Nuclear Magnetic Resonance Spectrum of Deamino-Lysine-Vasopressin in Aqueous Solution and Its Structural Implications

Transcription

1 Proc. Nat. Acad. Sci. USA Vol. 69, No. 11, pp , November 1972 Nuclear Magnetic Resonance Spectrum of Deamino-Lysine-Vasopressin in Aqueous Solution and Its Structural Implications P. H. VON DREELE, H. A. SCHERAGA*, D. F. DYCKES, M. F. FERGER, AND V. DU VIGNEAUD Department of Chemistry, Cornell University, Ithaca, New York Contributed by H. A. Scheraga and V. du Vigneaud, September 9, 1972 ABSTRACT The peaks in the proton nuclear magnetic resonance spectrum of deamino-lysine-vasopressin in aqueous solution at ph values between 3 and 5 were assigned to particular amino-acid residues by use of the results of transfer-of-saturation studies, NH-CaH and CaH- C'3H decoupling experiments, and other data. The conformation of deamino-lysine-vasopressin in water differs from that of lysine-vasopressin in the same solvent. This paper is part of a series (1-3) in which physical methods [primarily nuclear magnetic resonance (NMR)] are being used to obtain the structure of lysine-vasopressin (Lys-VP) and related compounds in solution, with a view toward the eventual elucidation of the relationship between structure and biological activity. In the present paper, a study of deaminolysine-vasopressin in aqueous media at ph values between 3 and 5 is reported, and similarities and differences between the conformational information available from the NMR spectra of deamino-lys-vp and Lys-VP are pointed out. MATERIALS AND METHODS Deamino-Lys-VP was a highly purified synthetic preparation (4) that possessed about 120 U/mg of rat pressor activity.t It has the following structure, in which the numbers indicate the positions of the individual amino-acid residues: r MPA-Tyr-Phe-Gln-Asn-Cys-Pro-Lys-Gly-NH2. The half-cys tine residue in position 6 is referred to as Cys-6. The proton NMR spectra were recorded on a 250 MHz spectrometer described elsewhere (5). Sample concentrations were 0.06 M in D20 or H20 that contained acetic acid. The chemical shifts in aqueous solution are downfield from an external standard of tetramethylsilane in CC14. The measured ph was corrected to pd, by use of the relation pd = ph when D20 was used (6). Spectra were obtained at ph about 5 to avoid (a) instability of the disulfide Abbreviations for amino-acid residues and protecting groups are in accordance with the IUPAC-IUB Tentative Rules [J. Biol. Chem., 241, (1966)]. Nomenclature for the description of the conformations of the polypeptide chain is in accordance with the IUPAC-IUB Tentative Rules [Biochemistry 9, 3471 (1970)]. All optically active amino acids are of the L-configuration. Deamino-Lys-VP, 1-deamino-8-lysine-vasopressin; J, coupling constant; MPA, 2-mercaptopropionic acid; NMR, nuclear magnetic resonance. * Requests for reprints should be addressed to Dr. Scheraga. t Pressor assays were performed on anesthetized male rates, as described in The Pharmacopeia of the United States of America (Easton, Pa.: Mack Publishing Co., 1970), 18th rev., p bond and (b) more rapid exchange of amide protons with those of water, at higher ph. Since an external rather than an internal reference was used, the data from the temperature studies were corrected for this difference by scaling the coefficient of temperature dependence of the chemical shift (d6/dt) to zero for the acetate peak. This sacrifice of proper referencing technique was made in order to safeguard against the possibility that the internal standard might react with or change the conformation of the deamino-lys-vp. Thus, attention should be paid only to the resulting relative (rather than absolute) values of db/dt. After the NMR measurements were completed, the deamino-lys-vp solutions were assayed biologically. No appreciable inactivation of the deamino-lys-vp was detected. RESULTS AND DISCUSSION Assignments of peaks by the spin-decoupling method Since the structural information contained in the NMR spectrum can be interpreted only when the peaks are correctly