64
R. R. SHOCP,H. T. MILES,AKD E. D. BECKER
crgy orders reported. This means that the cncrgy gained by the molecule in the formation of new bonds is more influential in determining the low-voltage spectra of these bisdifluoraminoalltancs than tho bond dissociation ordcrs in the original molecules.
Acknowledgments. The authors acknowledge the assistance of R. L. Hanson, the ONR Project llonitor,
who was able to provide the disubstituted difluornniino compounds, and to thank Ilk. Richard Lec for his nssistance in the construction and operation of the experimental equipment. Finally, the authors gratefully nclcnowledgc Drs. H. 34. Fales and G. W. A. llljlne of the National Institutes of Health for thier cooperation in obtaining the high-resolution mass measurements used inthisstudy.
Restricted Rotation about the Exocyclic Carbon-Nitrogen Bond
in Cytosine Derivatives by Regitze R. Shoup, H. Todd Miles, and Edwin D. Becker* Sational Institute of Arthritis and Metabolic Diseases, National Institutes of Health, Bethesda, Maryland (Receized August 0, 1071)
20014
Publication costs assisted by the National InstiLutes of Health
Rotational barriers of methyl-substituted amino groups in cytosine derivatives have been examined by nmr spectroscopy. Total line shape analysis of the spectra of dimethylamino derivatives gave activation energies in the range 15-18 kcal/mol. Qualitative measurements showed that the rotation in monomethylamino derivatives is restricted and that the predominant conformer (-95% at - 19’) has the Ni-methyl group syn to NB.
Introduclion It is generally assumed that the amino group hydrogens of the nucleic acid bases are essentially coplanar with the heterocyclic rings, both in hydrogenbonded complexes and in the unassociated molecules. The amino group is considered to have significant amide character as a result of contributions of such forms as a-c t o the resonance hybrid.
a
b
C
The question of rotational barriers for the amino groups of the bases is important to our understanding of the configurations and interactions of the polynucleotides. When the amino groups are unsubstituted (ie., R7 = R7’ = H) either coplanar conformation is capable of base pairing, and when R7 = R?’ = CH3 neither conformation is capable of pairing. A special interest, however, arises in the case of monoThe Journal
of
Physical Chemistry, Vol. 76, Yo. 1, 1972
methyl substitution (R,= H, R7’ = CH3) since in this case one rotamer is capable of pairing and one is not. Rotational restrictions which exist in the monomeric units would also exist in the polymers, but the equilibrium between the two rotamers may be changed by helix formation. It is therefore important to establish the equilibrium conformations of monomethylamino derivatives of the bases and obtain information on the rates of rotation. I n view of the amide character of the amino group in cytosine derivatives, the energy barrier for the rotation about the C-N bond should fall in the range commonly observed for amides ( i e . ! 15-25 kcal/mol) and should therefore be subject to study by nuclear magnetic resonance. We recently reported1 the observation of restricted rotation of the amino group in 1-methylcytosine in dimethylformamide solution. For this molecule, however, it was not feasible to obtain an accurate measure of the energy barrier by the use of total line shape (TLS) analysis because of the low signal/noise ratio and competitive chemical exchange processes. We (1) R. R. Shoup, E. D. Becker, and H. T.Miles, Biochem. Biophys. Res. Commun., 43, 1350 (1971).
65
RESTRICTED ROTATION ABOUT THE EXOCYCLIC C - S BOND present here the result of a measurement of the rotational barrier in the related dimethylamino cytosine derivatives I and 11, which are experimentally more tractable.
R
I n addition we demonstrate that the rotation in the mono-N-methyl derivatives 111, IV, and V is slow on the nmr time scale, and we present semiquantitative data on the equilibrium populations of the conformers a and b.
