3804
J . Phys. Chem. 1991,95, 3804-381 1 the values to be so alike at high concentration is indeed remarkable. There appear to be no interferometric results for comparison with our diaphragm cell results for aqueous CoSOe On the other hand, Han and Tang employed a diaphragm cell with a thin Gelman VERSAPORE membrane to determine the interdiffusion coefficient of aqueous C&04 at 25 0C.30 Their results exceed ours by about a factor of 2 at all concentrations; however, an exact comparison is difficult, since they did not specify the technique whereby they unfolded their diaphragm cell integral diffusion coefficient data to obtain values for D(E). Figure 3 shows a plot of our results for aqueous Ni(S03NH2)> throughout the conNote that D(P) is a linear function of centration range considered, which extends from dilute to near saturation. We fitted the data in Figure 3 to the empirical equation, 2
4
6
n
IO
I2
14
Js" (Y"')
Figure 3. Differential diffusion coefficient of aqueous Ni(S03NHJ2 at 25 & 0.01 O C . The straight line is a plot of eq 3.3.
data for MgS04 and ZnSO, were obtained by interferometry.sapB The remaining results are diaphragm cell determinations.I2 Since the infinite dilution equivalent conductance of all the cations are quite close, one might expect on the basis of eq 3.1 that all five salts might have similar diffusion coefficients as c 0; but for
-
(28) (29)
Rard, J. A.; Miller, D. G. J. Solution Chem. 1979, 8, 755. Albright, J. G.; Miller, D. G. J . Solufion Chem. 1975, 9, 809.
+
D(E) = D(O)[l a(r)'/2]
(3.3)
allowing D(0) and a to be least-squares adjustable constants. We found D(0) = 0.956 X cm2 s-' and a = -0.261 M-'i2. This value of D(0) appears at the head of Table IV.
Acknowledgment. This research was sponsored by the National Aeronautics and Space Administration through Grant NAGW-8 1 with the Consortium for Materials Development in Space at the University of Alabama in Huntsville. Partial support from the donors of the Petroleum Research Fund, administered by the American Chemical Society, is also acknowledged. (30)Han, K. N.; Kang, T. K. Mefull. Trans. B 1986, 17B. 425.
''0 NMR Relaxation Tlmes of the Protein Amino Acids in Aqueous Solution. Estimation of the Relative Hydration Numbers in the Cationic, Anionic, and Zwltterionic Forms Jiirgen hutenvein,* Ioannis P. Gerothansssis,+Roger N. Hunston,t and Martin Schumacherg Institut de Chimie Organique, UniversitZ de Lausanne. CH- I005 Lausanne, Switzerland (Received: August 22, 1989; In Final Form: September 17, 1990)
The I70NMR line widths of the a-carboxyl groups of the protein amino acids includin 4-hydroxyproline, sarcosine, N,N-dimethylglycine, and Omethyltyrosine were measured in aqueous solution at 40 O C ( I ! 0 enrichment 10 atom W). A linear correlation was found between the line widths and the molecular weights of the amino acids at the pH values 0.5, 6.0, and 12.5, which are characteristic of the three ionization states of the neutral amino acids. The slopes of the straight lines were independent of pH; however, the line widths at acidic pH were increased by 98 f 13 Hz relative to those at neutral or basic pH. Since the I7O quadrupole coupling constant is only weakly influenced by the protonation state of the amino acids, it can be concluded that the a-carboxylic group is hydrated by an excess of two molecules of water relative to the a-carboxylate group. The lifetime of the water association is far below the NMR chemical shift time scale. Of the oxygen-containing functional groups of the amino acid side chains, only the phenolate ion of tyrosine was found to form water complexes that are stable within the range of the correlation times. It is shown that the I7Oline widths of the a-carboxyl groups reflect the overall rotational correlation time of the amino acids. In contrast, the a-carbons are characterized by shorter effective correlation times presumably due to internal rotation of the C,-H vector.
Introduction The hydration of amino acids and peptides is a problem of utmost importance and is a prerequisite to the understanding of protein-water interactions. 1.2 Hydration phenomena are also essential in understanding, a t a molecular level, such factors as *Address correspondence to this author at the Organisch-Chemisches Institut, UniveniUt Monster, Orltans-Ring 23, D-4400 Monster, F.R.G. Present address: University of Ioannina, 451 10 Ioannina, Greece. *Present address: Ciba-Geigy S.A.. 1870 Monthey, Switzerland. 1 Present address: Lonza AG, 3930 Visp, Switzerland.
0022-3654/91/2095-3804$02.50/0
transport mechanisms, chemical kinetics, and thermodynamic properties Of the amino acids. It iS therefore not unreasonable that the hydration of amino acids has been extensively studied by a variety of physicochemical techniques including multinuclear (1) (a) Rupley, J. A.; Yang, P. H.; Tollin, G. In Wafer in Polymers; Rowland, S.P., Ed.; ACS Symposium Series No. 127: American Chemical Society: Washington, DC, 1089; (b) Pain, R. H. In Biophysics of Water; Franks, F., Mathias, S. F., Eds.; Wiley: New York, 1982; Section 1. (2) Halle, B.;Anderson, T.; Forstn, S.;Lindman, B.J. Am.Chem. Soc. 1981, 103, 500.
