N-H Hydrogen Bond in Nucleic Acid Bases
and ligand character and the interaction between ligands within these complexes. Acknowledgment. This research was supported by the National Science Foundation (CF 40894 and CHE 7605716). The authors express their thanks to Clifford Carlin for his assistance with the preparation of Figures 1-5.
References and Notes (1) J. Kahl, K. Hanck, and K. DeArmond, J . Phys. Chem., preceding paper in this Issue. (2) G. Kew, K. Hanck, and K. DeArmond, J. fhys. Chem., 79, 1828 (1975). (3) R. Lane and A. Hubbard, J . Phys. Chem., 77, 1401 (1973). (4) J. Kahl, K. Hanck, and K. DeArmond, J. phys. Chem., 82,540 (1978). ( 5 ) K. DeArmond and K. Hanck, to be published. (6) J. Kahl, Ph.D. Thesis, North Carolina State Unlversity, 1978. (7) C. Flynn, Jr., and J. Demas, J. Am. Chem. SOC.,96, 1959 (1974). (8) C. Flynn, Jr., and J. Demas, J. Am. Chem. Soc.,97, 1988 (1975).
The Journal of Physical Chemistty, Vol. 83, No. 20, 1979 2615
(9) R. Watts, J. Harrington, and J. Van Houten, J. Am. Chem. Soc., 99, 2179 (1977). (10) J. Kahl, K. DeArmond, and K. Hanck, J. Inorg. Nucl. Chem., 41, 495 (1979). (11) A. Vlcek, Rev. Chim. Mlner., 5, 299 (1968). (12) N. TokeCTakvciyan, R. Hemingway, and A. Bard, J. Am. Chem. Soc., 95, 6582 (1973). (13) R. Nlcholson, Anal. Chem., 37, 1351 (1965). (14) R. Wopschall and I. Shain, Anal. Chem., 39, 1514 (1967). (15) J. Demas, E. Harris, and R. McBrkle, J. Am. Chem. Soc., 99, 3547 (1977). (16) I. Hanazaki and S. Nagakua, 5uU. Chem. Soc.Jpn., 44, 2312 (1971). (17) T. Saji and S. Aoyagui, J. Elecfroanal. Chem., 63, 31 (1975). (18) S. Margel, W. Smtth, and F. Anson, J. Electrochem. Soc., 125, 241 (1978). (19) G. Kew, K. DeArmond, and K. Hanck, J. phys. Chem., 78, 727 (1974). (20) M. K. DeArmond, C. M. Carlin, and W. L. Huang, Inorg. Chem., In Dress. (21) N. Tanaka, T. Ogata, and S.Niizuma, Bull. Chem. SOC.Jpn., 46, 3299 (1973). (22) H. Caklararu, K. DeArmond, K. Hanck. and V. Sahlnl, J. Am. Chem. Soc., 98, 4455 (1976).
The N-H Hydrogen Bond. 2. Models for Nucleic Acid Bases J. N. Spencer,* Jeffrey E. Glelm, Charles H. Blevins, Robert C. Garrett, Fred J. Mayer, Johanna E. Merkle, Susan L. Smlth, and M. Louise Hackman Department of Chemistty, Lebanon Valley College, Annville, Pennsylvania 17003 (Received December 28, 1978; Revlsed Manuscript Received June 4, 1979) Publication costs asslsted by the Petroleum Research Fund
The pure-base calorimetric method has been used to determine enthalpies of formation for hydrogen-bonded complexes of pyrrole, indole, and imidazole with various bases. These enthalpies are compared to those obtained by other methods. Frequency shifts determined in CCld solvent have been used to find AH-Av relationships for N-H-eO and N-H-N adducts. From these relationships, hydrogen-bond enthalpies have been calculated for bases not determined by the pure-base method. The systems investigated have been used as models to calculate enthalpies of formation for the more complex nucleic acid base pairs. Comparison of enthalpies calculated from the model compounds is made to enthalpies determined by theoretical and other experimental methods.
Introduction This is the second article of a series reporting on calorimetric and spectroscopic investigations of N-H.-N and N-H-00 hydrogen bonds. These hydrogen bonds have particular significance for studies of molecules of biological interest, but, because of the difficulty of obtaining thermodynamic parameters for such weak hydrogen-bonded systems, few reliable data are available. The pure-base calorimetric procedure used for this study has several advantages over other calorimetric or spectroscopic means of investigation for the study of weakly hydrogen-bonded systems.
Experimental Section The calorimetric and spectroscopic procedures have been previously described, as have the purifications of most reagent~.l-~Aldrich imidazole was recrystallized from benzene, vacuum dried, and further dried over Pz05. Aldrich 99+ 5% N-methylimidazole was used without further purification. Dioxane was dried over NaOH and distilled from NaOH in an N2 atmosphere. Acetonitrile was allowed to stand over Pz06until no orange coloration of Pz05was noted and then distilled under N2 Indole was purified by vacuum sublimation. N-Methylindole was refluxed over CaO and distilled under N2. Aldrich pyrimidine was used without further purification. 0022-365417912083-2615$01.OO/O
For the spectroscopic studies the N-H stretch at about 3500 cm-l was monitored. Indole concentrations were about 0.006 M, pyridine and DMF concentrations were about 0.1 M, and N-methylimidazole concentration was about 0.03 M. A slight correction for overlap of indole and indole-DMF complex bands was made. Thermodynamic parameters were calculated from the temperature dependence of the equilibrium constant as previously reported.2 Frequency shifts were determined in CC14solvent for all systems. Base concentrations were varied to detect any possible dependence of the complex frequency on base concentration.6 Enthalpies of solution were determined by injecting about 0.5-5 mmol of acid into 200 mL of solvent. With the exception of imidazole in CHCl,, no concentration dependenciesof the enthalpies of solution were noted. The imidazole-CHC1, data were extrapolated to infinite dilution. The calorimetric pure-base approach to the determination of enthalpy changes is independent of the equilibrium constant, provided that the equilibrium constant is large enough to ensure complete complexation. According to the pure-base method, if a small quantity of hydrogen-bonding acid is injected into the base as a solvent, two contributions to the heat observed are involved: the heat due to hydrogen bonding and that heat 0 1979 American Chemical Society
2616
The Journal of Physical Chemistry, Vol. 83, No. 20, 7979
Spencer et al.