Experimental Section 1,7,7-Trimethylcytosine (I) was prepared from 1methyl-4-methoxypyrimidone-2 and aqueous dimethylamine.2 It was recrystallized from ethyl acetate with a small amount of methanol, yielding white prisms, mp 181-182". 7,7-Dimethyl-1-(2',3',4',6'tetraacetyl-wglucopyranosy1)cytosine (11) was prepared according to I I i l e ~ ,and ~ recrystallized from 95% ethanol, yielding white prisms, mp 283-284". Both compounds were dried in vacuo at 80" over PzOF, before use; 0.4 M solutions were prepared in the nmr tube by weighing 0.2 mmol of material and adding solvent to a predetermined height equivalent to 0.5 ml of solution. A small amount of tetramethylsilane (TMS) was added and the samples were thoroughly degassed by the usual freeze-t'haw technique and sealed under vacuum. 1,7-Dimethylcytosine (111) and 1,5,7-trimethylcytosine4 (IV) were prepared as described previously. 1,6,7-Trimethylcytosine was prepared from 4-methoxy1,6-dimethyl-1,2-dihydro-2-oxo-pyrimidine5 and aqueous methylamine, yielding white prisms, mp 248.5'. Anal. Calcd for C7HI1N30: C, 54.89; €1, 7.24; N, 27.43. Found: C, 55.42; H, 7.05; S,26.45. Nmr spectra of these compounds in dimethylformamide-& solution were recorded on a Varian HR-220 spectrometer. The nmr spectra of I and I1 were obtained as a function of temperature on a Varian A-60spectrometer. The spectra were recorded a t a sweep rate of 0.1 Hzlsec using a filter bandwidth of 0.4 He, conditions which
were demonstrated riot to causc distortion of the line shape.6 The rf power level was minimized to avoid saturation, and particular care was uscd in the adjustment of the phase at cach temperature. Several scans were recorded at each temperaturc to assure that temperature equilibrium had been reached. The temperature was measured before and after each set of spectra had been obtained using the standard methanol and ethylene glycol samples. New calibration curves7 were used for the relation between temperature and chemical shift of these samples. A more detailed discussion of the calibration curves is included in the following paper.* The rotational rate constants were obtained from the digitized nmr spectra by total line shape (TLS) analysis using the equation of Gutowsky and Holmg for an equally populated, uncoupled two-site system. Calculations of the rate constants were performed using a modified version of a least-squares fitting program by Jonas, et al.1° The necessary input parameters-chemical shift, Av, and line width, W , both in the absence of exchange-were observed to be temperature dependent. Av was obtained by extrapolation from values obtained a t lower temperatures, and W was obtained by interpolation between values observed above and below the region of exchange. Activation energies were obtained from a least-squares fit to the Arrhenius equation. A detailed discussion of the procedure and its attendant errors is reported separately.*
Rates of Rotation and the Rotational Barrier in Dimethylamino Cytosine Derivatives The activation parameters obtained for I and I1 by total line shape analysis are listed in Tablc I. Figure 1 shows the Arrlienius plots of the rate constants vs. inverse absolute temperature. The activation energies in Table I are listed without the customary statistical error limits, since we feel that systematic errors have not been eliminated to the extent that random errors dominate. Based on the analysis of the procedure presented in the following paper,* we apply confidence limits of about 1.5-2.0 kcaljmol to (2) G. W. Kenner, C. B. Reese, and A . R. Todd, J . Chem. Soc., 855 (1955). (3) H . T. Miles, J . Amer. Chem. Soc., 79, 256 (1957). (4) R . R . Shoup, H. T . Miles, and E. D. Becker, ibid., 89, 6200 (1967). (5) L. J. Rabinowitz and S. Gurin, ihid., 75, 5758 (1953). (6) A . Allerhand, H. S. Gutowsky, J. Jonas, and R. A . Meinser, ihid., 88, 3185 (1966). (7) A . L. Van Geet, Abstracts of the 10th Experimental NMR Conference, Pittsburgh, Pa., 1969; A . L. Van Geet, Anal. Chem., 42, 679 (1970). (8) R. R. Shoup, E. D. Becker, and M. McNeel, J . P h y s . Chem., 76, 71 (1972). (9) H. S. Gutowsky and C. H . Holm, J . Chem. Phys., 25, 1228 (1956). (10) J. Jonas, A . Allerhand, and H. S. Gutowsky, ibid., 42, 3396 (1965). The Journal of Phusical Chemistrg, Vol. 76, N o . 1, 1072
R. R . SHOUP,H. T. MILES,AKD E. D. BECKER
66
Table I: Activation Parameters for the Rotation of the Ilimethylamino Group in 1,7,7-Trimethylcytosine (I) and ~,~-l~~methy~-l-(2',3',4',6'-tetraacety~-~-~~ucopyranosy~)-cytos~ne (11) Compound
Bmb kcal/mol
Iczus, tec-1
123 48 1050 4.5
- 39
CDCls
17.6 15.1 15.7 11.5 17.3
CDaCN
18.1
8.8
22
Solventa
I I I I I1 I1
CDCla CDiCN
CDaOD
so*
90
(AV)O,c
IIZ
10 -3
e5
*
d
to,
OC
37
8.6 4.0 3.4 2.8 7.2 3.0
a All solutions are 0.4 M . Estimated error in E , is 1.6-2.0 kcal/mol (see text). Free energy of activation a t coalescence, accurate to 0.1 kcal/mol.
