Zn(II), 115mCd(II), 6oCo(II), and Mg(I1) Binding to Alkaline ...

THE JUURNAL OF BIULUGICAL CHEMISTRY

Printed in U.S.A. Vol. 258, No. 1, Issue of January 10, pp. 386495,1983

"Zn(II), 115mCd(II), 6oCo(II), and Mg(I1) Binding to Alkaline Phosphatase of Escherichia coli STRUCTURAL AND FUNCTIONAL EFFECTS*

(Received for publication, March 18, 1982)

Joseph E. Coleman, Ken-ichi NakamuraS, and Jan F. Chlebowskig From the Department of Molecular Biophysics and Biochemistry, Yale Uniuersity, New Haven, Connecticut 06510

Zn(II), Cd(II), Co(I1) and Mg(I1) binding to apoalkaline phosphatase of Escherichia coli and the relative stabilities of the resulting metalloenzyme complexes have been measured by equilibrium dialysis and metal exchange reactions using y-emitting isotopes of these metals. At millimolar concentrations of these metal ions the alkaline phosphatase dimer binds three pairs of metal ions (A, B, and C sites). One of these pairs dialyzes readily without detectable change in the structure or function of the enzyme (C site). Of the remaining two pairs, the binding affinity of both for Zn(I1) and Cd(I1) is increased by formation of the phosphoenzyme intermediates. Cd(I1) is bound less tightly to both A and B sites than Zn(II), and at pH 6.5 Cd(I1) is induced to bind to the B sites by formation of the phosphate complexes. Mg(lI), 5-10 mM, competes successfully with the IIB metal ions for the second or lower affinity pair of binding sites (B sites), although Mg(I1) is a relatively poor competitor on an equimolar basis, especially for Cd(I1). Binding of metal ions to the apoenzyme appears to be a cooperative process involving conformational changes in the protein which are not readily reversible. The initial binding of a pair of Zn(I1) or Cd(I1) ions to the apoenzyme is characterized by equilibrium constants of to 10" M for Zn(I1) and to M for Cd(I1). Following the cooperative binding of all three pairs of metal ions, however, re-establishment of equilibrium by dialysis indicates binding constants of <lo-' M for Zn(I1) and <loM6 M for Cd(I1) at the sites of greatest affinity (A sites). Binding of Mg(I1) or Cd(I1) to the B site, once the A site is occupied, increases the phosphorylation rate of the Cd(I1) enzyme by 20-fold. In the presence of saturating concentrations of Mg(I1) complete activity is restored to the apoenzyme by 2 Zn(I1) ions. In the absence of Mg(I1) as many as 6 Zn(I1) ions may be required before complete restoration is achieved. Roles for the A and B site metal ions in the catalytic mechanism are discussed.

The stoichiometry of Zn(I1) binding to Escherzchia coli apoalkaline phosphatase has been the object of a number of

* This work was supported by Grant AM09070-17 from the Na- tional Institutes of Health and Grant PCM 7912269 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 soleiy to indicate this fact.

f Current address: Hiroshima Women's University, Uzina, Hiro- shima, Japan, 734.

6 Current address, Department of Biochemistry, Box 614, Medical College of Virginia, VirGnia Commonwealth University, MCV Sta- tion, Richmond, VA 23298.

studies. The results show the stoichiometry of the rebinding of Zn(I1) to the metal-free apoenzyme to vary from 2 to 6 or more g atoms of Zn(II)/mol of dimer depending on pH and buffer conditions (1-3). The Zn(I1) content of the native enzyme reported in the literature has also varied between 2 and 4 g atoms/mol of enzyme dimer, depending on the mode of preparation (1, 4, 5 ) . The lower values appear to correlate with the use of ammonium sulfate precipitation in the absence of Zn(I1)-containing buffers. The average Zn(1I) content of enzyme released from E. coli by osmotic shock after treatment of the cells with EDTA and the cold water wash loaded directly onto a DEAE-cellulose column, is 2.7 to 3.5 g atoms/ mol (1).

