Volume 5 Number 11 November 1978 Nucleic Acids Research Zwitterionic character of nucleotides: possible significance in the evolution of nucleic acids M.Sundaralingam and P.Prusiner Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin- Madison, Madison, WI, 53706, USA Received 7 August 1978 ABSTRACT X-ray crystallography has shown that the free acids of adenoslne 5'- and 3'-monophosphates and of cytidine 5'- and 3'-monophosphates exist as zwitterions in the solid state with protonation of the adenine base at the N(l) site and of the cytosine base at the corresponding site N(3) and the phosphate group negatively charged. In this paper, evidence is presented for the zwitterionic character of the free acids of the monomeric nucleotides guanosine S'-monophosphate and inosine 5'-monophosphate with protonation of the base at the N(7) site of the imidazole moiety. INTRODUCTION X-ray crystal structures of the free acids of adenosine 5'- and 3'-monophosphates (AMP) (1-4) and cytidine 5'- and 3'-monophosphates (CMP) (5-7) have firmly established the zwitterionic character of these nucleotides. the protonation of the base is at N(l) and in CMP it is at the analogous pyrimidine ring site N(3). In AMP As a consequence of the zwitterionic character one of the phosphate hydrogen atoms is ionized, the phosphate group is negatively charged, and the base positively charged. It has been observed that the effect of protonation of the base and deprotonation of the phosphate results in significant distortions in their molecular dimensions (8,9). In particular, the bond angle and bond distances at the site of protonation of the base ring are noticeably enlarged compared to the unprotonated situation and significant differences are also seen around the pyrimidine ring. The influence of protonation goes even beyond the pyrimidine ring in that the exocyclic amino C-N linkage shows a tendency of being shortened in the protonated adenine and cytosine bases, thus indicating an enhancement in delocalization of the lone pair electrons of the amino nitrogen into the base ring. Similarly 1 ", the protonated P-O(H) distance (1.569 + 0.015 A is significantly T The values of the bond distances and bond angles given are averaged values of the six structures of 3'- and 5'-monophosphates of adenosine and cytidine mentioned at the beginning of the paper. Information Retrieval Limited 1 Falconberg Court London W1V5FG England 4375
longer than the unprotonated P-0~ distance and P=O distance (1.493 + 0.010 A). Also, the valence angles between the oxygens sharing the negative charge (~0-P=0 0=P-0~) is usually the largest one (117.7 + 1.5 ), the others being less than the tetrahedral angle (106.5 + 2.5 ) except one, the H(0)-P=O or (H)0-P-O~ angle, which is about 112.3 + 1.0". In this paper, we deduce the zwitterionic character of the free acids of guanosine and inosine 5'-monophosphate from the published X-ray structures where this important point was overlooked (see Fig. 1). We find that the evidence strongly points to the base protonation at N(7) of the imidazole moiety and consequent ionization of a phosphate proton. This conclusion is further corroborated by our determination of the crystal structure of the free 3',5'-cyclic inosine monophosphate monohydrate (cimp) in these laboratories (10,11), which exists in the zwitterionic form with N(7) protonation, and by the recent crystal structure determination of the free 3',5'-cyclic guanosine monophosphate monohydrate (12), which is also protonated at N(7) of the base. RESULTS AND DISCUSSION The mean bond distances and bond angles in protonated and nonprotonated hypoxanthine and guanine base rings of nucleosides and nucleotides are presented in the table. It can be seen that the difference in the geometries between the protonated and nonprotonated bases are large enough to establish the state of the base without the determination of the hydrogen atoms. The differences between the bond distances N(7)-C(8) and C(8)-N(9) as well as between the valency angles at C(5), N(7), and C(8) are much larger than the standard deviations from the mean. For instance, the C(5)-N(7)-C(8) angle is augmented by 4-5, while both the C(4)-C(5)-N(7) and N(7)-C(8)-N(9) angles are decreased by 4-5. Also, the N(7)-C(8) bond is about 0.02 A larger in protonated