alkaline phosphatase · 31p nmr signals from substrates and products of alkaline phosphatase have...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1985 by The American Society of Biological Chemists, Inc.

Vol. 260, No. 5, Issue of March 10, pp. 2875-2883, 1985 Printed in U. S. A.

Alkaline Phosphatase 31P NMR PROBES OF THE MECHANISM*

(Received for publication, October 11,1984)

Peter Gettins$, Mary Metzler, and Joseph E. Coleman From the $Department of Biochemistry, School of Medicine, Vanderbilt University, Nashville, Tennessee 37232 and the Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, Connecticut 06510

31P NMR signals from substrates and products of alkaline phosphatase have been adapted to measure the rates and product ratios for the hydrolysis and phosphotransferase reactions from pH 6 to 10. Below pH 8, glycerol is a poorer acceptor than H,O (glycerol phosphates:Pi = 0.5). Tris is a more effective acceptor below pH 8, showing a maximum acceptor efficiency at pH 8 (Tris phosphate:Pi = 2). Phosphotransferase efficiencies are in the order expected for the pK,s of the alcohol groups, Tris < glycerol C1, C3 < glycerol C2. Tris and glycerol induce chemical shifts in "'Cd(II) present at the A site but not the B or C sites of the metal triad present at each active center of Cd(II), alkaline phosphatase, suggesting that the alcoxides of the acceptors coordinate the A site metal and become the nucleophiles attacking the phosphoseryl residue (E-P) in the second step of the mechanism. The interaction is through the oxygen of Tris.

The transferase activity of the amino alcohol shows a bell-shaped pH dependency. Aliphatic alcohol acceptors show small increases in acceptor activity between pH 6 and 8, with 5-fold increases from pH 8 to 10 (at pH 10, glycerol phosphates:Pi = 2.5). "P NMR inversion transfer has been used to measure the koff for Pi dissociation from the noncovalent enzyme complex (E-P). For the Zn(II)4 alkaline phosphatase koff is essentially pH independent at -35 s-'. For Cd(I1) or Mg(I1) at the B site in place of Zn(II), kef i 1 s-'. C1- ion, which appears to coordinate the A site metal ion, enhances k,n, suggesting that both C1- and HPO2- can coordinate the A site metal ion in a 5-coordinate intermediate. pH control of the alkaline phosphatase mechanism appears to reside in the stability of E-P and not the dissociation of E-P, compatible with the hypothesis that the activity-linked pK. is that of a HzO molecule coordinated to the A site metal, which in the hydroxide form becomes the nucleophile attacking the phosphoseryl group (E-P).

Alkaline phosphatase from Escherichia coli forms two phos- phoenzyme intermediates, a noncovalent complex, E-P,' in

* This work was supported by National Institutes of Health Grant AM 09070 and National Science Foundation Grant PCM 76-82231. The 200-MHz NMR facility is partially supported by National Sci- ence Foundation Grants PCM 77-18941 and CHE 79-16210. The 400- MHz NMR facility is supported by Vanderbilt University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduer- tisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are: E . P, noncovalent product Michaelis complex; E-P, covalent enzyme phosphoseryl intermediate; AP, alkaline phosphatase; p-NPP, p-nitrophenyl phosphate.

which the product phosphate is coordinated to one of two Zn(I1) ions present at each active center and a covalent intermediate, E-P, formed by phosphorylation of serine 102 (1). High concentrations of both intermediates can be formed under a variety of conditions and this has allowed extensive 31P NMR studies of the equilibrium between these two intermediates (2, 3). We have recently extended the 31P NMR techniques applied to alkaline phosphatase to the measure- ment of substrate turnover which allows us to probe two specific features of the mechanism of alkaline phosphatase, namely phosphate transfer from E-P to acceptor alcohols and dissociation of the product phosphate.

The 31P chemical shifts of inorganic phosphate, p-nitrophenyl phosphate, and a variety of other phosphate monoesters formed by the enzyme from the respective alcohols are sufficiently different that the separate 31P signals can be followed on a time scale of minutes. Hence the progress of both the hydrolysis and transferase reactions can be followed directly in the NMR tube. The direct detection of phospho- transfer to a variety of acceptors has provided much new information about the transferase reaction. Using 'Wd NMR on cadmium-113 substituted enzyme species it has also been possible to investigate specific interactions between phosphate acceptor alcohols and the three metal ions at each active center.

A formulation of the alkaline phosphatase mechanism including both hydrolysis and phosphotransferase reactions is given in Scheme 1. It is probable that the transferase mechanism is symmetrical about E-P and involves both formation of an E. R'OP complex and dissociation of the product phos- phoalcohol as separate steps. This would then exactly parallel normal hydrolysis with R'OH taking the place of HzO as the attacking nucleophile.

The rate-limiting step for the hydrolysis reaction at low pH is dephosphorylation of E-P, while at high pH where the enzyme is maximally active, significant equilibrium concentrations of E-P disappear and dissociation of product from E.P becomes rate limiting (4,5). Since the 31P resonances of enzyme-bound E. P and free Pi are in slow chemical exchange and are well separated it is possible to apply the technique of 31P NMR inversion transfer to determine the rate constant

E + RO P

RO-

ROP- E-P k-2

R ' O H 3

SCHEME 1

k4 E . P e k-4 E t Pi

k - 4 '

-G EaR'OP- E t R ' O P

2875

2876 31P NMR of Alkaline Phosphatase

for phosphate dissociation, krr, when this falls within the limits of IO" s" to 100 s-'. Influences of different metal ion species, anions, and pH on the rate-limiting step can be readily determined by this method and are reported here.

The E-P + E . P equilibrium is pH dependent and for the native enzyme [E-PI = [E.P] at pH 5 (2). As the pH is raised the equilibrium shifts to the right in favor of E.P. Of the 2 mol of phosphate bound per dimer of enzyme at pH 6 ([enzyme] s 1 mM), only one remains bound at pH 8, while at pH 9 less than one is bound (2). The latter result, clearly observed in the NMR tube, appeared to be inconsistent with reported values of KI for phosphate at pH 8 which are in the range 1- 10 p~ (6-lo), although one report on the properties of the calf intestinal enzyme shows an approximately linear fall in pK1 with increasing pH in the range pH 7 to 10 (11).

