The Intramolecular Conformation of Adenosine S-Monophosphate in Aqueous Solution as Studied by Fast Fourier Transform ‘H and 1H-{31P[ Nuclear Magnetic Resonance Spectroscopy*

SUMMARY Nuclear magnetic resonance spectra of 5’-AMP were examined over a concentration range of 0.001 to 2.2 M in the Fourier mode in the lH and lH-( a1P) configuration and were analyzed by computer simulation. The principal conclusions with respect to the solution conformation of 5’-AMP are as follows. (a) The e’xocyclic C(S)-O(5’) C(4’)-C(S’) bonds are flexible, and preferentially exist in the gauche-gauche conformation with a “W” relationship across H(4’)-C(4’)-C(5’)-O(S’)-P(5’) in which the above atoms lie in one plane as in its crystal structure. (b) The ribose moiety exists as a flexible ring system undergoing interconversion between the *E and 2E conformations. The sugar ring conformation is strongly concentration dependent, being 63 f 10% *E and 37 f 10 % *E at concentrations below 0.01 M. At the population of the aE form at the of the 2E form, and at 2.0 M, aE is the preferred conformation. In the solid state the conformation of the sugar ring is aE for 5’-AMP.

the gauche-gauche conformation with a "W" relationship across H(4')-C(4')-C(5')-O(S')-P(5') in which the above atoms lie in one plane as in its crystal structure. (b) The ribose moiety exists as a flexible ring system undergoing interconversion between the *E and 2E conformations. The sugar ring conformation is strongly concentration dependent, being 63 f 10% *E and 37 f 10 % *E at concentrations below 0.01 M. At higher concentrations, the population of the aE form increases at the expense of the 2E form, and at 2.0 M, aE is the preferred conformation. In the solid state the conformation of the sugar ring is aE for 5'-AMP.
A detailed study on 5'-AMP was undertaken because of its ubiquitous nature in biochemistry. Not only is it one of the important building blocks of polynucleotides, it is also part of high energy compounds such as ATP and oxidation-reduction coenzymes such as NAD, NADH, FAD, and FADH; it even occurs as part of cofactors involved in glycogen synthesis such as adenosine diphosphoglucose. It is our belief that a thorough understanding of the dynamic three-dimensional solution geometry of 5'-AMP will enable one to understand the conformation of similar 5'-nucleotides. This information, in conjunction with the "concept of conformational rigidity" advocated by Sundaralingam should be of considerable help in unraveling the stereochemical basis of conformational biology.
*This work was supported by National Science Foundation grants GB28015 and B028015-091 and National Cancer Institute of the National Institutes of Health Grant CA12462-03.
Nuclear magnetic resonance spectroscopy has been used to study the conformation of 5'-AMP in solution (1-9). These studies have been conducted using continuous wave iH NMR spectroscopy and the conclusions drawn were based on concentration dependence of a few nonexchangeable protons in a narrow concentration range. Further, the complete intramolecular solution conformation of 5'-AMP was not known.
The availability of a large Fourier transform system in our laboratory enabled us to obtain both 'H and iH-(81P) magnetic resonance spectra of 5'-AMP, superior to any hitherto reported in the concentration range of 0.001 to 1.0 M. Computer line shape simulations enabled complete analysis of the spectra giving the most precise concentration dependence of the chemical shifts and coupling constants of all the nonexchangeable protons. These data are used to propose the intramolecular conformations of 5'-AMP in aqueous solution.

EXPERIMENTAL PROCEDURE
The *H and 1H-(31P)r NMR spectra were recorded at 100 MHz on a Varian HAlOOD snectrometer interfaced to a Dieilab FTS-3 Fourier transform data system. This svstem has a total memory of 128K and is capable of performing a maximum of 64K single me&ion (16 bits ner word leneth) or a 32K double m-ecision (32 bits per word length) transform and possesses an adequate dynamic range.
The frequencies for the proton (observing) channel and fluorine (lock) channel were derived from a Digilab lo-94 frequency synthesizer and a Digilab 409-2 pulser. Hexafluorobenzene in a l-mm capillary served as an external reference as well as providing the signal for the rgF lock. The internal reference was tetramethyiammonium chloride.
The sample temperature was 30.5". Those spectra where the phosphorus nuclei were decoupled were recorded with irradiation derived from a Digilab 50-30 PD plug-in amplifier.