structures than in neutral structures. The crystal structures of the free acids of both guanosine 5'-monophosphate trihydrate (5'-GMP) (13) and inosine 5'-monophosphate monohydrate (5 1 - IMP) (14) have been published. But, in neither case the original authors suspected the zwitterionic character of the nucleotide, since the hydrogen atoms were not located in these studies. In Fig. 2, the bond distances and angles of cimp, where N(7) is protonated, are compared with those reported for 5'-IMP (14) and with the average values of five unprotonated ionic structures. The values reported for 5'-IMP clearly compare better with those of cimp than with those of the unprotonated structure. Therefore, we conclude that the free acid of 5'-IMP is protonated at N(7) in the solid state. 4376
r 0 HO r o CH, o I ^o v H* OH H(OH) I OH I HIOH) HO P O CH, 0 P 0 o, OH HIOH) ^..i OH HIOH) Fig. 1. Zwitterionic character of the three common nucleotides of RNA and DNA: (a) adenoslne 5'-monophosphate with base N(l) site protonated; (b) guanoslne 5'-monophosphate with base N(7) site protonated; (c) cytidlne 5'-monophosphate with base N(3) site protonated. The fourth common nucleotlde, containing uracil and thymine in RNA and DNA respectively, is unusual in that it appears to exist In the neutral form (d). Subsequent to our findings, a preliminary report on the crystal structure of the free acid of 3',5'-cyclic guanosine monophosphate (cgmp) has appeared (12), and, as expected, these authors also find the cyclic nucleotide to exist as a zwitterion with base protonation at N(7) and the phosphate in the ionized state with a resident negative charge. In Fig. 3, the geometry of ccmp, where the base is protonated at N(7), is compared with that reported for 5'-GMP and with the average geometry of two unprotonated structures. Again, the geometry of 5'-GMP compares better with that of cgmp than with that of the unprotonated structure. Thus, the free acid of 5'-GMP is protonated at N(7) as is the free acid of 5'-IMP. As is observed In the structures of cgmp and cimp, charge neutralization requires that the phosphate group be negatively charged. Indeed, the geometry of the 4377
TABLE I_ I + G + No. of Structures N(l)-C(2) C(2)-N(3) N(3)-C(4) C(4)-C(5) C(5)-C(6) C(6)-N(l) C(6)-0(6) C(5)-N(7) *N(7)-C(8) *C(8)-N(9) N(9)-C(4) N(9)-C(l") N(l)-C(2)-N(3) C(2)-N(3)-C(4) N(3)-C(4)-C(5) C(4)-C(5)-C(6) C(5)-C(6)-N(l) C(6)-N(l)-C(2) C(5)-C(6)-0(6) K(l)-C(6)-0(6) *C(6)-C(5)-N(7) *C(4)-C(5)-N(7) *C(5)-N(7)-C(8) *N(7)-C(8)-N(9) *C(8)-N(9)-C(4) C(8)-N(9)-C(l') N(3)-C(4)-N(9) C(5)-C(4)-N(9) C(4)-N(9)-C(l') 5 1.352(7) 1.302(6) 1.359(8) 1.375(7) 1.430(3) 1.407(14) 1.218(11) 1.379(7) 1.304(3) 1.369(7) 1.375(4) 1.463(6) 125.6(8) 111.7(6) 127.6(6) 119.8(9) 109.6(9) 125.1(5) 129.0(7) 121.0(4) 129.1(8) 110.9(5) 104.1(4) 113.5(2) 105.8(2) 125.9(16) 126.7(1) 105.5(2) 127.6(16) 2 1.338(12) 1.321(9) 1.351(2) 1.376(17) 1.436(4) 1.420(11) 1.215(5) 1.372(2) 1.322(2) 1.344(14) 1.380(1) 1.494(15) 126.0(20) 111.3(16) 128.5(5) 119.9(9) 109.1(9) 125.5(5) 128.6(6) 122.3(3) 132.0(1) 108.1(8) 107.9(9) 109.7(3) 108.1(1) 127.1(18) 125.8(2) 106.2(2) 124.7(17) 2 1.367(3) 1.327(4) 1.358(4) 1.377(3) 1.419(1) 1.391(1) 1.236(2) 1.387(2) 1.306(6) 1.374(9) 1.375(4) 1.453(1) 124.3(1) 111.8(2) 128.1(3) 119.3(1) 111.5(2) 125.1(1) 127.9(2) 120.7(3) 129.9(1) 110.8(1) 104.3(2) 113.3(1) 106.1(5) 127.2(10) 126.3(5) 105.6(2) 126.6(4) 3 1.37(1) 1.35(2) 1.34(1) 1.38(1) 1.41(1) 1.40(1) 1.23(1) 1.39(2) 1.32(1) 1.35(1) 1.37(2) 1.49(1) 124(0.3) 112(0.3) 128(0) 120(1) 111(0.3) 125(0.3) 129(1) 120(1) 133(1) 107(1) 108(0.3) 109(0) 108(1) 126(1) 125(1) 107(1) 125(0.3) Comparison of the mean bond distances (A) and bond angles (degrees) In the base rings of protonated (+) and unprotonated lnoslne (1) and guanosine (G) nucleosldes/tides. The structures used for averaging are the following. For I, Inosine 5'-monophosphate monohydrate (24); inoslne (25); lnoslne dihydrate (26). For I, Inoslne 3',5'-cyclic monophosphate (27) and inosine 5'-monophosphate (14). For G, guanosine dihydrate (26). And for G, guanosine hydrobromide hemihydrate (28) and guanosine 5'-monophosphate (13). The geometrical parameters which are most markedly affected by N(7) protonation are noted by an asterisk (*). Average deviations from the mean In parentheses refer to the least significant digit. phosphate group in 5'-GMP is typical of a negatively charged phosphate. While the latter feature was noticed by Murayama and coworkers (13), they did not suspect protonation of the base to balance the charge. Thus, X-ray investigations have shown that three of the four comnon nudeotide building blocks 4378