TABLE I: Spectroscopic Results for Indole Complexes in CC1, --AH, kcal mol"
shown7that an inert reference solvent is not necessary for use in the pure-base method, provided that the reference solvent does not interact with the proton of the acid. Even though CHC13would be expected to hydrogen bond to the imine nitrogen of imidazole and N-methylimidazole, the interaction enthalpy should cancel when the difference ( N s A- A&M)ref is taken. Spectra of pyrrole and N-methylpyrrole in CC14 show two absorption maxima, with one slightly shifted toward longer wavelengths. The longer wavelength band is probably due to N-H-N or N-H-.n interactions of the pyrrole with N-methylpyrrole. Spectra of imidazole and N-methylimidazole in CC14 show an absorption band due to free imidazole and a broad absorption band 307 cm-l from the free band. No band attributable to imidazole interaction with the amine nitrogen or n electrons of N-methylimidazole is seen. Thus the conclusion of Tovrog and Drago8 that binding occurs only through the imine nitrogen of N-methylimidazole appears to be substantiated. Table IV lists a comparison of the enthalpies of hydrogen-bond formation obtained by various methods. For consistency only the pure-base enthalpies determined with CHC13 as the reference solvent are given. Also given in Table IV are the pure-base enthalpies corrected as previously describedl for differences in acid and model compound polarizabilities and cavity formation. Calculations based on the Schroeder-Lippincott potential function model for hydrogen-bonded systems have shown that a linear correlation between AH and Au can be expected only if proton donor and proton acceptor are similar: i.e., a correlation between the frequency shift and enthalpy for the N-H-N systems of this study should exist, but N-H-0 enthalpies should not fall on the same AH-Au plot. Nozari and Drago'O determined the enthalpy for complexes of pyrrole with triethylamine, pyridine, and MezSO and then calculated enthalpies for other bases from E and C parameters. A good AH-Au correlation was found for all systems. Figure 1 gives the relationship between
Au,
K,,,
cm-'
indole-pyridine
3.7 i 0.3a 4.15 3.55' 4.04' indole-DMF 3.5 t O.la 10.7 3.40' 9.85' indole-N-methylimidazole 4.0 i 0.1 18.6
257 271' 164 166' 273
a The error reported is the least-squares error in the slope of a In K vs. T1 plot. H. Dunken and H. Fritzsche, 2. Chem., 2, 379 (1962).
'
term which might occur if there were no hydrogen bonding. If a proper model compound is chosen, the hydrogenbonding heat term can be isolated. An inert reference solvent is chosen to correct for heats of solution of the acid and model compound.
Results The spectroscopic data are given in Table I along with the results of Dunken and Fritzsche for comparison. Table I1 lists the enthalpies of solution at infinite dilution for the various compounds of this work. The enthalpies of hydrogen-bond formation, AHHBf, calculated from the pure-base equation6 (eq 1)are given in Table 111. AHH,,f =
( N S A - rnSM)base - ( N S A - N S y r e f
(1)
In eq 1 A&, refers to the enthalpy of solution, the superscripts A and M refer to acid and model compound, and base and ref refer to the pure base and reference solvent, respectively. The enthalpies calculated from this equation will be referred to as pure-base enthalpies. In all cases the N-methylated compound was used as the model compound. For the pyrrole systems three different reference solvents (CC14,cyclohexane, CHC13)were used with nearly identical results. Because of the Emited solubility of indole and imidazole in C C 4 and cyclohexane, CHC13 was the reference solvent for these systems. It has previously been TABLE 11: Enthalpies of Solution at Infinite Dilutiona
-
AH,, kcal mol-' ( 2 9 8 K ) solvent CCl, cyclohexane CHCI, ethyl acetate DMF pyridine N-methylimidazole dioxane acetonitrile
pyrrole
+ 2.22
i 0.06' t 3.74 t 0.13' t 0 . 3 3 i 0.02
-1.00 -2.18 -2.01
i i t
imidazole
N-methylimidazole
indole
N-methylindole
t 5 . 0 1 i 0.01'
-2.24 i 0.14
t 2 . 9 3 i 0.03
-0.99 i 0.05
t 2 . 5 7 i 0.06 t 3 . 0 1 c 0.02 t 2 . 2 3 i 0.04
-0.17 i 0.01 t 0 . 1 8 i 0.04
-0.51 c 0.03 -0.42 i 0.02
-0.28 -0.11
N-methylpyrrole
0.02
0.08 0.05
-1.23 t 0.06 -0.32 i 0.02
t 0.38 i 0.02' i 0.06'
t1.85 -1.47 +0.06 -0.19 -0.06
-0.08
i
0.02
t 0.01 i 0.02 i
t
0.01
C
0.02
0.00
0.01
t 0 . 1 8 i 0.01
+4.66 c 0.20
+0.07 i 0.01
The error reported is the standard deviation of a single measurement. of AHvs. number of moles.
