Phosphate Binding to Alkaline Phosphatase

SUMMARY The metal ion dependence of a2P binding to Escherichia coli alkaline phosphatase has been studied by means of equilibrium dialysis. not phosphate sites with the addition tight binding of phosphate magnitude of this binding constant is affected by ionic strength, pH, and protein concentration.


SUMMARY
The metal ion dependence of a2P binding to Escherichia coli alkaline phosphatase has been studied by means of equilibrium dialysis. Whereas the apoenzyme does not bind phosphate (no sites with K < 5 X 10-5 M), the addition of 2 Zn(I1) cations per molecule induces the tight binding of 1 phosphate dianion (K = 6 X 1OF M). The magnitude of this binding constant is affected by ionic strength, pH, and protein concentration.
Of the first transition and IIB metal ions, Mn(II), Co(II), Zn(II), and Cd(H) all induce the tight binding of one phosphate per dimer, but only Zn(I1) and Co(I1) induce enzymatic activity.
The formation of the phosphoryl enzyme, as measured by treatment of the 32P-protein mixture with perchloric acid, is also metal ion-dependent.
The Zn(I1) enzyme forms the phosphoryl enzyme only at low pH, a maximum of 0.6 mole per mole of dimer being observed at pH 5. The apoenzyme forms no phosphoryl enzyme at any pH. In marked contrast, Cd(H) induces the formation of significant equilibrium concentrations of phosphoryl enzyme in the alkaline pH range, reaching a maximum stoichiometry of 1 mole per mole of dimer at pH 6.5. Isolation of a phosphorylated peptide from the Cd(I1) enzyme shows that the same seryl residue is phosphorylated in the Cd(I1) protein as in the native Zn(I1) enzyme. Mn(I1) and Co(H) also induce the formation of significant amounts of phosphoryl enzyme in the alkaline pH region. Phosphorylation of the Cd(H) enzyme is relatively slow, k = -3 x lo-" set-I, but once formed the Cd(I1) phosphoryl enzyme does not break down as shown by its failure to catalyze the exchange of I*0 from H2180 into inorganic phosphate.
The data suggest that the metal ion plays an important role in formation and breakdown of the phosphoryl enzyme as well as in the binding of phosphate. A great number of substrates, whose leaving groups differ widely, have the same K, and V,,,.
The K, of these esters differs little from the K i for the phosphate itself (l-5).
The value for K, and Ki is of the order of 1OW RI. Subst'antial evidence also exists for the formation of a phosphoryl enzyme intermediate (E-P). At low PI-I this covalent intermediate can be isolated and has been shown to be a phosphorylated serine side chain (6-11).
Kinetic and thermodynamic evidence also strongly supports the presence of this intermediate in the reaction pathway (12-15). Orthophosphate is not only a competitive inhibitor of the hydrolytic reaction but can be considered a virtual substrate since the enzyme catalyzes 180 exchange from water into phosphate.
Significant equilibrium concentrations of the phosphoryl enzyme intermediate are formed at low pH in the presence of orthophosphate (6,8,9,11,13 (19,20). We are now interested in determining what role the metal ions play in the catalytic action and whether the metal ion is directly involved at the active site.
The set of studies presented here was undertaken to determine whether the metal ions affect any of the intermediates in the hydrolytic reaction pathway and to investigate further the nature of these intermediates.
A preliminary account of this work has been given (21).

EXPERIMENTAL PROCEDURE
Reagents-All chemicals were reagent grade. Buffer solutions, HCl, and NaOH were prepared free of metal as previously described (19,22 The gel filtration separated a small peak at the void volume, constituting about 0.5 to 1% of the total protein applied.
The remaining protein was eluted just after the void volume as one symmetrical peak and had a constant specific activity across the peak. The dissolved crystalline protein has a specific activity of 3000 f 500 pmoles of substrate hydrolyzed per hour per mg of protein when assayed at 25" and contains an average of 3 g atoms of zinc per mole of alkaline phosphatase dimer at pH 8.
Preparation of Apo-and Me(N)-Alkaline Phosphatases-The apoenzyme was prepared by treating the enzyme with a suspension of Chelex 100 (200 to 400 mesh) (Bio-Rad) in 1.0 M Tris buffer, pH 8.0, as originally suggested by Cohn1 and also described by Csopak (25). In 2 to 3 hours the zinc content is less than 0.1 g atom per mole.
The (2) c where c is the concentration of free phosphate, ni is the number of identical independent sites having a binding constant' of Ri, and Vi is the average number of phosphates bound per mole of protein.