0 1991 American Chemical Society
"0NMR of Amino Acids in Aqueous Solution NMR,3-S IR spectroscopy$ partial molar volume^,^ ICR mass spectroscopy,* and theoretical calculations."l However, the hydration numbers obtained were quite controversial, and it is clear that there exists a need for new methodologies. I7O NMR relaxation parameters can be considered to be an excellent measure of solution interactions and molecular dynamics,12-14but the general utility of the "0 probe has not yet approached that of the I3C or ISN nucleus. It is however of great interest to use a nucleus, such as oxygen, that is located at strategic molecular sites and is directly involved in solute-solvent interactions. I7ONMR could be a most sensitive technique for the study of the transport of amino acids across cell membranesIs and their interaction with nucleotidesI6 and metal ions.I7 The sensitivity problem (0.037% I7O at natural abundance) can be alleviated by the use of isotopic enrichment, which is readily done for the case of amino acids.'*J9 Since the "0 nucleus has a spin quantum number of S/2, it relaxes essentially by quadrupolar interaction.m I7Ospin relaxation data can therefore be translated into dynamic information, provided that the nuclear quadrupole coupling constant (QCCs) of the amino acids can be obtained by independent means. Unfortunately, although there exists a large body of I7ONQR results on carboxylic acids collected by Poplett and Smith,z1-23no data are known for amino acids. Only very
(3) (a) WOthrich, K. NMR in Biological Reseurch. Peptides and Proteins; Elsevier: Amsterdam, 1976. (b) Deslauriers, R.; Smith, I. C. P. In Biological Magnetic Resonance; Berliner, L. J., Reuben, J., Eds.; Plenum Press: New York, 1980 Vol. 2, p 243. (c) Cohen, J. S.;Hughes, L. J.; Wooten, J. B. In Magnetic ResoMnce in Biology; Cohen, J. S.,Ed.; Wiley: New York, 1983; Vol. 2, p 130. (4) Witanowski, M.; Stefaniak, L.; Webb, G.A. In Annual Reports on N M R Spectroscopy; Webb, G. A., Ed.; Academic Press: New York, 1981; Vol. 1IB, p 58. ( 5 ) St. Amour, T.; Fiat, D. Bull. Magn. Reson. 1980, I , 118. (6) (a) Pominov, I. S.;Sidorova, D. R.;Khalepp, B. P. Zh. Srrukt. Khim. 1972, 13, 1084. (b) Hollenberg, J. L.; Ifft, J. B. J . Phys. Chem. 1982, 86, 1938. (7) (a) Shahidi, F. J . Solution Chem. 1983, 12, 295. (b) Millero, F.; LoSurdo. A.; Shin, C. J. Phys. Chem. 1978, 82, 784. (8) Locke, M.; McIves, R. J . Am. Chem. Soc. 1983, 105,4226. (9) (a) F h e r , W.; Otto, P.; Bcmhardt, J.; Ladik, J. Theoret. Chim.Acra (Berlin) 1981,60,269. (b) Bonaccorsi, R.; Palla, P.; Tomasi, J. J. Am. Chem. Soc. 1984,106, 1945. (IO) Port, G. N. J.; Pullman, A. Inr. J . Quantum Chem., Quantum Biol. SymP. . . 1974, I , 21. (1 1) Mezei, M.; Mehrotra, P. K.; Beveridge, D. L. J. Biomol. Strucr. Dyn. 1984, 2, 1. (12) Steinschneider, A,; Valentine, B.; Burgar, M. I.; Fiat, D. Magn. Reson. Chem. 1985,23, 104. (13) Gerothanassis, I. P.; Hunston, R. N.; Lauterwein, J. Helu. Chim. Acta 1982, 65. 1774. (14) (a) Petersheim, M.; Miner, V. W.; Gerlt, J. A,; Prestegard, J. H. J . Am. Chrm. Soc. 1983, 105, 6357. (b) Aime, S.;Gobetto, R.; Osella, D.; Milone, L.; Hawkes, G. E.; Randall, E. W. J. Chem. Soc., Chem. Commun. 1983,794. (c) Baltzcr, L.; Bccker, E. D. J. Am. Chem. Soc. 1983,105,5730. (d) Piculell, L.; Lindman, B.; Einarsson, R. Biopolymers 1984, 23, 1683. (15) (a) Ring, K.Angew. Chem., Int. Ed. Engl. 1970,9,345. (b) Sunamoto, J.; Iwamoto, K.; Mohri, Y.; Kominato, T. J. Am. Chem. Soc. 1982,104, 5502. (16) Porschke, D.; Ronnenberg, J. Biopolymers 1983, 22, 2549. (17) Gotsis, E. D.; Fiat, D.Magn. Reson. Chem. 1987, 25,407. (18) Steinschneider, A.; Burgar, M. I.; Buku, A.; Fiat, D. Inr. J . Peptide Protein Res. 1981, 18, 324. (19) Gerothanassis, 1. P.; Hunston, R. N.; Lauterwein, J. Magn. Reson. Chem. 1985, 23, 659. (20) (a) Roger, C.; Sheppard, N. In NMR and the Periodic Table; Harris, R. K., Mann, B. E.,Eds.; Academic Press: New York, 1978; Chapter 12, p 383. (b) Klemperer, W. G. Angew. Chem.. Int. Ed. E n d 1978, 17. 246. (21) Poplett,-I. Y. F.; Smith,J. A. S. J . Chem. Soc., Faraday Trans. 2 1979, 75, 1703. (22) (a) Suhara, M.; Smith, J. A. S . J . Magn.Reson. 1982,50,237. (b) Poplett, 1. Y. F.; Smith, J. A. S.J . Chem. Soc., Faraday Trans. 2 1981, 77, 1473. (c) Poplett, I. J. F.; Sabir, M.; Smith, J. A. S . J . Chem. Soc., Faraday Trans. 2 1981, 77, 1651. (23) Broanan, S.G. P.; Edmonds, D. T.; Poplett, I. J. F. J. Magn. Reson. 1981, 45, 451.
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3805 recently, solid-state I7O NMR spectroscopy was applied to polycrystalline L-alanine, and a QCC value of 6.6 MHz was estimated.24 The QCCs for carboxylic acids and their salts were found to cover a range between 5.4 and 7.5 MHz. Furthermore, Gready studied the influence of protonation and hydrogen bonding on the 170QCCs using ab initio calculations.2Svz6 Cheng and Brownz7calculated a value of 12.8 MHz for the C - 0 group of formic acid, compared to 10.5 MHz for the formate ion. It must be emphasized, however, that neither the NQR and NMR results from solids nor the theoretical results (which are related to the gas phase) can give a sure estimate of the QCC values in solution since different forms of inter- and intramolecular interactions may exist in the various environment^.'^*^* In previous w ~ r k ~ ~the J *I7O * ~line widths of several amino acids were observed to increase at low pH with respect to neutral pH. A similar, though smaller effect was observed for the I3C and ISN relaxation rates of g l y ~ i n e ~and - ~ llater for the zH relaxation rates of glycine and alanine32(early I3C relaxation times of glycine were reported to be independent on the P H ) . ~ In ~ 1982 we concludedI3 that further investigation is needed to decide whether the observed reduction in the I7O line width is exclusively due to an alteration of the QCC or whether there is a further contribution, e.g., due to a change in the molecular correlation time by a different hydration of the ionic forms (unfortunately, Van Haverbeke et al.32did not cite our conclusion correctly). In this paper we contribute to a further understanding of the 1 7 0 relaxation of amino acids in aqueous solution by attempting to quantify the number of bound water molecules. We have analyzed, for the first time, the I7ONMR spectra of the complete set of protein amino acids (carboxyl oxygens enriched to 10 atom % in I7O) and their a-carboxyl groups relaxation times. For comparative reasons, the I7O relaxation times of the uncommon amino acids ~-4-hydroxyproline,N-methylglycine (sarcosine), N,N-dimethylglycine, and 0-methyl-L-tyrosine were also measured. As in previous studies we have chosen controlled conditions of concentration, ionic strength, temperature and pH.13J9*34We have shown that the "0line widths of the amino acids depend upon the hydration of the carboxyl groups and that a pH-dependent change of the line widths is mostly indicative of a change in the number of hydrated water molecules and not in the QCC values. Also, the "C relaxation time of the a-carbon of glycine was remeasured as a function of pH, and the effective correlation times for the I3C, 2H, and I7O nuclei were reexamined in order to get a coherent picture of the overall and partial motions of the amino acids. Experimental Section Compounds. The preparation of the I7O-enriched amino acids (10 atom %) has previously been de3~ribed.l~ As a new compound the amino acid derivative 0-methyl-L-[I7O]tyrosine was synthesized: Omethyl-L-tyrosine methyl ester hydrochloride was obtained from BOC-Omethyl-L-tyrosine dicyclohexylamine salt (Bachem) by concomitant esterification of the carboxyl group and removal of the BOC-protecting group using methanolic HCl.3S (24) Goc, R.; Ponnusamy, E.; Tritt-Goc, J.; Fiat, D. Int. J. Peptide Protein Res. 1988. 31. 130. (25) Gready, J. E. Chem. Phys. 1981,55, 1. (26) (a) Gready, J. E. J. Am. Chem. Soc. 1981,103,3682. (b) Gready, J. E.; Bacskay, G.B.; Hush, N. S . Chem. Phys. 1982.64, 1. (c) Gready, J. E. J. Phys. &em. 1984,88, 3497. (27) Cheng, C. P.; Brown, T. L. J . Am. Chem. Soc. 1979, 101, 2327. (28) (a) Marsh, R. E.; Donoune, J. Adv. Protein Chem. 1%7,22,235. (b) Nahringbauer, I. Acta Chem. Scand. 1970, 24,453. (29) Valentine, B.; St.Amour, T.; Walter, R.;Fiat, D. J . Magn. Reson. 1980, 38, 413. (30) Saito, H.; Smith, 1. C. P. Arch. Biochem. Biophys. 1974, 163, 699. (31) Leipert, T. K.;Noggle, J. H. J. Am. Chem. Soc. 1975, 97, 269. (32) Van Haverbeke, Y.; Muller, R. N.; Van der Elst, L. J . Phys. Chem. 1984,88, 4978. (33) Pearson, H.; Gust, D.; Armitage, I. M.; Huber, H.; Roberts, J. D.; Stark, R. E.; Vold. R. D.; Vold, R. L. Proc. Narl. Acad. Sci. U S A . 1975, 72, 1599. (34) (a) Hunston, R. N.; Gerothanassis, I. P.; Lauterwein, J. Org. Magn. Reson. 1982.18, 120. (b) Gerothanassis, I. P.; Hunston, R.N.; Lautenvein, J. Helv. Chim. Acra 1982, 65, 1764. (35) Brenner, M.;Huber, W. Helv. Chim. Acra 1953, 36, 1109.
