Li, 31P, and H NMR Studies of Interactions between ATP, Monovalent Cations, and Divalent Cation Sites on Rabbit Muscle Pyruvate Kinase*

Transcription

1 ~~?- ~ THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society of Biological Chemists, Inc. Vol. 260, No. 25, Issue of November 15, pp Printed in d.s.a. Li, 31P, and H NMR Studies of Interactions between ATP, Monovalent Cations, and Divalent Cation Sites on Rabbit Muscle Pyruvate Kinase* Jeanne M. Van DivenderS and Charles M. Grishams From the Department of Chemistq, University of Virginia, Chartottesuille, Virginia (Received for publication, May 13, 1985) The interactions between ATP, monovalent cations, and divalent cations on rabbit muscle pyruvate kinase have been examined using Li, 31P, and H nuclear magnetic resonance. Water proton nuclear relaxation studies are consistent with the binding of Li+ to the K+ site on pyruvate kinase with an affinity of 120 mm in the absence of substrates and 16 mm in the presence of P-enolpyruvate. Titrations with pyruvate demonstrate that pyruvate binds to the enzyme with an affinity of 0.65 mm in the presence of Li+ and 0.4 mm in the presence of R+. Li+ nuclear relaxation rates in solu- tions of pyruvate kinase are increased upon titration with the metal-nucleotide analogue, Cr(H,O)ATP. Mn2+ EPR spectra were used to determined the distribution of the enzyme between the so-called isotropic and anisotropic conformations of the enzyme (Ash, D. E., Kayne, F., and Reed, G. H. Arch. Biochern. Bio- phys. (197j3) 190, ). Li-Cr distances of 5.6 and 11.0 A were calculated for the anisotropic and isotropic forms, respectively, in the absence or presence of pyruvate. When the divalent cation site on the enzyme was saturated yith Mg+, these distances increased to 6.7 and 9.5 A, respectively, regardless of the presence or absence of pyruvate. P nuclear relaxation studies with the diamagnetic metal-nucleotide analogue, CO(NH~)~ATP, indicated that addition of Mn2+ ion to the divalent cation site on the enzyme increased the longitudinal relaxation rates of all three phosphorus nuclei of the analogue. The 31P data indicate that the presence of pyruvate at the active site effects a decrease in the Mn-P distances, bringing Mn2+ and Co(NH&ATP closer together at the active site. The data also permit an evaluation of the role of the metal coordinated to the B-P and 7-P of ATP at the active site. Due to the large amount of active site structural information which has already been obtained for the rabbit muscle enzyme, pyruvate kinase, it is an ideal model system for the development of new spectroscopic techniques or for the ex- * This work was supported by National Institutes of Health Research Grant AM19419 and grants from the Muscular Dystrophy Association of America, the Research Corporation, and the University of Viriginia. The NMR and EPR instrumentation used in these studies was provided by grants from the National Science Foundation and the University of Virginia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Present address: Stauffer Chemical Co., Dobbs Ferry, NY Research Career Development Awardee of United States Public Health Service Grant NIH-AM ploration of new spectroscopic probes. Thus, in our laboratory, a long-standing interest in monovalent cation active transport across the mammalian plasma membrane led us to examine 7Li NMR as a probe of monovalent cation sites, first on pyruvate kinase (1) and later with kidney (Na+ + K+)-ATPase (2) and then with sarcoplasmic reticulum Ca2+-ATPase (3). Similarly, we have used the complexes of Cr(II1) and Co(II1) with ATP to characterize the interactions of divalent metals and ATP at the active sites of the ATPases (3, 4). We.have relied at times on the wealth of information which exists on the interactions of these probes with pyruvate kinase and other related phosphoryl transfer systems. It was in the interest of a further characterization of 7Li and the Cr(II1) and Co(II1) complexes of ATP as spectroscopic probes, and in order to address several ambiguities in the active site structure of pyruvate kinase, that we undertook the experiments described in this paper. Pyruvate kinase is a tetrameric enzyme of M, = 237,000, which catalyzes the reversible phosphorylation of adenosine diphosphate (ADP) from phosphoenolpyruvate (P-enolpyruvate) to form adenosine triphosphate (ATP). This enzyme requires both monovalent (K+) and divalent (M$+) cations for activity and recent studies indicate that the enzyme binds, and may require for activity, two divalent cations at the active site (5). Spectroscopic and kinetic studies have yielded a model of the active site in which 17 distances between relevant substrate nuclei and/or metal ions can be reconciled. The data are consistent with molecular contact between the y- phosphoryl phosphorus of ATP and the carbonyl oxygen of pyruvate, consistent with direct phosphoryl transfer. The enzyme-bound divalent cation appears to form second sphere complexes with the phosphoryl groups of P-enolpyruvate and ATP and may activate the transferred phosphoryl group indirectly through a water ligand. On the other hand, as pointed out by Mildvan et az. (6), several ambiguities remain. One of the these concerns the activating monovalent cation, normally K+. Studies with Tl+(7) and in our lab with 7Li+ (1) have demonstrated that the binding of P-enolpyruvate causes a substantial decrease in the separation of the monovalent cation site and the high affinity catalytic site for Mn2+ on the enzyme. Comparison of the 205Tl+ and 7Li+ data led us to suggest that the low activity of the enzyme with Li+ (2% of that with T1+ or K ) might be a consequence of a poor fit for Li+ at the active site, with a larger M+-M2+ separation in the case of Li+ than with T1+ (1, 7). This model was disputed by Ash et al. (S), but Raushel and Villafranca (9, 10) later showed that the activating abilities of the alkali cations were indeed inversely related to the M+-Mn2+ separations measured for these ions at the active site of pyruvate kinase, as we had originally suggested. Still, the relation of the monovalent cation site to the other moieties