Based on our previously proposed mechanism for dephosphorylation involving a Zn-coordinated hydroxide generated from the ionization of a metal-bound solvent, the pH shift in the E-P + E-P equilibrium reflects the increase in dephosphorylation resulting from the higher concentration of the nucleophile, ZnOH- (1-3, 12). The shift in the coupled equi- libria (E-P e E.P and E .P t= E + Pi), therefore, does not necessarily require the Pi dissociation rate (kff) to increase with pH in order to account for the change in rate-limiting step. If k,ff for Pi does not increase with pH, however, the fall in apparent binding of phosphate must involve the on rate. Inversion transfer has been used to separate these possibilities and to delineate other factors which modulate the rate-limiting step at alkaline pH, i.e. phosphate dissociation.

MATERIALS AND METHODS

Enzyme Preparation-Alkaline phosphatase was isolated from an overproducing strain of E. coli, CW3747, and purified as previously described (13). Enzyme concentrations were determined spectropho- tometrically at 278 nm using P'' = 0.72 (14) and a molecular weight for the dimer of 94,000, calculated from the amino acid sequence (15). The activity of the enzyme was determined by the hydrolysis of p - nitrophenyl phosphate in 1 M Tris-HC1 at pH 8, 22 "C, following release of p-nitrophenolate at 410 nm. Native enzyme has a specific activity of 2500 k 500 mol of substrate hydrolyzed/h/mg of enzyme. All buffer solutions were made metal free by extraction with dithizone in CCl,. Glassware was acid washed. Twice deionized water was used throughout. Apoenzyme was prepared by dialysis against 2 liters of 2 M (NHJ2S04, pH 9, with two changes of dialysate followed by 3 X 2 liters of desired buffer to remove ammonium sulfate (2). Ammonium sulfate was ultrapure grade (Schwarz/Mann). Concentration of enzyme samples for NMR was carried out at 4 "C in an Amicon ultrafiltration cell (made metal free by soaking in 1 mM o-phenan- throline) using a PM30 membrane. Preparation of Zn(II),AP and Cd(II)6AP was achieved by addition of stoichiometric amounts of stock solutions of ZnCl, or 1'3Cd(CH3COO),, respectively, to apoen-

enriched cadmium metal (U.S. Services, Summit, NJ). The hybrid zyme. The cadmium acetate was prepared from 95% isotopically

species Zn(II),Mg(II),AP and Zn(II),Cd(II),AP as well as the asym- metrically occupied Cd(II),AP were prepared as described elsewhere (3). Metal ion content of metalloenzymes was checked by flame atomic absorption using an IL157 spectrometer (Instrumentation Laboratories, Wilmington, MA).

31P N M R Spectra-Spectra were recorded on a Bruker CXP200 spectrometer operating at 80.9 MHz (Yale) or on a Bruker AM400 at 162.0 MHz (Vanderbilt). The former spectrometer was equipped with a broad band tunable probe, while a dedicated 31P probehead was used on the latter. In both cases samples of 1.8 cm3 were used in 10- mm tubes fitted with a Vortex plug and a co-axial capillary containing D,O for the external field-frequency lock. Broad band proton decou- pling of 2 watts was employed. Chemical shifts are reported relative to 1% phosphoric acid.

31P NMR Assay of Hydrolysis of p-NPP and Phosphotransferase Actiuity-The initial concentration of p-NPP was 20 mM in 0.01 M Tris acetate containing 10% D20 for the field-frequency lock. To this was added an appropriate quantity of enzyme to permit collection of 80 or 160 scans for each time point in the hydrolysis course. For the

P-NPP

A.

Pi

glycerol-0-P B.

6 5 L 3 2 1 PPm

FIG. 1. "P NMR (80.9 MHz) spectrum of substrate,p-nitrophenyl phosphate, and products, inorganic phosphate and 0- Tris phosphate (A) or glycerol phosphates (B) of alkaline phosphatase. The composition at the start of reaction was 20 mM p-nitrophenyl phosphate in 1 M Tris or 3 M glycerol, pH 8. To this was added alkaline phosphatase (5 X 10" M). The reaction was performed at 293 K. The spectra were collected after about 5 0 % hydrolysis and represent 80 scans. In the case of glycerol ( B ) the upper spectrum is a proton-coupled spectrum, while the lower is a proton-decoupled spectrum of the same sample. There was a small temperature difference between the two acquisitions due to dielectric heating in the latter that accounts for the small differences in chemical shift for the same resonances in the two spectra. Note that there appears to be a different nuclear Overhauser enhancement to the glycerol monophosphate resonances than to inorganic phosphate. Since changes in ratios of rates of formation are to be compared it is not necessary to make an explicit allowance for this, though this would be necessary for calculation of absolute rates of product formation.

phosphotransferase reactions 1 to 3 M acceptor alcohol (Tris or glycerol) was added to the above reaction mixture. Typical "P NMR spectra for two transferase reactions, one with Tris and one with glycerol as acceptor, are shown in Fig. 1, A and B, respectively, after

31P NMR of Alkaline Phosphatase 2877 produces the following relationship which relates the Pi resonance area of the inversion transfer spectra to time after inversion,

PEAK AREA

0 2 4 6 6 1 0 TIME

FIG. 2. Time course of product formation from alkaline phosphatase-catalyzed breakdown of p-nitrophenyl phosphate. The ordinate represents the peak area of the "P resonance of each product in arbitrary units and is proportional to product concentration. The abscissa is also in arbitrary units since only ratios of initial rates of formation are to be compared. 0, represents 0-Tris phosphate; 0, represents inorganic phosphate. A, results obtained at pH 6; B, results obtained at pH 9.

-20 min of reaction. The Pi signal represents the hydrolysis portion of the reaction. The most downfield peak in the glycerol reaction (Fig. 1B) represents the C(1,3) monoesters resonating at 4.35 ppm, while the C(2) monoester resonates a t 4.0 ppm as documented by the expected 1H-31P coupling in the undecoupled spectrum (Fig. lB, upper spectrum).

Each assay consisted of 10 or 20 points as a basis for determining initial rates and ratios of product formation. Fully relaxed spectra were obtained by using a 30" pulse angle and a repetition time of 5 a. A sweep width of 2000 Hz was used. Spectra were broad band decoupled using 2 watts power and were obtained on a Bruker CXP- 200 operating at 80.9 MHz for phosphorus. The temperature was maintained at 293 K. Quantitation of the two products, inorganic phosphate and 0-Tris phosphate (or glycerol phosphate), was performed by integration of the respective 31P NMR resonances. These were well separated from each other and from the substrate p - nitrophenyl phosphate (see Fig. 1).

Typical assays plotting the rise in Pi (hydrolysis) and the new ester (Tris phosphate) as functions of time are shown in Fig. 2 for assays at pH 6 (Fig. 2A) and pH 8 (Fig. 28). At acid pH values significant product inhibition was observed at times longer than 2 min, while at pH 8 and above very little product inhibition was observed which appears to correlate with the falling Ki value for phosphate (see below).

