which produces an increase in intensity when strong irradiation at the proton resonance frequency is introduced during the 13C nmr experiment. The NOE causes a maximum intensity increase of a factor of 2.988 when the I3C nuclei are undergoing purely dipolar I3C-’H relaxation (Kuhlmann er a/., 1970) and eq 1 is satisfied where T R is the correlation time for molecular rotation and WH and uc are the resonance frequencies (in radians per second) of lH and 13C, respectively. It is now generally accepted that, with currently available instrumentation, 0.01 hi is the lowest practical concentration for observing single-carbon resonances in natural-abundance spectra, when the maximum NOE factor of 2.988 occurs (Doddrell and Allerhand, 1971; Levy and Nelson, 1972). Theoretical considerations (Doddrell et a/., 1972) and experimental results (Oldfield, E., and Allerhand. A , , manuscript in preparation) indicate that the NOE may be nearly nonexistent for carbons on or near the backbone of a native protein, because T R is so large that eq 1 is not valid. Then the lowest concentration for observing single-carbon resonances in natural-abundance spectra is expected to be about 0.03 M (with 1 day or less of signal accumulation). This molarity represents a 45% solution for a protein of mol wt 15,000. Clearly, a practical approach to the detection of single-carbon resonances of proteins requires an increase in instrumental sensitivity. While proton nmr studies are usually made in sample tubes with an outside diameter of 5 mm, most existing I3C nmr instruments utilize tubes with a diameter of 12 or 13 mm. We have recently constructed a probe which uses spinning sample tubes with an outside diameter of 20 mm (1%” inside diameter). This relatively inexpensive development (Allerhand et al., 1972) produces an increase in sensitivity of about a factor of three with respect to commercial FT nmr equipment, in spite of the fact that we work at 15.18 MHz while most commercial FT nmr instruments operate a t 13C resonance frequencies above 20 MHz. Happily, the magnetic field inhomogeneity over our large sample volume (about 10 ml) is still sufficiently small (0.3 Hz) for high1336 B I O C H E M I S T R Y , V O L . 1 2 , N O . 7, 1 9 7 3 resolution studies. With the new probe, systematic studies of single-carbon resonances of proteins become practical. We report here the observation and partial assignment of numerous narrow single-carbon resonances in the aromatic region of the proton-decoupled natural-abundance 1 JC spectrum of native lysozyme a t 15.18 MHz. We show that all the narrow resonances arise from nonprotonated carbons, a fact that greatly simplifies the assignments. We also show that denaturation removes the large chemical-shift nonequivalence observed in native lysozyme. Experimental Section Materials. Hen egg-white lysozyme (EC 3.2.1.17) was obtained from Miles Laboratories (grade I, six times recrystallized, lyophilized, lot no. 7103). Tryptophan was purchased from Sigma Chemical Co. All other materials were reagent grade. Methods. SAMPLE PREPARATION. Lysozyme (3 g) was dialyzed in Spectrapor tubing No. 132680 (Spectrum Medical Industries), three times for 8 hr against 2.5 1. of water a t 4” , and then lyophilized. Lysozyme (2.05 g) was then dissolved in 8.5 ml of 0.1 M NaCl. For the undialyzed samples, 2.5 g of the commercial lysozyme was dissolved in 8.5 ml of 0.1 M NaCl (for spectra of native enzyme) or in 8.5 ml of a solution 0.1 M in NaCl and 6.2 M in guanidinium chloride. The pH was adjusted with hydrochloric acid. pH measurements were made before and after each spectral run with a Radiometer PHM 52 Digital pH meter equipped with a Radiometer GK2322C combined electrode. The sample was then passed through a 0.8-p Millipore filter. Identical I3C nmr spectra were obtained from dialyzed and undialyzed lysozyme, and also from a sample that had been passed through a column of Sephadex G-25 equilibrated with deionized water. It is safe to