assigned to particular protons in the molecule, a careful peak assignment is essential in order to obtain the structure of deamino-lys-vp in solution. We have accomplished this by spin-decoupling the NH from the CcH protons, and the C'H from the C0H protons, and then matching the chemical shifts and splitting patterns of the C"H protons to similar quantities observed for each amino-acid residue in small peptides (7) and in Lys-VP (3). The instrumental difficulties encountered in this approach were solved in a manner explained elsewhere (8), and the observed NH-CaH decouplings for deamino-lys-vp in H20 are shown in Fig. 1 and Table 1. The positions of CcH peaks (and their coupled NH peaks) were located by a blind-search irradiation under the H20 peak. C"H peaks located in this manner were subsequently observed at the same positions when the spectrum was obtained in D20 (pd = 5.1). The CaH-CH decouplings in D20 are shown in Fig. 2. Having established the decoupling relationships of the peaks, the assignments of the peaks to particular amino-a'cid residues were made in the following way. The NH triplet at 8.00 ppm corresponds to the amino acid having two C'H protons, so that it and the decoupling-related CcH peak at 3.55 ppm were assigned to glycine. The C'H peak at 3.72 ppm that is coupled to an NH at 8.00 ppm is a one-proton peak, which is coupled to the equivalent COHrproton peak at 1.68 ppm. Although the C"H2 resonances of Pro and Gln both occur in this region, the Pro does not have an NH peak. Therefore, the peaks at 1.68, 3.72, and 8.00 ppm were assigned 3322

2 Proc. Nat. Acad. Sci. USA 69 (1972) Deamino-Lysine-Vasopressin in Aqueous Solution 3323 TABLE 1. Peak assignment of the protons in the NMR spectrum of deamino-lys-vp in water Peptide NH CaH COH2 Amino NH acid type v* Hz appm V*Hz a ppm v* Hz a ppm Phe doublet t F756, , 2.67, L741, , 2.61 Tyr doublet [ [647, , 2.36 L [633, , 2.30 Cys-6 doublet [ [704, , 2.52 [ [690, , 2.58 Asn doublet [ L1o Gly triplet Gln doublet Lys doublet Pro * Downfield from an external standard (tetramethylsilane/ccl4) at 250 MHz, 300, and ph = 4.4. t Decoupling arrows point from the irradiating frequency to the frequency where a change is observed. to the COH2, COH, and NH protons of Gln, respectively. The C"H peak at 4.07 ppm is. coupled to nonequivalent COH2 groups at 1.56 and 1.94 ppm (the Gln or Pro region). Since Pro is not coupled to an NH peak and since it has no NH proton, the COH2 resonances at 1.56 and 1.94 ppm and one C"H resonance at 4.07 ppm are those of Pro. The COH2 protons of Lys are generally the highest-field COH2 resonances of all the amino-acid residues in this molecule, and therefore the COH2 resonances at 1.41 ppm and the related C"H resonance at 3.90 ppm, and the NH resonance at 8.13 ppm were assigned to Lys. The 2.66-ppm peak is coupled to the 1.34-ppm peak, and they correspond to the e-ch2 and 6-CH2 protons of Lys, respectively. The 1.68-ppm peak is coupled to the 3.4-ppm peak, and they correspond to the y-ch2 and 5-CH2 protons of Pro, respectively. The remaining amino acids that have to be assigned are those with an amide, an aromatic ring, or a heteroatom at the -y position (Asn, Phe, Tyr, Cys-6, and 2-mercaptopropionic acid). COH2 peaks of these residues occur between 2.2 and 3.2 ppm. COH2 resonances of Asn in the Lys-VP precursors were found (1) to be fairly equivalent and to occur at 2.48 ppm in Lys-VP in aqueous solution. As shown in Table 1, FIG. 1. Spin-decoupling of the 250-MHz NMR spectrum of deamino-lys-vp at 300 in H20 at ph = 4.4. The 4000-Hz spectrum is the undecoupled one obtained by irradiating offresonance from all proton frequencies. The other spectra show the various decouplings observed when irradiating at the frequency shown by each trace. There is an extraneous peak that appears at a frequency that is the sum of the H20 frequency and the irradiating frequency, and should not be confused with the spin-decoupling changes in the spectra. FIG. 2. Same as Fig. 1, but in D20 at pd = 5.1. The lowest spectrum is the undecoupled one.