A plot of (vi/c) against vi should yield a straight line with the ordinate intercept equal to nKi and the abscissa intercept equal to n;. If more than one class of binding sites is present, a Scatchard plot may show curvature. In this case the simplest (but not only) interpretation is to assume that there are i sets of sites which are noninteracting such that the following equation holds: Where applicable, the data were fitted by an appropriate choice of constants by Equation 3.

Determination of Acid-precipitable
Enzyme-bound Phosphate--h 0.25-ml aliquot of enzyme solution (~10~~ M in alkaline phosphatase dimer) was extracted by syringe from the dialysis bag used in the equilibrium dialysis procedure described above. This aliquot was injected rapidly into 0.5 ml of 70% perchloric acid and vigorously mixed. After 5 to 10 min, a O.l-ml aliquot of this solution was placed on a 2.4.cm disc of Whatman No. 3 filter paper.
The disc was then washed three times with cold 5% trichloracetic acid and once with cold 95% ethanol.
1. Scatchard plot,s of 32P binding to native alkaline phosphatase. 4, with 0.01 M Tris, pH 8: 0, 1 X 10e5 M alkaline phosphatase; 0, 1.5 X lo+ M alkaline phosphat.ase; 0, 2.78 X 10-S M alkaline phosphat.ase; 0, 1.1% X 10-j M alkaline phosphatase dialyzed against equimolar Zn(I1). The data are fitted as described nnder "Experimental Procedure" with the following constants: isO Exchange Studies-Stock solutions of 0.05 M inorganic phosphate were made by dissolving the required amounts of KHzPOl and K211P04 in H,r80, 1.52 atom To excess (Yeda Development Corporation, Rehovoth, Israel). Varying volumes of the dibasic and monobasic solutions were mixed to give solutions of pH values from 5.6 to 9.0. The final solutions were treated with Chelex 100 (Bio-Rad) to remove contaminating metal ions. Alkaline phosphatase (Zn(I1) or Cd(I1) enzyme), 100 ~1, was then added to 3-ml aliquots of the phosphate solutions to make a final enzyme concentration of 1.7 X 10d6 M. The reaction mixtures were incubated at 25" in a toluene atmosphere to prevent bacterial contamination. One-milliliter aliquots were removed at 24 and 48 hours, frozen, and lyophilized. After drying the samples by heating at 100" and storage over PC&, phosphate oxygen was converted into CO2 by heating under vacuum with guanidine I-ICI according to the method of Boyer and Bryan (31). The small amount of lyophilized protein present does not give rise to detectable COz under these conditions. The fractional amounts of the CO2 represented by masses of 44, 45, and 46 were determined on a mass spectrometer (Consolidated Electrodynamics, model 21-614). Atom per cent excess l*O present in the sample was calculated by standard methods (31).

RESULTS
Binding of Phosphate by Native Alkaline Phosphatase- Fig. 1 shows the Scatchard plots of phosphate binding to alkaline phosphatase. The plot of phosphate binding in absence of high salt shows pronounced curvature, which suggests the presence of at least two sites and possibly more (Fig. IA). The curvature can be fitted by assuming that there is one strong binding site of dissociation constant K1 = 1.1 x low6 M and a second weaker site & = 1.0 x 10e6 FIX. A slightly better fit for the data at high phosphate is obtained if it is assumed that two loose sites are present, K2 = 2.5 x 10e5 M. The data show well that there is one tight binding site. In the region of high phosphate the experimental error is large (32, 33) and except in one case the protein concentrations are such that saturation of the higher order sites cannot be achieved (see "Discussion"). The native alkaline phosphatase used in this experiment contains 3 g at Zn(I1) per mole of dimer; no extra precautions were taken to control the low levels of metal contamination. With the presence of excess zinc in solution, the results of phosphate binding are consistent with the general bulk of data and require no change in acceptable constants for the binding sites.
When the pH is decreased to 5.5, the Scatchard plot shows that alkaline phosphatase increases its affinity for phosphate at the tighter binding site (Fig. IA). Higher order binding sites still exist at this pH, but their affinity seems little changed from that at pH 8; 1<r = 6.8 x lo+ M for the tight binding site at pH 5.5.
Phosphate binding in the presence of salt shows that ionic strength significantly affects the dissociation constant of the tight binding site. In 0.1 M KCl, 1<r = 3.1 X 10' RI; in 0.5 M KCl, K1 = 6.6 x 10' M. Higher order sites with dissociation constants comparable to those present in the absence of salt are present as well in both cases (Fig. 1B).