Reference 1.
a
' Error in the intercept of a plot
TABLE 111: Enthalpy of Complex Formation Calculated by the Pure-Base Method
-AHHBf, kcal mol" ( 2 9 8 K) reference solvent CCl, pyrrole-pyridine -DMF -ethyl acetate -dioxane -acetonitrile indole-pyridine -DMF imidazole-pyridine -DMF -N-meth ylimidazole -acetonitrile
* 0.01
3.79 3.83 2.90 2.99 2.34
i
0.14
c 0.18 c 0.11
0.15 c 0.11 i
cyclohexane 3.84 0.25 3.88 ?: 0.29 2.95 ?: 0.22 3.04 i 0.26 2.39 i: 0.22 _+
CHCI, 3.75 3.79 2.86 2.95 2.30 4.23 4.15 4.42 4.51 5.02 2.66
0.10 0.14 i 0.07 ?: 0.11 i: 0.07 f
?:
f
0.12
c 0.12 i 0.21 ?:
i i
0.22 0.19 0.36
The Journal of Physical Chemistry, Vol. 83, No. 20, 1979 2617
N-H Hydrogen Bond in Nucleic Acid Bases
TABLE IV: Comparison of Pure-Base Enthalpies with Enthalpies from Other Methods -A
pyrrole-p yridine -DMF -ethyl acetate -dioxane -acetonitrile indole-pyridine -DMF imidazole-pyridine -DMF 4-methylimidazole -acetonitrile
pure base
pure base COP
3.75 3.79 2.86 2.95 2.30 4.23 4.15 4.42 4.51 5.02 2.66
3.7 3.6 2.9 2.8 2.1 4.1 4.0 4.3 4.3 4.8 2.5
H H ~kcal ~ ,mol-' c a1
E,C
AH-AV
AH-Avl
3.8,d 5.0e
4.9i 3.9i 3.0' 3.5' 2.6'
4.4k
4.7 3.5 2.7 3.3 2.7 5.0 3.8 5.4 4.0 5.6 3.1
spec 3.2,b 4.3c
1.92f 3.7,g 3.55h 3.5,8 3.40h
2.5'
4.Y 4.q 4.55 3.w
Corrected according to procedures given in ref 7; the estimated error in the correction procedure is i 0.2 kcal mol-'. J. A. Happe, J. Phys. Chem., 65, H. J. Wimette and R. H. Linnell, J. Phys. Chem., 66, 546 (1962). CCl, solvent, IR. S. N. Vinogradov and R. H. Linnell, J. Chem. Phys., 23, 93 (1955). CC1, sol72 (1961). Cyclohexane solvent, NMR. vent. e M. s. Nozari and R. s. Drago, J. A m . Chem. SOC.,92, 7086 (1970). Cyclohexane solvent. s. s. Mitra, J. Chem. Reference of footnote b, Table I. Calculated from the E and C Phys., 36, 3286 (1962). CCl, solvent. g This work. parameter equation. F. L. Slejko and R. S. Drago, J. A m . Chem. SOC.,95, 6935 (1973). j Calculated from E and C parameters for pyridine, DMF, and N-methylimidazole to obtain E A and CA for imidazole: E , = 3.03, CA 7 0.210. E , and CB for N-methylimidazole were taken from ref 8. Calculated from the constant base A H - A V relationship given in ref 10. Calculated from the A H - A U relationship given in ref 10. Frequency shifts from Table V. a
'
'
TABLE V: Pure-Base Enthabies and Freauencv Shifts - A H H ~ kcal ~ , mol-'
N-H.
100
300
200
*
*N
pyrrole-pyridine -acetonitrile -pyrimidine -N-methylimidazole indole-pyridine -pyrimidine -N-me thylimidazole imidazole-pyridine -N-methylimidazole -acetonitrile -pyrimidine
AV Figure 1. Pure-base enthalpies vs. frequency shift: (0)N-H. ..O=C; (A)N-H--N; (0)pyrrole-dioxane. AH in kcal mol-', Au in cm-'.
the pure-base enthalpies and frequency shifts in CC14for the systems of this study. It is apparent that two separate linear relationships apply. In fact, even the point for pyrrole-dioxane falls considerably off the line connecting AH and Au for N-H...O=C systems. Column 8 of Table IV gives AH calculated from the Nozari-Drago relationshiplo between AH and Au. Corrected and uncorrected pure-base enthalpies, frequency shifts, and enthalpies calculated from the straight-line equations of Figure 1 are given in Table V. Also included in Table V are enthalpies calculated from the AH-Au equations for several systems for which frequency shifts in C C 4 were determined. For the uncorrected pure-base enthalpies for N-H-N systems -AH = (0.0106 f 0.0009)Av + (1.51 f 0.2) (2) For the corrected pure-base enthalpies -AH = (0.0107 f 0.0005)Au (1.33 f 0.14) (3) The uncorrected pure-base AH-Au relation for N-H-.O=C systems is -AH = (0.0156 f 0.0011)Av + (1.68 f 0.16) (4) and the corrected relation is -AH = (0.0133 f 0.0011)Av + (1.87 f 0.16) ( 5 )
+
pure calcd AV,a pure base basea cor eq 2 eq 3 cm-' 3.75 2.30
3.7 2.1
4.0 2.3 3.4 4.2
3.8 2.1 3.2 4.0
232 72 174 253
4.23
4.1
4.2 3.7 4.4
4.1 3.5 4.3
257 203 273
4.42 5.02
4.3 4.8
4.6 4.8
4.4 4.6
291 307
2.66
2.5
2.6 4.1
2.5 4.0
105 245
- a H H B fkcal , mol-' N-H.. -O=C pyrrole-DMF -ethyl acetate indole-DMF imidazole-DMF
pure calcd pure base AV,b base cor ea 4 ea 5 cm" 3.79 2.86 4.15 4.51
3.6 2.9 4.0 4.3
3.8 2.9 4.2 4.4
3.7 2.9 4.1 4.2
135 75 164 176
Reference solvent is CHCl, for all systems. Frequency shift in CC1,: pyrrole, u 3498 cm-I; indole, v 3492 cm";imidazole, v 3485 cm-'. A v = i.10cm-'.