3806 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991
Lauterwein et al.
' I
TABLE I: "0 Line Widtlw of the a-Carboxyl Resonances of the Protein Amino Acids and Some Amino Acid Derivatives in the Different Ionization States' L,b Hz
amino acid GIY Ala Val
Ile Leu Ser Thr Pro AspC Glue
LYS Arg Asn Gln
Met CYS TtP Phe TYt His Sar8 N,N-dimethyl-Glyr 4-hydroxy-Prd O-Methyl-Tyr8
mol wt
pH 0.5
pH 6
pH 12.5
75.07 89.10 117.15 131.18 131.18 105.09 119.12 115.13 133.11 147.13 146.19 174.20 132.12 146.15 149.21 121.26 204.23 165.19 181.19 155.16 89.10 103.12 131.13 195.19
205 276 316 453 397 359 410 286
132 194 280 335 298 249 300 197 285 316 354 394 274 330 297 256 508 376 457f 359 170 198 275 525
137 190 290 345 297 244 305 260 279 332 349 397
d d 443 482 e e
397 e 602 485 558f 438 272 329 366 622
I
600.
L IHz) . LOO.
e
e 290
_.
e
470 369 580 338 190 240 310 490
,
50
100
,
~~
200
150 M W
F g u e 1. Plot of the I'O line widths, L, of the protein amino acids versus their molecular weights, M,: (A)pH 0.5; (A)pH 6. Four uncommon amino acids are also included (cf. Table I). Both functional relationship are linear with the correlation coefficients given in Table 11. Notice that the two lines are parallel with a line-width separation AL = 98 i 13 Hz
(lower trace).
1
0.1 M solutions (10 atom I"0)in H20containing 1 M NaCl and 0.0005 M EDTA; T = 40 OC. bLine widths at half-height, estimated errors < & : S I . Measured at 48.8 MHz to separate the a- from the 8-
";,,
6001
(y-)carboxyl resonances. dOverlapping resonances. 'Not measured because of degradation. 'Saturated solutions 5T, of the amino acids, and n = 100000. For measurements at low pH the WH-360 spectrometer was chosen in order to minimize the measuring time and thus I7O back-exchange with the solvent. Effects of NaCl on the 90° pulse length were taken into account. For the evaluation of the Tl's the peak intensities were used for a three-parameter nonlinear least-squares procedure. 13C NMR Spectra. Glycine with 90 atom % I3C enrichment at C, was purchased from Stohler AG, and solutions in D20(2.5 mL) were deoxygenated before use. I3CLongitudinal Relaxation Time Measurements. TI values were determinated by inversion-recovery on the CXP-200 instrument using an internal deuterium lock. IH selective decoupling has been employed at the resonance position of the methylene group of glycine in order to minimize the decoupler power (-0.3 W)and to avoid sample heating. The nuclear Overhauser enhancement factors were measured via inverse gated decoupling
*
1 50
, 100
150
200 M W
i 2. Plot of the I7O line widths of the protein amino acids (plus four uncommon amino acids) versus their molecular weights at pH 12.5. The correlation is linear with a correlation coefficient of 0.913. Notice that the value of tyrosine, L = 580 Hz, contrary to that of Omethyltyrosine, lies outside the 95% confidence limits.
F
with delay times greater than 10TI,necessary to ensure complete return to equilibrium of both 'Hand I3C magneti~ations.~~ Viscosity Measurements. Viscosities were determined at 40 OC using a Contraves Low Shear 30 rotational rheometer based on the Couette principle. Results and Discussion 1. Correlation of the I7OLine Width with Molecular Weight. The I7Oquadrupole coupling constant: The "0NMR spectra of the cy-carboxyl groups of the protein amino acids, including 4-hydroxyproline, sarcosine, N,N-dimethylglycine, and 0methyltyrosine were recorded a t 40 OC in aqueous solution at variable pH. The I7Oline widths of the resonances are reported in Table I. It can be seen that they increase with the bulk of the amino acid side chains and appear to be independent of the site of substitution (a-carbon or amino group). From a plot of the line widths as a function of the molecular weights of the amino acids a linear correlation was established for all three ionization states (Figures 1 and 2). Table I1 collects the results of the least-squares analyses showing fairly good correlation coefficients. The straigth line correlations are highly significant as ensured, (36) (a) Canet, D. J . Mag. Reson. 1976, 23, 361. (b) Opclla, S. J.; Nelson, D. J.; Jardetzky, 0. J . Chew. Phys. 1976, 64, 2533.