Inversion Transfer "P NMR Spectra-This was obtained by ap- plying a selective 180" pulse to the inorganic phosphate resonance followed by a short delay, T, and a 90 ' nonselective observation pulse. A delay of 8 s between scans allowed for recovery of equilibrium magnetization. The selective 180 pulse on the phosphate resonance was achieved by placing the carrier a t this frequency and using a DANTE pulse sequence (16) of several closely spaced short pulses.

to give band width and sideband separation that resulted in good The pulse separation and overall time of the pulse train were chosen

inversion with minimum perturbation of the other resonance(s) (see legends for details of individual spectra). By collecting spectra a t different values of T, the transfer of inversion of magnetization from the Pi resonance to the E.P resonance on the enzyme through chemical exchange could be followed from changes in resonance area (17).

Kinetic Analysis-The observables in the 31P NMR spectra are areas of the Pi and E . P resonances as a function of time, T, after inversion of the Pi resonance. In addition the peak areas of normal spectra can be used to determine the total concentrations of E. P and

treatment of the equilibrium Pi, which are invariant since the system is at equilibrium. Simple

k4

k-4 E.P e E + Pi

SCHEME 2

SCHEME 3

where [E. PI-, and [Pi] are the equilibrium concentrations of E. P and Pi, respectively. Pi, is the observed area of the Pi resonance expressed as conentration, k4 is the Pi dissociation constant, and t is the time after resonance inversion. In plots of P, against time, the data were fitted to this equation by optimal choice of k4.

ll3Cd NMR-Cadmium spectra were recorded on a Bruker CXP- 200 spectrometer, operating at 44.37 MHz. A sweep width of 10,OOO Hz, pulse angle of 30 O, and repetition time of 1 s were used. The temperature was 293 K. Chemical shifts are reported relative to 0.1 M Cd(C104),. Samples were 2 ml in a 10-mm tube fitted with a coaxial insert containing D,O for the field-frequency lock.

RESULTS

Phosphate Transfer Catalyzed by Zn(II), Alkaline Phospha- tuse-The ratio of rates of formation of inorganic phosphate and of 0-Tris phosphate and glycerol phosphates was determined over the pH range 6 to 10. Tris concentration was 1 M and glycerol concentration was 3 M. At each pH value three separate determinations were made and an average value taken. There are dramatic pH dependencies to the phosphotransferase reaction which vary with the nature of the acceptor. With Tris as the acceptor alcohol, a bell-shaped pH dependence is observed with a maximum at pH 7.5 followed by a rapid decrease in phosphotransferase at pH values above 8 (Fig. 3). In contrast, when glycerol is the acceptor, phosphotransferase activity is quite low at pH values below 8 (total glycerol phosphate:Pi = 0.5) but rises rapidly at the higher pH values (Fig. 3). At pH 9.5, the total glycerol phosphate:Pi = 2.5.

The 31P NMR method shows that all three OH groups of glycerol are used as acceptors (Fig. 1). An additional finding is that the relative acceptor efficiency of the C(1,3) uersus the C(2) OH groups changes with pH. At low pH the equivalent c(1,3) OHs are the preferred acceptor, while by pH 9.5 the C(2) OH is the predominant acceptor in the ratio 1.41, even though the C(1,3) OH group has twice the effective concentration (Fig. 3).

In the normal pH range (7-9) Tris is a considerably more potent nucleophile than aliphatic alcohols like glycerol in competing for solvent. Formation of Tris phosphate is 2-3 times faster than glycerol phosphate even when 3 M glycerol is used. In the extreme alkaline pH range, however, this advantage is lost. The sharp decline in Tris as an acceptor at pH values above 8 means that pH rate profiles for alkaline phosphatase determined in the presence of enough Tris to be a significant acceptor are composite curves and need to be reinterpreted (see "Discussion").

Phosphate Transfer Catalyzed by Co(II), Alkaline Phospha- tase-Hydrolysis of p-NPP by Co(II), alkaline phosphatase was also followed by 31P NMR over the pH range 6 to 10 under comparable conditions to those for the zinc enzyme except that somewhat higher concentrations of enzyme were necessary (-&fold) due to the lower activity of the Co(I1)- substituted species (18, 19). In no case was there detectable 0-Tris phosphate formation attributable to the cobalt enzyme. The presence of residual zinc to the extent of 1 to 2% gave a small amount of Tris phosphate. From the known


k-

Ly 1.5-

0 1.0-

0.5 -

0-

a

t- U 3

0 fx a

a a+ I I I

6 7 8 9 1 0 PH

FIG. 3. pH dependence of transferase and hydrolase products in the presence of 1 M Tris and 3 M glycerol. 0, ratio of initial rates of formation of Tris phosphate to inorganic phosphate; 0, ratio of initial rates of formation of glycerol C1, C3, and C2 monophosphate to inorganic phosphate; 0, ratio of initial rates of formation of glycerol C2 monophosphate to glycerol Cl,C3 monophosphate. Note that at pH 9.5 the C2 transferase product exceeds the C1 plus C3 product by the ratio 1.3:l.O. The reaction mixture in the case of glycerol contained initially 3 M glycerol in 50 mM Tris (solely as buffer and not at high enough concentration to result in competition with 3 M glycerol). Enzyme (10-7-10-8 M) was used depending on pH. The substrate concentration was 20 mM.

enhancement of activity by zinc alkaline phosphatase, allowance could be made for this. It must, therefore, be concluded that for the cobalt enzyme the ratio of 0-Tris phosphate to Pi formed is <0.1, since levels approaching this would have been readily measurable.* In contrast, significant phospho- transfer from p-NPP to glycerol catalyzed by the cobalt enzyme can be detected (data not shown, see below).

'I3Cd NMR, Effect of Tris and Glycerol on Cd(II)s Alkaline Phosphatase-Previous NMR and activity data, especially comparisons of the pH dependence of the E-P + E. P equilibrium between the slowly turning over Cd(I1)-substituted enzymes and the native Zn(I1) enzyme, have suggested that the nucleophile in the dephosphorylation reaction on the hydrolysis pathway is a Zn(I1)-coordinated hydroxide (2, 20). Of the three metal sites (A, B, and C) located at each active center, the ZnOH- appears to be at A site (1, 21).