Alkaline phosphatase has an absolute requirement for Zn(II), but in common with many phosphoryl transferases the Zn(I1) enzyme can be further activated by Mg(I1) (6). Studies of the reactivation of the apoenzyme by metal ions have also yielded variable results concerning the minimum stoichiometry required for restoration of full activity with values reported from 2 to 4 g atoms Zn(II)/mol of enzyme dimer required to restore maximal activity (2, 7-11). Part of this variability appears to derive from the fact that restoration of the "native" state from metal ions and apoenzyme is not an instantaneous process and is subject to other variables in addition to the Zn(I1) stoichiometry per se. The presence of Mg(1I) and phosphate are additional factors which affect the restoration of activity. The presence of Mg(I1) has been shown to poten- tiate the restoration of activity to the apoenzyme by Zn(I1) and the addition of Mg(I1) to enzymes containing 2 Zn(I1) or 2 Co(11) ions/mol of dimer has been reported to enhance the activity of the enzymes containing 2 metal ions to that observed with excess Zn(I1) or Co(I1) (3, 12). The native enzyme prepared by the usual procedures contains slightly more than 1 g atom of Mg(II)/mol of dimer and has been observed to bind -1.8 g atoms Mg(II)/mol at free Mg(I1) concentrations of 0.45 mM (3). The latter binding was cooperative with that of Zn(I1) at pH values below 9, Le. maximum Mg(I1) binding required the presence of Zn(I1).

Studies of metal binding to alkaline phosphatase have been most extensive for Zn(I1) and much less extensive for the other fist transition and IIB metal ions which bind to the active center. Even for Zn(II), data on metal binding as functions of a wide range of metal and protein concentrations in the presence and absence of Mg(1I) and phosphate are not available. Recent "3'P and ''%d NMR studies from this labo- ratory examining both the unliganded enzyme (*I3Cd NMR) and its phosphoenzyme intermediates ('3'P and ""Cd NMR) have shown significant changes in chemical shift of the active site "'Cd ion as well as the 'jlP resonance of the phosphoseryl and noncovalent phosphoenzyme intermediates depending on metal ion stoichiometry (13-15). The most striking finding

386

Metal Ion Binding to Alkaline Phosphatase 387

from the 31P NMR spectra was a shift of phosphate binding stoichiometry from 1 to 2 mol of bound phosphate/mol of enzyme as the Cd(I1) ion stoichiometry was increased from 2 mol/mol of dimer to 4 mol/mol of dimer (15). The NMR studies are distinguished from earlier investigations of phosphate binding stoichiometry by requiring protein concentrations > mM, i.e. at least an order of magnitude greater than those used in most previous studies of alkaline phosphatase. At these relatively high protein, metal ion, and phosphate ion concentrations, some sites may be saturated which are not saturated with either metal ion or phosphate at lower concentrations. Likewise, cooperative binding between first transition ions and Mg(I1) or phosphate may radically alter the NMR resonances. Hence, in order to interpret the '"P and ""Cd NMR signals precisely a complete understanding of the ther- modynamics of metal ion binding to alkaline phosphatase is required. The present studies using "'Zn, 'lSmCd, and "Co as radioactive labels in equilibrium dialysis studies were designed to obtain this basic information.