A S *, eu
AF * m , kcal/mol
7 .5
14.8 14.6 15.2 13.3 16.6 16.2
-0.5 0.2 -8.1
0.5 4.4 c
A F *~, d kcal/mol
14.9 14.6 15.1 12.7 16.5 16.2
Extrapolated from lower temperature values.
line shape studies. In such cases electron delocalization has been postulatedl7 to reduce the double bond character of the C-N bond in comparison with the simple amides. The activation energies found for I and I1 in this study are twice that (8.6 kcal/mol) reported by Martin and ReeseIs for 2',3'-O-isopropylidene-7,7-dimethylcytidine (VIa). Considering that the difference in
6-
5432I-
H a R=, R;= CH,
VIb R,= H; R:=CH,
0-I
-
- 2 ~ " " " " " " 1 " " " " '
3.0
3.5
4.0
4.5 IOYT
Figure 1. Arrhenius plot for the hindered rotation of the dimethylamino group in 0.4 M solutions of cytosine derivatives I and 11.
the activation energies. This uncertainty is considerably larger than the ones quoted in recent TLS studiesl1-l4 of restricted rotation in amides and results from the combination of a less favorable signal/noise ratio in the spectra of 0.4 M cytosine solutions, low coalescence temperatures, and relatively small chemical shifts between the dimethylamino protons. With one exception, the SO2 solution of I, which is discussed later. the range of activation energies is 15-18 kcal/mol. These values are somewhat lower than the ones recently obtained from total line shape studies of the restricted rotation in dimethylformamide14 (20.5 kcallmol) and dimethyla~etamidel~ (19.6 kcal/mol). The difference seems reasonable if one considers the heteroaromatic amines as an intermediate case between true amides and anilines. Only in special cases, (e.g., through substitution with several nitro groups) do the latter molecules show restricted r0tati0n.l~ The values observed for the cytosine derivatives are similar to those obtained for dimethylcarbamoyl chloride12 (16.9 lical/mol) and dimethyltrichloroacetamide16 (16.2 kcal/mol) in recent total T h e Journal of Physical Chemistry, Vol. 76, KO.1, 1972
0 0 H,C~CH,
activation energy between I and I1 is only of the order of 2-3 kcal/mol, it appears unlikely that a change in the sugar moiety can account for the difference between their results and ours. Since IIartic and Reese18 did not describe the method by which this number was obtained, however, and since it has been demon&rated6that the commonly used approximat)e methods for obtaining rate constants from nmr spectra usually give much too low activation energies, we feel the values obtained in the present study are more reliable. For further consideration of the results in Table I, we can focus on several properties: the coalescence temperature, t,, and the associated rate constant, IC,; the activation energy, Ea; and the entropy of (11) H. S. Gutowsky, J. Jonas, and T. H. Siddall, 111, J . Amer. Chem. Soc., 89, 4300 (1967). (12) R. C. Xewmann, Jr., D. Tu'. Roark, and J. Jonas, ibid.,89, 3412 (1967). (13) R. C. Neumann, Jr., and J. Jonas, ibid., 90, 1970 (1968). (14) M. Rabinowitz and A. Pines, ibid., 91, 1585 (1969). (15) J. Heidberg, J. A. Weil, G. A . Janusonis, and J. K. Anderson, J . Chem. P h y s . , 41, 1033 (1964). (16) R. R. Shoup, unpublished results. (17) M. T. Rogers and J. C. Woodbrey, J . P h y s . Chem., 66, 540 (1962). (18) D. M. G. Martin and C. B. Reese, Chem. Commun., 1275 (1967).