2 NMR Studies of Substrate and Ion Sites on Pyruvate Kinase at the active site of this enzyme is not well understood, and we have attempted to address this problem in the present work. Another of the remaining uncertainties raised by Mildvan et al. (6) concerns the conformation of the triphosphate chain of ATP at the active site. 31P NMR studies of the interaction of ATP with enzyme-bound Mn" (11) have provided some information on this point. However, it now appears that there are two divalent metal sites at the pyruvate kinase active site, one bound to the enzyme itself, and the other coordinated to enzyme-bound ATP (5). To date, there has been no examination of the influence of metal at the latter site on the conformation of the triphosphate chain of ATP in the active site. It has been shown that Co(NH,),ATP and Cr(H20)ATP bind to pyruvate kinase, replacing MgATP at the active site (5). In this paper, we use the p,y-bidentate Co(NHJ4ATP complex of ATP in a series of 31P nuclear relaxation studies to address this point. The results suggest that coordination of metal to the 8- and y-phosphates of ATP alters the conformation of the triphosphate chain at the active site. EXPERIMENTAL PROCEDURES Materials-Rabbit muscle pyruvate kinase was purchased from Boehringer Mannheim. Lactate dehydrogenase, glyceraldehyde-3- phosphate dehydrogenase, 3-phosphoglycerate kinase, adenosine-5'- triphosphate, pyruvate, NADH, phosphoenolpyruvate, glycerate 3- phosphate, Tris buffer, Dowex 50W-X8 ( mesh), and Dowex chelating resin were purchased from Sigma. Chelex 100 and AG 50W- X2 ( mesh, H+ form) were purchased from Bio-Rad, and D20 (99.8% D) was obtained from Aldrich. All other reagents were reagent grade or of the highest purity commercially available. General-UV visible spectra are obtained on a Hitachi double beam spectrometer with an attached strip chart recorder. ph measurements are made on a Radiometer PHM 26c ph meter. D20 is passed through a column containing Chelex 100 (TMA+-form') to remove trace metal impurities. Preparation of Co(NH&ATP-The starting material, carbonatotetraamine cobalt (111) nitrate, [Co(NH3)4CO3]NO3, has been prepared according to the method of Schlessinger (12). The preparation of Co(NH3)JTP from [CO(NH~)~CO~]NO~ and follows ATP closely the procedure described by Cornelius et al. (13). The product is divided into 0.1-mmol samples (usually 1 ml of 100 m ~ and ) stored at -20 "C. The purity of the complex is confirmed by UV and 31P NMR spectroscopy. The concentration of Co(NH3)ATP is calculated from the absorbance at 260 nm, using an extinction coefficient of 16,900 M" cm" (13). Cr(Hz0)+4TP was prepared as previously described and its concentration was determined using a molar extinction coefficient of 14,400 M-' cm" (14). Preparation of Pyruvate Kinase-Pyruvate kinase is obtained as a crystalline suspension in a 3.2 M (NH&S04 solution. The enzyme ( mg) is desalted by dialyzing at 4 "C against 100 mm Tris buffer, ph 7.5. The dialysis tubing is boiled thrice in EDTA and NaHC03. The enzyme is concentrated at 4 "C by vacuum dialysis to levels suitable for use in the NMR experiments ( p~ active sites). The enzyme is then dialyzed at 4 "C thrice in 500-fold excess of 100 mm Tris-C1, ph 7.5. The third dialysis contains 10 g of Chelex 100 (outside of the bag) to remove trace metal impurities. The enzyme concentration is determined by using its extinction coefficient at 280 nm (elem = 0.54) and its molecular weight (237,000 g/mol). Specific activities are usually units/mg. Preparation of Pyruvate Solution-TMA /pyruvate (0.2) is prepared in the following way: 12.5 g of cleaned Dowex 50W-X8-H+ ( mesh) is placed in a 50-ml Erlenmeyer flask. A second batch of Dowex (12.5g) is placed in a 50-ml filter flask. Na/pyruvate (1.1g) is dissolved in 10 ml of water, added to the Erlenmeyer flask, stirred for 15 min at 4 "C, and filtered by suction into the second flask. The resin is washed twice with 5 ml of water. The second flask is treated similarly. The resulting solution is neutralized to ph 7.5 with 10% TMA-OH, and the volume is brought up to 50 ml with water. Tris+/ pyruvate and Li+/pyruvate were prepared similarly, except that these were neutralized to ph 7.5 with Tris base, or ph 6.6 with 1 M LiOH, respectively. NMR Methods-The 7Li* NMR experiments were performed using a JEOL PS-100P Fourier transform spectrometer. Measurements were made at probe temperatures of 5 1 "C and 23 -t 1 "C.All studies were carried out in 10-mm NMR tubes. Using the carbon probe, a 5-kHz crystal filter, and lowering the magnetic field from 23.5 to 15.2 kg, the 'Li+ signal resonated at MHz. Lowering the field 12 h before the experiment produced a stable field. Unfortunately, lowering the field prohibits the normal use of the deuterium lock; hence, experiments were done unlocked. Usually only one transient was required for good signal to noise ratios. Spin-lattice relax- ation rates (1/Tl) were measured using a 180"-~-90" pulse sequence where the 90" pulse was approximately 20 ps. Null times were measured giving a Tl = null time/ln2. Samples having an initial volume of 1.0 ml contained 50 mm LiCl, 80 mm Pipes buffer, and pm enzyme sites. The 31P NMR spectra and relaxation rates are determined by pulsed Fourier transform methods at two frequencies, 40 MHz and 145 MHz. Low field spectra are obtained at 2.3 tesla on a Jeol PSlOOP Fourier transform spectrometer operating at 40.5 MHz, equipped with a field frequency lock on an internal deuterium resonance. The spectra are obtained in the time domain mode, and a signal averaging is performed with a 980A Texas Instruments computer dedicated to the spectrometer. Broad-band decoupling of the proton resonances (-15 W) is employed. The 90" pulse width is ps. The spectra are collected as 8K FIDs using single-phase detection. The sweep width of the spectra is 1250 Hz, and the FIDs are processed with an exponential filter to improve the signal to noise ratio. High field spectra are obtained at 8.5 tesla with a Nicolet Magnetics Corp. NT-36O/Oxford Spectrometer equipped with the 1280/293B data system. An internal *H lock and uninterrupted incoherent 'H decoupling at low levels (<1 W) are employed. The 90" pulse width at MHz is ps. The spectra are recorded using quadrature phase detection with the carrier frequency in the middle of the spectrum. Standard Nicolet software (NMCFT V#10221) is used to collect and process the free induction decays. The 1D spectra are collected as 8K FIDs which are zero-filled to yield 16K FT spectra. TI plots are collected as 4K FIDs and zero-filled to 8K before Fourier transforming. The sweep width of the spectra is k Hz, and exponential multiplication