MATERIALS AND METHODS

Enzyme Preparations-Alkaline phosphatase was isolated from E. coli (strain CW3747) as previously described (16). Enzyme concentrations were determined spectrophotometrically a t 278 nm with I?,'$ = 0.72 (17). For molar calculations a dimer M, of 94,000 was used based on the most recent data on the complete amino acid sequence (18). Enzyme activity was measured by the release of p-nitrophenol from p-nitrophenyl-phosphate (Sigma) in 1 M Tris-HCI, pH 8, 22 "C (16). The native enzyme had a specific activity of 2500 -C 500 units (micromoles of substrate hydrolyzed/h/mg of protein). All buffer solutions were prepared metal-free (1). Apophosphatase-Apophosphatase was prepared by a new method

designed by Dr. Peter Gettins to prepare the concentrated NMR samples and uses ammonium sulfate to remove Zn(I1) from the enzyme as described in detail in the following paper (19). This results in the rapid production, 5 days, of apoenzyme containing ~ 3 % of the original Zn and uncontaminated with chelating agents which have shown a tendency to bind to the apoenzyme (20, 21). Me(ZI,J I phosphatases were prepared by direct addition of stoichiometric amount,s of ZnClr, CdCL, CoS04, and MgSOd (spectrograde from Johnson Matthey Chemicals, Ltd., London) to the apoenzyme or by dialysis of apoenzyme against buffered solutions of these metals. If radioisotope labeling was desired aliquots of standard solutions were labeled with "'ZnCl~, ""CdCL, or '"CoCI? (solutions in dilute HCI) from New England Nuclear. Metal binding studies using the y-emitting metal isotopes utilized the method of small closed dialysis bags of 1 ml of volume dialyzed against 50 ml of dialysis fluid as described in detail by Coleman and Vallee (22). Dialysis tubing was Visking-Nojax casing which has been freed of metals (23). The dialysis bags were counted by placing the entire bag in test tubes which fit a Packard Model 5019 well-type y spectrometer. Metal ion solutions were adjusted to count between 3,000 and 10,OOO cpm and protein concentrations relative to metal ion concentrations were routinely adjusted such that 1 g atom/mol of protein-bound metal gave counting rates 50 to 100% above 1 ml of the labeling solution. Only occasionally, i.e. at the upper extreme of the concentration range (lo-" M ) , did 1 g atom of bound metal represent counts only 158 above the equal volume of labeling solution. Blank bags containing buffer alone were used to correct for both the labeling solution and nonspecific binding to the dialysis membrane. Labeling studies were performed in 0.01 M Tris-HC1-0.1 M NaCI. In order to check for solute partitioning caused by Donnan effects or microscopic precipitation a t high pH, controls using protein already saturated with nonradioactive metal were tested under a variety of conditions. No artifacts were observed except when using Cd(I1) at pH 9.0 under some conditions (see text). Metal analyses for Zn, Cd, Co, and Mg were performed by atomic absorption spectros- copy on an Instrumentation Laboratories Model 157 spectrophotom-

' The abbreviations used are: Me(II)2AP, Me(II),AP, or Me(II)&"I)2AP refer to alkaline phosphatase containing metal ions at. A sites alone, the same metal at A and B sites or Me(I1) at A sites and Mg(I1) a t B sites. Me(I1) = Zn(II), Cd(II), or Mn(I1). E-P is the phosphoryl enzyme while E . P refers to the noncovalent complex with inorganic phosphate.

eter. Metal concentrations in samples were adjusted to a range of 2 to 10 pM.

RESULTS

Cadmium binds less tightly to alkaline phosphatase than the native zinc ion and in the presence of excess metal ion has fewer total binding sites at comparable free metal ion concentrations. Thus, the differential binding affinities of the several pairs of potential metal binding sites on alkaline phosphatase are most easily demonstrated with Il5Cd(II) binding and these data will be presented first.

lIsCd Binding to Apoalkaline Phosphatase as Functions of Phosphate, Magnesium, pH, and Cd(II) Concentration- Many of the studies of Cd(I1) alkaline phosphatase, particularly NMR studies, have been performed at pH 6.5 (13). At pH 6.5 Cd(I1) is relatively weakly bound; slightly less than 2 Cd(I1) ions are bound per enzyme dimer at enzyme and Cd(I1) concentrations of 5 x M or below (Fig. 1A ). Equilibrium binding as the Cd(1I) concentration is lowered to M shows the apparent dissociation constant for this pair to be -5 x

M. The presence of 1 mM phosphate a t pH 6.5 (5 X M Cd(I1)) induces the binding of a second pair of Cd(I1) ions (Fig. lA). The bound phosphate a t this pH is present entirely as the phosphoseryl intermediate (E-P). A total of 4 Cd(I1) ions/enzyme dimer are bound and the dissociation constants for both pairs are 2 X 10"' M or less (Fig. 1A ).