Fig. 2. Bond distances and bond angles in the imidazole ring of (a) cimp (10), 0>) 5'-IMP (14), and (c) neutral hypoxanthine base (mean values of five structures, see Table). The similarity of the geometrical parameters of (b) to (a) indicates that the base of 5'-IMP is also protonated at N(7). In (a), the strong hydrogen bond between the anionic phosphate oxygen and the protonated N(7) nitrogen is shown. In (b) again, the N(7) participates in a donor hydrogen bond to a water oxygen atom. In the original paper (14), N(7) was acting as an acceptor instead, since N(7) protonation was not suspected. (AMP, CMP, GMP) of the nucleic acids (DNA and ENA) in the free acid form exist as zwitterions in the solid state. To date there is no data available on the crystal structures for either the free acid of thymldine 5'-monophosphate (TMP) or the related uridine 5'- monophosphate (UMP). TMP and UMP differ from the other common nucleotides in that they do not have sites available for protonation on the base ring. However, it may be yet possible for these bases to be protonated at one of the 4379
Bond distances and bond angles in the imidazole ring of (a) cgmf (12), (b) 5'-GMP (13), and (c) neutral guanine base (mean values of the two structures, see table). The similarity of the geometrical parameters of (b) to (a) indicates that the base of 5'-IMF is also protonated at N(7). Again in (b), a possible hydrogen bond between the protonated N(7) nitrogen and a water oxygen atom is indicated. This hydrogen bond was not discussed by the original authors (13). extracyclic carbonyl oxygen atoms. A precedent for this is illustrated by the crystal structure of the base derivative 1-methyluracil hydrobromide (15), where the uracil base 0(4) atom is protonated. The protonation here may be a result of the stronger acid HBr, nevertheless, it is of interest to investigate the free acids of these nucleotides. It may be that these two nucleotides show lesser or no tendency for zwitterion formation (since the oxygen is less basic than the ring nitrogen), but careful x-ray crystal structure analysis of these should provide an answer. Recently, the first crystal 4380
structure of an uncharged nucleotide has been reported, viz., that of 3'-UMP (16). In this structure, the phosphate group carries two protons and the uracil base is neutral. Therefore, it will be expected that the corresponding 5'-nucleotide will also be uncharged. Similarly, thymidine 5'-monophosphate will be expected to be uncharged. On the basis of nitrogen-15 NMR, it was suggested that N(l) of AMP, N(7) of GMP, and N(3) of CMP were strongly involved in hydrogen bonding to water (17). The effects seen on the chemical shifts can be due to the protonation of the respective nitrogen atoms in these nucleotides in solution as well. Similar effects were not observed with UMP and IMP, since there is no available ring nitrogen for protonation. This is consistent with the X-ray results on 3'-UMP. The zwitterionic character is not restricted to the mononucleotides. It would appear that the adenine cytosine, and guanine containing oligo- and polynucleotides would also exhibit zwitterionic characteristics in the free acid form. Indeed, this is observed in the structure of the dinucleoside monophosphate UpA (18,19) and of the trinucleoside diphosphate ApApA (20). However, under physiological conditions (neutral ph's and presence of metal ions), the nucleotide bases are not protonated and the phosphates are ionized with two negative charges. The resistance to protonation of uridine and thymidine 5'-monophosphates is paralleled by their poor complexing ability with metal ions. On the other hand, the purine 5'-nucleotides and cytidine 5'-monophosphate, which are readily protonated, are good complexing agents with metal ions (21). One may add that the favored protonation sites are the same as those where metal binding occurs for the guanine and cytosine bases; while, in the case of adenine, metal binding