In order to detect the possible preferential interaction of the alcohol acceptors with one of the three metal ions at the active center we examined the effect of the alcohols on the "3Cd(II)JP, since l13Cd(II) chemical shifts are very sensitive to ligand rearrangements or substitutions. The '13Cd NMR spectrum of phosphorylated Cd(IIh alkaline phosphatase at low pH consists of three resonances arising from cadmium in each of the three metal sites, A, B, and C (1, 20, 21). It has been shown previously that only A site is sensitive to C1- ion,

* In Zinc Enzymes (Spiro, T., ed) (1983) p. 166, John Wiley and Sons, New York and in Advances in Enzymology (Meister, A., ed) 55, p. 394, John Wiley and Sons, New York (1983) due to an error in a sentence, we quoted work by Gottesman, M., Simpson, R. T., and Vallee, B. L. (1969) Biochemistry 8,3776-3783 as showing that cobalt alkaline phosphatase has a greater transphosphorylation to hydrolase ratio than does the zinc enzyme when the opposite is true. The conclusion of their paper is that no detectable Tris-phosphate results from the cobalt enzyme at pH 8, by chemical determination of free phosphate. This is in agreement with the present NMR findings for a wider pH range.

i.e. C1- shifts only the A site resonance. A maximum of two C1- binding sites appear to be present (12). Since coordination of Tris through either amino or hydroxyl moieties should produce significant perturbation of the cadmium chemical shift, the effect of Tris on the '13Cd NMR spectrum of Cd(II)eAP was examined. Fig. 4, A and B, shows the effect of adding 0.5 M Tris to a 2 mM sample of Cd(II)&P in the E-P form at pH 6 containing no C1- ion. The A site resonance at 138 ppm is shifted upfield -28 to 110 ppm, whereas the B and C site resonances at 69 and 2 ppm, respectively, are unaffected. The C site resonance does, however, become much narrower (a reduction in line width from about 200 to 70 Hz). Addition of 0.1 M C1- shifts the A site resonance back downfield to 120 ppm, while not affecting those from B or C sites (Fig. 4C). A further addition of 0.4 M C1- moves the A site resonance to 141 ppm, only 9 ppm from the A site chemical shift in the presence of 0.1 M C1-, but in the absence of competing Tris (spectrum not shown).

The pK, of Tris is about 8.2 so that the predominant species at pH 6 in the above sample is the cationic Tris-NH;. If interaction of Tris with the A site cadmium is through nitro- gen there should be a much stronger interaction above the pK, of the Tris than below it. Fig. 5 shows the '13Cd NMR spectrum of phosphorylated Cd(1I)AP at pH 9 in 0.1 M C1- to which Tris to a final concentration of 1 M has been added. B and C site resonances are unaffected, while A site resonates moderately upfield at 125 ppm showing that Tris continues to compete for binding at A site but that C1- competition remains as efficient relative to Tris as it was at pH 6.5. The resonance at 141 ppm is due to cadmium stripped from some of the alkaline phosphatase dimers by the neutral Tris ligand that now competes successfully with the enzyme for coordination to cadmium. The equal areas of A, B, and C site resonances suggests that removal of cadmium by Tris is a

noCC ( A )

+0,5M Tris IB)

+0.1 M CC IC)

FIG. 4. "'Cd NMR (44.3 MHz) spectra of phosphorylated llSCd(II)e alkaline phosphatase at pH 6. A , in the presence of no C1-; B, after addition of Tris to 0.5 M; C, B after addition of C1- to 0.1 M. The enzyme concentration was 2.0 mM. Spectrum A represents 87,000 scans, while B and C are each 38,000 scans. The temperature was 293 K. A sweep width of 12,000 Hz was used. The labels A , B, and C above the resonances correspond to the respective metal- binding sites to which cadmium is bound.

31P NMR of Alkaline Phosphatase 2879

free

I I ( I t I , I 1 , " " ' " ' 150 100 50 0

PPm FIG. 5. "'Cd NMR spectrum of phosphorylated "'Cd(II)s

alkaline phosphatase at pH 9 in the presence of 0.1 M C1- and 1 M Tris. The resonance labeled free corresponds in position to that of unbound cadmium in buffer of the same composition. The enzyme- bound '13Cd resonances are at 125, 70, and 2 ppm for A, B , and C sites, respectively, indicating that Tris is still interacting.

TABLE I Chemical shifts of '13Cd resonances of ll3Cd(II)6 alkaline phosphatase

(E-P form) at pH 6 as a function of added glycerol

Glycerol Metal site

A B C

M 0 0.5 2.5

PPm 141.5 70 2 140.2 70 2 135.6 70 2

cooperative process so that the remaining metalloenzyme species still contain 6 eq of cadmium/dimer.

As a further probe of the interaction of acceptor alcohol with the A site metal ion, glycerol was added to a phosphorylated Cd(II)&P sample at pH 6 in the absence of C1-. The '13Cd NMR spectrum prior to addition of glycerol consisted of the expected three resonances from A, B, and C sites at 141.5, 70, and 2 ppm, respectively. Addition of glycerol to a concentration of 0.5 M produced a minor perturbation of A site of 1.3 ppm upfield, but left B and C unaffected (Table I). That this was a real effect was shown by further addition of glycerol to a final concentration of 2.5 M which caused a further upfield shift of 4.6 ppm on A site, without affecting B and C (Table I).

Determination of Phosphate Dissociation Rate, kff, for Zn(II)4 and Zn(II)&fg(II)2 Alkaline Phosphatases-It has long been observed that at alkaline pH the addition of Tris as an acceptor results in enhanced activity such that the original hydrolase (Pi as product) and new phosphotransferase (R'OP as product) activities were additive. Such a finding requires that the rate-limiting step for the hydrolysis reaction must be relatively unaffected, while a fast step, probably R'OP dissociation, can get rid of the new ester rapidly. These relation- ships are possible if the release of inorganic phosphate from E . P is the slowest step in the hydrolysis reaction path by a large factor.

31P inversion transfer can measure the latter step and how it is affected by reaction conditions. At pH 7.4, 31P NMR of 1 mM Zn(II), alkaline phosphatase shows resonances only from ESP and Pi, at 4.3 and 2.3 ppm, respectively (Fig. 6). The two signals show that slow chemical exchange of phos-

I 10

Zn4AP pH 7.4

I I I I I

0 6 4 2 0 PPm

FIG. 6. 81-MHz "P inversion transfer NMR spectra for Zn(II), alkaline phosphatase at pH 7.4. Selective inversion of the inorganic phosphate resonance, using a DANTE pulse train, gave an excitation band width of 40 Hz centered on the Pi resonance. After a delay, 7, a 90 nonselective pulse was applied and data acquisition started. The values of 7 in ms are given beside each spectrum. Each spectrum represents 2000 scans with a delay between each scan of 8 S.

phate between the two environments obtains at 81 MHz, i.e. krr = 1/r cc Aw (221, where r is the lifetime of E .P and Aw is the difference in chemical shift between the two resonances; thus, kOn < 130 s-l. This places an upper limit on the dissociation rate, which holds at all pH values examined, since two resonances are observed throughout the accessible pH range.