Pullin and Wernerll have given frequency shifts for indole and pyrrole with a variety of C=O proton acceptors. By use of eq 4 or 5, the uncorrected or corrected pure-base enthalpies for many systems having the N-H-.O=C linkage may be calculated. A relation between frequency shift and free energy is also given by these investigators so that estimates of entropy changes for N-H-O=C systems may also be made. The single-scale enthalpy equation proposed by Sherry and P ~ r c e l l may ~ ~ Jbe ~ used to provide estimates of the formation enthalpy of different acids with the same base. The equation -AH = CY& has been used to predict enthalpies of various alcohols with various bases. CY is defined where AHHref is the formation enthalpy for a as AH/AHHref, reference acid complex. If the corrected enthalpy for the
2618
The Journal of Physical Chemistry, Vol. 83, No. 20, 1979
Spencer et al.
TABLE VI: Corrected-Pure-BaseEnthalpies for Complexes Containing Linkages of Biological Interest complex
- A H , kcal mol-'
complex
- A H , kcal mol''
pyrrole-pyridine -pyrimidine -N-methylimidazole indole-pyridine -pyrimidine -N-methylimidazole imidazole-pyridine -pyrimidine -N-me thylimidazole N-methylaniline-pyridine -pyrimidine -N-me thylimidazole
3.7' 3.2b 4.0b 4.1' 3.5b 4.3b 4.3' 4.0b 4sa 2.1c
pyrrole-DMF -ethyl acetate -N,N-diethylacetamide indole-DMF -ethyl acetate -N, N-diethylacetamide N-methylaniline-DMF -ethyl acetate
3.6a 2.ga 4. If
4.0' 3.1f 4.4f 2.3e
1.6e
1.8d
2.3d
' Determined directly from corrected-pure-base data. Calculated from eq 3 and Au given in Table V. and 7. Estimated from the single-scale enthalpy e q u a t i ~ n . ' ~ *e ' Corrected-pure-base ~ data from ref 1. Au given in ref 11 and eq 5. pyrrole complex with pyridine is chosen as a reference, a may be calculated for the pyridine complexes with indole, imidazole, and N-methylaniline. The N-methylanilinepyridine complex enthalpy has been previously reported.' The best value is -2.1 kcal mol-' of complex. a determined for a series of acids with the same base is constant when a different series of bases is used. Thus a determined for the pyridine complexes with pyrrole, indole, imidazole, and N-methylaniline is the same as a determined for the same acids with other bases. If the a values determined for the bases pyridine, pyrimidine, and N-methylimidazole are used, it is possible to obtain reliable estimates of the hydrogen-bond enthalpy for N-methylaniline with pyrimidine and N-methylimidazole. The enthalpies so estimated are given in Table VI. It is possible by inspection of the data in Table V to make less sophisticated estimates for the enthalpy of complexation for N-methylaniline with pyrimidine and N-methylimidazole, but the single-scale enthalpy equation does seem to be established and estimates made by other means are insignificantly different from those made by the single-scale equation.
Discussion The pure-base enthalpies corrected for differences in acid and model compound are considered to be the most reliable, and all subsequent calculations will be made by using the pure-base-corrected enthalpies. For convenience of reference, the enthalpies that bear on the following discussion have been compiled in Table VI. The compounds studied in this work may be considered to be model compounds for more complex biological molecules. The data in Table VI may then be used to estimate hydrogen-bond enthalpies for systems of biological interest. One of the few direct experimental studies of the hydrogen bonding of nucleic acid bases is that of Kyogoku, Lord, and Rich.14 Adenine and uracil derivatives were studied by IR spectrophotometry in CHC1, solvent. Solvation effects due to specific interaction of CHC13with the bases studied may lead to calculated enthalpies much different from those obtained in less solvating media. The enthalpy change for the formation of the phenol-pyridine complex in cyclohexane has been reported3 to be -7.27 kcal mol-l while for the same complex in CHC1, the enthalpy is -5.10 kcal mol-l. Even larger differences are observed for phenol-Me2S0 complexes ,in cyclohexane and CHClP3 The self-association enthalpy of 9-ethyladenine was determined by Kyogoku et al.14 Only cyclic dimeric self-associated species were assumed to exist. Three structures are possible for the cyclic dimer of ethyladenine. In one structure (I) two hydrogen bonds from the amino group to the imidazole nitrogen are used, a second structure (11)has two H bonds from the amino group to
References 1 Calculated from
,c
a pyrimidine nitrogen, and a third structure (111)contains amino H bonds to the imidazole and pyrimidine nitrogens. Structure I should have a self-association enthalpy of twice the N-methylaniline-N-methylimidazoleenthalpy given in Table VI, Le., -4.6 kcal mol-'. Similarly, structure I1 would have a formation enthalpy of -3.6 kcal mol-' and structure 111, an enthalpy of -4.1 kcal mol-l. Kyogoku et al. report the self-association enthalpy to be -4.0 f 0.8 kcal mol-l. It is not possible to determine which structure is correct in CHC13solution and it is possible that all three coexist, but on the basis of enthalpy arguments, structure I is preferred. It does appear that the dimer is cyclic. The difference in enthalpy for the three structures reflects the greater basicity of the imidazole nitrogen over that of the pyrimidine nitrogen. The very good agreement between the enthalpy calculated from the model systems and the spectrophotometric data is surprising, especially for this dimeric species. Chloroform is known to interact strongly with pyridine; estimates in the literature range from -2.4 to -3.4 kcal rn01-'.'