170NMR of Amino Acids in Aqueous Solution
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3807
where the asymmetry correction has been ignored (the maximum error in the QCC from this source is less than 5% except for
exceptional cases where t > 0.5).42 For roughly spherical molecules, eq 4 predicts that the 170line widths are proportional to Mw, if the other parameters do not change significantly within the series of the amino acids studied. Indeed, the viscosity of the 0.1 M solutions was found to be independent of both substrate and ionization state (the viscosity of glycine in 1 M NaCl at 40 OC was found to be 7 = 0.89 f 0.02 CPat p H 0.5, 5.6, and 10.5 and independent of the addition of small quantities of EDTA). The densities of crystalline amino acids varied4' between 1.29 (leucine) and 1.66 (aspartic acid) g/cm3; however, there was no systematic variation with the molecular weight. Formal evaluation of the microviscosity factors from the van der Waals radii and applying the Gierer-Wirtz formula39gave, for example, a value off, = 0.22 for glycine. However, the well-known complexities of hydrogen bonding are not adequately re resented by such a simplistic model of solute/solvent boundary,' *39 and f, is expected to increase through the solvation of the amino acids. Edward,"' from experimental diffusion coefficients of several amino acidsM and the assumption that the carboxylate group of the amino acids is bounded to two water molecules with a consequent increase of approximately 37 A' in the van der Waals volume of the diffusing particle, evaluated values off, = 0.79 and 0.88 for glycine and alanine, respectively. Since in general the hydration of the amino acids is expected to be even more extensive (see below), the stick boundary conditionf, = 1 seems to be well approached for the amino acids in aqueous solution. The latter conclusion was adopted also without discussion by Steinschneider et al.I2 The linear correlations between L and Mw that were observed a t pH 0.5 and 6 (Figure 1) and, with the exception of tyrosine, also at p H 12.5 (Figure 2) are in support of the hydrodynamic model adopted.'* This strongly suggests that the oxygen QCCs of the amino acids a t a given ionization state change little from one compound to another (the standard deviation of the slope at pH 6, corresponding to that of the square of the QCCs, is f6%, Table 11; thus the scatter of the QCC's themselves is f3%). Furthermore, the identical slopes obtained for the different ionization states (Table 11) indicate clearly that the oxygen QCCs, within experimental error, are independent of pH. Equation 4 predicts that the intercept of the plot of the line widths versus Mw should be zero. However, we observed nonzero intercepts (Table 11) depending on the ionization state of the amino acids. They can be understood by assuming an effective molecular weight composed of the real molecular weight of the amino acids plus the weight of bound hydrate water (see below). Evaluation of the oxygen QCC of the amino acids was therefore made from the slope of the L versus Mw relation (eq 4). With an average p = 1.5 g/cm', we calculated under our solution conditions ( T = 313 K I] = 0.89 cP;fr = 1) a QCC of 7.5 f 0.25 MHz, in reasonable agreement with the value of 6.6 MHz (t = 0.55) estimated recently from 170NMR studies of polycrystalline Lalanine.2' We conclude that the oxygen QCC of the amino acids is independent of both the ionization and the degree of hydration of the carboxyl group. The influence of hydrogen bonding on the QCC has been estimated earlier by calculating the QCCs of both the monomer and the dimer of formic acid.25 Dimerization decreased the QCC of the c-0group by -1.7 MHz and increased that of the C-OH group by +0.8 MHz. There is also experimental evidence that the QCC of carbonyl groups is decreased by hydrogen However, in carboxyl groups that undergo rapid intermolecular proton transfer in aqueous solution," the mean effect would be smaller because of a partial cancellation of the QCC changes. It appears very difficult to transfer the
(37) Huntress, Jr., W. T. In Advances in Magnetic Resonance; Waugh, J. S.,Ed.; Academic Press: New York, 1970; Vol. 4, p 1. (38) Boer€, R.T.;Kidd, R.G.In Annual Reports on NMR Spectroscopy; Webb, G. A., Ed.; Academic Press: New York, 1982; Vol. 13, p 319. (39) (a) Giercr, A.; Wirtz, K. 2. Naturforsch. A 1%3,8,532. (b) Glasel, J. A. J. Am. Chem. Soc. 1969, 91,4569. (40)Edward, J. T. J. Chem. Educ. 1970,47, 261. (41) Noggle, J. H.; Schirmer, R. E. The Nuclear Overhauser E'ect, Chemical Applications; Academic Press: New York, 1971.
(42) Loewenstein, A. In Advances in Nuclear Quadrupole Resonance; Smith, J. A. S.,Ed.; Wiley: New York, 1983; Vol. 5, p 53. (43) Handbook of Chemistry and Physics; Weast, R. C., Ed.; Chemical Rubber Publishing Co.: Cleveland, 1982. (44) Longsworth, L. G. J. Am. Chem. SOC.1953, 75, 5705. (45) St.Amour, T. E.; Burgar, M.I.; Valentine, B.;Fiat, D. J. Am. Chem. SOC.1981, 103, 1128. (46) Jaccard, G.;Lauterwcin, J. Helv. Chim. Acta 1986, 69, 1469. (47) Kintzinger, J. P.In NMR.. Basic Principles and Progress; Diehl, P., Fluck, E., Kosfeld. R.,Eds.; Springer: New York, 1981; Vol. 17, p 26.
TABLE II: Statistics for a Least-Squares Linear Regression Analysis of Eq 4a
PH statistics
0.5
6
12.5
slop& 2.89 f 0.22 2.88 f 0.17 2.78 f 0.29 intercept" -5.4 f 11.2 30.0 f 6.6 20.7 f 11.6 0.963 0.913 corr coeff 0.954 no. of points 19 24 21d std dev of regression 34.7 27.0 41.1 OFitting of the line-width data was pcrformed via a linear equation of the type y = a(x + 6), where x = molecular weight of the amino acids and 6 = weight of the bound hydrate water (see section 4c). b f standard errors. CInterceptat the Mw axis. dIncluding value for tyrosine. for example, by Fisher's I transformation. Besides the line-width measured for tyrosine at high pH (Figure 2, see discussion below), all experimental points lie within the 95% confidence intervals of the least-squares lines. The line widths at neutral and basic p H were identical, having the same linear relationship within experimental error (Table 11). At pH 0.5, however, the straight line was parallel-shifted relative to that at pH 6, the line widths being increased by the constant amount of AL = 98 f 13 Hz (Figure 1). In diamagnetic solutions the I7O nucleus (electric quadrupole cm2) relaxes predominantly by the moment Q = -2.6 X quadrupole relaxation mechanismaZ0In the extreme narrowing limit, as would be expected for dilute solutions of amino acids, and assuming isotropic molecular reorientation, the expression for the rate of 170quadrupole relaxation is given by37
where x = $q,Q/h is the oxygen quadrupole coupling constant (QCC) and c = (qxx- qy,,!/qp is the asymmetry parameter. The qrz,qYpand qxxare the principal components of the electric field gradient tensor. T~ is the effective correlation time which characterizes the fluctuations of qrz at the oxygen nucleus. If TC in eq 1 can be identified with a single correlation time from overall molecular reorientation (T,,J, we can apply the expressionB for an isotropically tumbling rigid sphere in a medium of viscosity 7:
= Vmtlf,/kT (2) where V, is the molecular volume andf, is a microviscosity factor S1 depending on the relative sizes of solute and s0lvent.3~ V, can be estimated"' as Tmol
V,,,
0.74MW/N,p
(3)
where No is Avogadro's number and M, and p are the molecular weight and the density of the solute, respectively. Equation 3 implies that the amino acids in the crystalline state adopt a hexagonal compact structure occupying 74% of the available space in the lattice as spheres."'*" Inclusion of eq 1 results in a relationship between the line widths and the molecular weights of the form (4)
B
3808 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991