RESTRICTED ROTATION ABOUT
THE
EXOCYCLIC C-N BOND
activation, AS*. Of thcsr quantities t, is in general quit(. -~\;cll determined, and the ratc, IC,, a t coalescence can be dctcrmined with high accuracy. However, rate constants should be compared only at the same tcmpcrature, say 25" (298°K). The accuracy of lc298 then depends on the accuracy of E , and the magnitude of the difference between 25" and t,. In the present study the error in E , can be as large as 1.5-2.0 k ~ a l / m o l ,and ~ therefore the larger the extrapolation from the coalescence point the greater the uncertainty in lc2gs. The value of AS* is subject to the greatest uncertainty, since several systematic errors may be includcd in this quantity. Values of A S * very different from zero for systenis of the sort studied here are usually suspect and suggest the prrsence of systematic errors in the measurements. It is apparent that I in SOz gives values for all of these parameters that are quite different from those for the other systems. We defer further discussion of I in SO, to the section on solvent effects. The values of dS8and t, for I do not vary significantly among the other three solvents but are quite different from the values for 11; in fact, the values of k2g8 for I1 are more than an order of magnitude smaller than the values for I, and I1 has a correspondingly higher coalescence temperature. The origin of this pronounced effect of the substituent at the 1-position is not entirely clear. It seems most likely that it is attributable to the increase in double bond character of the G-NT bond as electrons are withdrawn from the cytosine ring by the more strongly electronegative sugar substituent and nonbonding electrons drawn from N7 into the CPN7 bond. The change in E , itself that would be expected to accompany this process is less easy to discern with the limited accuracy (1.5-2 kcal/mol) of E,. Alternatively, the difference in rates between I and I1 might be related to intermolecular solute-solute interactions, which are known to increase rotational barriers in some dimethylamides. However, we have no evidence of such specific interactions. (I1 does form a gel in CD3CN below NO", but the solution of I1 in CDCl3 remained nonviscous over the whole temperature range.)
Solvent Effects Table I and Figure 1 show that the rotational rate of I varies with solvent, increasing in the order CD30D < CDC& < CD3CN 99vo) in one conformation, almost certainly the one with the 7-methyl syn to N3. Steric considerations suggest that the more abundant conformers of 111 and V are also the ones with the N-methyl group syn to N3 ( i e . , IIIa and Va), in accord with studies of cis-trans isomerism in simple N-alkylamides. Mizushima21 showed, using (20) The width of the N H peak prevents detection at low concentration, and for the 7-methyl signal the spectral region is obscured by solvent signals and spinning side bands from the 1- and &methyl groups. A careful check of the frequencies of all small peaks was made t o identify spinning side bands (at w r t 9 6 Hz). (21) 5. Mizushima, "Structure of Molecules and Internal Rotation," Academic Press, New York, N. Y., 1954, p 18.
RESTRICTED ROTATION ABOUT
THE
69
EXOCYCEIC C-N BOND
at Figure 3. The 220-MHz proton nmr spectra of cytosine derivatives 111, IV (0.2 M ) , and V (0.1 171) in dimethylfor~namide-d7 21". The abscissa (6) is given in ppm relative to internal tetramethylsilane. Spinning side-bands ( a t -A95 Hz) are labeled P. Gain settings differed for different spectral regions.