is employed on the!fl data sets optimize to sensitivity, using a time constant, matched to the effective decay constant Tz*. Measurements are made at probe temperatures of 23 f 1 "C. Typical sample volumes for the NMR experiments are 1.0 ml (40.3 MHz) and 2.4 ml (145.7 MHz) containing approximately 20% DzO for the purpose of providing a field-frequency lock. Sample tubes (Wilmad 513-7PP), 10-mm outside diameter, were used with a vortex plug. The chemical shifts are expressed with reference to 85% H3P04 "as an external standard; positive values are downfield from the standard. For the one-pulse spectra, the rf pulse angle 8, for the optimum signal to noise ratio, is calculated according to the expression: cos 0 = where 7 and TI represent the repetition time and spin-lattice relaxation time, respectively (15). The longitudinal relaxation rates (l/tl) are measured using the inversion-recovery pulse sequence (18oO-7-90"). The proton relaxation rate experiments were performed using a variable frequency, pulsed NMR spectrometer. The magnet for this instrument is a Varian 4012A electromagnet with a 2100A power supply which have been modified for solid state operation. The rf components and pulse programmer have been designed and built by SEIMCO, and the frequency synthesizer is a PRD 7838 synthesizer. Measurements are made at a probe temperature of 23 f 1 "C. The experiment is carried out in 4-mm tube (0.1 ml) at 24 MHz. The relaxation rates are measured using the inversion-recovery sequence. Null times are measured giving a TI equal to the null time divided by ln2. NMR Samples-Just prior to performing the experiments, a sample of CoATP is thawed and passed through a column (0.4 x 7 cm) of ' The abbreviations used are: TMA, tetramethylammonium; Pipes, Chelex. The solution is adjusted to ph 7.5 with 1 M Tris-C1 (chelexed), 1,4-piperazinediethanesulfonic acid; FID, free induction decay; ph 7.4, and diluted to yield a solution containing 100 mm Tris-C1, Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. ph 7.5. The NMR samples contain the following components: 17-21

3 NMR Studksof Substrate and Ion Sites on Pyruvate Kinase mm CoATP, 80 mm Tris-C1 buffer, ph 7.5, 20% D20, and 0-25 p~ MnC12. Those samples which have enzyme contain 40 pm pyruvate kinase (in enzyme sites) and 10 mm KCl. For those experiments containing pyruvate, TMA/pyruvate is added to yield a final concentration of 17.6 mm. Theoretical Bask for Calculations-The calculations in this paper are based on previously described theory (16-25) and will only be summarized here. The paramagnetic contribution to the longitudinal relaxation rate, l/tlp, of a nucleus is given by: w ""5"- Tlp TI 7, + TIM where l/t1 is the observed relaxation rate in the presence of the paramagnet; l/tl(o) is the relaxation rate in the absence of the paramagnet; p is the ratio of the concentration of paramagnetic ion to the concentration of the observed nucleus; q is the number of nuclei in the coordination sphere of the Paramagnetic ion; thus, pq (pq = f ) is the mole fraction of the nuclei in the coordination sphere; l/rm is the ligand exchange rate between the bound and unbound form; and 1/T1~ is the relaxation rate in the coordination sphere of the paramagnetic ion. If 1/7,,, is fast compared to 1/Tm (;.e. conditions of fast exchange), the paramagnetic contribution to the relaxation rate, 1/TV, is given by the Solomon-Bloembergen equation: where and where r is the nucleus-paramagnet separation, T~ is the dipolar correlation time, and wr and w, are the nuclear and electronic Larmor frequencies, respectively. In order to use the above equations to calculate active site distances on an enzyme, it is necessary to make an estimate of the correlation time, T~. If the correlation time is dominated by the electron spin relaxation time, Tle, of the paramagnetic ion (Mn2+ in this case) then it is given by Ref. 17: 1 7" 47" "-B- T, w w.27.2 where B is a constant containing the electronic spins s and the zerofield splitting of Mn2+ and 7, is a time constant for transient symmetry distortions of the MnZ+ complex. This correlation time has two limiting cases. For case 1 W,"T," >> 1 In this case, Equation 4 reduces to T~ = w,27"/(2b), and the ratio of l/(ftlp) at two frequencies (u1 and v2) is given by In this case, XIe displays maximum frequency dependence, and T~ = T,, at ul. For case 2 w,"7,2 << 1 As can be seen by consideration of Equation 4, TIS = 1/(5&,) for this case and is independent of frequency. For the ratio of 1/(fTlp) at two frequencies, we obtain l/[ftlp(~l)] l/[ftlp(u~)] ~~~a r2ui7.2 Thus when Tu dominates T,, Equations 5 and 6 can be used together with measurements of l/(ftlp) at two frequencies to estimate 7,. Ternary Mn'+-Enzyme-Co(NH&ATP Complex-Under the experimental conditions employed here, Co(NH&ATP is present in both binary and ternary Mn2+ complexes. Thus, the observedparamagnetic relaxation. rate is a weighted average of the rates due to each complex of CoATP where [SM], [MES], and [SIT denote the concentrations of the binary (6) complex, ternary complex, and total Co(NH&ATP, respectively. (I/ ft&inary and (l/ft~jtemary are the relaxation rates of CoATP in the binary and ternary complexes. [SM] and [MES] are calculated from their respective dissociation constants (5). RESULTS Binding Constants for K+, Li+, and Pyruvate to Pyruvate Kinase-The interactions of Li+ with pyruvate kinase and the binding of pyruvate in the presence of Li+ were examined by following the rate of water proton longitudinal relaxation in solutions containing the enzyme, Mn2+, Li+, and pyruvate. Fast exchange of water protons at the single high affinity Mn2+ site on the enzyme provides an enhancement of the relaxation of bulk water in these solutions, as has been shown in previous studies (15). As reported previously (21), addition of K+ to solutions containing pyruvate kinase effects a decrease in e*, the enhancement of water relaxation. As shown in Fig. 1, a similar effect is observed with Li+. The affinity of K' estimated.from kinetic studies is 14 mm in the presence of Mn" (22) while the binding of Li+ obtained here (120 mm) is considerably weaker (Table I). This value compares with a value of 11 mm estimated from kinetic studies (1). In the presence of P-enolpyruvate, the affinities for both K+ and Li+ are increased dramatically. The value of KS for K+ compares well with that determined by Nowak and