occurs preferentially at N(7) instead of N(l). Metal binding to mononucleotides usually involve both the base and the phosphate. In some purine nucleotide-metal complexes, N(7) and the phosphate oxygens (of neighboring molecules) enter the coordination sphere; while in all cytidine 5'- monophosphate-metal complexes, N(3) and the phosphate oxygens (of neighboring molecules) bind to the metal. The purine-nucleotide-metal complexes present another pattern of coordination, where the metal binds only to N(7) with water molecules completing the coordination sphere. However, in the two uridine 5'- monophosphate-metal complexes reported to date, the transition metal binds exclusively to the phosphate (22,23). CONCLUSIONS It is interesting that three of the four common nucleotides of RNA and 4381
DNA (AMP, GMP, and CMP) exist a6 zwltterions, while the fourth one (IMP or THP) exists as an uncharged species. The property of the majority of the nucleotide building blocks to display zwitterionic character parallels the situation featured by the amlno acid building blocks of proteins, which without exception are protonated on the a-amino group with the ionization of the carboxyl group. The zwitterionic character of the monomeric building blocks of nucleic acids and proteins may have had an important bearing in the evolution and assembly of these two major classes of biological macromolecules under primordial conditions. The zwitterionic character of nucleotides is lost In the presence of metal ions which play the role of counterions Instead of the protonated base. Under this situation, the bases are free to participate In their genetic role (transcription, translation, and replication) of. complementary base pairing. This may be an explanation of the ubiquitous function of metal ions in nucleic acid processes. ACKNOWLEDGEMENTS We gratefully thank Dr. Eric Westhof for his invaluable assistance in the preparation of this manuscript. This work was supported by the National Institutes of Health (GM17398). REFERENCES 1. Kraut, J. and Jensen, L. H. (1963) Acta Cryst. 16_, 79-88. 2. Lin, H-Y. and Sundaralingam, M. (1971) unpublished results. 3. Sundaralingam, M. (1966) Acta Cryst..21, 495-506. 4. Neidle, S., KUhlbrandt, W., and Achari, A. (1976) Acta Cryst. B32., 1850-1855. 5. Sundaralingam, M. and Jensen, L. H. (1965a) J. Mol. Biol. 13_, 914-929. 6. Bugg, C. E. and Marsh, R. E. (1967) J. Mol. Biol. ^5_, 67-82. 7. Viswamltra, M. A., Reddy, B. S., Lin, G. H.-U., and Sundaralingam, M. (1971) J. Amer. Chem. Soc. 93_, 4565-4573. 8. Sundaralingam, M. and Jensen, L. H. (1965b) Acta Cryst. 13, 930-943. 9. Singh, C. (1965) Acta Cryst. 19_, 861-862. 10. Prusiner, P. (1974) Ph.D. Thesis, University of Wisconsin-Madison. 11. Sundaralingam, M. (1975) Annals N.Y. Acad. Sci. U.S.A. 2!25_, 3-42. 12. Druyan, M. E., Sparagana, M., and Peterson, S. W. (1976) J. Cyclic Nucleotide Res. 2_, 373-377. 13. Murayama, W., Nagashima, N., and Shimizu, Y. (1969) Acta Cryst. B25, 2236-2245. 14. Nagashima, N., Wakabayashi, K., Matzuzaki, T., and Iitaka, Y. (1974) Acta Cryst. B30, 320-326. 15. Sobell, N. M. and Tomita, K. (1964) Acta Cryst. 17., 122-125. 16. Srikrlshnan, T., Andrusz, S. M., and Parthasarathy, R. (1978) ACA Meeting Abst. PB8, Norman, Oklahoma. 17. Markowski, V., Sullivan, G. R., and Roberts, J. D. (1977) J. Amer. Chem. Soc. 99, 714-718. 4382
18. Rubin, J., Brennan, T., and Sundarallngam, M. (1972) Biochemistry 11, 3112-3129. 19. Sussman, J. L., Seeman, N. C, Kim, S. H., and Herman, H. M. (1972) J. Mol. Biol. 66, 403-421. 20. Suck, D., Manor, P. C, and Saenger, W. (1976) Acta Cryst. B32, 1727-1737. 21. Swaminathan, V. and Sundaralingam, H. Critical Review in Biochemistry, in press. 22. Fisher, B. E. and Bau, R. (1978) Inorg. Chem. 17_> 27-30. 23. Cartwright, B. A., Goodgame, D. M. C, Jeeves, 1., and Sfcapski, A. C. (1977) Biochim. Biophys. Acta 477, 195-198. 24. Rao, S. T. and Sundaralingam, M. (1969) J. Amer. Chem. Soc. 91, 1210-1217. 25. Munns, A. R. I., Tollin, P., Wilson, W. R., and Young, D. W. (1970) Acta Cryst. B26, 1114-1117. 26. Thewalt, U., Bugg, C. E., and Marsh, R. E. (1970) Acta Cryst. B26_, 1089-1101. 27. Prusiner, P., McAlister, J., Gross, S., and Sundaralingam, M. (1978), in press. 28. Tougard, P. and Chautot, J.-F. (1974) Acta Cryst. B30_, 214-220. 4383
43B4