As would be expected, both pairs of metal binding sites on the alkaline phosphatase dimer bind Cd(I1) more tightly at pH 8 than at pH 6.5, and at enzyme and Cd(I1) concentrations of 5 X M, 4 Cd(I1) ions are bound per dimer even in the absence of the phosphate ligand (Fig. 1B). At pH 8, two regions of Cd(I1) binding can be detected as a function of concentration, one pair with a dissociation constant, K , = 5 X lo-? M, and one pair bound -1OX less tightly, K,, = 5 X M (Fig. l e ) . The presence of 10 mM Mg(I1) at pH 8.0 prevents the binding of the pair with K d = 5 x M (Fig. IC), while the pair with a Kd = 5 X 10" M is undisturbed by Mg(I1).

The pH dependency of Cd(I1) binding from pH 6.5 to 9 and a t enzyme and Cd(I1) concentrations of 5 X M is shown in Fig. 1D. In the presence of the phosphate ligand the enzyme binds 4 Cd(I1) ions/dimer at all pH values tested, while in the absence of the ligand, binding rises from -1.5 Cd(I1) ions bound at pH 6.5 to -4 Cd(I1) ions bound per dimer at pH 8.5 to 9.0. One of the two pairs of Cd ions bound to the dimer is displaced at all pH values by 10 mM Mg(I1).

While the unliganded enzyme at pH 6.5 binds only 2 Cd(1I) ions/dimer at 5 X M free Cd(II), if higher enzyme concentrations (1.5 X M) are used in order to measure Cd(1I) binding at higher free metal ion concentrations, binding can be increased to 6 gm atoms/mol of dimer at lo-,' M Cd(1I) (Fig. 1A). These two additional pairs have binding constants between and M which suggest kinetic dissociation constants, kdis$, on the order of 10' to 10" s-I, the latter value applying if binding is assumed to be diffusion controlled, i.e. -loR M" s". Such exchange rates between bound and free Cd(1I) can easily be in the intermediate exchange range for ""Cd(11) NMR signals of exchanging species and can have dramatic effects on ""Cd(I1) NMR signals (19).

In the experiments of Fig. 1A measuring binding of "'Cd(11) to the apoenzyme as well as similar ones reported below, equilibrium was defined operationally by counting the dialysis bags every 24 h until no further increase in enzyme-bound radioactivity was observed in successive measurements separated by 24 h. Except where noted (see "Zn(I1) below), equilibrium was achieved in 24-48 h and test experiments carried to 7 days showed no further increase. The experimental points are plotted at the free metal ion concentration in

388 Metal Ion Binding to Alkaline Phosphatase

FIG. 1. l15Cd binding to apoalkaline phosphatase as a function of [Cd(II)], [HP04'-], [Mg(II)], and pH. A, enzyme-bound "'Cd/dimer versus [Cd(II)] at pH 6.5 (0); plus mM HPO,"

labeled a t 10W' M Cd(I1) (0). B, "'CCd/ (0); and following dialysis of samples

dimer versus [Cd(II)] at pH 6.5 (0) and pH 8.0 (0). C, "'Cd/dimer versu.s [Cd(II)] at pH 8.0 (0); plus 10 mM Mg(I1) (0). D, "'Cd/dimer versus pH with no addition (0) or in the presence of 1 mM HPO," (0) or 10 mM Mg(I1) (H); 5 x