Mildvan (21), while the value for Li+ is in agreement with the estimate from kinetic studies (1). The weak binding of Li+ and the low enhancement at high levels of Li+ raise the question of displacement of Mn2+ from the enzyme by Li+. That this is not the case was shown by us previously (1). In order to measure the affinity of the pyruvate kinase- Mn2+-M+ (M+ = K+ or Li+) complex for pyruvate, the appropriate solutions of enzyme and ions were titrated with pyruvate as shown in Fig. 2. The affinity for pyruvate is only slightly weaker in the presence of Li+ than in solutions containing K+ as shown in Table I. 7 Li + Nuclear Rehution Studies in Solutions of Pyruvate Kinase and Cr(H20).PTP"Measurements were made of 'Li+ longitudinal relaxation rates, l/t17 in solutions of pyruvate kinase as a function of added Cr(H20)ATP in order to determine the location of the monovalent cation site with respect to the bound paramagnetic Cr(II1) ion. Titrations were performed in the presence and absence ofmg2+ and pyruvate. Since pyruvate kinase is known to exist in two conformations, and since the distribution between these two states is known to be temperature dependent (23), all 7Li+ NMR studies were performed both at 5 and 23 "C. In the absence of enzyme, the relaxation rates of Li+ solutions ranged from s-l at 23 "C to 0.11 s-l at 5 "C. Addition of Mg2+ or of pyruvate in the absence or presence of enzyme caused no significant change in the relaxation rate. Addition of enzyme caused a diamagnetic increase in the relaxation rate, which was typical of the effects of dissolved macromolecules. In all cases, addition of Cr(H20)ATP to solutions of enzyme resulted in an increase in the longitudinal relaxation rate of 7Li+ until the enzymewas saturated with Cr(H20)4ATP, consistent with close approach of Li+ and Cr(1II) at the active site. A typical titration with Cr(H20)4ATP is shown in Fig. 3, and a summary of the titrations in the absence and presence of pyruvate and Mg2+ is presented in Table 11. As can be seen, the presence of pyruvate increases the values of l/t-, while Mg2+ effects a decrease in l/tw, both in the presence and absence of pyruvate. Mn2+ EPR Studies of the Distribution of Pyruvate Kinase between Isotropic and Anisotropic Conformations-Pyruvate kinase undergoes conformational changes under several con-

4 NMR Studies of Substrate and Ion Sites on Pyruvate Kinase '0 9 '0 L- '0 2 '0 0 '0 0 '02 0 ' '8 0 'r 0 '0 J U II z E TABLE I Binding of potassium, lithium, and pyruvate to pyruvate kinase Dissociation constant Complex of [ligand]" E.Mn.[K+] 12ob 14' E.Mn.[Li+] 120 * 10 E. Mn. [K+]. (P-enolpyruvate) 0.8 f O.0Sd E.Mn. [L1+]. (P-enolpyruvate) 16 & 2" E. Mn. K+. [pyruvate] 0.4 f 0.09 E. Mn. Li+. Iwruvatel C 0.10 Dissociation constant of [ligand] from indicated complex. Determined without divalent cations by difference spectroscopy. Determined with Mn2+ present by kinetic methods (22). Also determined in Ref. 21. e Also determined in Ref. 8. ditions (7, 21, 23): 1) the addition of either the required monovalent or divalent cation; 2) the binding of either P- enolpyruvate or pyruvate to the enzyme; and 3) variations of temperature. Reed and Cohn (23) showed using Mn(I1) EPR that the enzyme-bound Mn(1I) is very sensitive to ligand binding. Their work showed that extensive changes in the EPR spectrum of enzyme-bound Mn(I1) occurred upon formation of either pyruvate or P-enolpyruvate ternary complexes due to changes in the coordination geometry of Mn(I1). The isotropic electronic environment of Mn(1I) in the binary enzyme-mn complex can become highly anisotropic upon the formation of a ternary complex depending upon the nature of the ligand and the monovalent cation. The percentage of the anisotropic component in the spectrum falls in the same order as the activating abilities of the ions: K+ > Na+ > Li+. Ash et al. (8) used the Mn(I1) EPR spectra of pyruvate kinase-mn- (P-enolpyruvate)-Li complexes. to determine the percentage of anisotropic or "K-like" species in the presence of Li'. In order to determine the distribution of the enzyme between the anisotropic and isotropic conformations and in order to calculate Li-Cr distances from our 7Li+ relaxation data, we have examined the Mn(I1) EPR spectra of enzyme-mn-li complexes in the absence and presence of pyruvate and Co(NHa)rATP, a diamagnetic analogue of Cr(H,O),ATP. Fig. 4 shows the Mn(I1) EPR spectra of pyruvate kinase complexes with pyruvate and either K+ or Li+. The contributions to these spectra from the anisotropic or K+-like form and the isotropicform of the enzymewerecalculatedfrom these spectra using spectral subtraction and integration software described previously (24). The anisotropic conformation accounted for 92% of the Mn(I1) spectral intensity at 23 "C and only 45% at 5 "C. Calculation of Li-Cr Distances in Complexes of Enzyme, MP, and Pyruuate-The existence of two conformational states with, as Ash et al. (8) have shown, two different orientations for Li+ at the active site means that the observed relaxation rates for 7Li+ will contain a contribution from each species. Expressed in terms of the observed l/tim: 1/TlM(ob) = d TIM(1) f d TlMt2) where p1 and pz are the fractional populations of the two species (i.e. p1 + pz = 1 and l/tl~(l, and l/tim(z, are the characteristic relaxation rates of the two bound forms. Since the populations of the two forms are altered by vkying the temperature, we can use the observed relaxation rates to calculate the values of l/tim(l) and l/timcz) for the enzyme- Li-Cr(H,O)ATP-pyruvate complex with and without Mg+. Appropriate values of l/tl~ for the enzyme complexes with LI+ and Cr(HZ0)ATP are shown in Table 11. The calculation of Li-Cr distances in these complexes re-