~-~~The pyrimidine nitrogens are less basic than the nitrogen of pyridine but the imidazole nitrogen is more basic than the pyridine nitrogen. Strong specific interactions with the adenine nitrogens, which must be overcome to provide a site for hydrogen bonding, would be expected. The mixed ethyladenine-cyclohexyluracil system was also studied by Kyogoku et al.14 The associated adenine-uracil complex was assumed to be cyclic. Four different structures are possible. The amino group of adenine can hydrogen bond with the carboxyl oxygen of uracil at carbon 2 or carbon 4. For reasons to be discussed later, bonding at O4 is preferred and DMF will be used as the carbonyl-containing model compound. Uracil may bond through the N1 or N7 of adenine. The enthalpies of these structures may be estimated if a suitable model for the N-H bond of uracil can be found. The pK for N3 of uracil is 9.3,18the pK for pyrrole is 15,19and the pK for imidazole is 6.99.20 Indole is intermediate in acidity between pyrrole and imidazole, and the best choice for a model for the proton-donating ability of uracil would seem to be indole. The enthalpy for the uracil-adenine dimer with uracil N3 bonded to the pyrimidine nitrogen of adenine is -5.8 kcal mol-'. For the structure bonded through uracil N3 to the adenine imidazole nitrogen (N1-N3H) the enthalpy is -6.6 kcal mol-'. For both these structures the adenine amino group was considered to be bonded to the 4-carbon oxygen of uracil. The N1-N3H structure with thymine replacing uracil 'is the Watson-Crick structure found in DNA, N7--N3His the Hoogsteen structure. The Hoogsteen structure is predicted from the model compounds to be slightly more stable than the Watson-Crick
N-H Hydrogen Bond in Nucleic Acid
Bases
structure. Kyogoku et al.I4 reported the enthalpy for the adenine-uracil dimer to be -6.2 f 0.6 kcal mol-l. Binford and Holloway,21using calorimetric techniques, reported the AU association enthalpy in CHC13to be -6.2 f 0.3 kcal mol-l. The enthalpy of self-association of l-cyclohexyluracil was determined by Kyogoku et al.14 by assuming that only monomers and dimers existed in solution. The equations used to treat the spectral data would not allow determination of whether the dimer was cyclic or linear, The hydrogen-bonded interaction in this uracil is between the N-H bond of one pyrimidine moiety to the carbonyl of another. Indole bonding to ethyl acetate or DMF is the best model that can be obtained from the data in Table VI. Of the two bases DMF would seem to be the best choice. This leads to a N-H-O=C bond enthalpy of -4.0 kcal mol-l. If a cyclic dimer is formed, an H-bond enthalpy of -8.0 kcal mol-’ would be expected. N,NDiethylacetamide might be a better model for the carbonyl proton acceptor of uracil. Pullin and Wernerll give the frequency shift for the complex with indole to be 187 cm-l. This frequency shift used in eq 5 gives an N-H-O=C enthalpy of -4.4 kcal mol-l. Kyogoku et al.14 report the dimer enthalpy for the uracil to be -4.3 f 0.4 kcal mol-l, an enthalpy which corresponds to the formation of only one hydrogen bond as calculated from the model compounds. The reported frequency shift for the uracil dimer is 211 cm-l in CHC13.14 For alcohol systems the frequency shift in C C 4 compared to that in CHC13 ranges from 3-50 cm-l less. The dimer frequency shift for DMF in CCl, is 12 cm-l less than in CHC13. If the extremes are taken for an estimate of the uracil-uracil frequency shift in CC14, the enthalpy may be calculated from eq 5 to be between -4.0 and -4.7 kcal mol-l. Kyogoku et al.14assumed the uracil dimer to be cyclic. If this assignment is correct, the enthalpy in a relatively nonsolvating medium must be about twice that found in CHC13. In view of the good agreement between the model compound calculations for adenine and adenine-uracil complexes, the disagreement for the uracil dimer is surprising. The principal reason for considering the dimer to be cyclic was that the enthalpy was 1.5 times that of a N-H-O=C bond in CHC13.14 According to the references cited,24the bond for which this enthalpy was reported was the self-association bond in aCetyl-DL-nOrleUCine Nmethylamide. This compound does not contain a ring nitrogen but rather an amino nitrogen. The formation enthalpies for the two types of nitrogen do not seem to be comparable (Table VI). In addition, the cited value24for the enthalpy of this bond is -3.9 kcal mol-l. Evidently Kyogoku et al. read this as -2.9 kcal mol-l. Poulter and Frederick%@have reported that protonation a t O4 is considerably favored over protonation at O2 for uracil. The pK of DMA is -0.1927 and the pK of DMF would be expected to be similar. The pK for ethyl acetate is -5.1.27 The difference between the pK for ethyl acetate and DMF is of the same order as the difference in pK for O2 and O4 of uracil.26 Thus at least for comparisons of stability of bonding at Oz or 04,DMF seems a reasonable model for bonding at O4and ethyl acetate, for bonding at 0 2 . If the dimer of uracil is linear, bonding should be to the carbonyl oxygen at the 4-carbon position. The enthalpy calculated on this basis is -4.0 kcal mol-l which is in good agreement with that reported by Kyogoku et al. if the dimer is linear. However, there is additional evidence that the dimer may be cyclic. Iwahashi and Kyogoku28 have studied the 13C NMR spectra of l-cyclohexyluracil 6122923
The Journal of Physical Chemistry, Vol. 83, No. 20, 1979 2619
and have found a downfield shift in the Cz- and C4carbonyl group with increasing concentration. They further calculated that hydrogen bonding at C2 does not affect the electron density at C4and vice versa. The uracil derivatives were found to use bonding through the C4carbonyl group more often than the Cz-carbonyl group. Although C4bonding is preferred, reversed Watson-Crick and Hoogsteen bonding (through the Cz-carbonylgroup) still seems likely. If DMF and ethyl acetate are used for the model carbonyl groups and indole the nitrogen proton donor, the enthalpy of the cyclic dimer is calculated to be -7.1 kcal molT1. If Nfl-diethylacetamide is used for the O4carbonyl, the cyclic uracil dimer enthalpy is -7.5 kcal mol-l. The difference in H-bond enthalpy between bonding at Oz or O4 is about 0.9-1.3 kcal mol-l. Unless there is considerable strain in the assumed cyclic structure of the uracil dimer, the enthalpy reported by Kyogoku et al. seems