3 3324 Chemistry: Von Dreele et al. Proc. Nat. Acad. Sci. USA 69 (1972) this effect is not present and cannot be used to identify the Tyr peaks. Therefore, an analog of deamino-lys-vp was prepared, in which all the nonexchangeable protons in the Phe residue were deuterated. The data from this compound show unambiguously that the peak positions of Phe are 7.34, 4.28, 2.99, and 2.64 ppm. By elimination, we conclude that the peaks of Tyr are located at 7.74, 4.10, 2.56, and 2.33 ppm. FIG. 3. Saturation effects used in peak assignments. (C) and (F) are normal spectra of the aromatic CH and carboxamide NH region of deamino-lys-vp at 30 in H20 at ph 4.4. (A) Irradiation at the cis NH of the CONH2 group of Gln and Asn to locate the trans NH. (B) Irradiation of the H20 peak to locate the NH3 + peak of Lys. (D) Spectrum of the aromatic CH region in D20, in which the carboxamide peaks are no longer present. (E) Increase of the observing field to distinguish the aromatic CH from the carboxamide NH the 2.47-ppm peak is coupled to the 4.34-ppm peak, which is coupled to the 7.80-ppm peak. Therefore, the peaks at 2.47, 4.34, and 7.80 ppm were assigned to COH2, CaH, and NH of Asn, respectively. We have noticed that, when one compares the spectrum of an analog of vasopressin to that of the corresponding deamino compound in. [U-2J4]Me2SO, the deamino compounds shows a peak at higher field than the COH2 of Asn. This peak probably arises from the 2-mercaptopropionic acid moiety, since the replacement of an amino group by a proton would be expected to give rise to an upfield shift. The peak at 2.2 ppm that is coupled to a peak at 2.75 ppm are assigned to the 2-mercaptopropionic acid moiety. The lowest-field C"H occurs at 4.46 ppm in deamino-lys-vp, and is coupled to C0H2 at 2.79 and 2.55 ppm and to an NH at 7.74 ppm. For Lys-VP, the peaks at 7.74, 4.48, 2.80, and 2.50 ppm were assigned to Cys-6. Since these two sets of decoupling-related peaks correspond so closely, and since the C"H of Cys-6 has generally been observed to be the most downfield C"H peak, the peaks at 7.74, 4.46, 2.79, and 2.55 ppm were assigned to the NH, CaH, and C0H2 of Cys-6, respectively. The remaining two NH peaks belong to Phe and Tyr and are related by decoupling to the following peaks: 7.74 ppm (NH) to 4.10 ppm (C'H) to 2.56 and 2.33 ppm (COH2), and 7.34 ppm (NH) to 4.28 ppm (C'H) to 2.99 and 2.64 ppm (COH2). For Lys-VP, it was easy to discriminate between these two residues because the NH3+ group exerts an inductive effect on the amide group of the adjacent Tyr residue and increases its rate of exchange, thus leading to a broad NH peak that will exhibit transfer of saturation (3). Since the NH3+ group is absent in the deamino compound, Use of saturation effects in making peak assignments The method of transfer of saturation (9) has proved to be useful in identifying peaks corresponding to rapidly exchanging protons. If the rate of exchange of an exchangeable NH proton with water exceeds the rate of relaxation of the NH proton, then on irradiating the water peak the area of the peak corresponding to the NH proton will decrease because of the transfer of saturated protons from water. For Lys-VP, this experiment enabled us to identify the Tyr NH peak (3), whose rate of exchange at ph 4.6 was accelerated by the presence of the nearby a-nhi+ group. The same experiment was performed on deamino-lys-vp, by a tracesuperposition procedure, which eliminates the need for integration of the peak