5 mm NMR Studies of Substrate and Ion Sites on Pyruvate Kinase E* E" t [Pyruvate], mm [Pyruvate] ~ FIG. 2. PRR titrations measuring the effect of added pyruvate on the observed enhancement of pyruvate kinase complexes in the presence of potassium or lithium. The solutions contained A, 18 mm K+/Pipes buffer, ph 6.6, 55 mm KC1 and 46 X 10" M pyruvate kinase; B, 25 mm Li+/Pipes, ph 6.6, 50 mm Licl, and 48 X lo6 M pyruvate kinase. 0 s - P ffi N ID 6) v m 6 i & 4. nj ni d [CrATPl. mm FIG. 3. 'Li' nuclear relaxation titration measuring.the effect of Cr(H20)&TP on the l/tlp of Li+ in the presence of pyruvate kinase at 23 OC. The solution contained 80 mm Li+/ Pipes,.pH 6.6,56 mm Licl, 101 X M pyruvate kinase, and 19.6 mm k+/pyruvate. TABLE I1 Effect of Cr(HZ0)JTP on the relazation rates of 7Li+ in solutions containing pyruvate kinase l/ft, r' rb r, pl pz rl rz l.o *7 9* a Calculated under maximal frequency dependence of the correlation time. *Calculated under no frequency dependence of the correlation time, T~ = 1.63 X 10"os. e All experiments contained 80 mm Pipes, ph 6.6, 150 p~ enzyme, 50 mm LiCl. Experiments 1,3,5, and 7 were performed at 23 "c while 2, 4, 6, and 8 were performed at 5 "C. In addition, 1 and 2 contained no other substrates; 3 and 4 contained 19 DIM pyruvate; 5 and 6 contained 10 mm MgC12; 7 and 8 contained 10 mm MgCL and 18 mm pyruvate. quires a value for the dipolar correlation time, T,, for the Li- Cr interaction. From the frequency dependence of water proton relaxation rates in various en~yme-cr(h~0)~atp complexes, Gupta et al. (5) determined values of T= of X 10-l' s. Moreover, EPR spectra of free Cr(H20)4ATP show a line at g = 2.0 with a peak to peak line width of 650 G, setting a lower limit of 1 x 10-lo s on the electron spin relaxation time, T ~, of Cr(H20)4ATP. Since the dominant contribution to T, for IH20-Cr(III) interactions on pyruvate kinase is almost certain to be T,, the same correlation time is likely to pertain for the 7Li-Cr(III) interactions described here. Taking the values for l/tl~ and the value for T~ shown in Table 11, we obtain the distances for the various Li-enzyme complexes shown in Table P Nuclear Relaxation Studies of the Mn2'-Co(NHJ4ATP Complex-In order to examine the interactions between Mn2+ and CO(NH~)~ATP on pyruvate kinase, we first considered the binary Mn2+-Co(NH&ATP complex. 31P NMR studies of this complex have been described (25), but we have obtained results which differ significantly from the previous study. As we have recently shown,,t?,t-bidentate CO(NH~)~ATP is prone to small amounts of spontaneous breakdown, in which Co(II1) isreduced to paramagnetic Co(I1)over periods similar to those required for nuclear relaxation measurements with paramagnetic probes (26). It is possible that the higher values of l/tim (and consequently smaller values for Mn-P distances) obtained in the previous study (25) are a result of reduction of Co(II1) with time in those experiments. the In studies presented below, all solutions of CO(NH~)~ATP were bubbled gently with O2 just prior to NMR measurements, and each relaxation time measured here was obtained on a freshly prepared sample of Co(NH3)ATP. As we have shown, these procedures effectively prevent the reduction of Co(II1) to Co(I1) (26). We have observed no adverse effects due to O2 treatment of our NMR solutions in this manner (27), although more extensive O2 bubbling eventually causes broadening of the 31P resonances. As shown in Fig. 5, titration of solutions of CoATP results in increases in l/tl of the 31P resonances of the nucleotide in proportion to the concentration of added Mn2+. In order to

6 NMR Studies of Substrate and Ion Sites Pyruvate on Kinase Boo H, gauss FIG. 4. Electron paramagnetic resonance spectra at 23 OC of pyruvate kinase complexes with K+ and Li+. Solutions contained 3 mm pyruvate kinase, 50 mm TMA/Hepes, ph 7.5, 100 mm TMA/CI, 0.9 mm MnC12, and A, 50 mm KCl; B, 50 mm KCl, 21 mm pyruvate; and c, 50 mm Li+, 21 mm pyruvate. I I t - 0., Mn2+ (UM) FIG. 5. The effect of Mn(II) on the longitudinal relaxation rate of the phosphorus nuclei of Co(NH&ATP at 40 MHz (0) and 145 MHz (0, +, A, V). The solutions contained 20 mm Co(NH&ATP and 80 mm Tris, ph 7.5. Titrations at 145 MHz are shown for the a-p (A-isomer (A) and A-isomer (V)), 6-P (A-isomer (4) and A-isomer (m)), and y-p (0) resonances. examine the frequency dependence of the Mn2+-induced relaxation of the phosphorus nuclei of Co(NH&ATP, titrations were performed both at 40 and 145 MHz. Since the P, resonances of the A and A isomers of CoATP overlap at both these frequencies, it is not possible to compare relaxation parameters for the A and A diastereomers. On the other hand, the resonances for the P, and Po nuclei of the two isomers are sufficiently separated at 145 MHz to extract relaxation parameters for each of the A and A isomers. The assignments of the resonances shown in Fig. 6 were made (26) on the basis of two-dimensional chemical shift correlation NMR measurements and analysis of reaction mixtures containing CoATP, glucose, and hexokinase, an enzyme known to be specific for the A diastereomer of metal-atp complexes. As can be seen in Table 111, the P, of the A isomer of CoATPshows a substantially larger paramagnetic interaction with Mn2+ than is the case of the A isomer. Estimation of the Correlation Time, 7, for the Binary Mn2'- Co(NHJ4ATP Complex-The most direct method for determining the correlation time, rc, for the Mn2+-CoATP interaction is by measuring the frequency dependence of I/fTlp According to the Solomon-Bloembergen theory, significant dependence of l/ftlp on frequency should be detectable if T, -9:s -9:a a.s i.o PPM FIG. 6. Chemical shift correlation map (a) and ID spectrum (a) of Co(NH&ATP in the expanded region to ppm obtained using a 45O mixing pulse. The chemical shift correlation map was acquired using a six-step Jeener pulse sequence, 90"-tl-4-tz (digitize)-t,, where 4 - O", go", BO", and 270" represents the phase of the transmitter relative to the receiver. The period t, is incremented n times to create the 2D data block. Temperature = 25 "C. TABLE I11 Longitudinal relaxation rates and internuclear distances in the binao MnfII). CofNH&ATP COmDkX Atom (TIM)-' rc A P'7 9.8 A 80.7 A Po A p, A, A 520b 7.2 'Measured at MHz. ' Distances were calculated assuming no frequency dependence of the correlation time, re = 8.22 X 10"' s. Measured at MHz. is in the range of 1/u1. The ratio of l/t1~ values for the 7-P at MHz and MHz (Table 111) indicates that 1/ TIM for the Mn2+-CoATP complex is frequency dependent. The correlation time of 8.22 X 10"O s is calculated using Equation 6 and assuming no frequency dependence of re itself. This value of re can be thought of as an upper limit, due to the circumstances of the experiment. The longer time required for Tl measurements at 40 MHz leaves open the possibility of a greater systematic error at 40 MHz due to reduction of Co(1II) to Co(I1). Also we have compensated for pulse imper- A

