31p nmr studies of enzymatic reactions - semantic scholar · spin of 31p is vz, and the nmr spectra...

31P NMR Studies of Enzymatic Reactions

Mildred Cohn and B. D. Nageswara Rao 1

Department of Biochemistry and BiophysicsUniversity of Pennsylvania

Philadelphia, PA, USA

I. INTRODUCTION

Phosphate compounds are ubiquitous in living cells.Almost all of the chemical energy derived frommetabolic processes is stored in the form of adenosinetriphosphate (ATP)2 or other nucleoside triphosphates.Subsequently these triphosphates are utilized in allenergy-requiring functions of the organism such as thebiosynthesis of macromolecules, mechanical work, andtransport of metabolites. In addition, the phosphoryl-ation of neutral molecules such as glucose to formnegatively charged molecules such as glucose-6-phos-phate, ensures that the metabolite will be retained in-side the cell. Thus, for example, every reaction in theuniversal glycolytic pathway of glucose metabolism in-volves phosphorylated substrates and products. Lastly,phosphorylated compounds, including many nucleo-tides and particularly 3',5' cyclic adenosine andguanosine monophosphates, are important regulatorsof metabolism.

Phosphorus has a single natural isotope 31P that hasa nuclear spin and a substantial magnetic moment (ca40% of 'H). Among the isotopes of atoms that are com-monly present in biological molecules (H, C, N, O, P,and S), 31P combines a unique set of properties that

'Present address: Department of Physics, Indiana Univer-sity-Purdue University at Indianapolis, IN, USA

'The following abbreviations are used in this paper: ATP,ADP, AMP are adenosine-5'-tri-, di-, and monophosphate, re-spectively; CMP, cytosine-5'-monophosphate; UMP, uridinemonophosphate; Pj, inorganic orthophosphate; PPj inorganicpyrophosphate; P-, phospho; NAD, NADH, oxidized and re-duced forms of nicotinamide adenine dinucleotide; NADP,NADPH, oxidized and reduced forms of nicotinamide adeninephosphate; EDTA, ethylenediaminetetraacetate.

make it a useful probe to study these molecules by nmrspectroscopy. These properties are: (a) 100% abun-dance with a magnetic moment (compare with C, O,and S for which the abundant isotopes 12C, 18O, and 32Shave zero spin and are therefore nonmagnetic); (b) thespin of 31P is Vz, and the nmr spectra are thus free fromquadrupole broadening that might complicate the infor-mation available from the spectra (nuclei with spin >Vi, e.g., 14N, spin = 1, possess an electric quadrupolemoment that tends to broaden nmr lines due to interac-tions with fluctuating electric-field gradients at the siteof the nucleus in the molecule); (c) 31P nmr spectra ofspecific biological molecules can usually be observedin solution without interference from solvent, unlike theserious complications that may arise in observing 1Hspectra in H2O solutions (1). Furthermore, in complexeswith macromolecules such as enzymes, the protons ofinterest, unlike phosphorus, are usually obscured by themany protons in the spectrum arising from the macro-molecule. At the same magnetic field 31P nmr signalsoccur at a frequency 40% of that for 'H and are obtain-ed with a sensitivity of ca7% of that of 1H.

The possibility of readily distinguishing and assign-ing each phosphorus atom of ATP and of ADP as wellas the state of ionization of their respective terminalphosphates by 31P NMR was recognized and demon-strated in 1960 (2). Shortly thereafter (3) it was shownthat the metal chelates of ATP, which are generally theactive species in enzymatic reactions, could be distin-guished from the unliganded species by a significantchange in chemical shift of the /3-31P of ATP. T2 relaxa-tion rates of 31P resonances of ATP were affected byvery low concentrations of paramagnetic ions, in a spe-cific metal-dependent pattern.

With the advent of higher-frequency instruments and

38

Duplication of Bulletin of Magnetic Resonance, in whole or in part by any means tor any purpose is illegal.

Bulletin of Magnetic Resonance

Fourier transform nmr spectrometers, sensitivity wassufficiently increased so that 31P nmr studies of enzymesystems became feasible. Not all such studies will becovered in this review, but rather a sufficiently diversesample to illustrate the types of information relevant toenzyme structure and function that may be derived from31P nmr spectroscopy. In considering the mechanism ofan enzymatic reaction, it is first necessary to establishwhat the reactants and products are. Examples of thestraightforward use of 31P nmr to identify the productsof enzymatic reactions easily and directly will bepresented. The more subtle characterization of theanionic species, metal chelate species, anomeric orstereoisomeric form of phosphate substrates, products,or effectors that interact with the enzyme will be dis-cussed in some detail.

The parameters measured from nfnr studies are (1)chemical shifts (5), (2) indirect spin-spin coupling con-stants (J), (3) areas enclosed by the resonances, (4) linewidths (or spin-spin relaxation times T2), and (5) spin-lattice relaxation times (I,). The chemical shift of a nu-cleus represents change in its resonance frequency fora given value of the external magnetic field, due toshielding from its electronic environment; 8 is definedrelative to the resonance of 85% H3PO4. The magneticinteraction between nuclear spins mediated by theelectronic environment is observable in high-resolutionnmr as fine structure in the resonance. From the finestructure the strength of this interaction,, represented byJ, may be deduced. In studies pertaining to enzyme-sub-strate interactions of the kind mentioned above, inwhich nmr spectra of substrate molecules are observedin their enzyme-bound complexes, 8 and J may dependon the environment of the substrate on the enzyme.Measurements of 8 and J in suitably chosen complexesare thus potentially capable of providing informationon, e.g., stoichiometry, state of metal chelation, theionization state of the substrate, and possibly the con-formational changes in the substrate produced by in-teraction with the enzyme.

The application of 31P nmr as a nonperturbing methodfor determining equilibrium constants between phos-phorus-containing compounds (both for the overall re-action and the interconversion step between enzyme-bound species from the areas of the resonances) will bedescribed. The line width of a resonance is primarilydetermined by factors that govern the lifetime of thespin states. In particular if the nucleus changes its envi-ronment, and therefore chemical shift, by rate pro-cesses like E + S^E»SorE + P = £ • P, the linewidth changes may allow the measurement of the rateconstants in these processes. The effects observed onthe nmr line widths may be more dramatic if the par-ticular nucleus is present in an itinerant group as in an

equilibrium mixture of a phosphoryl transfer reactioncatalyzed by a kinase. The rates or limits of rates of dis-sociation of enzyme-substrate complexes and rates ofproduct formation at equilibrium have been evaluatedfrom 31P line widths. A theoretical consideration of nmrkinetics (T1,72) in terms of enzyme kinetics has beenpresented (4). Lastly, the disposition of substrates at theactive sites of a considerable number of enzymes hasbeen mapped based on distances between paramag-netic probes and phosphorus atoms of the substrate(s)estimated from the paramagnetic effect on 31P relaxa-tion rates (7", and T2). In many investigations of this type,the 31P data were supplemented by, or were supplemen-tary to, 1H or 13C nmr relaxation data or to data obtainedby other methods.

Isotopic substitution of the nuclei attached to theresonating nucleus produces characteristic and smallbut detectable changes in the resonance frequency,e.g., 31P resonance frequencies in 31P16CV and31F8O3

18O- differ by ca 0.02 ppm (5). The changes pro-duced by more than one substitution are approximatelyadditive. The nmr spectrum may thus be used to deter-mine the concentrations of the different isotopicallysubstituted species as in mass spectroscopy. By usingthe isotopic substitution as a label in a reaction with se-quential steps, it will then be feasible to design experi-ments to investigate intermediates in the reaction.

II. THEORETICAL AND EXPERIMENTALCONSIDERATIONS

A. Theoretical

The chemical shift 8 and the spin-spin coupling con-stant J are the two spectral parameters that may bemeasured from the different peak positions in the high-resolution nmr spectrum. If there is a change inchemical shift (or in any other nmr parameter) due to theformation of a complex with enzyme or with metal ion,the stoichiometry can be deduced for fast-exchangeconditions from the variation of the chemical shift (orthe relevant parameter) with concentration of one com-ponent in the complex. If the chemical shift of anucleus in a particular moiety on the substrate changeswith pH, a comparison of 8 vs pH of the enzyme-boundsubstrate with that in free solution is expected to shedlight on the environment at the binding site on the en-zyme. It might even be tempting to speculate, on thebasis of such a comparison, which ionic species mightbe the true substrate of the reaction, particularly whenthepH dependence disappears in the bound state. Duecaution must be exercised in doing this, however, since

Duplication of Bulletin of Magnetic Resonance, In whole or In part by any means tor any purpose is Illegal.

Vol. 1, No. 1 39

NOExchange

Figure 1. Effect of chemicalexchange on line shape andresonance position.

Slow

Intermediate

( A cu - A o i 0 ) / 2 = T ~

~ A . B

Fast «V,TB

the titration may be limited in pH range in the boundstate and furthermore, may be governed by entirely dif-ferent factors in the two cases being compared.

In general, when a chemical shift is observed in an£»S complex for a nucleus in the substrate, the shiftmay be due to both a change in environment and achange produced in the structure (or conformation) ofthe substrate induced by the enzyme. If a separation ofthese effects is made, it will be of considerable value tothe understanding of the enzyme mechanism if the ob-served shift can be interpreted in terms of structuralchanges. The same argument applies to spin-spin cou-pling constants, changes in which are likely to be en-tirely due to conformational changes produced in the£«S complex. In this respect 31P nmr suffers from aserious drawback in that no reliable theory applicableto molecules of biological interest is available either for6 or J that would allow meaningful correlations between6 and J measurements and molecular structure. Thetheoretical difficulties associated with 6 and J of 31Pcompounds were discussed by Letcher and Van Wazer(6, 7). Some empirical correlations of chemical shiftswith bond angles and torsional angles have been madeby Gorenstein and Kar (8). Since phosphorus com-pounds of biological interest are invariably phosphatederivatives, the four tetrahedrally arranged oxygensshield the 31P nucleus from the environment so thatshifts are small. If an O is replaced by S to form phos-phorothioates, then the 31P shift of the analogous com-pound is shifted downfield by 40-50 ppm (9).

We shall now briefly consider the effect of chemicalexchange on the line shape of a resonance. Detailedtheories for the calculation of these line shapes ap-plicable to simple spin systems are available (10). How-ever, since the systems of interest here are relativelycomplex, we shall only consider a few limiting condi-tions for which the line shapes can be qualitatively un-derstood and the relevant rates be quantitated. If UA anduBaxe the two resonance frequencies and A«o(/\) and A-wo(6) are respectively the corresponding linewidths in the absence of exchange in a two-site ex-change process, three limiting conditions are of in-terest, viz., (See Figure 1)

\ioA - ws| » TA~\TB~* (slow exchange)(intermediate exchange) (1)(fast exchange)

where TA and TB are the lifetimes in the two states, whenexchange takes place (11). Under slow exchange thetwo resonances would still be centered around UA and -wfl. but are broadened to Ao^A) and A«(B) respectively,such that

Aw(/\) - Awo(A) = TA"Aco(B) - Awo(8) = TB-'

Under fast exchange conditions a single resonance isobserved at an intermediate frequency (co) with interme-diate line width (Aw) given by

\UA-O>B\ = TA~\TB~'1

\<»A - <°8| « 7>f VB~1

(2)

W = PAAco = p pB

(3)

where PA andpe are the fractional populations of the ex-

40

Duplication of Bulletin of Magnetic Resonance, in whole or in part by any means for any purpose is illegal.


changing species. If the intermediate exchange condi-tion obtains, both frequency shift and line broadeningresult. It must be noted that the above is a simplifieddescription of the exchange effects and does not in-clude for example, effects of spin-spin coupling on theline shape. However, for the purpose of understandingthe results considered in this paper, the descriptionis adequate.

Relaxation measurements on 31P nuclei in the pres-ence of paramagnetic metal ions, usually replacing theobligatory divalent cation Mg2+ in enzymatic systems in-volving phosphoryl transfer and nucleotidyl transferreactions, have been used to gather structural informa-tion on the geometry of the active site. The details ofthe theory underlying paramagnetic effects on spin re-laxation in solutions and assumptions and limitationsinvolved, are too vast to be described in this article.Several authors have reviewed different aspects of thisproblem earlier (12-15). We shall only consider some ofthe results pertaining to 31P nmr studies to illustrate theoverall potentiality of the technique in deriving bio-chemically relevant information in these enzymaticsystems.