area, the decrease in area being visually observable. The low-field spectrum, obtained by irradiating first at 4099 Hz (off-resonance frequency) and then at 1099 Hz (H20 resonance frequency) and recording the two spectra on top of each other, is shown in Fig. 3B. As expected, since this molecule does not contain an N-terminal amino group, there was no decrease in the area of the peptide NH peaks. There was, however, a decrease in area in the region from 7.0 to 7.3 ppm. Since a similar experiment on Arg-containing peptides in aqueous solution has shown that the protons on the charged guanidino group exhibit transfer of saturation, it is likely that the decrease in area observed from 7.0 to 7.3 ppm serves to identify the presence of the e-nh3+ proton peak of Lys. We point out that it would not have been possible to locate the presence of this peak by any other method than transfer of saturation. An overlapping series of carboxamide and aromatic proton peaks is also present in this region of the spectrum. To distinguish between these two kinds of peaks we relied on the fact that their relaxation times are quite different-those of the NH protons being shorter because they are bonded to an atom that has a quadrupole. Therefore, an increase of the observing field strength to a very high power will preferentially saturate the peaks that have long relaxation times, i.e., aromatic peaks. The results of this experiment (Fig. 3E) indicate that the peaks at 6.46, 6.46, 6.67, 7.04, 7.10, and 7.18 ppm arise from CONH2 groups, while those at 6.38, 6.56, 6.86, and 6.97 ppm correspond to aromatic protons. By use of the splitting patterns in these spectra, it is possible to assign the peaks at 6.38 and 6.56 ppm to Tyr and the peaks at 6.86 and 6.97 ppm to Phe. These assignments can also be made from the spectrum in D20 (Fig. 3D) where the protons on the CONH2 group are replaced by deuterium. The three highest-field NH peaks of the carboxamide group correspond to the protons that are cis to the oxygen of the carbonyl group, and the lower-field peaks belong to the trans protons. The cis NH protons in Lys-VP in water occur at 6.44 ppm (Gln), 6.49 ppm (Asn), and 6.65 ppm (Gly-NH2), and the trans NH protons occur at 7.03 ppm (Gly-NH2), 7.08 ppm (Gln), and 7.18 (Asn) (3). By analogy with the

4 Proc. Nat. Acad. Sci. USA 69 (1972) Deamino-Lysine-Vasopressin in Aqueous Solution 3325 TABLE 2. Values of d6/dt, JNH-CaH, ONH-CaH, and O for deamino-lys-vp in aqueous solution at ph = 4.4 ds/dt* JNH-CaHt NH a ppm X 103ppm/0C Hz 0 deg O deg Phe :+115, , 75, -90, -150 Tyr t ±20, , 80, -85, -155 Cys t ±20, , 80, -85, -155 Asn , , -95, -145 Gln ±40, i , 100, -70, -170 Gly = Lys , , 85, -85, -155 * These are relative rather than absolute values, see Methods. t A change of ph from 4.4 to 3.2 did not alter the measurable values, although it did give a little better separation of peaks in some cases, permitting better measurements. t Peaks separate on raising temperature. Peaks separate on lowering ph. spectrum of Lys-VP in H20, the cis NH peaks in deamino- Lys-VP were assigned as follows: 6.46 ppm (Gln), 6.46 ppm (Asn), and 6.67 ppm (Gly). The trans NH peaks can be assigned directly from the cis NH by irradiating in turn each cis NH frequency and observing which trans NH peak will decrease in area. The success of this experiment depends on working at a temperature where the rate of rotation about the amide bond is fast enough to exceed the rate of relaxation of the NH proton in