7 14066 NMR Studies of Substrate and Ion Sites on Pyruvate Kinase fections which can lead to artificially short TI values at 145 MHz on our Nicolet spectrometer. On the other hand, such compensation is not possible on the JEOL PFT-100 spectrometer. Both these potential problems would lead to underestimates of TI values at 40 MHz and the net effect would be an overestimate of rc. Determination of the Distances in the Binary Mn2+- Co(NHA4TP Compkx. The bound-state longitudinal relaxation rates, l/ftlp, are calculated using a value of 15 mm for the dissociation constant of the Mn2+-CoATP complex (28). We have assumed in these calculations that free and bound CoATP are in fast exchange (ie. T~ << Tl~). This is justified by a measurement of the paramagnetic contribution to the T2 relaxation rate, l/ftzp, of2.7 X s-l, which greatly exceeds l/ftlp (354 s-l) and which sets a lower limit on the exchange rate. Since resonances for the a-p and p-p of CoATP are poorly resolved and strongly coupled at 40 MHz, individual relaxation rates could not be calculated at that frequency. Hence, the correlation time calculated for the y-p is used for all the phosphorus resonances in the distance calculation. The A and A diastereomers of CoATP have overlapping y- P resonances, and the distance calculation at the y-p is thus necessarily an average. The A and A isomers are sufficiently well resolved for the a-p and B-P resonances on the other hand that separate distances can be calculated for these nuclei. As shown in Table 111, the Mn2+-(/3-P) distances are the same for the two diastereomers while the Mn2+-(a-P) distances are clearly different. The distances calculated here are too large to permit inner sphere and probably even second sphere coordination of Mn2+ and the phosphates of CoATP. While this is somewhat surprising, it is interesting to compare our results with the 13C and 16N nuclear relaxation measurements of Levy and Dechter (29), who found a second metal site on AMP. Based on several Mn2+-carbon and Mn2+-nitrogen distances, they proposed inner sphere coordination of Mn2' by the N-7 nitrogen of the adenine ring. Likewise. it has been found by Martin and Mariam (30) that simultaneous coordination of N-7 and a-p of AMP does not occur. 31P Nuclear Relaxation Studies of Co(NHs)dTP in Ternary and Quaternary Complexes with Pyruvate Kinase-To determine the interaction of Mn2+ and the phosphorus nuclei of CoATP at the active site of pyruvate kinase, a series of 31P nuclear relaxation measurements similar to those with the binary complex was made. As for the studies with the binary complex, the NMR samples for the enzyme experiments were also bubbled with O2 prior to NMR measurements. The presence of this small amount of added O2 does not affect the activity of pyruvate kinase. The enzyme exhibits similar ac- tivities in the presence and absence of added 02. The addition of Mn2+ to solutions containing pyruvate kinase and CoATP resulted in a gradual increase in the relaxation rate, 1/Tlp, for all three phosphorus nuclei of CoATP (Fig. 7). In these experiments, it was not possible to saturate the enzyme with MnZ+, since at levels of p~ Mn2+, the 31P line widths had already begun to broaden to the point where line widths began to exceed the separation between lines so that individual resonances were too broad to resolve. The presence of pyruvate kinase clearly enhances the relaxation rates of all three phosphorus nuclei of p,y-bidentate CoATP, demonstrating the formation of a ternary enzyme- Mn-CoATP complex. Similar enhancements of relaxation rates are also observed in solutions containing pyruvate (data not shown, but see Table IV). Although pyruvate kinase activity is specific for the A diastereomer of the metal-atp complex (31), increases in the relaxation rates of the A diastereomer are also observed. CoATP does not bind tightly to r I 0' (5" ) o ' Mn** (um) FIG. 7. The effect of Mn(II) on the longitudinal relaxation rate of the phosphorus nuclei of Co(NH&ATP in the presence of pyruvate kinase at 40 MHz (0) and 145 MHz (0, W, +, A, T). The solutions contained 20 mm Co(NH&ATP, 80 mm Tris, ph 7.5, and 40 X 10- M pyruvate kinase, as well as 10 mm KCl. Titrations at 145 MHz are shown for the a-p (A-isomer (A) and A-isomer (A-isomer (+) and A-isomer (m)), and y-p (0) resonances. TABLE IV Longitudinal relaxation rates and internuclear distances in the ternary and quaternary pyruvate kinase-mn-substrate complexes Nucleus (TIMY ra rb C d c d c d r.n= S" A A A Ternary complex' (pyruvate kinase-mn-co(nh3)jtp) A pa A A pp h p," Quaternary complexesf (enzyme-mn-co(nh3)jtp-pyruvate) Pa PB P, Calculated under maximum frequency dependence of the correlation time, with 7, = 7.9 x lo-' s at 145 MHz. * Calculated under assumption of no frequency dependence of the correlation time, with 7, = 2.3 X lo-' s. e Calculated assuming KO = Calculated assuming KO = 150 pm. e Average for distances calculated under maximum frequency de- pendence of the correlation time. 'Co(NH3)JTP in the ternary complex and quaternary complex was 17 mm. [Pyruvate] in the quaternary complex was 17.6 mm. Other conditions were as described in the legend to Fig. 7. gall values shown were measured at 145 MHz. l/tl~ for P, at MHz was measured to be 1475 s-'. pyruvate kinase and under the conditions of these experiments, it is present in both the binary and ternary (enzyme) complexes with Mn2+. The measured relaxation rates reflect a weightedaverageof the relaxation rates for these two complexes. Using the appropriate dissociation constants, one can calculate the contribution of the ternary complex to the observed paramagnetic relaxation rates. A dissociation constant of 15 mm is used for the binary Mn2+-CoATP complex. Since p,+bidentate CoATP and B,y-bidentate Cr(H20).ATP can be assumed to display similar properties of binding with pyruvate kinase, the binding constants previously determined for Cr(H20).ATP (5) are used here in the calculations for the CoATP complexes. The values of T l obtained ~ from these calculations for the ternary enzyme-mn-coatp and quater-