low. However, as previously mentioned, solvation effects in CHC13solvent could be large. Hydrogen-bond formation by CHC13 to both carbonyl groups of uracil could occur. If this is the case, it is difficult to understand why the model calculations for AA and AU pairs agree so well with the spectrophotometric analysis. It is possible that mixtures of linear dimers formed by bonding to O2and O4 form. The 04-bonded dimer would be expected to predominate. Such a mixed dimeric system should show two absorption bands which are not evident in the published spectra. Because bonding a t the O2 position is weaker than bonding at the O4position, perhaps the CHC13solvent effectively competes for the Ozsite. The CHCl,-ethyl acetate enthalpy has been reported to be -2.5 kcal close to the N-H bonding enthalpy to Oz, estimated from the model compounds at -3.1 kcal mol-’. This explanation would also account for the 13CNMR shift for Cz observed in CHC13 solvent by Iwahashi and Kyogoku. However, as previously mentioned, CHC13 competition for the imidazole nitrogen in the adenine dimer would be expected to be comparable to that of the amino group. Such does not seem to be the case. D’Albis et al.30have studied derivatives of adenine and uracil in CHC13-H20 mixtures. They found that little or no competition occurs between the formation of base pairs and binding of water to the bases. They attribute this behavior to double-hydrogen-bondformation by the uracil carbonyl groups. Binford and HollowayZ1reported no significant difference in the association constant for AU in CHC13 or CHC13-ethanol mixtures. Alvarez and Biltonen31have also questioned the competitive efficacy of water with the hydrogen-bonding sites for pair formation for thymine in water. They propose that nucleic acid bases do not form all possible hydrogen bonds with the solvent. The suggestion of D’Albis et al.30of double-hydrogenbond formation may account for these observations on the competitiveness of the solvent for H-bonding sites. If CHC13solvates the carbonyl or nitrogen of nucleic acid bases, upon hydrogen-bond formation an effective solvation site must be “squeezed” out. The electronegative atoms comprising the hydrogen bond become more basic, thus providing a stronger site for CHC13 solvation and possibly canceling out solvation effects. It also seems likely that in certain cases the solvent may effectively compete for a hydrogen-bond site, as may be the case with uracil dimers. Binford and Holloway21did find a difference in enthalpy for AU in CHC13 and CHC1,-ethanol mixtures of 1.3 kcal mol-l. On the basis of the agreement of enthalpies calculated from models and the spectroscopic and calorimetric enthalpies reported by Kyogoku et al.14and
2820
The Journal of Physical Chemistry, Vol. 83, No. 20, 1979
TABLE VII: Comparison of Monopole Method to Model Compound Estimates of Base Pair Energetics base paiP AUe
I I1
AA
UU
AC GU
IV V I I1 I11 I I1 I11 I I1
I I1
GC
I
V,,b -6.07 -5.65 -5.59 -5.22 -3.48 -3.25 -3.55 -7.03 -6.56 -6.61 -5.06 -4.89 -8.47 -9.81 -15.03
AHC
H-bond contactsd
-5.9 -5.8 -6.6 -5.1 -4.6 -3.6 -4.1
N,H. *Oz,N,. * .N,H N,H. SO,, N,. *N,H N,Ha * no, N,. * aN,H N,H** -Oz,N,.* *N,H N,H***N,,N,**.N,H N,H.* *NL, N,.* .N,H N,H.* *N,, N,. * *N,H
-8.0
N,H..*O,,O,*..N,H N,H. * n o , , 0,. * .N,H N,H.* no2,0,.* *N,H N,H* *N,,N,. * .N,H N,H** -N,, N,. .N,H O,*.*N,H,N,H.**O, N,H** *04,0,.* .N,H
-6.2 -7.1 -4.1 -3.6 -7.1 -8.0 -7.4
9
*N,H,N,H** *N,, N,H* ‘ 0 ,
0,. *
9
The Roman numeral refers to the structure given by Nash and Bradley.33 In kcal mol-,. Calculated by monopole summing ref 33. In kcal mol-’. Estimated from model compounds. The numbering scheme follows that given in ref 20,pp 56-57, e Structure I for AU is reverse Hoogsteen, I1 is for Watson-Crick, IV is for Hoogsteen, and V is for reverse Watson-Crick. a
by Binford and Holloway,21the double hydrogen bond formed between acid, base, and solvent seems a likely possibility for some systems. In other cases, the solvent may compete for H-bond sites. These conclusions are of particular significance concerning the stability of proteins and nucleic acids in solvents capable of forming hydrogen bonds. The arguments advanced for models for bonding at O2 and O4 for uracil may be extended to estimate enthalpies for other nucleic acid base pairs. The enthalpy for the reverse Hoogsteen structure for adenine-uracil with bonding by the adenine amino group to O2 may be estimated by using the N-methylaniline-ethyl acetate bond enthalpy and the indole-N-methylimidazole bond enthalpy. Similarly, the enthalpy for the reverse WatsonCrick structure may be estimated from the N-methylaniline-ethyl acetate and the indole-pyrimidine bond enthalpies. The reverse Hoogsteen structure has an enthalpy of -5.9 kcal mol-l and the reverse Watson-Crick structure, an enthalpy of -5.1 kcal mol-’. The enthalpy of the Watson-Crick structure for guanine and cystosine may also be estimated. The cystosine amino group bonding enthalpy to O6of guanine will be assumed to be given by the N-methylaniline-DMF enthalpy. The guanine amino group bond enthalpy to O2of cytosine is taken to be that of the N-methylaniline-ethyl acetate bond. The indole-pyrimidine bond enthalpy is used for the guanine N1 bonding to cytosine N3 The enthalpy so estimated for GC is -7.4 kcal mol-I. Newmark and CantoP2in an NMR study of guanosine-cytidine in MezSO solvent found the enthalpy of association to be -5.8 kcal mol-’. The agreement between this enthalpy and that obtained by use of model compounds is quite reasonable for a strongly solvating medium such as Me2S0. Nash and Bradley33summed the electrostatic energy terms of the monopoles for all pairs of atoms for the four commonly occurring nucleic acid bases to compute the energies of various base pairs. In this type of calculation, the contributions due to hydrogen bonding are automatically included in the energy term. The orientation of base pairs corresponding to maximum hydrogen-bond formation gave potential minima. Even though purely
Spencer et ai.