the trans position, but not so fast that the cis NH and trans NH peaks coalesce. It further requires that the rate of rotation be faster than the rate of exchange of amide protons with water. This type of transfer of saturation experiment does not involve bond breaking, but rather rotation about a bond. In Fig. 3A, we show the results obtained by a trace superposition procedure, irradiating first at 2617 Hz (off-resonance frequency) and then at 1617 Hz (the frequency of the cis NH of Gln and Asn). The decrease in area of the peaks at 7.10 and 7.18 ppm serves to identify the trans NH of Gln and Asn although we cannot distinguish between them. The trans NH of Gly-NH2 is then the peak at 7.04 ppm. Since these peaks in Lys-VP (3) and deamino- Lys-VP correspond very closely, it is likely that the trans NH of Gln is at 7.10 ppm and that of Asn is at 7.18 ppm. This completes the assignment of the peaks in the proton NMR spectrum of deamino-lys-vp. Structural information available from the NMR spectra The NH-CaH coupling constant, JNHCaH, has been related empirically (10) to the angle, ONH-CGH, which can then be related to 4. The observed values of JNH-CH and the related angles for deamino-lys-vp are shown in Table 2. Since there was a good deal of overlapping in the spectra from which the data were derived, precise values of J could not be obtained in all cases. The values of JNHC0H of the residues in the tail of deamino- Lys-VP (6.5-7 Hz for Lys and MJNH-CaH = Hz for Gly) and the value of Jgem of Gly ( Hz) are similar to those observed for the tail of Lys-VP (JNH-0aH = 6.5 Hz for Lys and 2JNH-CrH = 11.5 Hz for Gly, and Jgem = 18.5 Hz). These facts suggest that a similar conformation is present in the tail of both compounds, with 4 of Gly being 0 or The value of A6 (ONH trans-sinh cis) of the Gly-NH2 group is similar for both compounds (AS = 0.37 ppm for deamino- Lys-VP and 0.38 ppm for Lys-VP) and is smaller than that of acetamide (AS = 0.75 ppm) (11) or the Asn carboxamide in Gly-Asn (A8 = 0.70 ppm) (12); the higher values of AS arise when the NH is not hydrogen bonded. These additional facts also point to a similar conformation in the tail of deamino-lys-vp and Lys-VP in water. Furthermore, the data are consistent with the possibility that this conformation includes a,b-turn involving a hydrogen bond from the trans NH of the CONH2 group to the C=O of Pro. This is the same,b-turn that was suggested (2) as a possible component of the conformation of Lys-VP in [U-2H]Me2SO. solution. The temperature dependence of the chemical shift (db/dt) of the carboxamide protons of deamino-lys-vp (in units of 10-3 ppm/ C) are as follows: Gly (cis NH = 6.2, trans NH = 5.3), Asn (cis NH = 5.2, trans NH = 6.1), and Gin (cis NH = 5.2, trans NH = 6.4). The values of db/dt of the peptide NH protons are shown in Table 2. The peptide NH peaks having the lowest values in deamino-lys-vp are Asn and Phe, while in Lys-VP they are Asn and Cys-6. While all the factors affecting d8/dt are not understood, one can say that a difference in the pattern of the temperature dependence of the chemical shifts implies a difference in conformation. However, it is not known whether this difference is one involving backbone hydrogen bonding or rotational averaging of a sidechain group that shields an NH. The chemical shifts of the numerous protons in the two molecules are remarkably similar, although not identical. The peptide NH protons of GIn, Asn, Phe, and Tyr and the C'H protons of