Spectra were obtained in the concentration range of 0.991 to gt tg m P IZL m model 320 pH meter; pD = pH meter reading + 0.4. The assignment of the various protons of the ribose and the exocyclic CH20POaa-group were made from coupling pattern, SIP decoupling, and computer line shape simulation.

AND DISCUSSION
Torsional Diastereomer.9 Constrained to C(6')-O(6') Bond of 6'-AMP-The minimum energy torsional isomers constrained to the C(6')-0(5') bond of 5'-AMP are shown by Newman projections in Structures I, II, and III. The magnitude of the coupling constant for coupling across slP and the C(5') protons should enable one to distinguish among the above three possibilities. This is because molecules containing the system H-C-O-P have been reported to show an angular dependence of spin-spin coupling similar to that found in H-C-C-H systems. The gauche' coupling is in the range of 3 Ha and the trans in the range of 21 Hz (11-16). One of the difficulties in determining the conformational preferences of the C(5')-O(5') bond in P-5'-nucleotides is that an unambiguous assignment by NMR of the 2 C(5') protons cannot be made at present and hence one cannot distinguish between the gauche'-trans' (Structure II, g't') and trans'-gauche' (Structure III, t'g') rotamers. By manipulation of Equations 1 to 3 in Hruska et al. (17) it has been shown (18-20) that the percentage of gauche'-gauche' (Structure I, g'g') conformer and * The term torsional diastereomer (10) is used to describe the various rotational isomers of a nucleoside or a nucleotide. the combined contributions from g't' and t'g' can be computed from Equation (1).
Fractional population of g'g' N_ (24 -2')/18 (1) where Z' = Ja~P + J6ep. Because of the uncerta.inty in the magnitude of pure bans and gauche couplings (11-16) there may be an error of at least 10% in the computed populations. Further, it should be pointed out that in deriving Equation 1 it is assumed that energy minima occur with dihedral angles centered near 60, 180, and 300" (Structures I to III), and that the fraction of time spent outside these minima, i.e. during interconversion, is small. The value of 2' in Equation 1 can be obtained from the 5' region of the spectrum. However, at concentrations above 0.1 M, an exact value of Z' cannot be obtained because the 5' region shows unusual broadening at higher concentrations and the derived data cannot be fitted unequivocally by line shape computer simulation. The observed values for Z' at 0.1, 0.05, and 0.005 M 5'-AMP (pD 8.0) are 9.0, 9.2, and 9.2 Hz, respectively. These values when used with Equation 1 indicate that 5'-AMP, on a time average basis, at biological pH exists predominantly (-80%) with a g'g' conformation (Structure I) about the C(5')-0(5') bonds with a small contribution (~20%) from g't' and t'g' conformers (Structures II and III). Further, this population distribution appears to be independent of concentration, at least at levels below 0.1 M. At concentrations above 0.1 M, as stated earlier, an unequivocal data fitting is not possible; however, the shape of the 5' region is inconsistent with any predominance of t'g' and g't' conformation. At pD 5.4,2' is 10.0 Hz. The change in phosphate ionization probably has some effect on the H-C-O-P Karplus relationship, but the observed value of 2' indicates the population could not have been greatly changed in going to the monoanion.
Torsional Diastereomers Constrained to C(4')-C(6') Bond of 6'-AMP-The minimum energy torsional diastereomers constrained to the C(4')-C(5') bond of 5'-AMP are shown by Newman projections in Structures IV, V, and VI. We have shown elsewhere (18-20) that the population distribution of the rotamers IV, V, and VI can be evaluated from the experimental sum J4'6' + Jga-. Because of reasons given earlier, one can by NMR methods evaluate only the contribution from the guuchegauche (Structure IV, gg) conformer and the combined contribution from gauche-lruns (Structure V, gt) and trans-gauche (Structure VI, tg) rotamers to the time average conformation constrained to the C(4')-C(5') bond. This can be computed from Equation 2, derived in Refs. 18 to 20 by manipulating Equations 1 to 3 in Hruska et al. (17).