8 NMR Studies of Substrate and Ion Sites on Pyruvate Kinase nary enzyme-mn-coatp-pyruvate complex are given in Table IV. Estimation of Correlation Times for the Ternary and Quaternary Enzyme Complexes-In order to provide estimates for the correlation times for the 31P-Mn2+ interactions in the ternary and quaternary enzyme complexes, we have measured the 31P nuclear relaxation rates at two frequencies, and MHz (Table N). In both cases, a significant inverse frequency dependence is observed. Equations 6 and 5, which assume no frequency dependence and maximal dependence of T,, respectively, were used to calculate the two extreme values of 7,. The values obtained in this way are shown in Table IV, and can be compared with valuespreviously obtained or estimated in other studies of pyruvate kinase-mn-complexes (9-11,32, 33). Determinations of Distances in the Ternary and Quaternary Enzyme Complexes-The distances between enzyme-bound Mn2+ and the 31P nuclei of bound CoATP can be calculated from Equation 2 and the values for l/ftlp (i.e. l/tl~) obtained here,provided that the observed relaxation rates are not dominated by the exchange term in Equation 1. That this is not the case is shown in two ways: 1) values for l/ftzp are substantially greater than for l/ftlp As shown in Table IV, the spin-spin relaxation rate for the y-p of CoATP of 3.8 X IO4 s-l is 2 orders of magnitude greater than the longitudinal relaxation rate of 358 s-l. 2) The longitudinal relaxation rates show a significant frequency dependence. As shown above, this is observed in the present case. Thus the assumption of the fast exchange condition would seem to be valid in the present case. The distances calculated for the ternary and quaternary enzyme complexes are shown in Table IV. The distances are consistent with a pyruvate-induced conformational change at the active site of pyruvate kinase. While such a conforma- tional change has been observed previously, the effect of this transition on the conformation of ATP at the active site has not previously been determined. DISCUSSION The NMR studies described here are consistent with the many previous active site structural studies on pyruvate kinase. The2e data suggest that Li+ at the monovalent cation site is 5.6 A from Cr(II1) at the ATP site, both in the presence and the absence of pyruvate. Addition of Mg2+ increases this distance slightly, perhaps due to electrostatic repulsion. Previous studies of enzyme-mg-pyruvate-cr(hzo),atp complexes indicate that the carboxyl and carbonyl carbons of pyruvate are 6.1 A from Cr(II1) at the active site. This distance is midway between the Li-Cr distances in the absence and presence of M$+. The Li-Cr distance in the presence of Mg2+ fits well into the model previously proposed (6) for the active site of pyruvate kinase, a portion of which involving Li+ is shown in Fig. 8. Binding of pyruvate induces a conform-ational change which decreases the Li-Mn distance from 8.3 A to 5.8 A (1). It has been suggested that the function of the monovalent cation in the pyruvate kinase mechanism is to assist in the orientation of the carboxyl group of pyruvate during the phosphoryl transfer reaction. This model is supported by comparison of Mn-pyruvate distances on the one hand, and Mn distances to T1+ (7), monomethyl ammonium ion (34), and Li+ (1,B-10). However, additional support for the binding of monovalent ions near the carboxyl carbon of bound pyruvate is provided by the present study. The present studies also clarify a discrepancy between the Li+-Mn2+ distance calculated in the absence of pyruvate or P- enolpyruvate (1) and the analogous T1+-Mn2+ separation re- FIG. 8. Active complex of pyruvate kinase, monovalent and divalent activators, Cr(H2O)*ATP, and pyruvate consistent with the 17 distances measured previously (6) and the data of Tables I1 and IV. a ported previously 7). The Li+-Mn2+ distance we calculated previously of 11.0 was obtained assuming that the enzyme was saturated with Li+ under the conditions of the NMR experiment, 50 mm Li+ and 2.1 pm pyruvate kinase sites. Our value for the affinity of Li+ for the enzyme in the absence of other substrates is 120 mm, a value substantially higher than the value we had obtained from kinetic studies of 11 mm. Applying the new KG for Li+ to a recalculation of the distance yields a value of 8.3 A for the Li+-Mn2+ separation. This value is es!entially the same as the value measured from TI+ NMR, 8.2 A (7). The EPR studies of the isotropic and anisotropic forms of the enzyme-mn2+-li+-pyruvate complex are similar to those reported for the enzyme-mn2+-li+-(p-enolpyruvate) complex. In both cases, the so-called anisotropic or K+-like form of the enzyme predominates at 23 C, while the fraction existing in the isotropic form is increased as temperature is decreased. The data presented here indicate that pyruvate produces slightly more of the K+-like conformation than does P-enolpyruvate (92% compared to 81% (8)). The Mn2+-P distances obtained in the present study for the binary Mn2+-CoATP complex are larger than those previously measured (25). This would appear to be a consequence of reduction of Co(II1) to paramagnetic Co(II), as we have described elsewhere (26). When we carried out the 31P NMR measurements with no precautionary treatments of our CoATP solutions, we obtained results essentially similar to the previous report. On the other hand, pretreatment of our Co(NH&ATP solutions either by O2 bubbling or by addition of NH: to prevent and/or reverse Co(II1) reduction, we obtain the results shown here. The results indicate that the Mn2+ interaction with CoATP does not involve inner sphere or even outer sphere coordination of the phosphates of ATP. This may be due to electrostatic repulsion of Mn2+ and Co3+, and may provide an explanation for the weak association between Mnz+ and CoATP (KO 2: 15 mm (28)). A more likely coordination site for a second metal in this type of complex has been suggested by Levy and Dechter (29), who have determined from 13C and 15N relaxation studies that the second metal in a M2ATP complex is coordinated to the N-7 nitrogen of the adenine ring on ATP. Proton NMR studies of the Mn2+-CoATP could possibly clarify this point. The 31P nuclear relaxation studies of the ternary and quaternary pyruvate kinase complexes provides the opportunity for several comparisons of active site structure. For example,