M ["'Cd(II)], 5 X IO-" M enzyme. Points are missing at pH 9 because of the accumulation with time of a dialyz- able ""Cd-containing species in the dialysis bag a t the very alkaline pH when P, and Mg(I1) and excess Cd(I1) are present. In A, B, and C, enzyme concentration was 1 X lo-' M in the lower concentration range, 5 X 10- ' M in the midcon- centration range, 2 X 10 -' M in the high concentration range. All buffers were 0.01 M Tris-0.1 M NaC1. 4 "C.

the surrounding 50-ml dialysate at the end of the dialysis as determined by the specific radioactivity of the final dialysate, except where noted. "Metal-free" buffer refers to buffer treated as described under "Materials and Methods" prior to the addition of the specific spectrographically pure metal ion under examination either by direct addition or by dialysis of labeled enzyme against metal-free buffer.

"'Cd(II,J t, Cd(II)2 Exchange at the Active Center of AZ- M i n e Phosphatase-Binding of a second pair of Cd(I1) ions induced by formation of the phosphoryl enzyme at pH 6.5 (Fig. 1A) suggests that the phosphate ligand alters protein structure in the immediate vicinity of one pair of Cd(I1) binding sites. Since the dissociation constant for both pairs is decreased (Fig. l A ) , the affinity of the first pair must also increase on phosphate binding. One way to assess these changes is to measure metal ion self-exchange rates, i.e. exchange of "'Cd(I1) on a fully metalated enzyme for nonradioactive Cd(I1) present in the dialysate at equimolar concentrations in a 50-fold volume excess. Since the stoichiometry of Me(I1) on the enzyme remains constant during such exchanges, the observed rates are primarily expected to be a function of the kinetic dissociation constant, hdlssr of the Cd(I1) ion from the enzyme complex. In the absence of phosphate at pH 6.5, the enzyme binds only one pair of "'Cd(I1) ions and these exchange with external Cd(I1) with a half-time of -10 h (Fig. 2). Mg(I1) (10 mM) a t pH 6.5 has a slight effect on this exchange, lengthening the half-time by -2 h (Table I). If the enzyme is first labeled with 2 '"Cd(I1) ions, placed in phosphate (1 mM) and Cd(II)(1 X M), the enzyme binds an additional pair of cadmium ions. Only the initial pair is labeled. Hence the self-exchange measures the exchange of the first pair in the presence of the second pair and the phosphoseryl residue. The half-life of this exchange increases dramatically to -70 h (Fig. 2). If both pairs of bound Cd(I1) ions are labeled by incubation with "'Cd(I1) in the presence of phos-

' Cd(1I) without an isotopic superscript will be used to designate either total cadmium or Cd(I1) solutions where no radioactive isotope is present, as in the exchange of an unlabeled solution of cadmium (Cd(I1)) for a labeled cadmium ("Cd(I1)) on the enzyme. While technically "'Cd is "'"Cd (designating a metastable isotope), "'Cd will be used in this paper.

4[1 3 + I mM Pi =-.-. ~ """ -" -=-

Q)

0

\

0.4 i 0.3 '\ . 0.2 \

0 10 20 30 40 50 60 70

HOURS FIG. 2. Exchange of enzyme-bound "'Cd(II) for equimolar

Cd(I1) in a 50-fold volume excess at pH 6.5. (0) "'Cd2AP (under exchange conditions this species exists only at pH 6.5; see Fig. 1); (0) '"Cd2Cd2AP in the presence of 1 mM P,; (H) "'Cd4AP + mM P,. Buffer was 0.01 M Tris-0.1 M NaCl, pH 6.5, 4 "C, 5 x 10 ' M enzyme, 5 x 10"' M Cd(I1).

phate, the exchange now divides into two distinct phases. One pair of "'Cd(I1) ions exchanges rapidly (complete exchange in -20 h), while the second pair exchanges with a half-time of -70 h, similar to the rate for the fiist pair added in the presence of phosphate in the previous experiment.