assume that no small molecule impurities contributed to the observed spectra of the undialyzed material. spectra were recorded at 15.18 MHz on a “home-built” Fourier transform nmr apparatus, which consists of a Varian 12-in. high-resolution electromagnet (14.2 kG), radiofrequency circuits and probe of our own design, a Nicolet 1074 instrument computer (4096, 18-bit words) for data acquisition, and a PDP-S/E computer (Digital Equipment Corporation) for data processing. The probe uses spinning sample tubes with a 20-mm outside diameter (Allerhand er ul., 1972). A spectral width of 250 ppm (3795 Hz) was used for recording the complete spectrum of native lysozyme, with a digital resolution of 1.85 Hz, which was imposed by the limited memory size of the Nicolet 1074 unit. The resolution in the unsaturated carbon region was increased twofold by using 125-ppm spectral windows. We would like to point out that after most of the research presented below was completed, we incorporated additional improvements into our equipment that have further reduced, by about a factor of three, the time required to obtain a given signal-to-noise ratio in spectra of proteins. TO facilitate a comparison with commercial equipment, we have determined that about 10 min of accumulation time is now required on our instrument to get a spectrum of lysozyme comparable to that reported by Chien and Brandts (1971), which was obtained after about 13 hr of accumulation time. In Figure 1 we show a spectrum of lysozyme (20% w/v in H20) obtained in less than 10 min of signal accumulation time. The signal-to-noise ratio is not sufficient for unambiguous detection of single-carbon resonances. In Figure 2 we show spectra obtained with more than enough accumulations to NMR SPECTRA. C A R B O N 1 3 N M R O F L Y S O Z Y M E permit detection of single-carbon resonances, if any resolved ones should be present. Results and Discussion Figure 2A shows the proton-decoupled 3C spectrum of native lysozyme (pH 4.08). Overall assignments can be made on the basis of reported 13C chemical shifts of amino acids (Horsley et al., 1970) and peptides (Glushko et al., 1972; Christ1 and Roberts, 1972). The region below 90 ppm upfield from CS2 contains the resonances of all unsaturated carbons. The saturated carbons resonate above 120 ppm. The aromatic region of the spectrum (about 37-85 ppm upfield from CS2) showed the most promise for observing resolved single-carbon resonances. Figure 2B shows the region of unsaturated carbon resonances of native lysozyme, recorded with twice the digital resolution of that in Figure 2A. Peaks in the range 13-25 ppm upfield from CS2 are carbonyl resonances. The strong resonance at 35.8 ppm in Figure 2B can be assigned to Cr of the 11 arginine residues of lysozyme. The rest of the unsaturated carbon region contains all the aromatic carbon resonances, i .e., those of the three phenylalanine, three tyrosine, one histidine, and six tryptophan residues, a total of 87 carbons. This region consists of numerous narrow resonances (some with a line width of 2 Hz or less) labeled 1-22 in Figure 2B, superimposed on a background of overlapping broad peaks. We present below experimental proof that the narrow resonances are those of the 28 nonprotonated aromatic carbons (Cy of Phe and His, Cy and Cr of Tyr, C y , C6%, C" of Trp). However, it is instructive to first discuss the theory of line widths in 13C spectra of proteins. spectrum of a protein, the only important contribution to the natural line width of a carbon bonded to one or more hydrogens is from I3C-lH dipolar relaxation (Allerhand et al., 1971b) given by (Solomon, 1955; Doddrell et al., 1972) If we consider a proton-decoupled where W is the line width in hertz, y c and YH are the gyromagnetic ratios of 13C and IH, respectively, N is the number of directly attached hydrogens, rCH is the C-H bond length, andf(rR) is defined by We have assumed that the molecule is undergoing isotropic rotation with a correlation