electrostatic forces were considered, Nash and Bradley described the complexes as “hydrogen-bondedcomplexes” because the N-H-.O and N-H-N contacts were essentially linear in most cases. Some stable configurations had bifurcated bonds. Nash and Bradley suggested that qualitatively correct geometries of base pairs but not energies ab initio could be obtained by simply drawing pairings in which such hydrogen bonds occur. Donohue and T r ~ e b l o o dhad ~ ~earlier obtained 29 possible H-bond schemes for N-methylated bases by using this procedure. Donohue’s results agreed qualitatively with those calculated by Nash and Bradley. Nash and Bradley suggested that estimates of N-Ha-0 and N-H-sN hydrogen-bond strengths could be used to estimate the relative stabilities of the Donohue-Trueblood configurations but cautioned that by using only the three atoms involved in the hydrogen bond many atom-atom terms in the energy would be ignored. It is now possible to carry out this analysis by using the model compounds of this work to estimate the relative energies of the base pairs studied by Nash and Bradley. This is, of course, exactly the procedure that has been followed for the base-pair enthalpies previously discussed. Table VI1 lists the energies calculated by Nash and Bradley for selected base pairs and the energies estimated by the model compounds of this work. The agreement between the calculated energies for AU and the enthalpies estimated from model compounds is quite good. Only in the Hoogsteen structure (IV) is there a large difference between the two values. The agreement with other systems is not as good, but relative estimates of stability are, on the whole, reasonable. The calculated UU energy is close to that estimated from model compounds if a cyclic dimer is assumed. This supports the earlier contention that the enthalpy of Kyogoku et al.14 reported’for UU may refer to the formation of a single hydrogen bond. The most serious disagreement is for the GC pair. This difference is too great to be attributed to any errors in model compound hydrogen-bond enthalpy estimates and suggests additional stability for GC considerably in excess of that provided by hydrogen bonds. Acknowledgment. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research. The authors also acknowldege Cindy College, a Project SEED student, for her assistance in many phases of this work.
References and Notes (1) J. N. Spencer, J. E. Gleim, M. L. Hackman, C. H. Blevins, and R. C. Garrett, J. fbys. Cbem., 82, 563 (1978). (2) J. N. Spencer, J. R. Sweigart, M. E. Brown, R. L. Bensing, T. L. Hassinger, W. Kelly, D. L. Housel, G. W. Reisinger, D. S. Reifsnyder, J. E. Gleim, and J. C. Peiper, J. fhys. Chem., 81, 2237 (1977). (3) J. N. Spencer, J. R. Sweigart, M. E. Brown, R. L. Bensing, T. L. Hassinger, W. Kelly, D. L. Housel, and G. W, Reisinger, J . fbys. Cbem., 80, 81 1 (1976). (4) J. N. Spencer, R. A. Hackman, R. S. Harner, S. L. Shoop, and K. S. Robertson, J. fbys. Chem., 77, 3103 (1973). (5) A. Allerhand and P. v. R. Schleyer, J. Am. Chem. Soc., 85, 371 (1963). (6) E. M. Arnett, E. J. Mitchell, and T. S. S. R. Murty, J. Am. Cbem. Soc., 98, 3875 (1974). (7) J. N. Spencer, J. E. Glelm, C. H. Blevins, R. C. Garrett, and F. J. Mayer, J. fbys. Cbem., submitted for publication. (8) B. S. Tovrog and R. S. Drago, J. Am. Cbem. Soc., 96, 2743 (1974). (9) W. R. Snyder, H. D. Schreiber, and J. N. Spencer, Spectrochim.Acta, Part A , 29, 1225 (1973). (10) M. S. Nozariand R. S. Drago, J. Am. Cbem. SOC.,92, 7086 (1970). (11) J. A. Puliin and R. L. Werner, Spectrocblm. Acta, 21, 1257 (1965). (12) A. D. Shew and K. F. Purcell, J. Am. Cbem. Soc., 94, 1853 (1972). (13) A. D. Sherry and K. F. Purcell, J. fbys. Chem., 74, 3535 (1970). (14) Y. Kyogoku, R. C. Lord, and A. Rich, J. Am. Cbem. Soc., 89, 496 (1967). (15) T. J. V. Flndlay, J. S. Keniry, A. D. KMman, and V. A. Pickles, Trans. Faraday Soc., 83, 846 (1967).