Binding of Phosphate by Apoalkaline phosphatase and Metalloalkaline phosphatases-Apoalkaline phosphatase binds no phosphate specifically (Fig. 2). Binding of phosphate, however, with a dissociation constant greater than 5 x 10" RI cannot be ill. L. Applebury, B. P. Johnson, and J. E. Coleman 4971 detected in these experiments. Upon adding stoichiometric amounts of Zn(I1) to the apoprotein the presence of a high affinity binding site can be restored.
Addition of 1 eq of Zn(I1) per mole of dimer restores a tight binding site with n1 = 0.5 and K1 = 5.9 X 10M7 M. At least one loose site, K2 S 5 X lop5 M, is also present which has a similar affinity to that present on the apoenzyme. Addition of 2 eq of Zn(I1) per mole of dimer restores one tight binding site with K1 = 5.9 x 10 M in addition to the site with K2 = 5 x 10M5 M (Fig. 2). Two zincs are necessary for the formation of the high affinity phosphate binding site. The binding of phosphate to alkaline phosphatase is also induced by Mn(II), Co(II), Cd(II), and Cu(II)z in addition to Zn(II), even though only the Zn(I1) and Co(I1) phosphatases show significant enzymatic activity (Table  I), Ni(I1) and Hg (I1) are not effective in inducing phosphate binding. The titration curve for phosphate binding by the Co(I1) and Cd(I1) proteins is given in Fig. 3. There are insufficient data for a representative Scatchard plot, but the titration curve again indicates there is one tight binding site and one or more loose binding sites. The binding constants for the high affinity binding site of the various metalloalkaline phosphatases are given in Table II Binding as Functions of pZ1-Alkaline phosphatase binds 1 phosphate anion specifically and one or more others at least an order of magnitude less tightly.
In order to determine how many of these phosphates can be covalently bound, the total phosphate binding was determined as a function of pH at a fixed phosphate concentration (10e5 M) and the fraction covalently bound was determined by treatment with perchloric acid (Fig. 4). The total phosphate bound remains constant at about 1.5 moles per dimer between pH 8 and 6 and then decreases to zero over the region pH 6 to 3 (Fig. 4A).
At pH 8 none of the total phosphate is covalently bound (Fig. 4B).
As the pH is lowered covalently bound phosphate is stabilized and the amount detectable increases to a maximum of 0.6 mole per mole of dimer at pH 5. Below pH 5 the amount of covalently bound phosphate decreases in the same manner as the total bound phosphate decreases (Fig. 4, A  The apoenzyme binds between 0.2 and 0.5 mole of phosphate per dimer over the pH range 3 to 8 (0.2 mole is expected to bind on the basis of a dissociation constant of 5 X lop5 M). None of this phosphate is covalently bound since no phosphoryl enzyme can be isolated from the apoenzyme at any pH (Fig. 4B).
At lo-5 M phosphate the amount of total phosphate bound to the Mn(II), Co(II), and Cd(I1) alkaline phosphatase as a function of pH is very similar to that bound by the Zn(I1) enzyme (Fig. 5A).
The decrease in total binding to these metalloenzymes takes place at higher pH values than for the Zn(I1) enzyme.
For all three of these metallophosphatases, the pH dependence of formation of the phosphoryl enzyme is radically different from that of the Zn(I1) enzyme.
Between pH 6 and 8, the Cd(I1) enzyme forms a large amount of phosphoryl enzyme. A maximum of 1 mole per mole of dimer is observed at pH 6.5 (Fig. 5B) .3 The Mn(I1) enzyme also forms significant amomlts of phosphoryl enzyme throughout the alkaline pH region. A maximum of 0.4 mole per mole of dimer is formed at pH 7.0. The Co(I1) enzyme forms only a small amount of phosphoryl enzyme with a maximum of 0.2 mole per mole of dimer at pH 6. In all three cases the decrease in phosphoryl enzyme at low pH parallels the decrease in total phosphate binding (Fig. 5).
2 Phosphate binding to Cu(I1) alkaline phosphatase is difficult 3 Partial hydrolysis of t.he 32P-labeled Cd(I1) enzyme in 0.2 N to interpret.
While Cu(I1) induces binding of phosphate signifi-HCl followed by chromatography of the hydrolysate on Dowex 50 cantly above that shown by t,he apoenzyme, a study of the de-yields large amounts of 32P-phosphoserine and a more slowly elutpendence of phosphate binding on phosphat,e concentration shows ing labeled peptide containing aspartic acid and alanine in the that binding is less tight than in the case of the manganese, cobalt, ratio 1:2 in addition to phosphoserine. These findings are comzinc, and cadmium enzymes.