time TR. Equation 2 is a fairly good approximation for the cy carbons of native lysozyme, but it applies to carbons on side chains only when internal rotations are much slower than the overall molecular reorientation. For native lysozyme, fluorescence polarization measurements yield a value of TR of 25 nsec (Irwin and Churchich, 1971; Rawitch, 1972) a t room temperature, and 13C spin-lattice relaxation studies yield 22 nsec at 40' (Allerhand, A., and Hailstone, R . K., unpublished results). If we choose the latter value, eq 2 predicts a line width of 36 Hz (2.4 ppm) a t 15.18 MHz for the cy carbons. The aromatic side chains are not expected to have fast internal rotations (Browne, D. T., Kenyon, G. L., Packer, E. L., Sternlicht, H., and Wilson, D. M., private communication) and thus a line width of this L , , , , I , , , , l , , , , l , , , , l 0 50 100 150 eo0 PPM FROM C S z FIGURE 1 : Proton-decoupled natural-abundance z3C FT nmr spectrum of hen egg-white lysozyme (about 20% w/v in 0.1 M NaCl, pH 4.0, 43") at 15.18 MHz in a 20-mm sample tube, recorded with a 250-ppm spectral window, 4096 points in the time domain, 1.09sec recycle time, and 512 accumulations (9.3 min total time). The signal-to-noise ratio is comparable to that in a spectrum obtained by Chien and Brandts (1971) after 13.3 hr of signal accumulation time on a commercial FT nmr instrument operating at 25.2 MHz. magnitude is also expected for protonated aromatic carbons. On the other hand, because of the inverse sixth-power dependence on carbon-hydrogen distances, the contribution to the line widths of nonprotonated carbons from 13C-lH dipolar relaxation will be reduced by at least an order of magnitude. Even when one includes other contributions, such as instrumental broadening and relaxation by chemical-shift anisotropy, the nonprotonated carbons are expected to have line widths of only a few hertz. This is indeed the case in the carbonyl region (Figure 2B). The resonance of the 11 arginine { carbons appears to be broadened by chemical-shift nonequivalence (a slight splitting is actually detected). We tentatively ascribed the sharp peaks 1-22 (Figure 2B) to the 28 nonprotonated aromatic carbons, and the broad features to the 59 protonated ones. This assignment was confirmed experimentally by means of noise-modulated, offresonance proton decoupling (Wenkert et al., 1969), which selectively broadens protonated carbons. The residual broadening from incomplete proton decoupling is proportional to the square of the pertinent carbon-hydrogen coupling constant (Ernst, 1966). One-bond 13C-'H coupling constants are larger than 120 Hz (Levy and Nelson, 1972), while long-range carbon-hydrogen coupling constants are smaller than 15 Hz (Levy and Nelson, 1972). The effectiveness of noise-modulated off-resonance decoupling for identifying resonances of nonprotonated carbons is illustrated in the aromatic region of the spectrum of 0.05 M tryptophan (Figure 3). Only the resonances of Cy, CJ2, and C" are sharp peaks in the noise-modulated off-resonance decoupled spectrum (Figure 3A). The protonated carbons are barely discernible broad humps. The corresponding fully proton-decoupled spectrum shows the expected eight sharp resonances (Figure 3B). The region of unsaturated carbons in the noise-modulated off-resonance proton-decoupled spectrum of native lysozyme is shown in Figure 1C. As expected, the region from 10 to 40 ppm upfield from CS, is superimposable with the corresponding one in the fully decoupled spectrum (Figure 2B). In the range 50-90 ppm, all the sharp resonances of Figure 2B are still sharp in Figure 2C, and can thus be assigned to nonprotonated carbons. The envelopes of protonated-carbon resonances in Figure 2B are broadened further in Figure 2C. It should be noted that only narrow B I O C H E M I S T R Y , VOL. 1 2 , N O . 7, 1 9 7 3 1337 A L L E R H A N D , C H I L D E R S , A N D O L D F I E L D I d 1: \Ju I I I I :/'..'L A too 150 200 250 _ _ _ _ _ 5.0 v