Fractional population of gg Lb! (13 -X)/10 ('4 where Z = J416f + Jq16n. Assumptions required in this derivation are the same as in Equation 1, except here one is dealing with the H-C-C-H bond system. A detailed error analysis of equations of the type in Equation 2 has been discussed by Blackburn et al. (21). It is estimated that there may be an error of 10% in the computed populations via Equation 2. The value of Z in Equation 2 can be obtained from the 4' and 5' region of the spectra. The unusual broadening of the 4' region of 5'-AMP at concentrations above 0.5 M prevents an exact determination of the value of Z above 0.5 M. In the concentration range of 0.005 to 0.5 M, the value of I; shows little variation and indicates that the population distribution of rotamers constrained to C(4')-C(5') is not significantly dependent on concentration, at least below 0.5 M. The observed value for L: is 6.5 Hz which when used in conjunction with Equation 2 indicates that 5'-AMP, on a time average basis, at biological pH exists predominantly (~657~0) with a gg conformation (Structure IV) about the C(4')-C (5') bond and with a significant contribution (-35%) from gt and tg conformers. At pD 5.4, I; is 6.1 Hz indicating phosphate ionization has little if any effect on the population about this bond. The above discussion does not mean that the conformations about the C(4')-C(5') and C(5')-0(5') bonds are totally independent of concentration; rather, the limited range of concentration in which the proton NMR data can be accurately analyzed does not show any detectable variation which is significant with respect to the errors in the calculations or the measurements.
Hall and co-workers (12,13,22) have shown that in phosphate esters such an in-plane "W" relationship will generate a maximum *J(1H-81P) of 2.7 Hz and this value reduces to zero when the planarity is destroyed.
We have observed a *Jvp of magnitude greater than 1 Hz in pyridine nucleotides (23)(24)(25) in 5 19) and in several mononucleotide components of DNA and RNA (19) and in such biologically important cofactors as adenosine diphosphoglucose, uridine diphosphoglucose, and o-glucose l-phosphate (26). We have argued, based on our treatment of 2Z and Z', that the observed 4J4,p in these compounds reflects the preference of the C(4')-C(5') and C(5')-0(5') bonds to orient gg and g'g', respectively.
Based on all of the above observations, we feel confident in presenting the observed *J4tp in 5'-AMP (1.4 to 1.7 Hz in the concentration range 0.1 to 0.005 M at pD 8, and 2.1 for 0.01 M at pD 5) as an important and additional piece of data to support out contention that the C(4')-C(5') and C(5')-0(5') bonds of 5'-AMP exist preferentially, but clearly not completely, in gg and g'g' conformations, respectively. The small difference in 4J41p between pD 8 and 5 may be caused by the change in phosphate ionization.

Conformation of o-Ribose Ring-Among
all the fragments of nucleosides and nucleotides the solution conformation of the n-ribofuranose system has probably received the maximum attention and controversy from NMR spectroscopists. Sarma and Mynott (23,28) have reviewed the status on this subject and have suggested that, in a qualitative sense, perhaps it is best to treat the ribose ring coupling constant data just in terms of two modes, viz. C(3') endo and C(2') endo undergoing interconversion via pseudorotation (Fig. 1). For 3E: the C (l'), C (2'), C (4')) and C (1')0 ring atoms lie approximately in a plane, and the C(3') is out of the plane and on the same side of the plane as the C(5') atom (endo). For $E: the C(l'), C (3'), C(4'), snd C(l')O ring atoms form a-plane with C(2') endo. The ribose rine of 5'-AMP is in eauilibrium between these two qualitative conformation types; the mechanism of their interconversion may be somewhat related to the pseudoarotational interconversion which takes place in cyclopentane (30). and J2f3t has been demonstrated before. In Fig. 3 we illustrate the 'H-lnlP) spectra of 0.5 and 0.005 M 5'-AMP along with computer simulations. Ionization of the phosphate had no detectable effect on the ring coupling constants.
Qualitatively speaking, the data are consistent with the existence of aE and 2E ( Fig.   1) conformations in equilibrium (23,28) and, at concentration levels higher than about 0.01 M, increases in 5'-AMP concentration cause a depopulation of *E and an increase in aE conformation.