Phe and Tyr exhibit differences of greater than 0.1 ppm between Lys-VP and deamino-lys-vp in water, While the difference for Tyr may be attributed to the absence of the a-nh3+ group in deamino-lys-vp, the differences for the other residues suggest that the disulfide ring portions of the two molecules differ somewhat in their conformations. The aromatic rings of Phe and Tyr in deamino-lys-vp appear to be stacked, as has been shown earlier for Lys-VP (13). The evidence for the stacking is the upfield shifts of the aromatic protons in these two compounds relative to the the same values for isolated amino acids. The difference in chemical shift between the CH meta and ortho to the OH of Tyr, which is = 0.18 ppm in deamino-lys-vp and = 0.22 ppm in Lys-VP, is 0.30 ppm for tyrosine itself (7). The aromatic protons of Phe appear at two different chemical shifts, which are = 0.11 ppm apart in deamino-lys-vp, = 0.16 ppm apart

5 3326 Chemistry: Von Dreele et al. in Lys-VP, and 0.05 ppm apart in the isolated amino acid (7). The changes in the peak positions in deamino-lys-vp and Lys-VP relative to the positions found for isolated amino acids are thought (13) to arise from the magnetically anisotropic character of each aromatic ring that shields and deshields the protons of the other when the two rings are brought together in a hydrophobic bond. Although the aromatic rings of the Phe and Tyr side chains appear to be stacked in both Lys-VP and deamino-lys-vp, the data indicate that the relative positions of the rings are different in Lys-VP from those in deamino-lys-vp. The twoproton aromatic peak of Phe is 0.06 ppm or 15 Hz more downfield in deamino-lys-vp than in Lys-VP, while the threeproton aromatic peak is essentially unchanged. The meta CH peaks of Tyr are 0.08 ppm or 20 Hz more upfield and the ortho CH peaks are 0.04 ppm or 10 Hz more upfield in deamino-lys-vp than in Lys-VP. Therefore, the positions of the aromatic protons in the shielding regions of the magnetically anisotropic groups differ in the two compounds. Since the nature of the residues has not changed, it appears that there is a difference in backbone conformation that prevents the aromatic rings from assuming the same conformation in both compounds. 1. Von Dreele, P. H., Brewster, A. I., Scheraga, H. A., Ferger, Proc. Nat. Acad. Sci. USA 69 (1972) M. F. & du Vigneaud, V. (1971) Proc. Nat. Acad. Sci. USA 68, Von Dreele, P. H., Brewster, A. I., Bovey, F. A., Scheraga, H. A., Ferger, M. F. & du Vigneaud, V. (1971) Proc. Nat. Acad. Sci. USA 68, Von Dreele, P. H., Brewster, A. I., Dadok, J., Scheraga, H. A., Bovey, F. A., Ferger, M. F. & du Vigneaud, V. (1972) Proc. Nat. Acad. Sci. USA 69, Kimbrough, R. D., Jr., Cash, W. D., Branda, L. A., Chan, W. Y. & du Vigneaud, V. (1963) J. Biol. Chem. 238, Dadok, J., Sprecher, R. F., Bothner-By, A. A. & Link, T. (1970) Abstracts of the 11th Experimental NMR Conference, Pittsburgh, Pa., April Glasoe, P. K. & Long, F. A. (1960) J. Phys. Chem. 64, McDonald, C. C. & Phillips, W. D. (1969) J. Amer. Chem. Soc. 91, Dadok, J., Von Dreele, P. H. & Scheraga, H. A. (1972) Chem. Commun., in press. 9. Forsen, S. & Hoffman, R. A. (1963) J. Chem. Phys. 39, Bystrov, V. F., Portnova, S. L., Tsetlin, V. I., Ivanov, V. T. & Ovchinnikov, Y. A. (1969) Tetrahedron 25, Liler, M. (1971)'J. Magn. Resonance 5, Feeney, J., Roberts, G. C. K., Rockey, J. H. & Burgen, A. S. V. (1971) Nature New Biol. 232, Deslauriers, R. & Smith, I. C. P. (1970) Biochem. Biophys. Res. Commun. 40,