In practically all cases where paramagnetic probeswere used in 31P nmr studies, the following assump-tions seem to be implicit, (i) Solomon-Bloembergenequations are valid for the paramagnetic contribution tospin-lattice (7"IM~1) and spin-spin relaxation (T2M'1 rates,(ii) Contributions to the observed relaxation rates of thenucleus of interest from other mechanisms (referred toas diamagnetic effects) or due to the ligand nuclei dif-fusing through the paramagnetic environment withoutbeing bound (referred to as outer sphere contributions)are either neglected or accounted for by suitable con-trol measurements, (iii) Only a small fraction (ca0.01-1%) of the ligand complexes are bound by theparamagnetic probe, (iv) The exchange times relevant tothe different complexes undergoing exchange {TM) aresmall compared to r ^ a n d T2M- ^ s^e 2 » 1. where os

is the epr frequency and r2e is the spin-spin relaxationtime of the ion; further the isotropic part of the hyperfineinteraction between electron and nucleus, A, is smallenough such that this interaction does not contribute toTiM significantly. The anisotropic part of the hyperfineinteraction is indistinguishable from the through-spacedipolar interaction between the electron spin and thenucleus and should also be small in order to ensure thatthe latter is the exclusive source of T\M- Under these as-sumptions, the distance r between the ligand nucleusand the paramagnetic probe may be written as

r = +̂ rc2 J1/6

(4)

where C is a constant that includes the gyromagnetic

ratio of the nucleus and the magnetic moment of theparamagnetic ion, w/ is the nuclear Larmor frequency,and TC is the correlation time modulating the dipolarinteraction. TC includes the reorientation of the enzyme-substrate-metal complex, the electron relaxation time inthat complex, and the exchange time between differentcomplexes. T2M'X cannot be simplified to this formunless additional assumptions are invoked requiringboth the shift in the resonance frequency and A to benegligible. These requirements are quite stringent. If7"IM and TC are reliably estimated, r may be evaluated.Because of the exponent 1/6 in Eq (4), inaccuracies inthe estimation of T-\M and rc are fortunately scaleddown in the determination of r.

In the study of paramagnetic effects on substratenuclei bound to enzyme, it is the measurement of T^'mdifferent EMS complexes that is of primary interest. Themany experimental and theoretical considerations re-quired for ensuring the validity of Eq (4) are discussedby McLaughlin et al (16) and Mildvan and Gupta (15). Anecessary condition is a reliable estimation of rc, whichmay be made from measuring T^M as a function of fre-quency or from the ratio of T-\MIT2M'< the former is likelyto be more reliable because of the stringent conditionsrequired for simplifying the equations for Taw- It may beseen from Eq (4) that if the values of T-\M are reliablyestimated for different nuclei in a complex, the ratios ofthe distances of these nuclei from the paramagnetic iondo not depend on TC and are, therefore, independent ofinaccuracies in rc measurement provided a single corre-lation time is appropriate for all the nuclei.

B. Experimental

The design of experiments that would minimizeambiguities in the interpretation of data from 31P nmrstudies of enzymatic reactions often requires optimiza-tion of parameters from both nmr and biochemicalpoints of view. The constraints placed upon achievingsuch an optimization due to factors such as availabilityof appropriate nmr facilities or required quantity of en-zymes of high purity perhaps contribute to the fact thatwhen more than one research group investigates aproblem even the experimental results are oftenat variance.

Commercial nmr spectrometers available for 31P nmrrange in their operating frequencies from ca 24 MHz toca 146 MHz. A few other instruments operating athigher frequencies up to 240 MHz are becomingavailable. From the consideration of sensitivity and formeasuring small chemical shifts, higher operating fre-quencies are desirable. However it is not infrequent thatanisotropy of chemical shift contributes to relaxation of

Duplication of Bulletin of Magnetic Resonance, In whole or in part by any means for any purpose is illegal.

Vol. 1,No. 1 41

31P nuclei in some compounds (17). The line width dueto chemical shift anisotropy increases as the square ofthe frequency, and depending on the fractional contri-bution of this mechanism to the overall 31P line widths,part or all of the advantage of choosing a higher operat-ing frequency may be lost by the increased line widths(18). In some cases the increased line width could ob-scure fine structure in the resonances due to spin-spincouplings or small chemical shifts.

Concentration of the compound of interest is one ofthe most important parameters for the purpose of opti-mization. Quite often it may be necessary to work at lowconcentrations of 1m/W due to limitations from factorssuch as solubility, unavailability of sufficient material,or artifacts arising from higher concentrations thatwould make the results misleading or irrelevant. Fromthe point of view of nmr it is better to work at highconcentrations that would increase the signal to noiseratio per unit time utilized for signal averaging. The timefactor may be especially important if the samplechanges with time or if there are processes taking placethat alter the particular aspect under investigation. Atthe highest operating frequencies available for 31P nmr,it is presently feasible to work with concentrations inthe vicinity of 1mM that require accumulations rangingfrom 10 minutes for sharp resonances to a few hours forbroad resonances.

In studies of enzymatic reactions where the deter-mination of nmr parameters of enzyme-bound com-plexes is usually the ultimate goal (whether it is achiev-ed by directly observing enzyme-bound complexesthemselves or by deducing the changes caused in theparameters of substrates in free solution due to thepresence of the enzyme in the sample), a knowledge ofthe dissociation constants and exchange rates perti-nent to the complexes being studied becomes crucial.In addition, the purity of the enzyme and in some casesthe past history of the particular enzyme preparationbecome important in the reproducibility of the results.

In experiments dealing directly with line widths orthose in which small chemical shifts that are compara-ble to line widths are to be measured (e.g., the 18Oisotope shifts of phosphate resonances), especially incompounds like ATP or other phosphorus compoundsthat carry substantial negative charges in the vicinity ofthe 31P nucleus in the molecule, it is of paramount im-portance to remove paramagnetic cationic impuritiesfrom the sample by suitable extraction procedures (19)or by the addition of chelating agents when such an ad-dition does not interfere with the experiment. Paramag-netic metal ions at concentrations 10"5 times that ofATP cause significant broadening of the 31P resonancesof ATP. Impurities at this level may sometimes arisefrom the glass of the sample tube itself.

III. EXPERIMENTAL STUDIES OFENZYME SYSTEMS

A. Identification of Products

In a number of enzymatic reactions involving phos-phates, as in nonenzymatic reactions, the most directmethod of identifying and quantitating the chemicalspecies produced has been to use 31P nmr spectros-copy as an analytical tool. For the phosphocarrier pro-tein HPr, 31P nmr (20) confirmed 1H nmr in identifying thephosphorylated protein (a histidine phosphorylated atthe N1 position) enzymatically produced from P-enolpy-ruvate as the donor in the phosphotransferase systemof Staphylococcus aureus. In contrast, the chemicallysynthesized product is phosphorylated at N3 of the his-tidine of the protein. The assignment was made on thebasis of chemical shift and pH titration curves of modelcompounds, N1- and N3-P-histidine. Another example isthe identification of the products of the hydrolysis ofphospholipids in human serum high-density lipopro-tein-3 catalyzed by a-phospholipase A2 (21). The two di-astereoisomers of adenosine-5'-O-(1-thiotriphosphate)and also of the corresponding diphosphate, which areepimeric at Pa, can be distinguished by their respectivechemical shifts (9, 22). It was thus demonstrated (22)that the phosphorylation of adenosine-5'-phosphoro-thioate catalyzed by adenylate kinase is stereospecific,leading to the formation of diastereoisomer A of adeno-sine-5'-O-(1-thiodiphosphate). The identification of the31P chemical shift of the diastereoisomeric product uri-dine-5'-(1-thiodiphosphate)-glucose A has been used toestablish stereochemical inversion at Pa during thecourse of the reaction of uridine-5'-(1-thiotriphosphate)B with glucose-1-phosphate catalyzed by UDP-glucosepyrophosphorylase (23).

B. Covalently Bound Phosphates

The interpretation of 31P nmr parameters is simplifiedwhen the phosphate of interest is covalently bound tothe enzyme, since a multiplicity of exchanging speciesis avoided. When dissociable phosphates are bound tothe enzyme, then the interpretation of chemical shifts,line widths, or paramagnetic effects on relaxation ratesis often complicated by equilibria between free andbound species without complete averaging as well asby the possibility of more than one binding site.

Six enzyme systems have been investigated by 31Pnmr where phosphates or derivatives thereof are cova-lently bound to the enzyme. In two cases, alkaline phos-phatase of Escherichia coli and rabbit muscle phospho-glucomutase, the phosphorylated enzymes in the form

42

Duplication of Bulletin of Magnetic Resonance, in whole or in part by any means lor any purpose is illegal.


of serine phosphate esters located at the active site areintermediates in the respective reaction pathways. Thefirst enzyme is a dimeric zinc metalloprotein of molecu-lar weight 100,000 that catalyzes the hydrolysis of phos-phate monoesters and other phosphate compounds.The second enzyme, a monomer of molecular weight67,000, requires a divalent metal ion for activity andcatalyzes the interconversion of glucose-1-phosphateand glucose-6-phosphate. The 31P chemical shift of themodel compound for these protein phosphates, serinephosphate, is -0.4 and -4.0 ppm for the mono- and dian-ion respectively (pKa = 5.8) (24). For E-P(phosphogluco-mutase) the 31P shift is the same as that of the dianionicform of free serine phosphate, but unlike the latter, theE-P chemical shift does not vary with pH in the range of5.4 to 8.6 (25). On the other hand, the chemical shift ofthe serine phosphate residue of the E-P form of alkalinephosphatase is unusually low, -8.5 ppm, and is also in-dependent of pH (26-28). The unexpected value of the31P shift in alkaline phosphatase has been ascribed to astrained configuration of the phosphate (26) and to anarrowing of the O-P-0 angle by an estimated 2° (27).The absence of a pH dependence of the serine phos-phate for both enzymes is a fairly common characteris-tic of enzyme-bound phosphates (covalently or ionicallybound) as will become abundantly clear in the course ofthis review.

A number of enzymes interact with the coenzymepyridoxal phosphate (pyridoxal-P) by covalent bond for-mation of a Schiff base (aldimine) with the e-aminogroup of a specific lysine on the protein. Two such en-zymes have been examined by 31P nmr, aspartate trans-aminase (29) and glycogen phosphorylase (30-33). Pyri-doxal phosphate binds covalently at the active site ofaspartate transaminase, and the Schiff base formedparticipates in the reaction with the amino acid to yieldthe corresponding keto acid and pyridoxamine phos-phate. The chemical shift in enzyme-bound forms ofboth compounds is insensitive to pH while for the com-pounds free in solution, 6 changes between the mono-and dianion by 3.7 ppm. The 31P shift of the covalentlybound pyridoxal-P on this enzyme yields little informa-tion since it is insensitive to the binding of substrate(glutamate) or inhibitors (2-methyl aspartate or2-oxoglutarate).

There are two enzymes investigated by 31P that canexist in phosphorylated or dephosphorylated forms at aregulatory site rather than at the catalytic site of the en-zyme. Glutamine synthetase from Escherichia coli, adodecameric enzyme that catalyzes the reaction ofL-glutamate, ammonia, and ATP to form L-glutamineand ADP, is regulated by adenylylation of a particulartyrosyl residue in each subunit to form a phosphodi-ester. The 31P chemical shift of this phosphodiester (34)

is 4.0 ppm, and the line width is ca 22 Hz. A line width ofca 23 Hz was calculated using a value of 2 X 10~7 s forTft, and the dipolar interaction with two protons of thetyrosyl -CH2OP and the single P-OH were considered asthe dominant contribution to relaxation. In a later sec-tion calculations of distance from measurements of the31P relaxation due to the paramagnetic Mn(ll) bound tothe enzyme will be described.

The enzyme glycogen phosphorylase a is unique inthat it has two different types of covalently bound phos-phates involved in the regulation of its activity. Both theactive tetrameric form a (subunit 96,000 daltons) and thedimeric form b, which is inactive without added effec-tors, contain one pyridoxal-5'-phosphate per subunitcovalently bound in the form of a Schiff base to the e-amino group of a particular lysine. Removal of pyri-doxal-P inactivates the enzyme. Furthermore, the inac-tive dimer, phosphorylase b, is converted to the active aform by phosphorylation of one particular serine resi-due per subunit. The enzyme has been investigated by31P nmr by two groups, the Oxford group (30,33) and theWurzburg group (31, 32). The work has largely centeredon pyridoxal-P since its role is not yet understood. Twofeatures of the 31P resonance of the bound pyridoxal-Phave been studied, the ionic species and pH depen-dence and the effect of binding Iigands at other sites ofthe enzyme on the chemical shift and line shape. Busbyet al (30) found that each phosphate subunit ofphosphorylase b gave rise to two signals that were notrapidly interconvertible. The signals were ascribed totwo conformational states of the protein. Their linewidths increased with increasing frequency. The reso-nances could be resolved at 36.4 MHz, and the twopeaks were assigned to di- and monoanion peaks re-spectively. This assignment would require that protona-tion could only occur on one of two slowly interconvert-ing forms. Addition of the activator adenosine mono-phosphate (AMP) shifts the peak mostly to the monoan-ion position; any dianion present would be obscured bythe AMP resonance. The inhibitor glucose-6-P shiftspyridoxal-P mainly to the dianion position. However,there is no relation between the active forms and thechemical shift since in phosphorylase a, which is an ac-tive form resulting from phosphorylation of serine, theshift corresponds to the dianion shift, but in the activeform of phosphorylase b, activated by AMP, thepyridoxal-P shift corresponds to the monoanionic form.The underlying assumption that the state of ionizationof phosphates is the single variable that causes achange in the chemical shift is highly questionable.