Determination of
Micelle Aggregation Numbers
The Journal of Pbysical Chemistty, Vol. 83, No. 20, 1979 2621
(16) 0.L. Bertrand and T. E. Burchfield, Anal. Calorlmetry, Proc. Symp., 3, 283 (1974). (17) L. G. Hepler and D. V. Fenby, J . Chem. Thermodyn.,5 , 471 (1973). (18) W. Guschlbaver, “Nucleic Acid Structure”, Springer-Verlag, New Yak, 1976. (19) C. R. Noller, “Chemistry of Organic Compounds”, 3rd ed., W. B. Saunders, Phlladelphia, Pa., 1965, p 658. (20) H. A. Sober, Ed., “Handbook of Blochemlstry”, 2nd ed., CRC Press, Cleveland, Ohio, 1970. (21) J. S. Binford and D. M. Holloway, J. Mol. Biol., 31, 91 (1968). (22) E. Osawa and 2 . Yoshida, Spectrochim. Acta, Part A , 23, 2029 (1967). (23) L. J. Bellamy, K. J. Morgan, and R. J. Pace, Specfrocblm. Acta, 22, 535 (1965). (24) G. C. Pimental and A. L. McClellan, “The Hydrogen Bond”, W. H.
Freeman and Co., San Francisco, Calif., 1960. (25) G. D. Frederick and C. D. Poulter, J. Am. Chem. SOC.,97, 1797 (1975). (26) C. D. PouRer and 0.D. Frederick, TetrahedronLeft.,26,2171 (1975). (27) D. D. Perrin, ”Dissociation Constants of Organic Bases In Aqueous Solution”, Butterworths, London, 1965. (28) H. Iwahashi and Y. Kyogoku, J. Am. Chem. Soc., 99, 7761 (1977). (29) G. R. Wiley and S. I. Miller, J. Am. Chem. SOC.,94, 3287 (1972). (30) A. DAlbis, M. P. Wickens, and W. B. Gratzer, SiOpo&mrs, 14, 1423 (1975). (31) J. Alvarez and R. Biltonen, Biopolymers, 12, 1815 (1973). (32) R. A. Newmark and C. R. Cantor, J . Am. Chem. Soc., 90, 5010 (1968). (33) H. A. Nash and D. F. Bradley, J. Chem. Phys., 45, 1380 (1966). (34) J. Donohue and K. Trueblood, J. Mol. Biol., 2, 363 (1960).
Isopiestic Compositions of Aqueous Ionic Surfactant Systems as a Measure of Preferential Interactions. Application to the Determination of Micelle Aggregation Numbers by Equilibrium Ultracentrifugation Daryl A. Doughty Department of Energy, Bartlesville Energy Technology Center, Sart/esvllle, Oklahoma 74003 (Received January 29, 1979) Publication costs assisted by the U.S. Department of Energy
The determination of reliable micelle aggregation numbers for ionid surfactants by means of equilibrium ultracentrifugation requires a correction for the preferential interactions which occur in multicomponent charged systems. Isopiestic distillation experiments on solutions containing the volatile solvent and the nonvolatile supporting electrolyte and varying amounts of the ionic surfactant lead directly to the desired correction for preferential interactions. Results for the surfactant, sodium dodecyl sulfate (SDDS), in various NaCl(aq) background solutions show that a substantial correction, increasing with increasing NaCl concentration,is required. Results on the surfactant, sodium decylsulfonate (SDS), show a similar correction above the critical micelle concentration (cmc). Data for SDS below the cmc show a much different trend from that above, a condition not evident with SDDS because of the much lower cmc. If the interaction is interpreted as arising strictly from micelle charge, values for the micelle charge can be obtained. Results for SDDS indicate a fractional charge ranging from 0.356 for SDDS in 0.1 m NaCl(aq) to 0.418 in 0.3 m NaCl(aq). Sedimentation equilibrium experiments on SDDS in NaCl(aq) solutions gave aggregation numbers in good agreement with published values though generally higher. These higher results are consistent with the more reliable corrections for charge effects or preferential interactions available from the isopiestic distillation experiments. The aggregation numbers for SDDS increased with increasing NaCl concentration, consistent with published results. Precision density measurements, using a magnetic float densimeter having a precision of 1ppm, were used to obtain accurate values for the partial specific volumes (L?) of SDDS in NaCl(aq) and water for use in the ultracentrifuge calculations. The results show a slight increase in 0 with NaCl concentration. Values for the cmc obtained from the density measurements are in excellent agreement with published values. A model relating the increasing fractional charge on the micelle to decreasing micelle surface area to micelle volume ratio as a function of NaCl concentration is presented.
Introduction Ultracentrifugation has been used to determine micellar weights and aggregation numbers of surfactant micelles and also can give information about micelle polydisper~ i t y . l - ~However, the ionic nature of the surfactants currently of interest to enhanced oil recovery research presents difficulties as far as the interpretation of results is concerned. These difficulties arise from the charged nature of the surfactant micelles and the resulting influence of this charge on the distribution of the various solution components in the sample during ~entrifugation.~ Correcting for these effects is a major obstacle to the routine application of ultracentrifugation to the study of ionic surfactant systems. My discussion will be restricted to sedimentation equilibrium ultracentrifugation because of the more rigorous theoretical foundation which exists This article not subject to
for the interpretation of sedimentation equilibrium data obtained from experiments on systems containing charged particles and/or several components. The system of interest will be an aqueous surfactant system containing an ionic surfactant together with a low molecular weight, uni-univalent electrolyte having an ion in common with the surfactant. In all discussions and equations which follow, I will use the standard convention of identifying solution components by subscripts with the subscript 2 referring to the ionic surfactant, the subscript 3 referring to the uni-univalent supporting electrolyte, and the subscript 1referring to the primary solvent, water in this case. Also, the equations as presented will apply strictly only under conditions of constant temperature and assumed incompressibility of the solution. The first condition may be controlled by the experimenter and the
U.S.Copyright. Published 1979 by the American Chemical Society