At pH 9.0 Il5Cd(II) t, Cd(I1) exchange studies of the 4 Cd(I1) species can be carried out both in the presence and absence of phosphate ligand, since 4 Cd(I1) ions are bound in both cases (Fig. 1D). If there is not rapid intramolecular exchange of the Cd(I1) ions on the enzyme surface, then it should be possible to differentially label the two pairs of sites by adding either the first or the second pair as "'Cd. The results in the presence and absence of phosphate are shown in Fig. 3. In both cases the two pairs of Cd(1I) complexes with the enzyme are differentiated. The first pair exchanges slower than the second, and both pairs are stabilized by phosphate.


TABLE I "'Cd(II)AP c) Cd(II), "Zn(II)AP fs Zn(II), and "'Cd(II)AP ++

Zn(II) exchange in alkaline phosphatase Alkaline phosphatase, 5 x 10 ' M; Cd(II), 5 X M; 0.1 M NaC1-0.01 M

Tris HC1 Enzyme composition

'I 'Cd - Cd

1. "'Cd, AP 2. "'Cd&d2AP + rnM P, 3. "'Cd2Mg2AP + 10 mM

4. 'l5Cd4AP + mM P, Mg'+

5. "'Cd2CdyAP 6. "'Cd2Mg2AP + 10 mM

7. "'Cd2CdyAP 8. Cd,"'Cd,AP 9. "'CCdyCd2AP + mM P,

10. Cd2"'Cd2AP + mM P, 11. "'Cd,AP

12. ""C&AP + mM P,

Mg"'

"'Zn tf Zn 13. "'Zn4AP

14. ""Zn4AP + mM P,

15. "'Zn,AP

16. "'ZnzZn2 + mM P, + 10 mM Mg'+

tf Zn I I'Cd 17. "'Cd2AP 18. "'Cd2Mg,AP + 10 mM

Mg'+ 19. "'Cd4AP + mM P,

20. "'Cd,AP + mM P, + 10 mM Mg'+ identical

6.5 6.5 6.5

h S I X 10'

10 1.0 70 0.14 12 0.83

6.5 2 rates -5 2.0 -70 0.14

8.0 15 0.67 8.0 20 0.50

9.0 21 0.48 9.0 13 0.77 9.0 42 0.24 9.0 25 0.40 9.0 2 rates 21 0.43

13 0.77 9.0 2 rates 42 0.24

25 0.40

6.5 2 rates 18 0.56 loo0 0.01

6.5 2 rates >loo0 <0.01 >loo0 t O . O 1

8.0 2 rates 24 0.42 840 0.012

8.0 2500 0.004

6.5 6.5

12 0.83 248 0.04

6.5 2 rates nearly -1200 0.008

6.5 2 rates nearly -900 0.011 identical

Defined as in Ref. 24. It can be assumed that the general order of magnitude of k,, reflects primarily k<~,+

The exchange half-time approximately doubles for both sites in the presence of phosphate. At pH 9.0 the phosphate is present almost exclusively as the noncovalent complex (E. P) (19).

The exchange reaction for the "'Cd,AP species in the presence and absence of phosphate is shown in Fig. 4. The nonlinear semi-log exchange plots can be broken into two linear exchange plots whose half-times correspond within the error of the measurement to those observed for the first and the second pair of "'Cd ions measured separately by the experiments graphed in Fig. 3. Cd(I1) exchange at the Cd(I1) binding sites of alkaline phosphatase under a variety of conditions are tabulated in Table I along with estimates of the kinetic exchange constants of the complexes, hrxrh6,nge.