A more restricted estimate of kff was previously made from the exchange contribution to the Pi line width at pH 7, possible since E . P and Pi are in fast exchange on the relaxa- tion time scale (23). This gave a value of 24 s-l, in good agreement with k,.,, though it was pointed out that the line width may be influenced by nonspecific binding and was not precisely reproducible from one sample to another. For this reason and also for the greater range of rate constants to which the method is sensitive, transfer of inversion of magnetization was chosen to determine kff in all the experiments reported here.

A representative set of 31P NMR inversion transfer spectra for Zn(II)* alkaline phosphatase is shown in Fig. 6. There is rapid loss of negative area for the Pi resonance and loss of positive area for the E .P resonance as chemical exchange scrambles the phosphorus nuclei whose bulk magnetization was inverted at t = 0 and which were originally present only as free inorganic phosphate. Since the Tl values are greater than 1 s, they contribute little to the intensity changes seen here in times <0.1 s (24), and corrections for Tls of free and bound phosphate have not been explicitly included in Scheme 3 but must be made when inversion rates are slow (see below).

Data at pH 7.4 together with theoretical curves for a range of values of k4 are shown in Fig. 7. The family of curves shows that k4 can be determined by inversion transfer to better than 30%. Best fit values of k4 for ZXI(II)~ alkaline phosphatase in the pH range 5.7 to 9.0 are summarized in Table 11. While


- z ' o r " - 7

- 0. .

I I I I I

0 10 20 T ms

FIG. 7. Time course of transfer of inversion for Zn(II),AP at pH 7.4 based on area of Pi resonance from data in Fig. 6. Solid circles are the areas at specific values of T , and the lines are those calculated from Equation 2 using the values of & indicated on each line.

TABLE I1 Diasociatwn rates of inorganic phosphate from alkaline phosphatase

f k , , J as a function of RH DH k d

u-' x 10"

Zn(II)rAP 5.7 3.3 f 1 5.9 6.0 f 2 6.5 3.0 f 1 7.1 2.5 f 1 7.4 1.5 -t 1 7.7 2.0 f 1 8.8 3.5 f 1

7.2 2.5 f 1 8.0 1.8 f 1

there is a spread of values that is greater than the expected experimental error, k4 shows no consistent change with pH and has a value of -30 s-'.

Mg(I1) is an activator of alkaline phosphatase in addition to the Zn(I1) ion (or ions) at the active center (4, 25). Since both NMR (1, 3, 20) and the crystal structure of alkaline phosphatase (21) show that there are two additional metal- binding sites, B and C, located 4 and 7 A, respectively, from the typical Zn(I1) metalloenzyme site designated A, Mg(I1) effects could involve either one or both of these sites. The Zn(I1) ion at A site is responsible for coordination of the phosphate group in the E -P complex and probably also in E-ROP (2, 20). It also generates metal bound OH- required for dephosphorylation of the phosphoseryl intermediate (1, 2). The metal ion at B site is not essential for phosphorylation of serine 102, although it accelerates this step (1,26). The effect of Mg(I1) occupancy of B and/or C site on the rate-limiting dissociation of E.P was, therefore, tested. The following enzymes (Zn(II)dP, Zn(II)4Mg(II)2AP, and Zn(II)2Mg(II)2AP) were prepared and the Pi + E . P interconversion determined by 31P inversion transfer. In the first case A and B sites are occupied by Zn(I1); in the second A and B

by Zn(II), C by Mg(I1); and in the third A by Zn(I1) and B by

While the chemical shift of the E . P species in Zn(I1)hP is shifted from 4.3 to 3.4 ppm by the addition of Mg(I1) to C site, the k4 is relatively unaffected (Table 11). On the other hand, the presence of Mg(I1) in the B site in Zn(II)2Mg(II)2AP has a much more significant effect on the properties of the enzyme. The "P chemical shift of E.P moves upfield to 1.8 ppm (Fig. 8). The latter chemical shift is even upfield of inorganic phosphate at this pH, but the two species are still in slow exchange on the chemical shift time scale (Fig. 8). The most striking change, however, is shown by the inversion transfer experiment where transfer between Pi and E P has slowed by 30- to 40-fold compared to the ZII(II)~ species shown previously above in Fig. 4. The calculated k, is 1 f 1 s-l (Table 111). Since 100 ms rather than 10 ms are required for significant transfer to take place, the data have been corrected for the contribution of Tl. Since k, is the same order of magnitude at Tl it is difficult to achieve a precision greater than +1 s-'.

The slow rate of phosphate dissociation from E -P formed by Zn(II)2Mg(II)2AP did not seem at first compatible with the demonstration by our laboratory and others that the Zn(II)2Mg(II)2 or Co(II)2Mg(II)2 enzymes are almost fully active in the usual assay system (26,27). The reason for this discrepancy became clear when the Zn(II)2Mg(II)2AP was assayed against p-nitrophenyl phosphate in the NMR buffer containing 0.01 M Tris. The turnover was only 300 mol of substrate/h/mg enzyme, Le. -10% of the activity of the native enzyme in 1 M Tris. On the other hand, when RO- release by the Zn(II)2Mg(II)AP was measured in 1 M Tris (0.5 M c1- as anion), turnover numbers from 1600 to 2500 mol/h/mg were observed as reported before (26). Using the "P NMR assay, we confirmed the very slow rate of turnover of the Zn(II)2(MgII)2AP in 0.01 M Tris with phosphate as the only product. In 1 M Tris-HC1, pH 8, the total rate of product formation in the NMR assay was enhanced by over 10-fold with half the product being Tris phosphate and half Pi. This

Mg(I1).

E.P

, I l l l l ~ l . ' L

6 4 2 0 2 PPm

FIG. 8. 81-MHz "P inversion transfer NMR spectra for Zn(II)pMg(II),AP at pH 9. The value of T , the time between the selective 180 pulse and the nonselective 90 pulse, is given beside each spectrum. Note that the E.P resonance (1.8 ppm) is to higher field than that of Pi (2.9 ppm) for this species.