One may attempt to describe the sugar pucker in quantitative terms using the concept of pseudorotation (30)(31)(32)(33)(34)(35)(36)(37). Altona and Sundaralingam (36,37) argue that accurate quantitative information about the conformational dynamics of the ribose ring The spectra at the two concentrations reflect the change in ribose coupling constants. A visual comparison of the two spectra further demonstrates that as one moves from lower to higher concentration levels, the chemical shifts of C(l')H and C(2')H move to higher fields, C(3')H does not show much perturbation, and the C(4')H and C(5')Hz chemical shifts move to lower + fields. The scale is given upfield from (CH,),NCl which is resolved into a triplet indicating the quality of resolution of the 8K Fourier-transformed spectra.
can be obtained from two basic parameters, P and rm. P is the phase angle of pseudorotation and r,,, is the amplitude of pucker. The magnitude of P, which in theory could be from 0 to 360", enables one to determine the exact conformation of the sugar and that of 7m gives information regarding the degree of pucker. Inspection of Fig. 3 in Ref. 36 shows that an arbitrary P value of 0" corresponds to a symmetrical twist conformation of C(3') endo-C(2') ezo($T); a P value of 9" corresponds to an unsymmetrical twist conformation of C(3') endo-C(2') exo(a7'J with C(V) as the major pucker; a P value of 18" corresponds to a symmetrical envelope conformation of C(3') endo ( Fig. 1) and so on. Altona and Sundaralingam (36,37) have observed that in the solid state the sugar conformation of nucleosides and nucleotides fall in two ranges separated by a barrier of 2 to 4 Cal, I&. P = 0 to 36" and 144 to 180". The range 0 to 36" comprises z8T, 8T2, aE, aT4, and 48T conformations. The range 144 to 180" comprises 12T, *T1, *E, 2T3, and a2T conformations. These two classes are, respectively, called N and S type conformations. They suggest that in solution a dynamic equilibrium exists between the above two classes and this is only a modified form of the idea of a *E + 'E equilibrium introduced by Sarma and Mynott (23) and Hruska (38). From the experimentally determined coupling constants J1t2r, J2#3#, and Jatlp, it has been argued (36,37) that one should be able to extract the following pseudorotational parameters which describe the ribose ring conformation in pure quantitative terms: PN, Ps, r,,,N, r,,&', %N, '%A'$ where P and rm represent the phase angle of pseudorotation and 4757 the amplitude of pucker and the letters N and S stand for the two types of pucker.
This approach, which prima facie sounds precise, has a number of pitfalls.
1. Granted, exact quantitative relationships exist among endocyclic torsion angles, P and rm, for the pseudorotational behavior of eyclopentane, and all endocyclic torsion angles which occur in the entire pseudorotational pathway can be exactly obtained by maintaining a constant rm while varying P. However, there is no evidence that the same quantitative relationships will exactly hold true for the projected pseudorotational interconversion of an asymmetrically substituted ribose of a nucleoside or nucleotide; there is likely only a qualitative resemblance.
2. The pseudorotational treatment (36,37) relies on the basic Karplus equation (JHH = A cos2 $I -B cos 9 + C) to distinguish accurately between small changes in dihedral angles; we have reported earlier that an adjusted, more complex form of the equation is required to obtain the conformational dynamics of the ribose ring (23).
3. Any attempt to utilize the Karplus equation to extract precise quantitative information on conformational equilibria will be fruitless unless the values of the constants A, B, and C are precisely determined for the system. The Altona-Sundaralingam method of obtaining the above constants has serious defects. Using crystalline data for a large number of nucleosides and nucleotides, the authors determined average torsion angles between carbon, oxygen, and nitrogen atoms (hydrogen atoms cannot be located accurately) for both N and S conformations, and then assumed that in the Newman projection the carbonhydrogen vectors are centered in the middle. An error of only 2.5" between two such vectors may produce an error up to 5" in the corresponding hydrogen-hydrogen vector 4. These solid state angles (4 values) were then assumed to be equal to the average solution $J angles of a collection of nucleosides and nucleotides, and from the average coupling constants Altona and Sundaralingam derived A and B values of 10.5 and 1.2, respectively. They further assumed that the value of C is zero. Using this procedure (36,37) 4. Even if all of the above objections were not taken into account, the conservatively estimated experimental error of ho.1 Hz in coupling constants for the computer-simulated spectrum (Fig. 3) gravely affects the predicted pseudorotational parameters (36,37). An error of ho.1 Hz means an uncertainty of 0.4 Hz in the sum Jr**< + J3<4' and 0.2 Hz in J2#3>, when included in the pseudorotational calculations (36,37), typically produces an uncertainty of &2' in r,,, and &7" in P. Consequently, if the pseudorotational computations yielded a value of P of say 18", it would not mean that the conformation of the sugar must be *E, but rather it may be aT2, aE, or aT4 (see Fig. 4 of Ref. 36). Further, the calculated r,,, values for the N and S conformations typically differ by 3", yet for a cyclopentane pseu-dorotational scheme, they should be the same. Here having only three couplings to solve five unknowns is a problem (36).