The complexities of this system are explored in thepapers of Feldman and collaborators (31,32). A detailedstudy of pH effects on the chemical shift of pyridoxal-Pand its Schiff base, free in solution and bound to rabbit

Duplication of Bulletin of Magnetic Resonance, In whole or in part by any means for any purpose is Illegal.

Vol.1,No.1 43

muscle glycogen phosphorylase in the presence ofvarious effectors, led to the following conclusions.1) Pyridoxal-P exists in three forms when bound tophosphorylase b. 2) Form 1,6 = 0 ppm.pH independent,is observed for phosphorylase b with no substrateanalog present and is ascribed to the fully protonatedform. 3) Form II, observed in the presence of arsenate(substrate analog) or the anions SOf and Ch, 8 = -0.37ppm, below pH 7, has the same pH dependence as themodel Schiff base. 4) Form III, 6 = -3.81 ppm, is observ-ed for the active form of phosphorylase b, i.e., in thepresence of AMP or its thio analog, AMPS, and for theactive phosphorylase a; this shift corresponds to that ofthe dianionic form of the model Schiff base. The dis-crepancy between the results of the two groups ofinvestigators has no obvious explanation althoughsome of the experimental data of the second group areeasier to interpret because of the use of the thioanalogof AMP to avoid overlap of the resonances from AMPand pyridoxal-P.

The last example of a covalent enzyme-phosphate isthat of an enzyme inactivated by reaction with an inhib-itor, diisopropylfluorophosphate, to form a phosphatebond with the active site serine residue. The proteolyticenzyme a-chymotrypsin (a monomer of 25,000 molecu-lar weight), which normally forms an acyl intermediatefrom a peptide substrate, forms an inhibited phosphatetriester derivative (DIP) by reaction of the active siteserine-195 with diisopropylfluorophosphate. Gorensteinand Findlay (35) observed two 31P peaks for DlP-chymo-trypsin at ca 0 and -2.1 ppm respectively. The authorsascribed the peaks to two interconvertible forms of theenzyme since they reported that the pH-dependent ratioof the two peaks was reversible. However, Bock (36) andMarkley (personal communication) both ascribe thedownfield peak (ca -2 ppm) to an artifact; the former in-vestigator removed this peak upon dialysis of the en-zyme, while the latter observed it only with aged en-zyme. Reeck et al (37) report observation of one peak,the upfield one, at +0.52 ppm at pH 8.0. The peakobserved by Bock (36) at 0.3 ppm, pH 7.2, was shifted to3.3 ppm upon denaturation of the enzyme with 5Mguanidine HCI.

Reeck et al (37) addressed the question of the dif-ference in active site structures between the enzymeand its almost inactive precursor, chymotrypsinogen,which differ by a short peptide, a difference too subtleto be discerned in X-ray analysis. The DIP derivatives ofthe two proteins were compared. The 31P chemical shiftof the zymogen was 2.46 ppm with a line width of 3 Hz,and under the same conditions the shift for DIP-a-chy-motrypsin was 0.52 ppm and the line width was 2 to 3times larger. The known structural difference betweenthe two resides in the inability of the NH group of

glycine-193 to form a hydrogen bond to the phosphateof DIP in chymotrypsinogen; in DIP-a-chymotrypsin, NHgroups of serine-195 and glycine-193 form hydrogenbonds to phosphate. It is not clear whether the down-field shift (ca 2 ppm) is due solely to hydrogen bondformation.

C. Interaction of Substrates,Inhibitors, and Effectors

with Enzymes

In this section we shall cite examples of the changein nmr parameters (5, J, T2, etc) of phosphorus-contain-ing substrates, inhibitors, or effectors brought about bythe presence of various types of enzymes. Of the largeclass of enzymes that catalyze the hydrolytic cleavageof phosphate compounds, two will be considered, alka-line phosphatase (Escherichia coli), which acts onmonosubstituted phosphates, and ribonuclease A (bo-vine pancreas), which acts on the diester phosphates ofribonucleic acid and analogous model compounds. Asecond group of enzymes, the kinases, which transferthe terminal phosphoryl group of ATP to various accep-tors including arginine, creatine, AMP, 3-P-glycerate,and pyruvate, will be discussed. The last group of en-zymes to be discussed are those that have phosphatesubstrates but that do not involve transfer of phos-phoryl groups but rather hydrogen transfer, glucosyltransfer, or other reactions.

1. Hydrolytic Enzymes

Alkaline phosphatase was investigated in many labo-ratories between 1973 and 1978, and although there aremany areas of agreement in the 31P nmr studies, thereare serious disagreements both in experimental resultsand interpretation. The native enzyme is a dimeric zincmetalloprotein that binds inorganic phosphate andforms a phosphoryl ated enzyme E-P (serine phosphate)from any phosphate substrate or the product Pi. Asdiscussed above, it has been found (26-28) that E-P thatis stable at low pH and is unobservable at pH >6.5, hasa most unusual chemical shift, -8.5 ppm, considerablydownfield from any phosphorylated amino acid. Severallines of evidence have been presented for the assign-ment of E-P. 1) The Cd form of E-P does not dephos-phbrylate, and the peak intensity of the Cd form at -8.5ppm is not diminished even at pH 7.1 (26, 27); for nativeZn-enzyme intensities of the -8.5 ppm peak and ofanother peak at -3.5 ppm (E»Pj) vary inversely with pH(26-28), consistent with known interconversion of E-P toE»Pj. 2) No multiplet structure is observed in E-P, butproton decoupling narrows the line for both ZnE (26) and

44

DupMcatton of Bulletin of Magnetic Resonance, In nrfiofa or In part by any means lor any purpose Is Illegal.


CdE (28) as expected for serine phosphate. The largeshift of serine phosphate in alkaline phosphatase couldbe explained by a 2° narrowing of the O-P-O angle (27),which might arise (26) from a hydrogen bond acting inconcert with a covalent bond to serine to mimic astrained cyclic ester. In the apoenzyme (no metal ion)the covalently bound 31P is observed at -6.5 ppm (26,28).

It is in the noncovalent binding of Pj, particularly thestoichiometry of Pj binding, that there are some contra-dictory reports. Csopak and Drakenberg (38) were thefirst to observe the binding of Pj at pH 8 to apoenzymeand the Zn2 enzyme.

Subsequently two groups of investigators (27, 39)found 1.5-2.0 P/dimer, but another group found only 1P/dimer (28, 40). Also the latter group finds negativecooperativity in binding of Pj (40) as evidenced not onlyfrom 31P nmr but from 113Cd, which yields one peak fortwo Cd2+ in unliganded E but two resonances in thedimer containing one P; only one active site can bephosphorylated. Some of the discrepancies may resultfrom the mode of preparation of the enzyme as in-dicated by Hull et al (27) and more recently by Bock andKowalsky (39). The latter investigators showed that aform of the enzyme with a Zn content of 2.4 Zn/E thathad been obtained by partial removal of Zn by dialysisfrom a native preparation (4.9 Zn/E), retained 100%enzymatic activity, bound 1.4 P/E, and had one bound(E» P) resonance peak, -3.5 ppm at pH 7.8. On the otherhand, a form of the enzyme containing 2.4 Zn/E that hadbeen reconstituted from apoenzyme was only 10% ac-tive and had a broadened resonance peak at -6.0 ppm,pH 7.5. Hull et al (27) had also concluded that the "purg-ing" procedure yielded a different conformational formof E»P at alkaline pH since the high-activity form had aresonance at -3.5 ppm and the "purged" low-activityform had a resonance at -4.2 ppm.

All investigators agree that paramagnetic metalloen-zymes make E-P and E»P unobservable [Co2t (26-28, 38),Mn2+(28, 38, 41)] and that the metal ion is consequentlyclose to the bound P. Estimates of the distance varyfrom <3.3A (26), which places the metal in the firstcoordination sphere, to 7.3A (41), which would place itin an outer coordination sphere. The distance of Mn toP in the E-inhibitor complex (p-amino benzyl phosphate)was estimated to be 7.7A (28). However, all distance cal-culations have many assumptions and are conse-quently approximate. Since excess Pj with theparamagnetic metalloenzymes always appears as asharp line, it is concluded that the rate of exchange be-tween free and bound Pj is slow (26-28,38). Hull et al (27)analyzed in detail the line widths of the various E»Pjcomplexes and of an enzyme-inhibitor complex (2-hy-droxy-5-nitrobenzylphosphonate) and compared thederived rates with known kinetic constants. All in-

vestigators agree that the dissociation of E»Pj is a ratelimiting step in the reaction.

In conclusion, the studies on alkaline phosphatase il-lustrate both the usefulness of 3tP nmr and difficultiesencountered due to poorly understood, subtleconformational changes in the enzyme. The mostserious area of disagreement resulting from the 31Pstudies of alkaline phosphatase is the stoichiometryboth of metal binding and of phosphate binding to thedimeric enzyme. The discrepancy may be due in part tothe previous history of the enzyme, and hopefully, futureexperiments will resolve the apparent contradictions.

The 31P nmr studies of ribonuclease have focused onthe pH dependence of chemical shifts in free andbound inhibitors with the objective of determining theionic form of the enzyme-inhibitor species. Earlier workhad not led to a consensus. On the basis of pioneeringwork on 1H nmr titration data and the pH-binding con-stant curve, Meadows et al (42) concluded that the mon-onucleotides are bound in the dianionic form to a dipro-tonated active site. On the basis of the same bindingcurve taken in conjunction with kinetic relaxationstudies, Anderson et al (43) concluded that the mononu-cleotides are bound as monoanions to a monoproton-ated active site. From 31P nmr data, Gorenstein et al (44)claim that both mono- and dianionic forms of the mono-nucleotide inhibitors bind to ribonuclease. Haar et al(45) found that the pH titration curves of the mononu-cleotide-enzyme complexes tested, including 2'-AMP,3'-AMP, the isomeric cytidine phosphates (2', 3', 5), andmodified forms 2'-CMP oxide and 2'-deoxy-3'-CMP, aswell as 2'-AMP and 3'-UMP could only be analyzed interms of two pK values except for 2'-CMP, which couldbe explained with one pK, and 5'-CMP where the two pKvalues were too close to be resolved. Gorenstein et al(44) also found that two pK values were needed to ex-plain the pH titration curves of all the mononucleotidesthat they investigated, 2'-,3'-, and 5'-CMP and 3MJMP.The primary difference between the two groups is thatHaar et al (45) proposed that the second pK derivedfrom 31P data is due to the protonation of histidine-12(mistakenly referred to as 119), because similar pKvalues are derived from 1H nmr titrations. The data ofGorenstein et al (44) yield different pK values, and theyclaim greater accuracy for their method, namely deter-mining chemical shifts of inhibitor-enzyme complexesby titration and extrapolation under fast exchange con-ditions rather than the direct method with a single 1:1complex. The direct method assumes complete com-plex formation at all pH values, and the criticism of thevalues of Haar et al (45) resides in the lower associationconstants at extremes of the pH range. On the otherhand, acceptance of the calculations (44) of the fourmicroscopic ionization constants, i.e., KpH(EH), KpH(E),


Vol. 1,No. 1 45

KEH(P) from the macroscopically observedpK values, rests on the assumption that only the ioniza-tion state of the phosphate determines its 31P chemicalshift. Before the controversial aspects of the interpreta-tion of these data can be resolved, the data from the dif-ferent laboratories must be in agreement. It is obviousthat the 31P nmr experiments on the two hydrolytic en-zymes, alkaline phosphatase and ribonuclease, haveanswered several important questions but have openedup several new ones.