TABLE 111 D&o&tion rates of inorganic phosphate from substituted alkaline

phosphatases, together with values for Zn(II)lAP and Cd(II)eAP for comparison

Enzyme PH ks 5-1

8.8 35 8 8 -1 9 1.8 9 -15 9 -2 9 <1 9 10 9 <1 9 6

D

~~ ~

a In the presence of 1 M C1-, the k , , ~ for phosphate is too fast to measure by inversion transfer; however, the resonances of E . P and Pi are still in slow exchange (measured at 162 MHz). Using the highest value of k,,ff reported here as a lower limit, k.,e must fall between the values of 60 and 260 s" in the presence of 1 M C1-

+ 1 M NaCI

T ms

2oor

- 15 1 0 5 0

+ 0.5 NaCl E- P

15 10 5 0

PPM FIG. 9. 162-MHz inversion transfer spectra of

containing 2.6 eq of total phosphate. The samples were in 10 mM Tris acetate and contained 0.5 or 1 M NaC1. The separation between selective 180 O and nonselective 90 e pulses applicable to each spectrum is indicated on the left-hand scale with the spectra plotted at the level corresponding to the correct T value.

"'Cd(II)&P (A) a d ( " s ~ d ~ " s C d a ) ( - - ~ - ~ ) ~ (B) at pH 9

acceleration involves both the transferase enhancement and an enhancement of phosphate dissociation due to the C1- (see below).

Phosphate Dissociation from Cd(II.. Derivatives of Alkaline Phosphatase-A number of previous studies measuring the E . P e E + P equilibrium (2, 20) as well as 31P saturation transfer (23) has suggested that replacement of Zn(I1) by Cd(I1) to form Cd(II)&P drastically reduces Pi dissociation

from E .P. This was confirmed by an inversion transfer experiment on Cd(II)&P. Very little inversion transfer occurred even at T = 100 ms, compatible with a k4 ( k d of 4 a-' estimated from the previous saturation transfer experiment (23) (see Table 111).

The hybrid Zn(II)2Cd(II)2 alkaline phosphatase is particularly interesting in relationship to the separate functions of A and B site metal ions in phosphate turnover. The B site Cd(I1) induces the unusual downfield shift of the 31P resonance of E . P, even though the A site metal ion is the native Zn(I1) ion (3). The 31P chemical shift of 12.66 ppm for E.P of the hybrid is particularly well separated from the Pi resonance at 2.7 ppm (see Fig. 9 below). Using delays between the selective 180 O pulse and the 90 O observation pulse up to 0.4 s, a value of k4 of -2 s-' was calculated. Table I11 summarizes the values of k4 for the two hybrid metalloenzyme species as well as giving data for Zn(II)4 and Cd(II), enzymes for comparison.

Effect of Chloride on the Phosphate Dissociation Rate- From the previous study using transfer of saturation (23) we estimated that the homogeneous Cd(II)&P had a phosphate dissociation rate of -1 s-' (Table 111). This places an upper limit on the turnover of the Cd(I1) enzyme. A separate study using V 1 , 31P, and '13Cd NMR suggested that anions like C1- can interact with the A site metal ion in addition to the phosphate of E -P, thus forming a 5-coordinate complex at the A site (12). Such a model could account for the "salt"- induced enhancement of activity known to be a characteristic feature of alkaline phosphatase (28), since the coordination of a second negatively charged species, i.e. the anion, in a 5- coordinate complex would potentiate the departure of the phosphate ligand.

In order to test this model, the effect of chloride addition on the Pi + E .P interconversion rate was examined by inversion transfer on three different enzymes, Cd(II)&P, (C~(II)A)~AP, and Zn(11)2Mg(II)2AP. We chose these three because these enzymes have initial k4 values that are 1 s-' or less; hence an increase in the rate of phosphate dissociation on the addition of C1- could easily be detected in the inversion transfer spectra. The substantial increase in inversion transfer rate observed on the addition of NaCl to the cadmium enzymes is illustrated in Fig. 9. Significant fall in the ampli- tude of the E .P resonance has occurred by 100 ms in each case (compare to Fig. 8). For all three enzymes k4 increases close to 10-fold on the addition of 1 M NaCl (Table 111). In the case of ZII(II)~M~(II)~AP k,n increases from 1.8 to 15 s-' (Table 111).

We have previously described the unusual Cd(I1) alkaline phosphatase resulting from the phosphorylation of one monomer of (C~(II)A)~AP. Phosphorylation of the latter species causes slow migration (subsequent to phosphorylation) of the A site Cd(I1) of the opposite unphosphorylated monomer to the B site of the phosphorylated monomer to form (Cd(II)*cd(II)B)(--A- -B)AP in which one monomer is devoid of metal ions (20). In the present context we use this species only to illustrate the effect of C1- on an isolated A-B site, unaffected by reaction at the other monomer of the tightly coupled dimer (Fig. 9B).

DISCUSSION

The dissociation rate of the phosphate product from alkaline phosphatase, -30 s-', is essentially independent of pH between pH 6 and 9 (Table I). This rate constant, k , ~ or k4, is approximately equal to the steady state kc, at pH 9; hence product dissociation is the rate-limiting step for hydrolysis at alkaline pH where the enzyme is active. At low pH, i.e. pH

2882 31P NMR of Alhline Phosphatase

5.5, the rate constant for dephosphorylation of the covalent phosphoseryl intermediate, E-P, is so slow, 0.23 s-' (29), that dephosphorylation of E-P takes over as the rate-limiting step. The shift in the rate-limiting step between dephosphorylation and dissociation of product as a function of pH appears to be controlled by a group at the enzyme-active site whose pK, is reflected in the pH-activity profile, pI(app 7.5 to 8 (2, 29). Based on kinetic, metal substitution, and NMR experiments we have previously proposed that this group is a Zn-OHz at the A site which ionizes to ZnOH- at alkaline pH (1-3). This pK. shifts to the range 9-10 in the Cd(1I) enzymes (2). The ZnOH- is proposed to be the nucleophile catalyzing the dephosphorylation of the seryl phosphate intermediate, E-P, in the second step of the mechanism; hence the ionization state of the coordinated water determines whether dephosphorylation is fast and phosphate dissociation takes over as the rate- limiting step.