That these Cd(I1) binding sites are shared by other transition and IIB metal ions binding to alkaline phosphatase and that one pair is shared by Mg(I1) is illustrated by the set of order of additions experiments listed in Table 11. The com- peting metal ions were added first and then the enzyme dialyzed against 5 X M "'Cd(I1) to label the remaining binding sites. The stoichiometry of the ""Cd(I1) label was determined at 24 h and after 30 days. The latter stoichiometry reflects the slow exchange of "'Cd(I1) with the occupied metal binding sites. Addition of 2 Zn(I1) ions and phosphate at either pH 6.5 or 8.0 blocks one pair of Cd(I1) sites. Four Zn(I1) ions

and phosphate block both pairs of Cd(I1) sites at both pH values, somewhat more effectively at pH 8. The fractional binding observed reflects a minimal exchange in 24 h. Without phosphate this exchange is more rapid. Mg(I1) (10 mM) added to the 2 Zn(I1) or 2 Mn(I1) enzyme blocks a t least 50% of the second pair of Cd(I1) binding sites. The combination of 2 Mn(I1) 2 Mg(I1) and phosphate is particularly effective. This is in contrast tq 2 Mn(I1) and phosphate which does not appear particularly stable, since over 3 g atoms of "5Cd(II) are bound within 24 h, while 2 Zn(I1) ions in the same sys- tem completely block one pair of Cd(I1) sites, even after 30 days (Table 11). This must reflect the relative stabilities of the mixed metal-enzyme-phosphate complexes (see "Dis- cussion").

"Zn Binding to Apoalkaline Phosphatase as Functions of Phosphate, Magnesium, pH, and Zn(Z4 Concentration- "Zn(II) binding to the apoenzyme (5 x M) in the presence

.L +Pi 0.3 a

0.2

2

I

Cd2AP--Cd

t I /2 \ = 13 hr.

0.2 I I I I

0 I O 20 30

HOURS FIG. 3. Differential exchange following differential labeling

of "'Cd-binding sites in alkaline phosphatase at pH 9.0; (above) in the presence of 1 mM Pi (below) in the absence of Pi. "'CdzCd?AP (O), Cdr"'Cd2AP (0).

1 0 T , , , , , I O , , , , / , A B

"'Cd. AP-Cd Exchange pn90

"'Cd. A P + mM P, -Cd Exhangs pH 90

0 10 20 30 40 50 60 70 0 2 I ' L I l '

0 10 20 30 40 50 60 70

FIG. 4. '"Cd(II) tf Cd(I1) exchange in "'Cd4AP, pH 9.0; A, no addition; B, plus 1 mM HP04'-. Conditions as in Fig. 2. The exchange of the 4 metal species can be fit by two first order exchange processes shown in the lower parts of the figures. These two first order exchanges sum to the solid line drawn through the initial experimental points.


TABLE I1 Competition between Cd(II), Zn(II), Mn(It), and Mg(llJ for metal

binding sites of alkaline phosphatase Stoichiometric additions were made to the apoenzyme as indicated

in the left column. One-ml aliquots were then placed in 1-ml dialysis bags and the samples labeled with "'Cd(I1) or "'Zn(I1) as described under "Materials and Methods." Conditions: 5 X lo-$ M Me(II), 5 X lo-' M enzyme, 0.01 M Tris HCI-0.1 M NaCl, 4 "C.

g atom Cd/ Additions to apoenzyme Additions to medium dimer

24 h 30 days

A. "'Cd binding, pH 6.5

Apo + 2 Zn + 1 mM P, 1 mM P, 2.14 2.0 Apo + 2 Zn + 10 mM MgL+ 1 mM P,, 10 mM Mg" 1.47 2.2

A p o + 2 M n + l m ~ P , 1 mM P, 3.10 3.6 Apo + 2 Mn + 10 mM Mg'+ 1 mM P,, 10 mM Mg" 0.69 0.44

A p o + 4 Z n + l r n ~ P , 1 mM P, 0.64 1.4

APO 1 mM P, 4.1 4.1

+ 1 mM P,

+ 1 mM P,

~

B. "'Cd binding to alkaline phosphatase, pH 8.0 APO 1 mM P, 4.2 4.2 Apo + 2 Zn + 1 mM P, 1 mM P, 1.75 2.4 Apo + 2 Zn + 10 mM Mg2+ 1 mM PC, 10 mM Mg" 1.05 1.4