2. Phosphoryl Transfer Enzymes

A substantial portion of the 31P nmr work on phos-phoryl transfer enzymes has been performed on en-zyme-bound substrate complexes. The enzymesstudied are arginine kinase from American lobster tailmuscle (46, 47), rabbit muscle creatine kinase (48), carpand porcine muscle adenylate kinase (49, 50), rabbitmuscle pyruvate kinase (51), and 3-phosphoglyceratekinase from yeast (52). All these reactions are of thetype

ATP + S =: ADP + SP (5)

where S is the second substrate, and all require adivalent cation normally Mg2+ as an obligatory compo-nent; pyruvate kinase requires the presence of amonovalent cation K* in addition to Mg2+. In the ex-periments on enzyme-bound substrates, the enzymeconcentrations chosen (3 to 5 mM) were in sufficient ex-cess of those of the substrates (2 to 4 m/W) so that 80 to90% of the substrate is in the enzyme-bound species ifthe relevant dissociation constants are below ca 200/iM.This condition is easily met for the substrates of the en-zymes listed above except for some of the substrates ofcreatine kinase and pyruvate kinase. The results ofthese experiments include measurements of nmr spec-tral parameters, chemical shifts, and spin-spin couplingconstants involving the 31P nuclei in the substrates, andequilibrium and kinetic parameters pertaining to the in-terconversion of reactants and products on the surfaceof the enzyme. We consider first the results on individ-ual enzymes and later make a comparison of some ofthe results for different enzymes.

For arginine kinase (53), S in Eq'(5) is arginine. Figure2A shows the 31P nmr spectrum of an equilibrium mix-ture of the arginine kinase reaction established withcatalytic quantities of enzyme (46). The signals of thesix phosphate groups contained in the reactants andproducts are labeled. The -y-P (ATP) signal is a doubletdue to spin-spin coupling with the 31P nucleus in (3-P(ATP). The a-P (ATP) is also a doublet due to couplingwith (3-P (ATP). The /3-P (ATP) signal is a 1:2:1 tripletbecause of simultaneous spin-spin coupling with a-P(ATP) and ?-P (ATP) with coupling constants that are in-

(ATP)

Equilibrium Mixture

[Nucleotide] / [E ] = 4000

P-Arg/3-P(ADP)

a-P(ATP)

a-P|(ADP)

(ATP)

Equilibrium Mixture

[Nucleotide] / [E] = 0.97

P-Arg*/3-P (ADP)t y.p(ATP)

/3-P(ADP)

P-Argy-P(ATP)

a-P(ATP « ADP)

B+EDTA

/3-P(ATP)

' 10 2

1 14 6 8

CHEMICAL

1—| i

10 12

SHIFT

1 ' .14 16

(ppm)

—\—' 1 1 '18 20

Figure 2. 31P nmr spectra of the arginine kinase reaction(pH = 7.25; T - 12°C): A) equilibrium mixture of overall reaction,catalytic concentration of enzyme; B) equilibrium mixture ofenzyme-bound substrates and products; C) same as B withEDTA added to chelate Mg2+ and stop reaction, no chemicalexchange.

distinguishable. The /3-P (ADP) and a-P (MgADP) signalsare also doublets due to their mutual spin-spin coupl-ing. The a-P signals of ATP and ADP are both ratherbroad due to a small spin-spin coupling of the 31Pnucleus in a-P with the two 5' protons. From the areasunder the different signals in Figure 2A, the catalyticequilibrium constant

Keq = [S,] [S2]

may be readily evaluated. At pH 7.2 and T = 12°C, K^= 0.1 for arginine kinase.

The spectrum in Figure 2B is obtained by setting upan equilibrium mixture of the reactants and products ofarginine kinase such that all the substrates are bound


46 Bulletin of Magnetic Resonance

to the enzyme. It can be readily shown that most of theline broadening in Figure 2B is caused by the participa-tion of almost all the components of the system in thereaction most of the time. By adding EDTA to the sam-ple of Figure 2B to sequester the Mg2* from the enzyme-bound complexes, the reaction was stopped. It is clearfrom Figure 2C that binding to the enzyme in the ab-sence of reaction does not cause excessive broadeningof 31P signals. It may also be seen from Figure 2C thatthe 31P chemical shifts of the different phosphategroups, with the exception of /3-P (ADP), remain essen-tially unaltered by binding of the substrates to the en-zyme. The position of /3-P (ATP) at a higher field than inFigure 2B is due to the removal of Mg2+; the same phe-nomenon occurs for free ATP. Furthermore the equilib-rium of the reaction of enzyme-bound substratesappears to favor P-arginine much more than is the casefor catalytic levels of the enzyme. (Compare Figure 2Bwith 2A.)

The three limiting conditions of exchange defined inFigure 1 readily explain the line shapes in Figure 2B.The a-P (ATP) s a-P (ADP) exchange is in the fast-exchange limit due to the small chemical shift betweenthe signals, and a single resonance is obtained for boththese signals. The /3-P (ATP) - /3-P (ADP) exchange is inthe slow-exchange limit because of the large chemicalshift between the respective signals. The 7-P (ATP) ~P-(P-arginine) falls in the intermediate-exchange conditionaccompanied by both a line broadening and a frequen-cy shift.

From the line width of the /3-P (ATP) signal in Figure2B relative to that in Figure 2C, the lifetime of theE'MgATP-arginine complex in the reaction can bereadily determined (cf Eq 2). Furthermore from the areasunder the superposed a-P signals of ATP and ADP andthe isolated /3-P (ATP) signals, the ratio of the concentra-tions [E«MgATP-arginine] and [E«MgADP»P-arginine]may be obtained. Thus the equilibrium constants andexchange rates associated exclusively with the inter-conversion step of the reaction E»S,»S2 — E»P,»P2

are obtained from the experiment in a rather straightfor-ward manner.

The ability to isolate and monitor exclusively the stepof interconversion of the reactants and products on thesurface of the enzyme is one of the attractive featuresof the 31P nmr of enzyme-bound substrates of thesereactions. It must be noted, however, that the accuracyin the determination of the different parameters,equilibrium constants, and exchange rates is primarilygoverned by the feasibility of having the substratespresent only in the active complexes, e.g., in the case ofFigure 2B all enzyme-bound complexes should be ineither of the two forms E»Mg ATP* arginine orE* MgADP* P-arginine. Departure from this condition

leads to the presence of a substantial fraction ofenzyme-bound complexes that do not undergo the reac-tion. In such cases the interconversion rates deter-mined from the spectra may only be taken as lowerlimits of the rates.

A comparison of the interconversion rates with theoverall rates of the reaction obtained under the sameconditions, by enzymatic assay, allows one to concludewhether the interconversion step is the rate-limitingstep for the reaction. The equilibrium constants of theenzyme-bound reactants and products are of physio-logical relevance in situations where the enzyme andsubstrate concentrations are comparable. For the spe-cific case of arginine kinase the equilibrium constant

_ [E* MgADP* P-arginine] _eq ~ [E*MgATP*arginine]

is an order of magnitude different from K^. The inter-conversion rates in either direction are an order ofmagnitude faster than the overall rate of the reaction,indicating that the interconversion is not the rate-deter-minating step of the arginine kinase reaction.

The pH dependence of the chemical shift of /3-P ofenzyme-bound MgADP (47) was compared to that offree MgADP, which has a pKa of ca 6.0. However, thepKg of |8-P (MgADP) bound to arginine kinase is ca 7.5.Furthermore, the variation of chemical shift with pH isopposite in direction to that in free MgADP. This varia-tion suggested the possibility that the pH dependenceof the chemical shift of MgADP bound to the enzymemay actually be due to the titration of one of the aminoacid residues on the protein in the proximity of the /3-Pof bound MgADP. The pKa of the pH profile of the enzy-matic activity of arginine kinase is about ca 7.35. Theamino acid residue in the proximity of /3-P (MgADP),which governs the apparent pKa of this group on theenzyme, may have a crucial role in the activity of theenzyme.

Experiments have also been performed on the transi-tion state analog complexes E* MgADP* NO;»arginineof arginine kinase (47). It was shown that when all thecomponents of the above complex are present thechemical shift of /3-P (MgADP) in the complex is in-termediate between those of E* MgADP and E*MgATP.

The 31P nmr results of enzyme-bound substrates ofcreatine kinase (54) are broadly similar to those on argi-nine kinase (48). The only significant chemical shift oc-curs for the /3-P of MgADP, the equilibrium on the en-zyme favors greater phosphocreatine production com-pared to the overall reaction, the interconversion ratesof enzyme-bound reactants and products are not ratelimiting. (See summary of chemical shifts presented inTable I.) Arginine kinase is considered to be the inverte-

Vol. 1,No. 1


47

Table 1.31P Chemical Shifts of P-enolpyruvate and itsComplexes with Pyruvate Kinase at pH 8.0.

Added Components

None

Pyruvate kinase

Pyruvate kinase + K*

Pyruvate kinase + K+ +

Pyruvate kinase + K+ 4

Pyruvate kinase + K+ +(equilibrium mixture)

Mg2t

ADP

ADP + Mg2+

5 (ppm)

0.9

0.2

0.2

-2.1

-3.1

-3.6

brate analog of oreatine kinase and the similar behaviorpattern is consistent with this idea. However the chem-ical shift of the /3-P of MgADP bound to creatine kinaseis independent of pH in the range 6.0 to 9.0, in contrastwith the result for arginine kinase discussed earlier.

The chemical shift of the /3-P (MgADP) in the transi-tion state analog complex increased in the direction ofMgATP in experiments with creatine kinase in the samemanner as for arginine kinase. The change in chemicalshift was observed with nitrate as well as formate ions.Furthermore, the /3-P (MgADP) resonance shows a split-ting in the broad resonance indicating a conformationalheterogeneity of the bound nucleotide at the active site(55). This effect was observed more clearly for creatinekinase than for arginine kinase.

In the case of creatine kinase, the dissociation con-stants of the different substrates from their enzyme-bound complexes are rather large, leading to the pres-ence of substantial concentrations of substrate mole-cules not present in the enzyme-bound form in these ex-periments. The analysis of the experiments, althoughsimilar, is not as straightforward as in the case ofarginine kinase.

The adenylate kinase reaction ATP + AMP ^ f2ADP(56) is particularly well suited for study by 31P nmr sinceall four substrates contain 31P nuclei. Of particular in-terest in the case of this reaction was the attempt toestablish the distinction, if any, between the acceptorand donor ADP molecules that are converted to ATPand AMP respectively in the reaction and to pinpointthe possible role of Mg2* in effecting such a distinction(49). The presence of two identical ADP molecules assubstrates provides an additional chemical exchangeprocess that is unique to this enzyme. This exchangebetween the acceptor and donor ADP molecules (see

Figure 3) is unrelated to the chemical exchanges occur-ring during the adenylate kinase reaction, but should beconsidered in any detailed interpretation of the spectra.

The key experiment that provided the clue for thepossible distinction between the donor and acceptorADP is illustrated in Figure 3. The 31P nmr spectrum ofan equilibrium mixture of enzyme-bound substrates ofporcine adenylate kinase is obtained after ca 4 h of nmrsignal averaging as shown in Figure 3A. In contrast tothe arginine kinase experiments, the spectrum in Figure3A is difficult to explain, particularly since noresonances are seen either in the region where AMPresonates (ca -4 ppm) or in the region for /3-P of MgATP(ca 19.0 ppm). Note the presence of a small Pi peak at ca-1.5 ppm indicating that some ATP hydrolysis occurredduring the signal accumulation.

Attempts to obtain a signal to noise ratio better thanthat in Figure 3A by signal averaging for ca 12 h on thesame sample yielded the spectrum shown in Figure 3Bwhich appears quite different from that in Figure 3A.The Pj resonance is considerably enhanced, there is anAMP resonance, and there are now two resonances inthe region of the /3-P resonance of ADP (4-7 ppm) in-stead of one at 6.5 ppm as in Figure 3A. The appearanceof the spectrum in Figure 3B may be explained bynoting that as the irreversible ATP hydrolysis proceedsconcomitantly with the adenylate kinase reaction, theconcentration of ATP is progressively depleted whilethe concentrations of AMP and Pj increase. Theresonance of /3-P (MgADP) is in slow exchange with thatof /3-P of MgATP in the adenylate kinase reaction as inthe arginine kinase reaction. As the ATP is progressive-ly depleted a greater fraction of MgADP would befound in abortive enzyme-bound complexes, e.g.,E»AMP«MgADP complexes. The MgADP resonance isthen effectively narrowed and becomes observable.

The above explanation implies that in Figure 3A theresonance of MgADP is too broad to be observedbecause of chemical exchange in the presence of suffi-cient amounts of ATP in the sample. To verify this inter-pretation, a small quantity of ATP was introduced intothe sample of Figure 3B and the 31P nmr spectrum wasrecorded again. In the resulting spectrum, shown inFigure 3C, the MgADP resonance is significantlybroadened. There should then be an appreciable con-centration of ATP in the sample of Figure 3C althoughno resonance is observable in the vicinity of 19 ppm.Apparently, this resonance is still too broad to be ob-servable when turnover is occurring. However, additionof a sufficient quantity of EDTA to the sample of Figure3C to remove Mg2+ from the reaction complexes, stopsthe reaction. The spectrum of the resulting sampleshown in Figure 3D does exhibit an unmistakable /3-P(ATP) resonance.