From the ratio of transphosphorylation to hydrolysis products obtained in the presence of 3 M glycerol (Fig. 3), it is clear that there are different pH ranges over which the three hydroxyl moieties (OH of H20, C(1) and C(3)-OH of glycerol, and C(2)-OH of glycerol) are effective as acceptors of the phosphate group from serine 102. At acid pH HzO is the preferred acceptor; at neutral and moderately alkaline pH values the primary alcohol groups at C(1) and C(3) become effective and at more alkaline pH values the secondary alcohol group at the C(2) position becomes the most effective (Fig. 3). This order, increasing to higher pH, is also the order of the pK,s of the acceptor OH groups. Even though the apparent pK, values would be lowered by coordination to A site metal ion, the same order would be expected to hold. Based on this model the pH preference of the acceptors is understandable in terms of the available concentrations of the coordinated nucleophiles at each pH. At high enough pH the C(2) product exceeds that of the C(1) plus C(3) product (Fig. 3), which suggests that there may be a steric preference of the enzyme active center for the secondary alcoxide, offset at low pH by its higher p&.

Additional factors must account for the bell-shaped pH curve describing the formation of Tris phosphate (Fig. 3). The decline of the Tris phosphate product a t alkaline pH may reflect the loss of the proton from the amino group with a pK, of 8.2. Removal of the positive charge might result in some rise in the microscopic pK, of the hydroxyl which would reduce the effective concentration of coordinated alcoxide at a given pH, but the drop in transferase activity at high pH appears far too dramatic to be accounted for by this mechanism alone. The importance of the amino group has been noted elsewhere, since amino alcohols have been emphasized as the only really effective acceptors in the alkaline phosphatase reaction (4). The NMR method of detection shows, however, that at high pH the nonamino alcohols may catch up with the amino alcohols (Fig. 3).

The pH dependence of the transferase activity with Tris as the acceptor is counter to earlier statements in the literature that transferase activity is relatively constant with pH (4). Extensive data obtained with the old method of colorimetric determination of phosphate, however, do not exist and are probably subject to some imprecision especially as a function of pH. The rapid drop in transferase activity above pH 8 means that pH rate profiles done in the presence of significant concentrations of Tris are not a true reflection of the apparent pK, of activity. pH rate profiles in the absence of Tris are much closer to theoretical sigmoid curves expected of a single ionization on the enzyme surface (29).

Another feature which must be taken into account in inter-

preting the transferase mechanism is that in the case of alkaline phosphatase the nucleophile is coordinated to the A site metal ion (Fig. 4). Hence the donor properties of the alcoxide oxygen as ligand to the metal ion also enter as factors determining the relative efficiency. In small coordination complexes the alcoxides are generally weaker ligands than the oxygen of water and hence tend to be easily displaced in aqueous media (30). At the alkaline phosphatase site they seem to be relatively more effective in displacing solvent at the Zn(I1) ion (1-3 M uersu 55 M for approximately equal participation in the nucleophilic reaction).

Replacement of the native Zn(I1) ions with Co(I1) leads to a dramatic reduction in the specific activity of the enzyme to 10% of the native activity (19,29). The pathway of hydrolysis via the covalent (E-P) and noncovalent (E. P) intermediates appears to remain the same, however, including release of the inorganic phosphate product as the rate-limiting step. For the Co(I1) enzyme there is no detectable equilibrium concentration of E-P at alkaline pH (13) which requires that k-,/k3 << 1 for the equilibrium, E-P E. P., Data on "0 exchange out of HP"0:- catalyzed by Co(I1) alkaline phosphatase show that k-, = 3k4, where k4 is the rate constant for HP0:- release from E. P (31). Since dephosphorylation of E-P is much faster than rephosphorylation from E.P, i.e. k3 >> k-,, then k3 >> 3/24, and k4 must be rate limiting unless phosphorylation from ROP is extremely slow which it is not (19).

The loss of the Tris alcoxide as an effective nucleophile when Co(I1) is present at A site is unlikely to be due to a less effective reduction of the p& of the Tris hydroxyl, since no transphosphorylation is seen at all up to pH 10, i.e. 2 pH units higher than the pH at which the Tris phosphate:Pi product ratio reaches a maximum for the Zn(I1). It could be that coordination of Tris in a nonproductive manner through the amino group takes place more readily, thereby tending to eliminate the alcoxide as a phosphate acceptor. That such a mechanism may operate is suggested by the observation of detectable transfer to glycerol catalyzed by the Co(I1) enzyme, although we have not measured the absolute rate.

There is also a well-demonstrated mechanistic difference between the Co(I1) and Zn(I1) enzymes which might possibly be invoked to explain some of the loss of phosphotransferase activity. Loss of "0 from HP"0:- catalyzed by alkaline phosphatase shows that each dephosphorylation of E-P by the Zn(I1) enzyme results in loss of Pi to the medium; hence there is loss of only a single "0 at each turnover, and a simple statistical loss of l80 is observed (31). For the Co(I1) enzyme, on the other hand, significant rephosphorylation of Ser 102 takes place at each turnover; hence more than one "'0 (2 to 4) are lost at each cycle (31). One could argue that the significant probability of rephosphorylation in the Co(I1) enzyme also applies to E + R'OP as well as E. HOP and could, therefore, increase the likelihood that the final dissociating species would be HOP rather than R'OP because of the much greater concentration of HOP compared to R'OH.

E-P disappears as a significant equilibrium species by pH 6.5 for the native zinc enzyme, reflecting the fact that phosphorylation from E - P , k-,, is sufficiently slow that a 10-fold rise in the concentration of the nucleophile, ZnOH-, between pH 5.5 and 6.5 is sufficient to remove E-P as a detectable equilibrium species. In the hydrolysis of ROP, the relatively

The rate constants are named in accordance with Scheme 1, the overall enzyme reaction. For clarity we have not considered the E + Pi + E . P E-P equilibrium as a separate reaction as has been done in some previous literature. Identical numbering for the rate constants describing phosphorylation from HOP and phosphorylation from ROP leads to unnecessary confusion.


slow dissociation of phosphate, -30 s-’ (Table 11), simply takes over as the rate-limiting step as the concentration of nucleophile rises. The additive enhancement of RO- release by acceptor (Figs. 1 and 2) must mean that the alternate phosphotransferase pathway, where the R’O- alcoxide coordinated to A site (Fig. 4) is the nucleophile, must involve a much more rapid release of R’OP from E. R’OP than Pi from

By pH 8.0 when most of the enzyme is in the ZnOH- form, the binding of phosphate must compete with the OH- anion for a coordination site which could well explain the fall in the binding constant for phosphate above pH 8 detected by the equilibrium 31P NMR measurements. In this formulation of the mechanism the coordinated OH- must be in the position to initiate an in line attack on the phosphorus of the adjacent seryl phosphate; hence the ZnOH- would not be expected to be present in the E . P complex which would explain the lack of influence of the activity-linked pK, on the phosphate dissociation rate (Table 11). The ZnOH- + HPOf + Zn-HPOZ- + OH- competition, however, would be expected to decrease the apparent k,, and hence increase the apparent dissociation constant, Kit.,,, as documented by 31P NMR (2).