results for a static system to the complex kinetic situation in solution. However, small changes of a few percent of the QCC as a result of different substitution effects and/or hydration of the carboxyl groups of the amino acids, inside the estimated limits of error, cannot be excluded. 2. Anisotropic and Internal Motion. The linear correlations established in Figures 1 and 2 seem to discount a major contribution to oxygen relaxation from anisotropic motions and internal rotation of the amino acids. However, because of the scatter in the data it is unclear whether there is systematic deviations from linearity especially at pH 12.5 and 0.5. Therefore, it is not possible to distinguish isotropic motion from these other two possibilities, and so isotropic motion was assumed. However, independent argument for the isotropic motion of the amino acids comes from the similarity of the I7O line widths for the isomers alanine and sarcosine with different sites of methyl substitution and also from that for molecules with comparable molecular weight but different shape (e.g., Asp, Leu, Ile, and 4-OH Pro; Table I). Parbhoo and Nagy48 obtained confirmation for the isotropic motion of glycine by combined I F and IH relaxation time measurements. Assuming stochastic diffusion49of the carboxyl group about the Ca-Co bond and assuming that the largest principal axis of the field gradient tensor of the carboxyl groups is oriented 60' with respect to the internal rotational axis,33 the internal correlation time should be at least 3 times that of the overall molecular reorientation time because of the linearity of the plots. 3. I7O Line Widths at High pH. The case of tyrosine: In a previous pH titration of glycine," we showed that the increase of the 170line width at high pH arises from t r a m of paramagnetic metal ions and that it can be eliminated by addition of minor amounts of EDTA.I3 In fact, under our standardized conditions containing EDTA we can confirm here (Table I) that amino acids with primary amino group (with the exception of tyrosine) have similar I7O line widths a t pH 6 and 12.5. Amino acids with secondary or tertiary amino groups, Le., proline, sarcosine, and N,N-dimethylglycine, exhibit slightly larger I7Oline widths in the anionic state. Furthermore, the linear dependence of the 170line widths on the molecular weight was found identical at both pH values (Table 11). The a-carboxyl resonance of tyrosine experiences a strong broadening with L = 580 Hz, relative to an expected value of L = 440 Hz from the regression line, and therefore falls outside the 95% confidence limits (Figure 2). The increased line width of tyrosine at high pH cannot be explained by an exchange broadening, since we found that the line width was independent of the magnetic field. The possibility of self-association of tyrosine could also be excluded since no concentration dependence of the I7O line widths was detected. Line broadening due to paramagnetic ions can be ruled out since progressive addition of EDTA up to 10 times the standard concentration did not influence the observed line width. In conclusion, we attribute the high pH broadening of tyrosine to a strong increase in hydration of the hydroxyl group upon deprotonation. The deprotonation (pK, a! 10.2) leads to an increase in the electronic charge density at the oxygen lone pairs of the phenolate ion and, as a consequence, allows strong binding of (probably two) water molecules. Strong argument in favor of this interpretation comes from the fact that the line width of the carboxyl groups of 0-methyltyrosine was unchanged on going from neutral to high pH (Table I). The present finding parallels the large hydration numbers determined in proteins for the deprotonated side chain of tyrosine.50 The I7Oline widths are not sensitive to the state of the charge at the a-amino groups of the amino acids. This means that either the hydration is not significantly changed on deprotonation of the NH3' group or that the complexes are too weak to be detected. The latter conclusion seems to be supported by the I7O line width behavior of proline, sarcosine, and N,N-dimethylglycine, which (48) Parbhoo, B.;Nagy, 0. B. J. Phys. Chem. 1985,89, 239. (49) (a) Woessner, D. E. J. Chem. Phys. 1%2,37,647. (b) Woessner, D. E.; Snowden, B. S.,Jr.; Meyer, G.H. J. Chem. Phys. 1969, 50, 719. (50) Kuntz, 1. D.; Kauzmann, W. Ado. Protein Chem. 1974, 28, 239.
Lauterwein et al.
TABLE III: l70Spin-Lattice and Spin-Spin Relaxation Time of Glycine as a Function of pH'
HZ0
1.44
2.43
2.35
1.42
2.31
2.32
2.18 2.07 1.25 2.13 2.08 D20 1.22 "0.1 M solutions containing 1 M NaCl and 0.001 M EDTA;T = 40 OC. bFrom inversion-recovery; estimated error i5%. 'From line widths at half-height; estimated error i 5 % . dpH could not be lowered further because of I7O back-exchange during the TI experiment.
is normal with respect to the linear correlations (Figures 1 and 2) although the number of hydrogen-bonded water molecules should be reduced in these N-substituted amino acids.5' Also, no influence on the I7Oline widths was seen upon titration of the t-amino group of lysine and the guanidino group of arginine. Finally we note that the I7O line widths of serine and threonine follow the linear correlations at both pH 6 and 12.5. Obviously, the acidity of these hydroxyl groups is not sufficient to produce a comparatively strong hydration. The same observation was made for serine- or threonine-containing cyclic dipeptide^.^^ 4. Increase of the 170Line Widths at Low pH. Upon protonation of the a-carboxyl groups of the amino acids the linear correlation of the I7O line width with molecular weight was preserved (Figure l), and the slopes of the two straight lines a t pH 0.5 and 6 were identical within error (Table 11). However, as already emphasized, a general increase in the line widths was observed on going from neutral to acid pH (AL = 98 f 13 Hz; Figure 1). Three explanations for the parallel line-width plots at neutral and acid pH are possible: (a) an additional relaxation contribution other than that from the quadrupolar mechanism, (b) intra- and intermolecular associations, or (c) a change in the hydration state of the carboxyl groups and therefore of the effective molecular weight of the amino acids at low pH. The eventualities of a contribution from these mechanisms are discussed below. ( a ) 170Relaxation by mechanisms other than quadrupolar: First of all we shall consider the influence of intermolecular proton-exchange processes. For the case of rapid exchange between different sites, but before the complete sharpening of the resonance, T2 (but not T I )should depend on the magnetic field according to53
where TM-' and T2B-1are the relaxation rates in the two environments A and B, f A and T~ are the residence times, and PA and PBare the populations of the respective sites. We have measured the I7O line width of several amino acids at different magnetic fields (4.7 and 8.5 T) and have also determined the T Ivalues of glycine a t pH 1 and 6 (Table 111). Since neither a field dependence of T2or a difference between T I and T2 was observed, we can conclude that exchange processes play a negligible role on the I7O relaxation times. The same conclusion was obtained earlier for the high pH range of glycine." Several other relaxation mechanisms must be examined for an eventual contribution to the I7O line widths. Spin-rotational relaxation, as concluded earlier from I3C and ISN N M R analyses,31,33can be assumed to be of no importance at least for the larger amino acids. Relaxation due to the rotational modulation of the chemical shift anisotropy can be excluded due to identical line widths at the two different magnetic fields. Chemical exchange of the carboxyl protons could also effect the I7O relaxation because of the I7O,IH scalar spin-spin inter(51) King, E. J. Acid-Base Equilibria; Pergamon Press: London, 1965; Chapter 7. (52) (a) Deslaurier, R.; Grzonka, 2.;Schaumburg, K.;Shiba, T.;Walter, R. J. Am. Chem. SOC.1975,97, 5093. (b) Levy, G. C.; Kumar, A.; Wang, D. J. Am. Chem. SOC.1983, 105, 7536. (53) Pople, J. A.; Schneider, W. G.; Bernstein, H. J. High-resolution Nucleur Mugnetic Resonunce; McGraw-Hill: New York, 1959; Chapter 10, p 218.