9 14068 NMR Studies of Substrate and Ion Sites on Pyruvate Kinase it has not previously been possible to examine the quaternary complex of enzyme, Mn2+, ATP, and pyruvate, since this, complex is a functional one, and breaks down to products much faster than the time required for NMR measurements. In the present case, using CoATP, which is not a functional substrate for pyruvate kinase, we have been able to examine the effect of pyruvate on the conformation of the ternary enzyme-mn2+-coatp complex. As seen in Table IV, pyruvate effects a decrease in the Mn-P distances for this complex, bringing Mn2+ and CoATP closer together at the active site. This is a different result than that obtained by Sloan and Mildvan (11), who found no effect of oxalate, a suggested transition state analogue of pyruvate, on the geometry of bound ATP in the enzyme-mn2+-atp complex. Our conditions differ from those in this latter study, in that we have used CoATP and pyruvate, while Sloan and Mildvan (11) used ATP and oxalate. Since neither of these complexes is a functional one, it is difficult to say which of these represents a more accurate model for pyruvate kinase. One can at least suggest that the presence of a second metal (coordinated to ATP) alters the structure of bound ATP, while pyruvate brings the Mn2+ and the nucleotide closer in the quaternary complex. The individual functional roles of the two divalent cation sites at the active site of pyruvate kinase have not yet been clearly defined. It wouldbe interesting in this context to compare the structures of the enzyme-mn-atp complexes in the presence and absence of an ATP-coordinated metal. The one-metal complex (using ATP instead of CoATP) has been examined by Sloan and Mildvan (ll), who calculated Mn-P distance! to the and y-phosphorus nuclei of 5.1, 5.0, and 4.9 A, respectively. At first glance, these distances would seem to be quite different from those we calculate here. However, it should be pointed out that the value for the correlation time rc used in the distance calculations in this previous report is perhaps the shortest value on record for a pyruvate kinase-mn2+ complex, 5 X 10"' s. Table V presents a survey of the values estimated by various methods for pyruvate kinase-mn2+ complexes. The values for re range from 5 X 10"" s to 9.4 X s. One can reasonably ask whether the correlation times for various Mn2+ complexes of this enzyme can actually vary by a factor of 19, or whether these numbers represent some degree of error in either the methods for measuring T= or the theory of paramagnetic interactions in macromolecular systems. In any case, in the absence of more sophisticated means of dealing with these questions, we can perhaps best compare two sets of data from different laboratories by using the same correlation time in both sets of calculations. Even this approach is fraught with problems, however. As defined (16), the correlation function, f(tc), depends both on rc and on the frequency of the NMR experiment. Even when rc is frequency independent, experiments at two different frequencies result in two different values for TABLE V Correlation times estimated or measured for complexes of Mnz+pyruvate kinase Reference Method 7, Sloan and Mildvan (11) Frequency dependence 5 X 10"' for 'H of ATP James et al. (32) Frequency dependence 7-26 X 10-l' of water protons Raushel and Villafranca Comparison of 6Li and 37 X lo-'' (9,10) Reuben and Cohn (33) "Li Frequency dependence 94 X 10-l' of water protons S f(tc). On the other hand, T, itself can be frequency dependent, and thus the estimation of T~ by measurement of the frequency dependence of 1/T, is best performed in the frequency range in which the value of T, is to be used for distance calculations. The data of Sloan and Mildvan (11) were obtained at 40.5 MHz, while our studies were performed at 40.5 and at MHz. The value of T, we obtain from the frequency dependence of l/ftip of phosphorus, assuming no frequency dependence of rc, of 2.3 X lo-' s is perhaps a reasonable one to use for comparison of the two data sets. Recalculation of the data for the enzyme-mn2+-atp complex usiqg this value for rc gives Mn-P distances of 6.4,6.3, and 6.1 A for the 01-, 6-, and y-phosphorus nuclei of ATP. These values are well within the limits of error for our values in the pyruvate kinase-mn2+- CoATP-pyruvate complex. If any conclusion can be drawn from this comparison, it would be that any changes in the conformation of ATP due to Co(II1) coordination are masked by the error limits in the calculations, at least a part of which is the uncertainty regarding the appropriate value of 7,. Regardless of the choice of a correlation time, there would appear to be a small effect of Co(1II) on the conformation of ATP which we may infer by considering each set of data independently. If only relative distances are considered, the errors in the distances in Table IV may be considered to be quite small. If the error in l/ftlp is lo%, then the relative errors in the calculated distances, due to the sixth root dependence of Equation 2, are? 1.6%. The distances we calculate in Table IV are thus consistent with a bent conformation for the triphosphate moiety of CoATP. The data for the enzyme-mn- ATP complex in the absence of the second metal are consistent either with an extended conformation of the triphosphate chain or with a rotation of the chain so that the plane defined by the bent chain is perpendicular to the Mn-P vectors. In either case, it is clear that the coordination of Co(II1) effects a significant change in the conformation of the triphosphate chain of bound ATP. This conclusion is drawn from consideration of relative distances only, and thus is independent of one's choice of a correlation time. REFERENCES 1. Hutton, W. C., Stephens, E. M., and Grisham, C. M. (1977) Arch. Biochem. Biophys. 184, Grisham,C. M., and Hutton, W. C. (1978) Biochem. Biophys. Res. Commun. 81, Stephens, E. M., and Grisham, C. M. (1979) Biochemistry 18, Klevickis, C., and Grisham, C. M. (1982) Biochemistry 21, Gupta, R. K., Fung, C. H., and Mildvan, A. S. (1976) J. Biol. Chem. 251, Mildvan, A. S., Sloan, D. L., Fung, C. H., Gupta, R. K., and Melamud, E. (1976) J. Biol. Chem. 251, Reuben, J., and Kayne, F. J. (1971) J. Biol. Chem. 246, Ash, D. E., Kayne, F. J., and Reed, G. H. (1978) Arch. Biochem. Biophys. 190, Raushel, F. M., and Villafranca, J. J. (1980) J. Am. Chem. SOC. 102, Raushel, F. M., and Villafranca, J. J. (1980) Biochemistry 19, Sloan, D. L., and Mildvan, A. S. (1976) J. Biol. Chem. 251, Schlessinger, G. (1960) in Inorganic Syntheses (Rochow, E. G., editor Vol. 6, pp , McGraw Hill, New York 13. Cornelius, R. D., Hart, P. A., and Cleland, W. W. (1977) Inorg. Chem. 16, Dunaway-Mariano, D., and Cleland, W. W. (1980) BiochemistSy 19, Ernst, R. R., and Anderson, W. A. (1966) Rev. Sci. Znstrum. 37,