5-H
I-cy
7-cy
Figure 4. The 220-MHz proton nmr spectra of 111, IV, and V in dimethylformamide-d7 at -19'. Spectral parameters are the same as in Figure 3. Note in the spectra of I11 and V the small peaks to the low-field side (by 0.03-0.16 ppm) of the major peaks labeled 5-H, 1-CH3, and 6-CH3. T h e Journal of Physical Chemistry, Vol. 76, A'o. I , 197%
R. R. SHOUP, H. T . MILES,AKD E. D. BECKER
70 Raman and infrared spectra, that N-methylacetamide (VIII, R = R' = CH3) exists mainly as the trans isomer. R, dC-N,H ,R'
R, C-N:
0"
R'
=
trans CIS LaPlanche and Rogers22 found from nmr spectra that the rotation in simple N-alkylamides is strongly restricted and that only in N-alkyl formamides could some (-20%) cis isomer be detected. Their spectra of higher N-alkylamides (VIII: R = CH,, C2H,, etc.) showed only one N-alkyl signal, and they concluded from solvent shifts and coupling constants that the conformation was 100% trans. Nore recently, Barker and B o u d r e a u ~ , ~and ~ Sandstrom and UppstromZ4 observed that an aqueous solution of N-methylacetamide contains 3% of the cis isomer. Inasmuch as a proton in the 5-position in the cytosine derivatives I11 and V can be expected to contribute an amount of steric interference similar to that of the acetyl methyl group in N-methylacetamide, the observed 4% and 5% abundance of IIIb and Vb are in qualitative agreement with the data reported for N-methylamides. We recently observed that the rotation of the unsubstituted amino group in 1-methylcytosine is also restricted, and a rate constant k 11 138 sec-' was obtained at room temperature for a 0.02 M solution in dimethylformamide-d7. Thus there is qualitative agreement between this rate and the rates obtained for the dimethylamino compound I at 25" in chloroform ( k = 95 sec-l), methanol (IC = 50 sec-l) and acetonitrile ( k = 122 sec-I). No rotational rates have been obtained for the monomethylamino derivatives I11 and V, since the low abundance of one isomer makes total line shape analysis quite inaccurate. However, closer inspection of the room temperature spectra of I11 and V reveals that low-intensity broad bands are present in the regions expected for observation of the less abundant isomers (see Figure 3). The rotation of the methylamino group is consequently still slow on the nmr time scale at this temperature.
T h e Journal of Physical Chemistry, Vol. 76, No. 1, 1078
This observation is in obvious contrast to the conclusions drawn by Martin and R ~ c s c , 'who ~ stated without presenting data that the rotation of the mcthylamino group of VIb is fast a t -60" and above. It seems lilcely that thcy actually obscrved the predominant conformer and failed to detect the less abundant conformer.
Conclusion The present study has shown that the rotation about the exocyclic C-N bound in cytosine derivatives is restricted by an energy barrier E , rv 15 ltcal/moi. It was further concluded that the 7-N-monomcthylsubstituted derivatives exist predominantly in the configuration with the 7-N-methyl group syn to N3. These results may help to explain the interaction behavior observed for 7-N-methylated polycytidylic acids. Brimacornbe and ReeseZ5 found that polgcytidylic acid completely substituted with 7-N-methyl groups (poly 7N-lleC) showed no detectable interaction with polyinosinic acid (poly I), whereas the copolymer (60:40) of C and 7N-RIeC did interact with poly I. The stoichiometry was suggestive of pairing to 7N-MeC as well as to C residues. From the present data on cytosine monomers one can estimate a lifetime of the order of low3see for the conformation with the amino proton in the position capable of base pairing. This lifetime may be too short to allow interaction of fully N-methylated poly C, but long enough, once helix formation has been initiated by unsubstituted C residues, to allow pairing also of the N-methylated C residues. The energetic advantage of helix formation may then be enough to overcome the steric interference between the N-methyl group and the 5 proton. (22) L. A. LaPlanche and M. T. Rogers, J . A m e r . Chem. SOC., 86, 337 (1964). (23) R. H. Barker and C. J. Boudreaux, Speetrochim. Acta, A23, 727 (1967). (24) J. SandstrGm and B. Uppstrom, Acta Chem. Scand., 21, 2254 (1967). (25) 11. L. C. Brimacornbe and C. E. Reese, J . Mol. Diol., 18, 529 (1966).