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3809
I7O NMR of Amino Acids in Aqueous Solution action.S3 Line broadening occurs when the rate of proton exchange l / r e (where T~ is the average lifetime of either state) approaches the order of magnitude of the coupling constant J (hertz). Then 1/ TZw = r2J%,
I
I
zwitterionic
anionic
(6)
with the contribution to the longitudinal relaxation being negligible.53 The measure of the T I and T2values of glycine at both pH 1 and 6 did not however result in a difference in the two relaxation modes (Table HI). In addition, measurements in D20 have shown a similar increase in their low pH liiewidth as observed in H 2 0 , although '.IoDis reduced by -6.5 with respect to lJoH (in fact, in D20 the relaxation rates are increased relative to H 2 0 by a factor 1.15, independent of pH, which corresponds to the change in viscosity; Table 111). This observation has already been made for acetic acid,13 and thus the proton-exchange rates are such that PT: CC 1. No contribution can be expected from scalar relaxation. In conclusion, it can be confirmed that the quadrupolar interaction governs exclusively, and independently of the pH, the I7O relaxation behavior of the amino acid carboxyl groups. (b) Intra- and intermolecular association: We have previously shown" that neither an intra- or an intermolecular interaction exists for glycine between the negatively charged carboxylate group and the positively charged a-amino group in the zwitterionic state. This conclusion can now be extended to the other amino acids because of the linear dependence of the I'O line widths on molecular weight (Figure l). The same conclusion can also be made for the amino acids at high pH since the I7O line widths are equal to those at neutral pH as long as the influence of paramagnetic impurities has been eliminated (Figure 2). The situation at low pH appears more complex since both the I7O line widths and the spin-lattice relaxation rates of the a-carbons increase, although by differing amounts (cf. Tables 111and IV and discussion below). However, intermolecular interactions between the amino acids can be excluded since no concentration dependence of either the I7O chemical shifts19 or the line widths is observed in aqueous solution. The IsN NMR analysis of the pH dependence of the T I and nuclear Overhauser values of histidine in aqueous solution revealed that it undergoes aggregation around the pK, (-5.5) of the imidazole side chain by forming dimeric structures between protonated and deprotonated ring systems.54 Our results suggest that this interaction does not exist, since the 170line widths of histidine behaved absolutely normal (Table I), although 170line widths are very sensitive to dimerization and/or oligomerization processes, as we have recently shown for N-protected amino acids in organic solution.s5 In addition, the I7Ochemical shift of the carboxyl group of histidine was found to be in the expected pHdependence range.I9 ( c ) Hydration of the amino acids as detected by I7O NMR spectroscopy: Since all relaxation contributions other than quadrupolar have been eliminated and since the oxygen QCC of the amino acids at neutral and acid pH changes very little, the difference in the I7Oline widths at the two ionization states can be explained only by a change in the rotational correlation time of the amino acids (cf. eq 1) and hence in their effective molecular weights (cf. eq 4). Obviously, the degree of hydration of the amino acids would be expected to change in the two ionization states. The observed I7O line widths (Table I) imply that the protonated form of the carboxyl group of the amino acids is forming a larger number of Yobservable"hydrogen bonds than the deprotonated form. To produce an overlap of the two straight lines at pH 0.5 and 6,the effective molecular weight of the amino acids at low pH must be increased by 35 f 10 (see also the difference in the intercepts at the Mw axis, Table 11), a result that suggests the complexation of an additional two molecules of water on protonation of the carboxyl groups of the amino acids. These hydrated
-
(54) Blomberg, F.; Maurer, W.; ROterjans, H. J . Am. Chem. Soc. 1977, 99. 8149. (55) (a) Hunston, R. N.; Gcrothanassis, I. P.; Lautemein, J. J. Am. Chem. Soc. 1985, 107, 2654. (b) Gerothanassis, I. P.; Hunston, R. N.; Lauterwein, J., manuscript in preparation.
cationic
Figure 3, Model for the hydration sites of the a-carboxyl group of alanine in the different ionization states. The disposition of the water molecules is obtained from the calculations for formic acid and the formate ion by Port and Pullman.'o
complexes must have lifetimes that are smaller than the NMR chemical shift time scale but presumably larger than the rotational correlation time scale (2-10 ps at 40 "C) to reorient as proper units.s6 We want to point out that the larger hydration number found for the carboxylic groups of the amino acids does not necessarily mean that the hydrogen bonds are also stronger. On the contrary, it is generally that water forms stronger hydrogen bonds with the carboxylic acid ions than with the neutral molecules, and larger hydration numbers of neutral carboxylic acids have been observed by light ~ c a t t e r i n g partial , ~ ~ molar volume determinations,' and 2HNMR spectro~copy.~~ The reduction in solvation of the amino acids at neutral pH has been assumed to be the result of an intramolecular ionic interaction between the COO- and NH3+g r o ~ p s . l ~However, s~~ we have given arguments that such an interaction is unlikely (see ref 13 and above). Figure 3 shows a schematic representation of the hydration sites for the a-carboxyl group of alanine in its three ionization states. The solvation in the zwitterionic state is adopted from Port and Pullman,Io who studied the formate ion-water interaction as a prototype of the carboxylate group. The authors identified three energetically favorable hydration sites, two equivalent sites on the carboxylate oxygens at the exterior of the ion and one water bridging the two oxygen atoms (Figure 3). However, the bridging structure of the type O--H20--O- is not supported by Monte Carlo computer simulations and seems unfavorable from an entropic point of view." Formic acid was used as a model for the hydration of the unionized carboxyl group.1° The five most favorable modes of interaction with water including a water molecule hydrogen bonded to the hydroxyl hydrogen of the carboxyl group are shown in Figure 3. Clearly, hydration numbers of 5 (or 4) and 3 (or 2), with the experimentally determined difference of 2,appear to be the most probable conclusions for the carboxyl groups of the amino acids in the corresponding ionization states. Spisni et al." have assessed the hydration of alanine and proline by the analysis of the carboxyl I7O chemical shifts in water/ DMSO mixtures. The authors suggested hydration numbers of 3 and 2 at acid and neutral pH, respectively. Data for the 170 chemical shifts of carbonyl-H20 hydrogen-bonding interactions are now available, and shift differences of -52 ppm have been found for the bonding of two water molecules to the carbonyl of acetone,59amides,60and uridines.61 However, the data appear less certain for carboxylic acids. Since the I7O chemical shifts of hydroxyl groups are less sensitive to hydrogen bonding than those of carbonyl groups and are usually in the opposite directi or^,^^^^ quantification of the hydration numbers on the basis of these shifts appears to be less reliable compared to that of amide and peptide bonds.5 Nevertheless, the result of Spisni et a1.58is (56) Laszlo, P. Prog. N M R Specrrosc. 1979, 13, 257. (57) (a) Alms, G. R.; Bauer, D. R.; Brauman, J. I.; Pecora, R. J . Chem. Phys. 1973,59, 5321. (b) Bauer, D. R.; Brauman, J. I.; Pecora, R. J . Am. Chem. Soc. 1974.96.6840. (58) Spisni, A,'; Gotsis, E. D.; Fiat, D. Biochem. Biophys. Res. Commun. 1986, 135, 363. (59) Reuben, J. J . Am. Chem. Soc. 1969, 92, 5725. (60) Burgar, M. I.; %.Amour, T. E.; Fiat, D. J. Phys. Chem. 1981,85. 502. .
(61) Schwartz, H. M.; MacCcss, M.; Danyluk, S. S. J . Am. Chem. Soc. 1983,.105. 5901.
3810 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991
TABLE IV: "C Spin-Lattice Relaxation Times and Nuclear Overbauaer Enhancement Factors of ClyciM1)c,,p as a Function of
PH conditions* pD 1.28 pD 6.0
TI: s
NOE"
3.94 t 0.10 3.99 0.12 5.19 t 0.19 5.01 f 0.12
2.98 3.06 3.03 2.81
*
'90% enrichment. M solutions in D20 containing 1 M NaCl and 0.001 M EDTA: T = 40 t 1 O C . C T windependent ~ experiments were performed to examine the reproducibility of the data. "Estimated error < i s % . 8According to the relation@pD = pH + 0.4 this should correspond well to the conditions for ''0 relaxation times measurements.
close to ours from I7O line widths, with the exception that a difference of one molecule of hydrated water was obtained in the two ionization states instead of the two molecules proposed by us.