Changes in the state of aggregation patible with the previous report of the amino acid sequence surof the enzyme induced by Cu(I1) have also been observed (M. L. rounding the active serine isolated from the native Zn(I1) enzyme, Applebury, unpublished observation). The results are shown in Table III. Throughout this range of concentrations ouly a single site per dimer can be phosphorylated in the Cd(I1) enzyme. The time dependence of phosphorylation of this site by the Cd(I1) enzyme at pH 6.5 is shown in Fig. 6 with lop5 M enzyme and 10e4 M phosphate.
Under these conditions phosphorylation is relatively slow, with a half-time of approximately 3 min. The pseudo-first order rate constant for phosphorylation of the Cd(I1) enzyme calculated from this data is -3 x 1OP set-I. This is the rate constant for phosphorylation, since as shown below formation of the Cd(I1) phosphoryl enzyme is essentially an irreversible process. The ability of Zn(I1) and Cd(I1) phosphatases to catalyze the exchange of 180 from H&80 into inorganic phosphate is shown in Fig. 7  change at a constant rate between pH 5.6 and 7.5. The rate of exchange decreases slowly but not dramatically between pH 7.5 and 9. In marked contrast Cd(I1) phosphatase does not catalyze detectable I80 exchange at any pH (Fig. 7). The rate of 180 exchange observed for the Zn(I1) enzyme (Fig. 7) shows that the over-all process leading to l*O exchange is a relatively slow one. In the presence of 1.5 atom per cent excess H2180, increase of 180 in the phosphate oxygens by 0.4 atom per cent excess requires 48 hours.
Calculation of the Pi e HOH exchange according to Boyer and Bryan (31) shows that the rate constant for exchange by the Zn(I1) enzyme under the conditions of Fig. 7 is -0.2 set-l at pH 6.5.  (Fig. 2 and Table  II).
A high affinity binding site may be restored upon the addition of metal ions to the apoenzyme. The presence of 2 metal ions is required for the tight binding of one phosphate to alkaline phosphatase with a dissociation constant of 6 X lop7 M.
Readdition of 1 eq of zinc generates a binding site with a dissociation constant of 6 X 10e7 M, but with only 0.5 mole of V bound per dimer (Fig. 2 and Table II).
Since a stoichiometry of 0.5 cannot exist at the molecular level, an acceptable interpretation requires that, in the presence of phosphate and 1 metal ion per dimer, 50% of the molecules are dimers containing 2 Zn(I1) cations and 1 phosphate anion, while the remainder of the protein is present as zinc-free apoalkaline phosphatase. Such a model suggests that the binding of 2 zinc ions to the apoenzyme will be cooperative in the presence of phosphate. Measurements of the zinc dissociation constants in the presence of phosphate have not been reported, but some evidence exists that the zinc-protein complex is stabilized in presence of phosphate (11). At alkaline pH the predominant species in the case of the Zn(I1) enzyme is the noncovalent complex ( Fig.  4). At pH 5, the stable covalent intermediate is the prominent species (Fig. 4B).
The interpretation of binding data for the native three-zinc enzyme and for the enzyme in presence of excess zinc is more difficult.
The binding data indicate that more than one phosphate binding site is present and Scatchard plots are distinctly nonlinear (Fig. 1 This tight binding site is also the one which appears to be Below pH 5 both the total bound phosphate and the amount of covalently bound phosphate decrease (Fig. 4). Previous studies have shown that below pH 5 the enzyme undergoes significant changes in tertiary and quaternary structure. The Zn(I1) ions dissociate from the dimer and below pH 4 the dimer dissociates into monomers which unfold to random coiled polymer chains (20,39,40 proteins, the absence of enzyme activity camlot be attributed to the lack of a phosphate binding site ( Fig. 3 and Table  I).
In addition to binding phosphate, both the Cd(I1) and Mn(I1) enzymes form the phosphoryl enzyme (Fig. 5) which is implicated as an important step in the catalytic reaction pathway (8, 12-15).3 The distinguishing feature of these proteins compared to the active metallophosphatases is the high stability of these phosphoryl enzymes at the higher pH values. This stability is so great for Cd(I1) that 1 mole of covalently linked phosphate per mole of enzyme can be isolated at pH 6.5 and 0.75 mole at pH 8.0 (Fig. 5).