It may also be seen from Figure 3D that the two 0-P(ADP) signals of Figure 3B merge into a single peak.This result, along with several other experiments, wasused to show that a single resonance is obtained for thetwo /3-P (ADP) groups due to fast exchange between theidentical ADP molecules at the acceptor and donorsites on the enzyme. The rate of this exchange isseverely reduced in the presence of Mg2+ leading to twodistinct resonances in Figure 3B. The evidence stronglysuggests that MgADP does not bind at the donor ADPsite. Furthermore, independent experiments on ATPbound to the enzyme in the presence of varying concen-trations of Mg2+ show that while ATP or MgATP bind atthe ATP site, only metal-free ATP binds at the AMP site.It has also been shown that the optimal Mg2+ concentra-tion for the reaction is [Mg2+] = [ATP] + 1/2[ADP], i.e.,one metal ion per reaction complex. It is quite evidentthat in addition to facilitating catalysis of the reactionas in other kinases, the Mg2+ provides the distinctionbetween acceptor and donor ADP molecules onthe enzyme.

The equilibrium constant Keq for the adenylate kinasereaction at catalytic enzyme concentrations determinedfrom 31P nmr is ca 0.4 at 4°C and pH = 7.0; K'eq =[E»ADP* MgADP]/[E»AMP» MgATP] is estimated to beca 1.5. The equilibrium is thus appreciably shiftedtowards ADP in the enzyme-bound species. Theinterconversion rates in either direction are an order of,magnitude faster than the overall rates of the reactionand are, therefore, not rate limiting.

Diadenosine pentaphosphate (Ap5A, a symmetricalmolecule) is a potent inhibitor of the porcine adenylatekinase reaction. The distinction between the acceptorand donor ADP bound to adenylate kinase establishedin the experiments described above, suggested that inthe enzyme-inhibitor complex Ap5A may not be symme-trical. This asymmetry is clearly depicted in the 31P nmrspectrum of the E»Ap5A complex compared with thespectrum of Ap5A or Mg«Ap5A free in solution (seeFigures 4A, 4B, and 4C). In the spectrum of the E»Ap5Acomplex (Figure 4C) the resonances of 1-P and 5-P(ca 11 ppm) are resolved and those of the three middlephosphate groups 2-P, 3-P, and 4-P are also distinctfrom each other. The resonance of 3-P is in the middleof the three upfield resonances. The resonances of 2-Pand 4-P, on either side of the 3-P resonance, are broad-ened because 2-P and 4-P can interchange when Ap5Adissociates from the E«Ap5A and binds again. This ex-change preserves the identity of 3-P. Addition of Mg2+ tothe sample of Figure 4C highly accentuates the asym-metry of the enzyme-bound Ap5A in the E»MgAp5A com-plex (Figure 4D). It has been shown (57) that only onemetal ion binds readily to an Ap5A molecule. All fivephosphates in this complex resonate at clearly distinct

E + ATP + AMP + MgCL (<- 4hrs)

P i /3-P (ADP)y-P(ATP)

APP + PPAMg

i\nAP + PPPAMg

I2hrs)

/3-P(ADP)

o-P

AMP B + ATP

AMP

r-p a-P(ATP) ( A D P )

fi-P(ADP)

C+EDTA

/3-P(ATP)

i•8

i i-4 0 4 8 12 16 20

CHEMICAL SHIFT (ppm)24 28

Figure 3. 31P nmr spectra of an equilibrium mixture of sub-strates and products fully bound to adenylate kinase (porcine;T = 15 °C; pH = 7.0): A) after 4 h of accumulation; B) same after12 h; separate resonances for each ADP bound to enzyme; C)same as B with addition of exogenous ATP; D) same as C withEDTA added to stop reaction as in Figure 2C.


Vol. 1,No. 1 49

AP5A (Free)

0-P's and y-P

a-P't

APSA(Enzyme bound)

I ' l ' l ' l ' l ' l ' l ' l6 10 12 14 16 18 20 22

1 I ' I ' | • ' "l2 4 26 28 30

CHEMICAL SHIFT (ppm)

positions. The 2-P and 4-P resonances are no longersignificantly broadened compared to the 3-P resonance.Since as mentioned earlier the enzyme has a preferredsite for metal-bound nucleotides, the exchange of 2-Pand 4-P (and 1-P and 5-P) in the E»MgAp5A complex can-not proceed with the same ease as before the additionof Mg. This exchange is now slower because an addi-tional step has been added; after MgAp5A dissociates

Figure 4.31P nmr spectra of Ap5A and MgAp5A free and boundto adenylate kinase (porcine; T = 4 °C; pH = 8.0): A) free Ap5A; B)free MgAp5A; C) Ap5A bound to adenylate kinase; D) MgAp5Abound to adenylate kinase; E) same as D with Mn(ll) added(1:50:: Mn:Mg).

from the enzyme, the metal ion may shift from one sideof Ap5A to the other before MgAp5A reassociates withthe enzyme again. Attempts to assign 2-P and 4-Punambiguously by the addition of Mn2+ to the sample ofFigure 4D (see Figure 4E) have not yielded definitiveresults. The chemical shift of ca 7 ppm between the 2-Pand 4-P resonances in Figure 4D is one of the largest 31Pchemical shifts observed thus far for enzyme-boundsubstrates (noncovalent) and is comparable to theshifts observed in alkaline phosphatase (26).

Of the various kinases studied thus far adenylatekinase is the only enzyme that is found to produce a sig-nificant change in the spectral parameters associatedwith ATP in different enzyme-bound complexes. Thetwo 31P spin-spin coupling constants (a-P and 7-P to /3-P),which are normally equal (ca 20 Hz) differ by ca 4 Hz inthe E»ATP complex. Furthermore the /3-P resonance inthe E»MgATP complex is shifted downfield by ca 1.5ppm.

Brown and Ogawa (58) performed a saturation trans-fer experiment on equilibrium mixtures of the adenylatekinase reaction to determine some of the rate con-stants. In these experiments the magnetization of thesuperposed a-P resonance of ADP and ATP is invertedby a selective 180° pulse, and the transfer of this satura-tion to the AMP resonance was monitored at various in-tervals after the inversion (see Figure 5). The rate of thistransfer of saturation depends on the rates of intercon-version of ADP and AMP as well as the rates of disso-ciation of AMP and ADP from their enzyme-bound com-plexes. The analysis of these data along with line widthand K'eq measurements allows the evaluation of all therate constants. The rates obtained by this method agreewith the results of bound substrate experiments, butK'eq values do not. The technique of saturation transferhas the potential of being very useful in assignment ofresonances as well as measurement of rate constants.

The enzyme 3-phosphoglycerate kinase (59) has anequilibrium constant strongly in favor of formation ofATP and 3-P-glycerate from ADP and 1,3-bis-P-glycerate,Keq = 3 X10"4.31P studies of the equilibrium mixture ofthis reaction with enzyme-bound substrates and pro-ducts indicates K'eq = 1. The 31P experiments also pro-vide distinct evidence for ATP binding to the enzyme attwo sites; at one of the sites, probably the noncatalyticsite, Mg2+ binds to the E«ATP complex much moreweakly than to the other. The resonances of ATP orMgATP are not shifted significantly by binding to the



-5 0 5 15 2 5 ppm

-5 0 5 15 2 5 ppm

Figure 5. An inversion transfer sequence for the a-31P of ADP and ATP in the adenylate kinase reaction.

enzyme. Significant upfield chemical shifts from freeMgADP were observed for the /3-P in the E»MgADP com-plex. Of the kinases studied thus far, 3-P-glycerate kin-ase is the only enzyme that shows an upfield chemicalshift of the bound nucleotide 31P resonances. Chemicalshifts (ca 1.5 to 2 ppm) were also observed between thefree and enzyme-bound resonances of 3-P-glycerate and1,3-bis-P-glycerate. Sulfate ion has a pronounced effecton the 31P resonances of the equilibrium mixture ofbound substrates (60, 61). The chemical shift of theresonance from 3-P-glycerate changes back to its valueoff the enzyme, the equilibrium is shifted towards ATP,and the interconversion rate is reduced.

Among the four substrates of pyruvate kinase (62),ATP, ADP, and pyruvate bind the enzyme rather poorly;their dissociation constants are in the millimolar range.P-enolpyruvate binds the enzyme very tightly with adissociation constant of 50 ̂ M. With enzyme concentra-tions in the range of 4-5 mM and substrate concentra-tions in the range 2-4 mM, appreciable fractions of thesubstrate are not in the enzyme-bound form. The analy-sis of 31P nmr results obtained on this reaction is sub-ject to the limitations arising from this condition. Inspite of this limitation, an appreciable amount ofqualitative information is obtained from the study.

At catalytic concentrations of the enzyme the pyru-vate kinase reaction has an equilibrium constant

- [ADP] [P-enolpyruvate] _~ [ATP] [pyruvate] =

1 f V1 U

strongly in favor of ATP. A striking result of the 31P nmrexperiments on bound substrates was that an equilib-rium mixture with enzyme concentrations in excess ofthe substrate concentrations yielded an equilibriumconstant K'eq s 1.

The equilibrium constant of enzyme-bound sub-strates and products was also studied for another re-action catalyzed by pyruvate kinase (63): ATP +glycolate ~ ADP + P-glycolate. For this reaction Keq >ca 50 in favor of ADP and P-glycolate. Once again K'eq

= 1. Thus for two reactions catalyzed by pyruvatekinase for which the values Keq differ by ca 105, bothvalues of K'eq are about unity. Furthermore, for all thekinases studied thus far K'eq is ca 1 regardless ofwhether Keq is strongly in favor of one side of the reac-tion (as in 3-P-glycerate kinase and pyruvate kinase) oronly moderately so (as in arginine kinase, creatinekinase, and adenylate kinase). A value of K'eq of unityimplies that the free-energy change in the interconver-sion step is very small so that the overall free-energychange in the reaction occurs in the remaining steps ofthe reaction, probably the association and dissociationsteps. Enzyme systems other than kinases exhibitsimilar behavior (64, 65). Whether this result proves tobe a general feature in enzyme catalysis awaits moreextensive data.

The phosphate groups of ATP and ADP do not showany significant chemical shift upon binding the enzyme.The largest chemical shift upon binding the enzyme oc-


Vol. 1,No. 1 51

curs for the 31P resonance of P-enolpyruvate. As shownin Table 2, the change in 5 from the dianionic form ofP-enolpyruvate in solution increases progressively aseach subsite of the active site is progres-sively occupied. However, 5 is not affected by theobligatory monovalent ion K*, but is strongly affectedby Mg2* although Mg2* is not directly coordinated toP-enolpyruvate.

The titration of the chemical shift of 0-P of enzyme-bound ATP with Mg2* shows that more than two Mg2+ e-quivalents are required for saturation, indicating ametal binding site on the enzyme in addition to that onATP at the catalytic site (66-68) and possibly a secondbound ATP that has a much weaker affinity for Mg2* (69).No effects that may be attributed to the obligatory K*ion were observed on the 31P chemical shifts or line wid-ths. Gupta and Mildvan (68) performed 31P experimentson the pyruvate kinase system with ca 1.5 mM enzyme-site concentration and equimolar ATP or ADPconcentrations. These experiments were interpreted asevidence for a second metal binding site on the enzymein addition to that on the nucleotide.

Summary of Chemical Shifts: The chemical shifts ofATP and ADP in enzyme-bound complexes of differentkinases are summarized in Table 1. ATP chemical shiftsare practically unaffected by binding to the enzymeswith one exception, viz, j8-P of MgATP bound toadenylate kinase. /3-P (ADP) is the only nucleotideresonance that shifts on all enzymes except pyruvatekinase. While there is a low field shift in most cases, anupfield shift occurs on 3-P-glycerate kinase. Thus the31P chemical shifts of nucleotide-kinase complexes donot reveal a systematic change that might signify grosssimilarities in the environment or conformation (or both)of the phosphate chain in these complexes. The 31P-31Pspin-spin coupling constants in the enzyme-bound com-plexes of ATP and ADP are equal to those of thenucleotides free in solution within experimental error,with the exception of ATP in its complex with adenylatekinase for which a change of ca 25% is observed.