In contrast to acceptor alcohols interacting with A site, anions like C1- also interacting with A site have a direct influence on the phosphate dissociation rate (Fig. 9 and Table 111). Previous ‘I3Cd and 31P NMR data show that C1- and HPOi- can bind to A site simultaneously (12). This finding suggested that the A site metal ion is probably 5-coordinate with two coordination positions occupied by solvent molecules in the unliganded state. In high anion concentrations both may be replaced by anions. One of these monodentate positions becomes occupied by phosphate in E . P.

The catalytic mechanism of alkaline phosphatase thus appears to involve a metal coordination site for ROPOg- at which the negative charge cloud is adequately neutralized, allowing the strategically placed hydroxyl of serine 102 to attack the phosphorus in the E . ROP complex. Transfer to form Ser-OP must involve little change in free energy. Indeed the ease with which alcoxides, R’O-, coordinate A site metal ion and form R’OP supports an equilibrium constant very near 1 for the overall equilibrium ROP + E-Ser-0- $ E-Ser- OP + RO- in the absence of OH-. In the absence of significant concentrations of ROH or R’OH, OH- becomes the exclusive nucleophile in the second step, and the inorganic phosphate product as E . P becomes the most stable species.

The stability of E .P , a controlling feature of the mechanism, appears subject to several forms of regulation in addition to the adjacent anion effect. The nature of the B site metal appears to control the dissociation of phosphate, both Cd(I1) and surprisingly Mg(I1) at this site dramatically slow- ing the rate (Table 111). The dramatic effect of C1- in returning the phosphate dissociation rate to a higher value when B site metal ion is Mg(I1) may relate to the large charge/radius ratio of this small metal ion.

When the enzyme is in the ZnOHz form at low pH, phosphate binds much tighter and the equilibrium is shifted in favor of E-P, so much so that at pH 5 and below even the slow phosphorylation of serine 102 by inorganic phosphate,

E-P.

0.2 s” (23, 29), exceeds dephosphorylation and E-P becomes the predominant species. The E. P s E-P equilibrium may be governed throughout the pH range primarily by the concentration of the nucleophile [ZnOH-], since it controls the dephosphorylation of E-P and competes with Pi in the E . P complex.

REFERENCES 1.

2.

3.

4.

5.

6.

7.

8.

9. 10.

11.

12.

13.

14.

15.

16.

17. 18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Coleman, J. E., and Gettins, P. (1983) Adv. Enzymol. 55, 381-

Gettins, P., and Coleman, J. E. (1983) J. Biol. Chem. 258, 408-

Gettins, P., and Coleman, J. E. (1984) J. Biol. Chem. 259,4991-

Reid, T. W., and Wilson, I. B. (1971) in The Enzymes (Boyer, P.

Hull, W. E., Halford, S. E., Gutfreund, H., and Sykes, B. D.

Garen, A., and Levinthal, C. (1960) Biochim. Biophys. Acta 38,

Schlesinger, M. J., and Barrett, K. (1965) J. Biol. Chem. 240,

Lazdunski. C.. and Lazdunski. M. (1966) Biochim. Bio~hvs. Acta

452

416

4997

E., ed) Vol. 4, pp. 373-416, Academic Press, New York

(1976) Biochemistry 15,1547-1561

470-483

4284-4292 . . .

113,551-563 . I

Wilson. I. B.. and Davan. J. (19651 Biochemistrv 4.645-649 Levine,‘D., &id, T. W., A d Wilson, I. B. (1969 BLhernistry 8,

Chappelet-Tordo, D., Fosset, M., Iwatsubo, M., Gache, C., and

Gettins, P., and Coleman, J. E. (1984) J. Biol. Chem. 269,

Applebury, M. L., Johnson, B. P., and Coleman, J. E. (1970) J.

Malamy, M. H., and Horecker, B. L. (1964) Biochemistry 3,

Bradshaw, R. A., Cancedda, F., Ericsson, L. H., Newman, P. A., Piccoli, S. P., Schlesinger, M. J., Shriefer, K., and Walsh, K. A. (1981) Proc. Natl. Acad. Sci. U. S. A. 78,3473-3477

Morris, G. A., and Freeman, R. (1978) J. Magn. Reson. 29,433- 462

Led, J. J., and Gesmar, H. (1982) J. Magn. Reson. 49,444-463 Applebury, M. L., and Coleman, J. E. (1969) J. Biol. Chem. 244,

Gottesman, M., Simpson, R. T., and Vallee, B. L. (1969) Bio-

Gettins, P., and Coleman, J. E. (1983) J. Biol. Chem. 258, 396-

Wyckoff, H. W., Handschumacher, M., Murthy, K., and Sowad-

Dwek, R. A. (1973) NMR Biochemistry, p. 37, Oxford University

Otvos, J. D., Alger, J. R., Coleman, J. E., and Armitage, I. M.

Chlebowski, J. F., Armitage, I. M., and Coleman, J. E. (1977) J.

Bosron, W. F., Anderson, R. A., Falk, M. C., Kennedy, F. S., and

Coleman, J. E., Nakamura, K., and Chlebowski, J. F. (1983) J.

Anderson, R. A., Kennedy, F. S., and Vallee, B. L. (1976) Bio-

2374-2380

Lazdunski, M. (1974) Biochemistry 13,1788-1795

11036-11040

Biol. Chem. 245,4968-4976

1893-1897

709-718

chemistry 8,3776-3783

407

ski, J. M. (1983) Adu. Enzymol. 56,453-479

Press, London

(1979) J. Biol. Chem. 254, 1778-1780

Biol. Chem. 252,7053-7061

Vallee, B. L. (1977) Biochemistry 16,610-614

Biol. Chem. 258,386-395

chemistry 15,3710-3716 Wilson, I. B., Dayan, J., and Cyr, K. (1964) J. Biol. Chem. 239,

Chlebowski, J. F., and Coleman, J. E. (1974) J. Biol. Chem. 249, 4182-4185

7192-7202 Cotton, F. A., and Wilkinson, G. (1980) Advanced Inorganic

Bock, J. L., and Cohn, M. (1978) J. Biol. Chem. 253,4082-4085 Chemistry, p. 158-162, John Wiley and Sons, New York

alkaline phosphatase · 31p nmr signals from substrates and products of alkaline phosphatase have...

Documents