5. Indirect Determination of the Oxygen Quadrupole Coupling Constant. To evaluate the QCCs of the amino acids in aqueous solution, one can apply an alternative approach6zthat determines the effective correlation time T~ from I3C TI relaxation studies. If the measured TI value is exclusively from the dipolar relaxation mechanism, and assuming isotropic reorientation, then (7)
where rcH is the vibrationally averaged C-H bond distance, assumed to be 1.09 A, N is the number of attached hydrogens, and K, a constant, is 2.14 X 1O1O s-l. It is usually accepted3sb3that TC evaluated from the C, relaxation time can be identified with the overall reorientation time, 7,,,,,of ,, the amino acids (for further discussion, see below). On the other hand, T~ rmolwas assumed for the reorientation of the carboxyl oxygens, since both anisotropic and internal motions of the carboxyl groups appear to be without important contribution. Clearly this approach, to be consistent, needs identical solution conditions in the I7O and 13C NMR experiments. 13CTI measurements were performed for glycine[I3C,] in D 2 0 at neutral and acidic pH in the presence of 1 M NaCl (Table IV). Maximum nuclear Overhauser enhancement factors were obtained for the &-carbon (Table IV) indicating that the relaxation is dominated by dipolar interaction with the directly attached methylene protons. From the value of TI(I3C) = 5.10 s obtained at pD 6 from the average of three independent measurements, a value of T~ = 4.6 ps was calculated (eq 7). To enable a direct comparison, the "0line widths of glycine were measured also in D 2 0 (L= 150 Hz a t pD 6; Table 111). Then, with rc = 4.6 ps a QCC value of 10.4 MHz was obtained from eq 1. It is difficult to give a realistic estimate of the error of determination of the QCC. Assuming that the approach of identical correlation times for the I3C and 170relaxation is correa, the error is given by the precision of the relaxation times (both T I and T2 ca. 43%) but also by the sensitivity of the 4 factor to variations in the C-H bond lengthb5(eq 7). Taking into account an error of k0.005A in rCH,the QCC of glycine was determined with an accuracy of approximately *5%. However, the value of 10.4 MHz is clearly different from that value determined above from the Stokes formula (eq 2), and it was therefore necessary to pursue the investigation. A reexamination of the assumption of identical correlation times for both the C, and 0 nuclei of the amino acids seemed to be necessary. In the past years several effective correlation times for the C, atom of glycine at neutral pD have been determined (62) (a) Kintzinger, J. P.;Lehn, J. M.J. Am. Chem. Soc. 1974,%, 3313. (b) London,R.E. In Magnetic Resowme in Biology;&hen, J. S.,Ed.; Wiley:
New York, 1980. (63) Deslauriers, R.;Paiva, A. C. M.;Schaumburg, K.; Smith, I. C. P. Biochemistry 1975, 14,878. (64)Glasoe, P. K.; Long, F. A. J . Phys. Chem. 1960,64, 188. (65) Komorski, R. A.; Levy, G. C. J. Phys. Chem. 1976, 80, 2410.
Lauterwein et al. by T I measurements (5,'O 8.0,)35.8,48 and 5.5& ps; concentration and temperature variable). The deuterium relaxation times of the C, deuterons of glycine gave rC = 4.6 ps in the zwitterionic state.32 All these results are very similar to the value (4.6 f 0.5 ps) evaluated in the present work (eq 7). However, they are much smaller than those values obtained from eq 2. Nery and Canet,& from longitudinal relaxation studies of 13C in uniformly enriched glycine, evaluated a larger correlation time for the C,-Co bond (rm = 17 f 4 ps) than for the C,-H bond (TCH 5.5 f 0.3 ps). Assuming rotational diffusion of the C-H vector around the C,-Co a correlation time for internal motion rin,= 2.0 ps can be determined. We conclude that the identification of rcc with T~ is certainly a better approximation than the assumption TCH =. 1 , 7 Although there are relatively large uncertainities in the effective correlation times of Nery and Canet, we assume that under our solution conditions rCHmust be increased by a factor of approximately 3 to give a realistic rmO1 value. In this way, T~ = 13.8 ps, perfectly compatible with the values obtained from the Stokes formula. 6. Comparison of the 170,I3C,and 2H Relaxation Times of Glycine at Neutral and Low p H . It is clear that the analysis of the "0line width changes on transition from the zwitterionic to the cationic state of the amino acids has to be supported by a simultaneous consideration of different atoms at different positions in the same molecule. We obtained a ratio of 1.68 f 0.12 of both the TI and T2I7Orelaxation times of glycine at pH 1 and 6, for both HzO and D 2 0 solutions (Table 111). However, the ratio of the relaxation times of the a-carbon of glycine, under identical solution conditions, was only 1.28 f 0.06 (Table IV). We think that the precision of the latter result is superior to earlier 13C TI measurements which showed an even smaller variation as a function of pH.30-33 Furthermore, the relaxation times of the methylene deuteriums of [2Hs]glycine have been reported,32and despite the strange profile of the titration curve a ratio of the relaxation at pH 0 and pH 7 of 1.24 f 0.12 (after correction for a viscosity change) was evaluated. Obviously, the TImeasurements of the a-carbon, and similarly the a-deuteriums, of glycine show a much smaller difference at neutral and acid pH than the I7O relaxation times and are therefore less of an indication of a hydration dependence on the ionization state of the amino acid. In N M R studies of the rotational motion of amino acids, systematic uncertainties in the molecular parameters can cause large errors in the estimation of the effective correlation In quadrupole relaxation studies, such as by 2H or 170NMR, inappropriate values for the nuclear QQC values may be retained since pure liquid-phase values are lacking (see above). However, we have shown above that the I7O QCC of the amino acids are only weakly influenced by their ionization state. Van Haverbeke et al.32 also assumed that the deuterium QCC of glycine is insensitive to pH. I3C relaxation measurements can be handicapped by the lack of an exact knowledge of the molecular geometry. The crystal structure of the zwitterionic form of glycine has been determined by neutron diffraction@ (C,H bond distances equal 1.090 A, within the standard deviations). However, no C,H distances are really known for the protonated carboxyl group. It is indeed conceivable that the C,H distance is shorter a t pH 1 than a t pH 7. Due to the inverse sixth-power dependence of the dipolar relaxation rates on distance (eq 7) a 4% change in the average bond length will result in a 25% change in TI. A possible interpretation of the discrepancy of the relaxation time ratios of glycine obtained by either I7O NMR or 13C/2H N M R spectroscopy could be based on the results of Nery and Canet.& These authors, at neutral pH, found a larger correlation time for motion of the C,-Co vector than for that of the C,-H vector (see above) and concluded that the latter undergoes rapid internal rotation around the C-C axis. On the other hand, we can assume from the present work that the motion of the principal (66) (a) Nery, H.;Canet, D.J. hfugn. Reson. 1981,42,370. (b) Nery, H.; Canet, D.; Toma, F.; Fermandjian, S.J. Am. Chem. Soc. 1983, 105, 1482. (67) Eliasson, B.; Larsson, K. M.;Kowalewski, J. J . Phys. Chem. 1985, 89. 258. (68) JBnsson, P. G.; Krick, A. Acta Crystullogr. 1972,828, 1827.