In order to form observable equilibrium concentrations of the phosphoryl enzyme at any pH, kz must be greater than lcmz (k-& << 1). This condition is met by the Cd(I1) enzyme. While phosphorylation of the Cd(I1) enzyme is relatively slow, lcz = 3 x 1O-3 set+ at pH 6.5 (Fig. 6), dephosphorylation simply does not occur over a measurable time (Fig. 7). The failure to form the Cd(I1) phosphoryl enzyme completely at pH 8 may relate to the presence of an even more stable Michaelis complex as postulated in the case of the Zn(I1) enzyme (14). At pH 5 to 5.5 the Zn(I1) enzyme not only forms significant equilibrium concentrations of the phosphoryl enzyme, but catalyzes its turnover with a rate constant of -0.2 set-l (Fig. 7). This suggests that phosphorylation of the Zn(I1) enzyme must occur faster than for the Cd(I1) enzyme pictured in Fig. 6. This is confirmed by our attempts to measure the phosphorylation rate of the Zn(I1) enzyme.
By 10 set (the shortest time conveniently measured by our techniques), the phosphorylation of Zn(II) enzyme at pH 5.5 is complete.
Thus Cd(I1) must also have an affected by ionic strength and pH ( by guest on July 8, 2020 http://www.jbc.org/ effect on the phosphorylation rate in addition to its dramatic effect on dephosphorylation. Sincethe maximum catalytic constant observed for the catalysis of "0 exchange into phosphate by the Zn(I1) enzyme is 0.2 set-1, considerably slower than catalysis of phosphate ester hydrolysis, phosphorylation of the enzyme from HOP must be at least an order of magnitude slower than phosphorylation from ROP. The rate constant for phosphorylation of the enzyme by 4methylumbelliferyl phosphate has been determined to be 6.1 se6 at pH 6.34 (41).
The lack of activity or low activity6 of the Cd(I1) and Mn(I1) enzymes may be explained by their failure to catalyze the breakdown of the phosphoryl enzyme at significant rates. These changes in the rates of phosphorylation and dephosphorylation induced by different metal ions may relate either to intrinsic chemical properties of the cation or to slight changes induced in the conformation of protein side chains located at the active center.
The decrease in total bound phosphate and phosphoryl enzyme for all of the metallophosphatases at the lower pH values (Figs. 4 and 5) may be adequately accounted for by the loss of metal ion and dissociation of the protein into subunits (20,39,40).
The observed stoichiometric features of this protein raise interesting questions about the necessary structural organization of the active protein.
The dimer binds between 2 and 4 metal atoms at neutral pH (18-20, 44, 45) and in the presence of excess metal ions or higher pH it will bind even more (20). Spectrophotometric titration of the apoenzyme with Co(I1) shows only two 6 Mn(II) alkaline phosphatase shows very low activity, but it is significantly greater than the activity shown by the zinc-free enzyme (Table I).
This may represent significant activity since the phosphoryl enzyme formed at. neutral pH by the Mn(I1) enzyme does not appear as stable as that formed by the Cd(I1) enzyme (Fig. 5).
highly specific Co(I1) binding sites of unusual and apparently distorted coordination geometry (46). Two metal ions are necessary for activity (17,18,47), and 2 metal ions are necessary for the formation of one highly specific phosphate binding site (Fig. 2).
We have observed more than one phosphate binding site per dimer, but are able to measure the formation of only one phosphoryl serine per dimer. With the Cd(I1) alkaline phosphatase, the phosphoryl enzyme may be observed at alkaline pH at which the protein is not so susceptible to quaternary changes and loss of metal ions. At high phosphate concentrations at which more than 1 mole of total phosphate is bound, a maximum of one phosphate is incorporated into the protein (Fig. 5, Table III), suggesting that only one site is involved in catalysis at any given instant. Initial x-ray diffraction data on crystalline alkaline phosphatase indicate that the unit cell contains three dimers with the monomer as the asymmetric unit (see "Appendix"). It is tempting to assume that the 2-fold rotation axis between monomer units of the protein observed in the crystal is a 2-fold axis relating the two subunits of a functional dimer.
For this symmetry, unique single sites per dimer would be present only along the 2-fold axis. All other sites would exist as identical pairs.
Location of the active center on the 2-fold axis could explain the apparent presence of a single phosphate binding site and the apparent maximum phosphorylation of only a single seryl residue in the presence of 2-fold molecular symmetry.
On the other hand, interference between closely adjacent sites or negative cooperativity between more distantly separated identical sites as postulated by Koshland (51, 52) would also explain the observation of a single high affinity binding site.