We have used our data for 5'-AMP to calculate* its solution pseudorotational parameters.
For the S conformation of 5'-AMP, the resulting mean amplitude of pucker r,,, is 38" and the phase angle Ps is 175". For N, r,,, is 35" and PN is 5". Taking the above uncertainty into account, it would mean that 5'-AMP may exist as 3T2, 32Z', or *T3 conformations on the S side and as *Ta, 28T, or aTz conformations on the N side, i.e on the N side the atoms C(2') and C(3') may be symmetrically puckered or they may be unsymmetrically puckered with either C(2') or C(3') undergoing major puckering.
It is obvious that the pseudorotational treatment (36,37) does not enable one to pinpoint the exact conformational status of a sugar ring. 5. In the pseudorotational treatment (36,37) one must assume J213t of N and S conformations to be equal; experimentally a spread of 1 Hz in J21aj values is observed upon inspection of data from several nucleosides and nucleotides.
It is possible that this approximation (36,37) may be valid. However, we caution, if a difference of about 1 Hz exists in the value of J21al between pure N and pure S conformations, this alone causes un-certainty8 of about 6" in rn and 10" in P.
It is clear from the above discussion that there is inaccuracy and uncertainty in the pseudorotational treatment of coupling constants.
The pseudorotational approach of Altona and Sundaralingam enables one to compute the population of the percentages of N and S conformations from the experimental coupling constants because such a treatment yields the magnitude of the various coupling constants in the pure N and S conformations. It is necessary to incorporate an uncertainty of 1 Hz in these calculations which in turn generates an error of &lo% in the computed populations.
For example, 5'-AMP data so treated indicate that at all concentrations below 0.01 M, the ribose ring conformation exists 63 f 10% *E(S), the remainder being aE(N).
As the concentration increases, the fraction of *E(S) decreases and *E(N) correspondingly increases. At high concentrations, i.e. 2.0 M, an estimated value of J1f2t of less than 4 Hz indicates the ribose ring is greater than 50% aE(N) conformation.
The computed *E population of 5'-AMP at concentrations below 0.1 M, using the simple Karplus equation as employed by Schleich et al. (39), Hruska et al. (40), and Smith and Jardetzky (41) vary from 63 to 73%.
In concluding we wish to point out that the description of the sugar pucker in terms of the precise values of P and rm as an entity in the pseudorotational itinerary is very articulate and elegant with respect to the solid state data and most certainly an improvement over the traditional description of sugar pucker based on the best 3-or 4-atom plane. However, the pseudorotational concept, when extended to describe the dynamic solution a Computer output of the calculation of PN, Ps, r,,,N, r,,,S, JI,V N, JI'~ 8, Ja'a'S, J:'d'N for given values of Js'g' and 7m are deposited in the JBC data repository.
These data are available as JBC Document No. 73M-1369 in the form of one microfiche or 30 pages. Orders for this material should specify the title, authors, and reference to this paper and the JBC Document number, the form desired (microfiche or photocopy), and the number of copies desired. Orders should be addressed to The Journal of Biological Chemistry, 9656 Rockville Pike, Bethesda, Maryland 20014, and must be accompanied by remittance to the order of the Journal in the amount of $2.56 for microfiche or $4.59 for photocopy.
It should be pointed out that the Altona-Sundaralingam equations become inoperative when attempts are made to compute conformer populations for a sum J1,2, + Ja*r* less than 9.5 Hz coupled with Jz,a* of 5.0 to 5.5 HZ. This is simp!y because the value of 72 should be smaller than or equal to rm. conformation of the sugar ring, loses precision and accuracy because one has to solve for five unknowns from three coupling constants via a Karplus equation and several assumptions of questionable validity.
It is our belief that the pseudorotational treatment of coupling constants is neither better nor worse' than the traditional Karplus approach in describing the ribose ring conformation in solution. Finally, we wish to restate that in solution the ribofuranose system exists as an envelope of continuously exchanging conformations.
For the sake of simplicity in description it may be treated as a two-state 2E e aE equilibrium.
See Ref. 42 for a discussion of sugar-base torsion.