3811
J. Phys. Chem. 1991,95, 381 1-3815 axis of the oxygen field gradient tensor can be identified with T ~ Then, at low pH, the increase of the hydration number of glycine should cause a proportional increase in T ~which , is clearly larger than the observed increase in the effective correlation time T C H , provided that the internal rotation is not subtantially down' The observation of a very similar ratio at pH 7 and 1 for the 2H and I3Crelaxation times of the methylene group of glycine supports the present explanation since both nuclei have the same angular dependence and experience the same kind of motion.
.
Acknowledgment. This research was supported by the Swiss National Science Foundation. We thank Martial Rey for iterative calculations. Registry No. Gly, 56-40-6; Ala, 56-41-7;Val, 72-18-4;Ile, 73-32-5; Leu, 61-90-5;Ser, 56-45-1;Thr, 72-19-5;Pro, 147-85-3;Asp, 56-84-8; GI,,, 56436-0;L ~56-87-1; ~ , Arg, 74-79-3;Asn, 70-47-3;Gin, 56-85-9; Met, 63-68-3;cys,52-90-4;Trp, 73-22-3;Phe, 63-91-2;Tyr, 60-18-4; His,71-00-1; Sar, 107-97-1;Hyp, 51-35-4;Me-Sar, 1 1 18-68-9;Tyr(Me),
6230-11-1; I'O, 13968-48-4.
Carbon Dioxide Clathrate Hydrate Epitaxial Growth: Spectroscopic Evidence for Formation of the Simple Type-I1 COP Hydrate Fouad Fleyfel and J. Paul Devlin* Department of Chemistry, Oklahoma State University, Stillwater, Oklahoma 74078 (Received: September 21, 1990)
It is known that small atomic and diatomic species such as argon, krypton, nitrogen, and oxygen preferentially form a structure-I1 clathrate hydrate through occupation of the structure-I1 small (and lar e) cages. There have apparently been no similar reports indicating that small triatomic molecules having a diameter of 5.4 or less also disobey the classical clathrate hydrate size-structure rule by forming a structure-I1 hydrate. However, an exception has now been recognized in FT-IR studies of C02clathrate hydrate metastable thin films prepared by using epitaxial cryogenic methods. For the type-I C02simple hydrate, grown epitaxially to the ethylene oxide clathrate hydrate under kinetic control, an occupation of both the small and the large cages is indicated by C02v,-band peak frequencies of 2347 and 2335 cm-', respectively. More surprisingly, the small C02molecules, with a maximum van der Waals parameter of 4.7 A, are also capable of "stabilizing" the type-I1 hydrate through interactions within the small cages. This is evident from a single v j absorption band at 2345 cm-I for the simple hydrate grown epitaxially to the structure-I1 clathrate hydrate of tetrahydrofuran at 150 K. Thermal effects on the spectra of these simple hydrates, attributed to the onset of rotational motion of the guest molecules within the large cages, are also examined. For example, a transition dipole coupling of the C02guest molecules of neighboring cages, apparent from a comparison of the bands of the isotopomers of the naturally abundant 12C and ')C isotopes, is disrupted by warming.
f
Introduction X-ray diffraction,'J NMR, and dielectric analysi~'~have been used to study the structure, formation, and molecular motion of ethylene oxide (EO) and tetrahydrofuran (THF) clathrate hydrates. It was concluded that the simple clathrate hydrates of these two molecules obey the classical size-structure rules, with the guest molecules occupying the appropriate cages according to their size (Le., THF molecules occupy the large cages of a type-I1 hydrate, and EO molecules occupy the large and, to a lesser degree, the small cages of a type-I hydrate). Recently though, it has been shown that small atomic and diatomic molecules such as argon, krypton,6 oxygen, and nitrogen'** do not follow the Glew, D. N.; Rath, N. S.J. Chem. Phys. 1966, 44, 1710. (2) Mak, T. C.W.; McMullan, R. K. J . Chem. Phys. 1965, 42, 2732. (3) (a) Davidson, D. W.; Ripmtester, J. A. In Inclusion Compounds II; Atwood, J. L., Davies,J. E. D., MacNicol, D. D., Eds.;Academic: New York, 1984;Vol. 3. (b) Gough, S.R.; Whalley, E.; Davidson, D. W. Can. J . Chem. 1968,46, 1673. (c) White, M. A.; MacLean, M. J. Phys. Chem. 1985,89, 1380. (4)Davidson, D. W.; Davies, M. M.; Williams, K.J. Chem. Phys. 1964, 40. . -,3449. - . .- . (5) Davidson, D. W. In Water, a Comprehemioc Treutisc; Franks, F.,Ed.; Plenum: New York, 1973;Vol. 2, Chapter 2. (6)Davidson, D. W.; Handa, Y. P.; Ratcliffc, C.1.; Tsc, J. S.;Powell, B. M.Nafure 1984, 311, 142. (7)Davidson, D. W.; Gough, S.R.; Handa, Y. P.; Ratcliffc, C.I.; R i p mtester, J. A.; Tse, J. S.J . Phys. (Frunce) 1987,Colloque C1, Suppl. _ . to No. 3, 48, Cl-537. (8) Nakahara, J.; Shigesato. Y.; Higash, A.; Hondoh, T.;Langway, C . C. Philos. Mag. B 1988, SI, No. 3. 421. (1)
0022-3654/91/2095-3811$02.50/0
clathrate hydrate classical size-structure rule but favor a type-I1 structure, confirming what was claimed by Holder and Manganiello in 1982.9 Also, it has been noted that the methane hydrate, containing only minor amounts of a contaminant, tends to grow as a type-I1 rather than type-I hydrate.'O Infrared spectroscopy can also be an important method of study of the clathrate hydrates. In pioneering work, Bertie acquired both X-ray diffraction and infrared data for the same bulk-grown EO hydrate sample." By determining the structure as well as the absorption spectrum of a single sample, he established a bridge that relates the two methods. Later, jointly with Devlin,I2 he reinvestigated the FT-IR spectrum of the EO clathrate hydrate, prepared using a cryogenic thin-film vapor-deposition technique, and established this approach as a viable option in the formation and study of the clathrate hydrates of many polar molecules. Richardson et al. took advantage of the simplicity of this new technique to extend the FT-IR studies to the clathrate hydrates of additional gases including THF." Though it proved difficult to grow thin films of the simple clathrate hydrates of the small nonpolar polyatomic molecules such as ethane, methane, cyclopropane, carbon dioxide, and acetylene, it was possible to form their mixed hydrates if the deposit contained a small amount of ~
~~
(9)Holder, G. D.; Manganiello, D. G. Chem. Eng. Sci. 1982,9, 37. (10)Sloan, E.D. Paper Presented at 69th Annual GPA Convention, 1990. (11) (a) Bertie, J. E.; Othen, A. D. Can. J . Chem. 1972,5I,1159. (b) Bertie, J. E.; Othcn, A. D. Can. J. Chem. 1973,51, 3641. (12) Bertie, J. E.;Devlin, J. P. J . Chem. Phys. 1983,78, 6340. (13) Richardson, H.H.; Woolridge, P. J.; Devlin, J. P. J . Chem. Phys.
1985,85,4387.
0 1991 American Chemical Society