3. Other Enzymes

In this section enzymatic reactions with phosphate-containing substrates are included, but the phosphategroups are not cleaved or transferred in these reactions.For example, in hydrogen transfer reactions catalyzedby dehydrogenases, several aspects of the interactionof the pyrophosphate moiety of the oxidized and re-duced forms of the coenzymes NAD or NADP and of theadditional 2-phosphate of the latter have been investi-gated. For dogfish lactate dehydrogenase (70) whereNAD is known from X-ray crystallography to interactwith an arginine residue of the enzyme, the 31P reso-nance of the bound coenzyme shifts downfield ca 1.9ppm from the free coenzyme position, but for lobstermuscle glyceraldehyde-3-phosphate-dehydrogenase inwhich a lysine residue is involved in the interaction, the31P resonance of the bound coenzyme shifts upfield.The lactate dehydrogenase complexes with reducedand oxidized coenzymes respectively yield similar shiftsand line shapes, which is rather surprising since thebinding constant for NADH is a thousandfold greaterthan for NAD. At high enzyme concentrations, twopeaks are observed in spectra of NADH; one in slow ex-change between bound and free forms and the other infast exchange. The latter persists at low enzyme con-centrations. The existence of two aggregated forms ofthe enzyme is invoked by the author to explain all theobserved shifts of NADH. Aggregation can be a prob-lem at high enzyme concentration, but can also bechecked independently should interpretation of the nmrresults necessitate it. Aggregation is apparently not aproblem with the glyceraldehyde-3-phosphate-dehydro-genase where both free NAD and NADH are in slow ex-change with the bound forms independent of enzymeconcentration.

Some interesting results have been found in the inter-action of NADP and NADPH with Lactobacillus caseidihydrofolate reductase (71). More recently (72) the inter-action of dihydrofolate reductase with 2'-AMP (the P-ad-

Table 2.31P Chemical Shifts of Free and Enzyme-Bound Nucleotides at pH 7.0 (ppm from 85% H3PO4).

Enzyme

None (-Mg)( + Mg)

Arginine kinaseCreatine kinaseAdenylate kinasePhosphoglycerate kinasePyruvate kinase

*Second ADP (Mg) visible

Mg

a-P

10.89.9

11.011.010.211.010.0

under conditions of slow

ADP

0-P7.55.93.33.83.5*, 6.77.55.7

exchange

a-P

11.010.811.010.910.711.010.9

Mg ATP

/3-P

21.819.219.419.017.819.419.2

r P

7.35.65.65.46.16.05.5

Duplication of Bulletin of Magnetic Resonance, in whole or in part by any means lot any purpose is illegal.


Table 3.31P Chemical Shifts of Oxidized and ReducedCoenzyme NADP and NADPH

in the Dihydrofolate Complexes(Reference, 6 = 0. 50 mM KH2PO4, pH 8.0).

Species

NADPHE-NADPH

NADPE-NADP

5 (ppm)

2'-P-0.47-2.66

0.22-2.72

Pyrophosphate13.78*

13.94 16.4714.15 14.4714.32 16.23

*The two 31P resonances of the pyrophosphate moietyhave the same chemical shift.

enosyl moiety of NADP) has also been investigated. Thechemical shift data are summarized in Table 3. Twosets of signals are observed for enzyme plus NADPHbecause of slow exchange. The 2'-P shift in the enzymecomplex is independent of pH in the range between 4.5and 7.5 and has a value of 1.7 ppm downfield from thefree dianionic species. The pyrophosphate shifts and -Jppof bound NADPH are also independent of pH. In ad-dition to the asymmetry introduced in the chemicalshifts of the two 31P peaks of the pyrophosphate moietyof NADPH upon binding to enzyme (see Table 3), anasymmetry is introduced in the 31P-1H coupling con-stants as well.

The bound form of 2'-AMP is in fast exchange withthe free form. Furthermore the pH dependence of theshift indicates that both mono- and dianionic forms canbind although the latter binds 16 times more strongly.The pKa is lowered from 6.0 to 4.8 upon binding. Thesimilarity of the 2'-P shift in the bound forms of NADPand NADPH suggests binding at the same site and thesame state of ionization. The increased downfield shiftof 2-P in the coenzymes (ca 1.5 ppm) over that in bound2'-AMP indicates environments are different. Theauthors suggest that it may be a change in P-O-P angle,yet a thousandfold increase in binding between NADPand NADPH is not accompanied by a change in 2'-P en-vironment as reflected in chemical shift.

For two other enzymes, glycogen phosphorylase dis-cussed earlier and triose phosphate isomerase, whichcatalyzes the reversible isomerization of dihydroxyace-tone-P to glyceraldehyde-3-P, the question of the chemi-cal species that interacts with the enzyme was address-ed. With glycogen phosphorylase b for the effector glu-cose-6-P, which exists in solution as 40% a-anomerand 60% /3-anomer each with a characteristic 31P reso-nance, it was found that only the a-anomer resonancewas affected in the presence of enzyme. Thus, earlierdeterminations of binding constants using total

glucose-6-P concentrations were in error. For triosephosphate isomerase, the question raised was whetherthe hydrated form of dihydroxyacetone-P is a substrateor inhibitor of the enzyme (73). The answer is neither; on-ly the 31P resonance of the keto form broadens (from 1to 14.6 Hz) upon binding and also shifts downfield 0.3ppm. The keto and hydrated forms are in slow ex-change, and the dissociation constant of the hydratedform is at least ten times that of the keto form. Thebound and free species of the keto form of dihydroxy-acetone-P are in fast exchange (74), and the pKg valuesare about the same. The line width depends on pH, andthe authors conclude that the rate being detected,/CS2X10V1 is the rate of ionization. The inhibitorglycerol-3-P behaves like the substrate, but the inhibitor2-P-glycolate behaves differently since it is in slow ex-change and its chemical shift is not pH dependent.

A further investigation of the binding of the inhibitor2-P-glycolate (75) revealed that the chemical shift is thesame as that of the dianionic form of the inhibitor freein solution and does not vary between pH 5.5 and 8.5;pKg (free) = 6.4. Since the 13C resonance of carbon-1 ofbound 2-P-glycolate showed no change between pH 7.4and 8.5 (although the free form shows a small effect atpH 6.5, pKa = 3.5), the authors conclude that the boundform is neither P2~ nor COO~P" but a trianion and sug-gest that the enzyme is protonated in the binding pro-cess. The conclusion is once again based on the ques-tionable assumption that the chemical shift in thebound form is identical to that of the correspondingionic species in the unbound form.

D. Detection of Intermediates: Oxygen-18 Shiftof 31P Resonance

The recent finding of an observable shift of the 31Presonance when 18O is substituted for the normallybonded 16O (5, 76) led to the initiation of new types ofmechanistic investigations using 31P nmr. As pointedout by Cohn and Hu (76), the resolution of all five 31Presonances in a randomized sample of 50% 18O-phos-phate (five species: 18O0l1,2,3,4) not only yields the ratio of18Q/16O, but the 18O-labeled phosphate can serve as atracer in following reactions involving phosphategroups. Obviously the fate of phosphate oxygen can al-ways be followed and the fate of phosphorus in allcases except under those unusual conditions when thephosphate oxygen exchanges with solvent water. Con-sequently it is now possible with 31P nmr to follow label-ed phosphorus or labeled oxygen continuously as re-actions proceed.

The magnitude of the isotopic shift is small (0.0206ppm per 18O for the dianion of inorganic orthophos-


Vol. 1,No. 1 53

Table 4. Effect of ia0 on the 31PChemical Shifts of ATP.

P of ATPa

a

7

Oxygen Positiona-fi bridge

a-|3 bridgenonbridgep-y bridge

(3-7 bridgenonbridge

AS (ppm)per 18O atom

0.017

0.0170.0280.016

0.0160.023

phate) and varies depending on the chemical environ-ment or the particular oxygen. For example, as shownin the compilation of Table 4, the isotopic shift pro-duced by the bridge oxygens on the p-P of the poly-phosphate chain of ATP (18O) is much smaller than thatproduced by the nonbridge oxygens (77). The ability toresolve these small shifts for 18O may be a limitation onthe usefulness of this probe. The shift between the 16O4and 18O4 species can be resolved easily at 24.3 MHz (5).At high frequency (145.7 MHz) there is no difficulty in re-solving 31P18O4 from 31P18O316O, permitting direct quan-titation from peak heights but such direct quantitationis not possible at 24.3 MHz. Successful resolution andquantitation was effected at 40.5 MHz (78) by the use ofa Curve Resolver set for Lorentzian curves. In this sameinvestigation (78), the nmr method for determining rate

constants of a phosphate (18O)-H216O exchange reac-tion was found to be in good agreement with the con-ventional noncontinuous mass spectrometric analyticalmethod for following the kinetics of the reaction. Itshould be pointed out that narrow lines are essential forthe resolution of 31P-18O shifts. Paramagnetic impuritiesmust be sedulously avoided or eliminated by chelatingagents. If line broadening due to chemical exchangeduring the course of the reaction being investigated ob-scures the isotopic shift, then the reaction can no long-er be followed continuously; useful spectra can only beobtained by stopping the reaction at each time point.

Several types of enzymatic reactions involving phos-phates have been probed with the isotopic (18O) shift of31P including an ADP-Pj exchange reaction catalyzed bypoiynucleotide phosphoryiase (5), an ATP-PPjexchangereaction catalyzed by valyl tRNA synthetase (78), anATP-ADP scrambling reaction catalyzed by pyruvatekinase (80), and phosphate (18O)-H2O exchange cata-lyzed by inorganic pyrophosphatase (5), myosin ATPase(80), alkaline phosphatase (81), acid phosphatase (78),and fructose-1,6-diphosphatase (83). Information maybe derived from two parameters: 1) the change of 18Ocontent with time, which yields the rate of the reaction,and 2) the distribution of the 18O among the variousphosphate (18O) species at any one point in time. Thelatter yields the site of bond cleavage or may estab-lish the existence of an undetectable intermediate or

Figure 6.31P nmr spectrum of the equilibrium mixture of the ADP-Pj exchange in the poiynucleotide phosphoryiase reaction. Initialspecies ADP (16O) and Pj (93.4% 18O); ADP:PJ::3:2.

a-P(ADP)

0T2 4 6 8

CHEMICAL SHIFT (ppm)io I2

lI4



0 * 0A-0-P-0-P-«-F>-0

i i i0 * 0

+ Vol0

A-O-P-Val6 6

OA-O-P-Val

6? t

0-P-»-P-0I I

0 •

11 I0 0 0

A-0-P-»-P-»-P-0i i i0 • 0

Val0

A-O-P-ValiO

Figure 7. Partial reaction of Val tRNA synthetase with ATP (/3-18O3).

establishes the ratio of rates of intermediate steps inthe overall reaction under observation.

The catalysis of a phosphate (18O)-H216O exchange re-

action by alkaline phosphatase, inorganic pyrophos-phatase, myosin ATPase, and fructose-1,6-diphospha-tase had been previously established. The catalysis ofthe exchange reaction by human prostatic acid phos-phatase was established only recently with the 31P nmrmethod (78). The question of 180 species distribution inthe PO4 (

18O4,3,2,1. o) as the reaction progresses was ad-dressed in all the studies. It has been shown by Eargleet al (83) by mass spectrometric analysis of phosphate(18O) derivatives at each time point in the alkaline-phos-phatase-catalyzed reaction that the distribution of 18Ospecies was random. It can be inferred that the rate ofdissociation /c., of Pj from the enzyme is rapid comparedto k2, the rate of formation of phosphorylated enzymeE-P, k2 as shown in the equation

E + = E»P =ft., k.2

E-P

This result was confirmed by 31P(18O) nmr (81) for thenative Zn-enzyme, but the Co-enzyme exhibited a differ-ent nonrandom pattern. In the latter case (81), the ratioof k2lk-i is 3 ±0.5. An extreme case was observed byWebb et al (80) with myosin subfragment 1, where com-plete exchange of bound Pj occurs before release of Pjfrom the enzyme so that from the initial P18O4, the onlyproduct observed is P18O4, and no mixed 16O18O speciesare observed during the course of the exchange. On theother hand, the acid phosphatase (78), like the nativealkaline phosphatase, yields a random distribution ofproducts. The fructose-1,6-diphosphatase (82), like theCo-alkaline phosphatase, is intermediate yielding anon random distribution of 18O species with kjk^ = 2for the Mn(ll)-activated reaction, and the ratio is 1.4

o o *A-0-P-0-P-«-P-0

or

0 0 01 I I

A-0-P-«-P-«-P-0i i i0 * 0

Val

Val

when Mg(ll) is the activator. In all cases, the rate of thePj-H2O oxygen exchange was also determined.

The study of the exchange reaction between Pj (18O4)and ADP (16O) catalyzed by polynucleotide phosphory-lase (76) exemplifies the use of the 18O shift to labelphosphorus. The rate of the exchange reaction can bedetermined and the appearance of an a-P(16O3

18O)- and a/3-P(18O4)-ADP species as shown in Figure 6 proves thatthe site of bond cleavage occurs between the a-P andthe o-jS bridge O (product A-O-P-18O-P-18O3).

Valyl tRNA synthetase is known to catalyze the par-tial ATP-PPj exchange reaction in the presence of va-line. The fate of 7-labeled ATP (three nonbridge 18Oatoms) and of /3-labeled ATP (two nonbridge, one 0-ybridge 18O) in the presence of valine and the adenyltransferring enzyme were investigated as a function oftime. The overall rate of appearance of the <8O from 7-phosphate to the ^-phosphate is a function of the rateof formation of the valyl-AMP intermediate and enzyme-bound PPj, the dissociation of PPj from the enzyme toform symmetrical PP (/3- and 7-P equivalent), and the re-versal of these steps as indicated in Figure 7. The ap-pearance of 18O from 7-P (18O) in the a-0 bridge positionconfirms the site of cleavage of ATP between the a-Pand a-/3 bridge oxygen which had previously beenestablished in the overall reaction. When the /3-P-labeled ATP (18O3, no 18O in a-(3 bridge) was followedunder the same conditions, the rate of approach of 18Oto its equilibrium value at the a-/3 bridge was about 20%greater than for the 7-P oxygen. With the /3-P(18O3)-labeled ATP as substrate, not only is there a contribu-tion to the a-/S bridge position from the dissociation andreversal of the unbound symmetrical PPj but also fromthe rotation of the nonbridge 18O of the enzyme-boundPPj to the bridge positive, i.e., scrambling.


Vol. 1,No. 1 55

An experiment with pyruvate kinase, a phosphoryl-transferring enzyme with |3-P(18O, no 18O in the /3—/bridge)-labeled ATP (79) showed that scrambling occur-red in the presence of the substrate pyruvate or an in-hibitor, oxalate, and was eliminated by phosphoenolpy-ruvate. The authors favor a metaphosphate intermedi-ate as the explanation of this phenomenon rather thana phosphorylated enzyme. Nmr spectroscopy is current-ly being applied (84) to the type of 18O scramblingexperiment introduced by Midelfort and Rose (85) todetect intermediates, i.e., bond cleavage of ATP by i8Oscrambling in ATP in a partial reaction catalyzed by en-zymes such as glutamine synthetase, where no otherpartial reactions can be detected because the ADPformed does not dissociate from the enzyme.

E. Active Site Structures

We shall now consider a few examples in which 31Prelaxation measurements in the presence of paramag-netic metal ions were used to deduce structuralinformation about the active site. Very often the para-magnetic relaxation data are collected for more thanone nucleus on the substrate, and distance measure-ments are grouped together to build a model that per-mits conclusions that are biochemically significant.One of the initial applications of this method was topredict a possible structure of the creatine kinase ac-tive site (16, 86). Measurements of distances of Mn(ll)from protons of creatine in the E»MnATP«creatine com-plex indicated that these distances are nearly the sameas in the abortive E«MnADP»creatine»NO3 transitionstate analog complex. This led to the conclusion that inthe equilibrium mixture on the enzyme the Mn(ll) ionmay be chelated to the a- and j3-phosphate groups ofATP.

A major part of the work in recent years on derivingstructural information on active sites of enzymes usingparamagnetic probes was contributed by Mildvan andco-workers. In all cases studied by this group, distancesof different nuclei of the substrates were calculated onthe basis of changes in their relaxation parameters inthe presence of enzyme at concentrations muchsmaller than those of the substrates, which necessi-tates extrapolation to the enzyme-bound form and valid-ating the condition of fast exchange under the experi-mental conditions used.

1. Pyruvate Kinase

Rabbit muscle pyruvate kinase was studied at somelength using Mn(ll), Co(ll), and CrATP as paramagnetic

probes. This enzyme represents a complicated examplesince there are two divalent cations per active complex.Distances between the paramagnetic atom and select-ed protons and phosphorus nuclei of ATP and of P-enol-pyruvate and carbon atoms and protons of pyruvate(87-92) were calculated. From these distances it wasconcluded that the phosphoryl groups of ATP andP-enolpyruvate are located in the second coordinationsphere of the enzyme-bound divalent cation. The role ofthis metal ion (in addition to that chelated to thenucleotide) for the activation of the pyruvate kinasereaction is unclear.

2. DNA Polymerase I

Thymidine-5'-triphosphate (dTTP) bound to DNApolymerase I from Escherichia coli was studied usingMn(ll) as a paramagnetic probe (93); 31P and 1H relaxa-tion rate measurements were made. A significant con-clusion made by the authors was that direct coordina-tion of Mn(ll) occurs only with -y-P of dTTP on the en-zyme, whereas coordination to a-P, /3-P, and -y-P occursfor Mn(ll)-dTTP complex free in solution. This feature in-dicates a possible distinction in the role of the divalentcation between a nucleotidyl transfer and phosphoryltransfer (94). The enzyme-bound Mn»dATP complex wasalso studied. It was suggested that the conformation ofthe enzyme-bound Mn»dTTP is similar to that requiredfor Watson-Crick pairing.

3. (Na+ + K+)ATPase

Membrane-bound (Na+ + Reactivated ATPase fromsheep kidney was investigated. With Mn(ll) as the para-magnetic probe (95), epr, proton, and 20STI relaxation ef-fects were used in conjunction. It was suggested that inthe presence of Na* the phosphate monoanion inter-acts with enzyme-bound Mn(ll) but that in the presenceof K+ the phosphate dianion interacts. Thus protonationof an enzyme-bound phosphate would convert a K+-binding site to Na+-binding site. The interaction of ananalog, methyl phosphonate, with enzyme-bound-Mn(ll)complex was shown to lead to a second-spherecomplex.

4. Phosphoglucomutase

To determine the distance between the bound metalion and the active-site serine phosphate of phospho-glucomutase from rabbit muscle, 31P nmr relaxationrates were measured (25, 96). Paramagnetic effects on31P line width (7"2) and spin-lattice relaxation (f,) due toNi(ll) were used to deduce a distance (4-6 A) suggestinga second sphere interaction between enzyme-bound

56



Ni2+ and 31P. Ni(ll) was used because of its weak para-magnetism compared to Mn(ll) since it was felt that adirect coordination between Mn(ll) and enzymic phos-phate would not be observable. The 31P spectrum of theserine phosphate at the active site of the phosphory-lated enzyme has been discussed in the section on en-zymes with covalently bound phosphates.

5. Inorganic Pyrophosphatase

Recently Hamm and Cooperman (97) made a detailed*H and 31P study of Pj binding to yeast inorganic pyro-phosphatase. Two phosphate sites per subunit withvery different affinities (dissociation constants 0.24 mMand 18 mM) were detected. The enzyme also requirestwo divalent cations with disparate affinities per sub-unit. Using Mn(ll) as a paramagnetic probe it wasshown that the weaker Mn(ll) site is far from both phos-phate sites, which are in close physical proximity. Thetighter Mn(ll) site is at a distance consistent with sec-ond sphere coordination with the weaker phosphategroup.

6. Glutamine Synthetase

The 31P spectrum of the AMP-tyrosine moiety on thisenzyme was discussed earlier in the section on en-zymes with covalently bound phosphates. Two metalions are required per subunit, one bound directly to theenzyme and another bound to the nucleotide in thereaction. Villafranca et al (34) used paramagnetic ef-fects on 13C and 31P nmr signals from (2-13C) AMP-ad-enylylated enzyme to measure the paramagnetic ef-

fects due to bound Mn(ll) ions. Analysis of the data ledthem to deduce the spatial relationship of the two Mn(ll)ions with respect to the regulatory, covalently boundAMP. Furthermore, distances between the adenylylgroup and bound Co(ll) ions were established utilizingCo(ll)-induced quenching of the fluorescence emissionspectrum of the e-ATP-adenylylated enzyme. The pic-torial representation of the spatial relationships derivedby these two methods are shown in Figure 8. The re-sults are in good agreement.

7. Other Enzymes

Some of the other enzymatic systems studied by themethod of paramagnetic probes an 31P nmr include (a)enolase (98), (b) RNA polymerase (99), (c) fructose di-phosphatase (100), (d) transcarboxylase (101), and (e)alkaline phosphatase (26, 28, 41).

(a) In the case of enolase (which catalyzes the revers-ible hydration of P-enolpyruvate to 2-phospho-o-glyc-erate), ternary complexes of enolase-Mn(ll) with P-enolpyruvate, an active analog of the reaction (CH2-P-enolpyruvate) and three inactive analogs (D-P-lactate,L-P-lactate, and P-glycolate) were studied to show thatMn(ll)-31P distances of 5.5 A to 6.2 A are consistentwith a second sphere coordination.

(b) For RNA polymerase the effects of Mn(ll)-enzymeon adenylyl (3'-5') uridine (ApU) bound at the initiationsite and on ATP bound at the elongation site were usedto conclude that the activating divalent cation bindstightly at the elongation site.

(c) 31P relaxation studies were used to detect a fruc-

Figure 8. Spatial relations between the metal ion sites and the adenylyl groups of glutamine synthetase.

DISTANCES FROM METAL IONS TO THE ADENYLYL SITE

NMRN H 2

FLUORESCENCE _ N H

* AloA -O-P =


Vol. 1, No. 1 57

tose-diphosphatase-Mn(ll)-fructose-1-phosphatebridge complex.

(d) In transcarboxylase paramagnetic effects on 31Prelaxation and 'H relaxation were used to obtain infor-mation on the conformation of enzyme-bound propionylcoenzyme A.

(e) The results on alkaline phosphatase were brieflydescribed in the section on hydrolytic enzymes.

IV. CONCLUDING REMARKS

It was known for nearly two decades that many bio-logical molecules possess characteristic 31P nmr spec-tra, and some of these spectra are sensitive to impor-tant parameters like pH, metal chelation, and so on,making 31P nmr a useful technique for studying thesesystems. However, the instrumentation required for per-forming experiments that can provide biochemicalinformation of significance became available only dur-ing the past four or five years. A large volume of re-search work was published in the past few years, andmore papers are appearing every month. The present ar-ticle covers some examples of the diverse aspects ofenzymatic reactions studied by this technique and isnot meant to be an exhaustive survey of all the workdone in this area. It is also beyond the scope of thisreview to discuss the mechanistic implication of thenmr results.

The volume of the research being done with 31P nmrand the variety and depth of information being extract-ed on biochemical systems testifies to the promise themethod holds. The information obtained on enzymaticreactions includes changes in environment at the activesite, chemical identification of substrates and products,dissociation constants, exchange rates, interconver-sion rates, and active site structural parameters. It maybe noted, however, that this information is not usuallyobtained in a routine manner and often requires carefuland subtle analysis of data, keeping in mind thephysical principles of nmr. This requirement is especial-ly stringent in the analysis of dynamical features of nmrspectra, line widths, and relaxation effects that are cen-tral to the determination of active site structures byparamagnetic effects, for example. Every analysis ofthese effects is contingent upon the validity of certainassumptions about correlation times, exchange times,dissociation constants, and so on. Inadequate exami-nation of the consequence of these assumptions oftenleads to structural data of questionable value and con-sequently tenuous biochemical conclusions. Added tothis is the fact that biochemical preparations vary inpurity, leading to slightly different results (even by thesame group) each time the experiments are performed.

The above uncertainties in analysis and experiment docontribute to some of the discrepancies and disagree-ments in the literature.

One of the problems for which 31P nmr promised to bethe most appropriate technique is the question of therole of the cations in biochemical reactions of ATP, inparticular the divalent cations in phosphoryl-transferand nucleotidyl-transfer reactions. In spite of the datacollected by substituting activating paramagneticcations (Mn(ll), Co(ll), etc) and in several other experi-ments on diamagnetic enzyme-bound complexes ofsubstrates, equilibrium mixtures, and transition-stateanalogs, the answer in a number of reactions remainselusive. The availability of thiophosphate analogs of nu-cleotides (102) may shed fresh light on the issue. Corre-lation of the stereospecificity of the a or j3-S diastereo-mer with different metal activators (103) with the 31P nmrspectra of enzyme-bound thioanalogs of ATP may wellprovide a definitive answer.

The present state of the art in nmr instrumentationproduces what may be referred to as the "1-mM barrier"for 31P nmr studies. Few experiments are feasible atconcentrations appreciably lower than 1 m/W. This isperhaps the single, most serious limitation in using thetechnique more widely in biochemistry. The second lim-itation related to 31P nmr, especially in relation to someof the work described in this article, is the unavailabilityof a reliable theory to relate the chemical shifts andspin-spin coupling constants to structural features ofthe phosphorus-containing biological molecules. It hasbeen frequently pointed out in the course of this review,for example, that a value of the chemical shift for anenzyme-bound substrate cannot be used to specify theanionic species since many shifts arising fromunknown perturbations are far dowhfield in enzyme-bound substrates from the lowest shift attainable forthe substrate free in solution. Progress in overcomingthese two limitations would undoubtedly enhance theutility of this already powerful technique by orders ofmagnitude.

References

1A. G. Redfield, NMR Basic Principles and Advances, (P. Diehl, G.Rack, and R. Kosfield, Eds.) Vol. 13, Springer-Verlag, Gottingen, 1976.

2M. Conn and T. R. Hughes, J. Biol. Chem. 235, 3250 (1960).3M. Conn and T. R. Hughes, J. Biol. Chem. 237,176 (1962).4A. G. Redfield, in Methods in Enzymology, (C. H. W. Hirs and S. N.

Timasheff, Eds.) Vol 49, Part G, Academic Press, New York, 1978, pp253-269.



5M. Conn and A. Hu, Proc. Natl. Acad. Sci. (U.S.A.) 75, 200 (1978).6J. H. Letcher and J. R. van Wazer, in Topics in Phosphorus

Chemistry, (M. Grayson and E. J. Griffith, Eds.) Vol. 5, IntersciencePublishers, New York, 1967, pp. 75-168.

7J. R. van Wazer and J. H. Letcher, in Topics in PhosphorusChemistry, (M. Grayson and E. J. Griffiths, Eds.) Vol. 5, IntersciencePublishers, New York, 1967, pp. 169-226.

8D. G. Gorenstein and D. Kar, Biochem. Biophys. Res. Commun.65, 1073 (1975).

9E. K. Jaffe and M. Cohn, Biochemistry 17, 652 (1978).10C. R. Johnson, Adv. Magn. Reson. 1, 33 (1965).11J. A. Pople, W. G. Schneider, and H. J. Bernstein, High Resolution

Nuclear Magnetic Resonance. McGraw-Hill, New York, 1959, pp.218-225.

12A. S. Mildvan and M. Cohn, Adv. Enzymol. 33,1 (1970).13R. A. Dwek, NMR in Biochemistry, Clarendon Press, Oxford, 1973.14T. L. James, NMR in Biochemistry, Academic Press, New York,

1975.15A. S. Mildvan and R. K. Gupta, in Methods in Enzymology (C. H.

W. Hirs and S. N. Timasheff, Eds.) Vol. 49, Part G, Academic Press,New York, 1978, pp. 322-358.

16A. C. McLaughlin, J. S. Leigh, and M. Cohn, J. Biol. Chem. 251,2777 (1976).

17T. C. Farrar and E. D. Becker, Pulsed Nuclear MagneticResonance, Academic Press, New York, 1971.

18H. Gueron and R. G. Shulman, Proc. Natl. Acad. Sci. (U.S.A.) 72,3482 (1975).

19F. F. Brown, I. D. Campbell, R. Henson, and C. W. J. Hurst, and R.E. Richards, Eur. J. Biochem. 38, 54 (1973).

20M. Gassner, D. Stehlik, O. Schrecker, W. Hengstenberg, W.Maurer and H. RUterjans, Eur. J. Biochem. 75, 287 (1977).

21E. B. Erasure, T. O. Henderson, T. Glonek, N. M. Pattnaik, and A.M. Scanu, Biochemistry 17, 3934 (1978).

22K. R. Sheu and P. A. Frey, J. Biol. Chem. 252, 4445 (1977).23K. R. Sheu and P. A. Frey, J. Biol. Chem. 253, 3378 (1978).24C. Ho, J. A. Magnuson, J. B. Wilson, N. S. Magnuson, and R. J.

Kurland, Biochemistry 8, 2074 (1969).25W. J. Ray, A. S. Mildvan, and J. B. Grutzner, Arch. Biochem.

Biophys. 184, 453 (1977).26J. L. Bock and B. Sheard, Biochem. Biophys. Res. Commun. 66,

24 (1975).27W. E. Hull, S. E. Halford, H. Gutfreund, and B. D. Sykes,

Biochemistry 15,1547 (1976).28J. F. Chlebowski, I. M. Armitage, P. P. Tusa, and J. E. Coleman, J.

Biol. Chem. 251,1207 (1976).29M. Martinez-Carrion, Eur. J. Biochem. 54, 39 (1975).^S. J. W. Busby, D. G. Gadian, G. K. Radda, R. E. Richards, and P. J.

Seeley, FEBS Lett. 55, 14 (1975).31K. Feldman and E. J. M. Helmreich, Biochemistry 15,2394 (1976).32K. Feldman and W. E. Hull, Proc. Natl. Acad. Sci. (U.S.A.) 74, 856

(1977).33M. K. Battersby and G. K. Radda, FEBS Lett. 72, 319 (1976).M J . J. Villafranca, S. G. Rhee, and P. B. Chock, Proc. Natl. Acad.

Sci. (U.S.A.) 75, 1255 (1978).MD. G. Gorenstein and J. B. Findlay, Biochem. Biophys. Res. Com-

mun. 72, 640 (1976).^ J . L. Bock, Ph.D. Thesis, Albert Einstein College of Medicine, New

York, 1976.37G. R. Reeck, T. B. Nelson, J. V. Paukstelis, and D. D. Mueller,

Biochem. Biophys. Res. Commun. 74, 643 (1977).M H . Csopak and T. Drakenberg, FEBS Lett. 30, 296 (1973).39J. L. Bock and A. Kowalsky, Biochim. Biophys. Acta 526, 135

(1978).40J. F. Chlebowski, I. M. Armitage, and J. E. Coleman, J. Biol. Chem.

252, 7053 (1977).

41R. S. Zukin, D. P. Hollis, and G. A. Gray, Biochem. Biophys. Res.Commun. 53, 238 (1973).

42D. H. Meadows, G. C. K. Roberts and O. Jardetzky, J. Mol. Biol. 45,491 (1969).

^D. G. Anderson, G. G. Hammes, and F. G. Walz, Biochemistry 7,1637(1968).

""D. G. Gorenstein, A. M. Wyriwicz, and J. Bode, J. Am. Chem. Soc.98, 2308 (1976).

45W. Haar, J. C. Thompson, M. Maurer, and H. RUterjans, Eur. J.Biochem. 40, 259 (1973).

46B. D. Nageswara Rao, D. H. Buttlaire, and M. Cohn, J. Biol. Chem.251, 6981 (1976).

47B. D. Nageswara Rao and M. Cohn, J. Biol. Chem. 252, 3344(1977).

48B. D. Nageswara Rao and M. Cohn, unpublished results.49B. D. Nageswara Rao, M. Cohn, and L. Noda, J. Biol. Chem. 253,

1149(1978).'"B. D. Nageswara Rao and M. Cohn, Proc. Natl. Acad. Sci. (U.S.A.)

74,5355(1977).61B. D. Nageswara Rao, F. J. Kayne, and M. Cohn, J. Biol. Chem.

(in press).52B. D. Nageswara Rao, M. Cohn, and R. K. Scopes, J. Biol. Chem.

253, 8056 (1978).53J. R. Morrison, in The Enzymes, (P. D. Boyer, Ed.) Vol. 8,

Academic Press, New York, 1973, pp. 457-486.MD. C. Watts, in The Enzymes, (P. B. Boyer, Ed.) Vol. 8, Academic

Press, New York, 1973, pp. 384-456.55E. J. Milner-White and D. S. Rycroft, Biochem. J. 167, 827 (1977).M L Noda, in The Enzymes, (P. D. Boyer, Ed.) Vol. 8, Academic

Press, New York, 1973, pp. 279-306.57N. C. Price, G. H. Reed, and M. Cohn, Biochemistry 12, 3322

(1973).MT. R. Brown and S. Ogawa, Proc. Natl. Acad. Sci. (U.S.A.) 74,3627

(1977).59R. K. Scopes, in The Enzymes, (P. D. Boyer, Ed.) Vol. 8, Academic

Press, New York, 1973, pp. 335-352.eoM. Larsson-Raznikiewicz and J. R. Jansson, J. R., FEBS Lett. 29,

345 (1973).61R. K. Scopes, Eur. J. Biochem. 85, 503 (1978).82F. J. Kayne, in The Enzymes, (P. D. Boyer, Ed.) Vol. 8, Academic

Press, New York, 1973, pp. 353-383.63F. J. Kayne, Biochem. Biophys. Res. Commun. 59, 8 (1974).64H. Gutfreund, Prog. Biophys. and Mol. Biol. 29, 161 (1975).65D. R. Trentham, Biochem. Soc. Trans. 5, 5 (1977).66A. S. Mildvan and M. Cohn, J. Biol. Chem. 240, 238 (1965).67R. K. Gupta, R. M. Osterling, and A. S. Mildvan, Biochemistry 15,

2881 (1976).68R. K. Gupta and A. S. Mildvan, J. Biol. Chem. 252, 5967 (1977).6 9J. L. Wyatt and R. F. Colman, Biochemistry, 16, 1333 (1977).70M. Blumenstein, Biochemistry 14, 5004 (1976).71J. Feeney, B. Birdsall, G. C. K. Roberts, and A. S. V. Burgen,

Nature (London) 257, 564 (1975).72B. Birdsall, G. C. K. Roberts, J. Feeney, and A. S. V. Burgen, FEBS

Lett. 80, 313 (1977).73M. R. Webb, D. N. Standing, and J. R. Knowles, Biochemistry 16,

2738 (1977).7 4 I . D. Campbell, R. A. Kiener, S. G. Waley, and R. Wolfenden,

Biochem. Soc. Trans. 5, 750 (1977).7 5 I . D. Campbell, R. B. Jones, P. N. Kiener, R. E. Richards, S. G.

Waley, and R. Wolfenden, Biochem. Biophys. Res. Commun. 83, 347(1978).

76O. Lutz, A. Nolle, and D. Staschewski, Z. Naturforsch, Teil A. 33,380 (1978).

77M. Cohn, in NMR and Biochemistry S. Opella and P. Lu, Eds.,Marcel Dekker, New York, in press.


Vol. 1,No. 1 59

78R. L Van Etten and J. M. Risley, Proc. Natl. Acad. Sci. (U.S.A.) 75,4784 (1978).

79S. D. Lowe and B.S. Sproat, J. Chem. Soc. (in press).80M. R. Webb, G. G. McDonald, and D. R. Trentham, J. Biol. Chem.

253, 2908 (1978).81J. L. Bock and M. Conn, J. Biol. Chem. 253, 4082 (1978).82T. R. Sharp and S. J. Benkovic, submitted for publication.83D. H. Eargle, V. Licks, and G. L. Kenyon, Anal. Biochem. 81,186

(1977).84L. T. Smith and M. Conn, unpublished results.85C. F. Midelfort and I. Rose, J. Biol. Chem. 251, 5881 (1976).86J. S. Leigh, Thesis, University of Pennsylvania, Philadelphia, 1971.87E. Melamud and A. S. Mildvan, J. Biol. Chem. 250, 8193 (1975).88D. L Sloan and A. S. Mildvan, J. Biol. Chem. 251, 2412 (1976).89C. H. Fung, A. S. Mildvan, A. Allerhand, R. Komoroski, and M. C.

Scrutton, Biochemistry, 12, 620 (1973).9*T. James and M. Conn, J. Biol. Chem. 249, 3519 (1974).91R. K. Gupta, C. H. Fung, and A. S. Mildvan, J. Biol. Chem. 251,

2421 (1975).92A. S. Mildvan, D. L Sloan, C. H. Fung, R. K. Gupta, and E.

Melamud, J. Biol. Chem. 251, 2431 (1975).

93D. L. Sloan, L A. Loeb, A. S. Mildvan, and R. J. Feldman, J. Biol.Chem. 250, 8913 (1975).

94A. S. Mildvan and C. M. Grisham, Structure and Bonding (Berlin)20,1 (1974).

95C. M. Grisham and A. S. Mildvan, J. Supramol. Structure 3, 304(1975).

96W. J. Ray and A. S. Mildvan, Biochemistry 12, 3733 (1973).97D. J. Hamm and B. S. Cooperman, Biochemistry 17, 4033 (1978).98T. Nowak, A. S. Mildvan, and G. L. Kenyon, Biochemistry 12,1690

(1973)." B . L. Bean, R. Koren, and A. S. Mildvan, Biochemistry 16 3322

(1977).100S. J. Benkovic, J. J. Villafranca, and J. J. Kleinschuster, Arch.

Biochem. Biophys. 155, 458 (1973).101C. H. Fung, R. J. Feldman and A. S. Mildvan, Biochemistry 15, 76,

(1976).102F. Eckstein and R. S. Goody, Biochemistry 15, 1685 (1976).103E. K. Jaffe and M. Cohn, J. Biol. Chem. 253, 4825 (1978).



31p nmr studies of enzymatic reactions - semantic scholar · spin of 31p is vz, and the nmr spectra...

Documents