Apo + 4 Zn None 0.88 1.6 Apo + 4 Zn + 1 mM P, 1 mM P, 0.12 1.7

+ l r n M P ,

g atom Zn/mol dimer

24 h 120 h C. "'Zn binding to alkaline phosphatase, pH 6.5

APO None 4.05 APO 1 mM P, 3.92 Apo + 10 m~ Mg2+ 1 mM P,, 10 mM Mg2' 2.48 Apo + 2 Cd None 1.60 2.30 Apo + 2 Cd + 1mM P, + 1 mM PI, 10 mM Mg2+ 2.25 4.20

10 mM Mg2' Apo + 4 Cd None Apo + 4 Cd + 1 mM P, 1 mM P,

0.84 2.30 0.91 1.20

FIG. 5. EsZn(II) binding to apoalkaline phosphatase as a function of [zn(n)l, [ H p 0 4 2 - 1 , P " ) I , and pH. A, enzyme-bound 'j5Zn(II)/dimer uersus pH, no addition (0); plus mM HP04" (0); 5 x M Zn(II), 5 X M enzyme. B , enzyme-bound =Zn(II)/dimer uersus pH; plus 10 mM Mg(I1) (W), identical data was observed without Mg(I1); (0) plus I mM P,; (0) plus I mM P, and 10 mM Mg(I1); 5 X M Zn(II), 5 X lo-' M enzyme. C, enzyme-bound ""Zn(II)/dimer uersus [Zn(II)] pH 6.5, (0) no addition; (0) plus 1 mM P,; (m) plus 1 mM P, and 10 mM Mg(I1). Corresponding Cd(I1) data in the absence of P, (-); plus P, (-.-). D, enzyme-bound "Zn(II)/dimer uersus [Zn(II)], pH 8.0. (0) initial labeling; (O), following dialysis of samples labeled a t 5 X M Zn(I1) against metal-free buffer; points are plotted at final free Zn(I1) concentration. (0) two samples were further dialyzed against 10 mM Mg(I1). Conditions as in Fig. 1.

of a 50-ml volume excess of 5 X M "Zn(I1) as a function of pH is shown in Fig. 5A. Even in the absence of phosphate ligand, 4 g atoms of Zn(II)/mol of dimer are bound at pH 6.5. This stoichiometry remains constant to pH 7.5. At higher pH further binding occurs and -5 g atoms of Zn(II)/mol of dimer are bound at pH 9.0. While the Zn(I1) stoichiometry at pH 6.5 is not altered by mM Pii, the binding of extra Zn(I1) at alkaline pH begins at lower pH and reaches -6.5 g atoms/mol of dimer at pH 9.0. Enzyme-bound phosphate is almost exclusively in the form of the noncovalent complex over this pH range in the case of the Zn(I1) enzyme (19). Addition of 10 mM Mg(I1) has little effect on the "Zn(I1) binding stoichiometry at any pH (Fig. 5 B ) . On the other hand, if the phosphate ligand is added along with 10 mM Mg(II), the Zn(I1) stoichiometry is depressed by -2 g at.oms/mol of dimer a t all pH values (Fig. 5 B ) . The Zn(I1) binding occurring at higher pH does not appear to be influenced by Mg(II), since the binding curve moves upward toward higher pH values in the same manner, but begins at a lower stoichiometry (Fig. 5B). The Zn(II), above 4 g atoms/mol of dimer, dialyzes very quickly (see below).

At pH 6.5 a pair of Zn(I1) ions bind to apoalkaline phosphatase with a binding constant of -5 X 10" M, estimated from the Zn(I1j concentration required for binding of an average of 1 Zn(I1) ion/dimer (Fig. 5 C ) . The second pair binding at pH 6.5 has a binding constant of -5 X M. In contrast to Cd(I1) binding at pH 6.5, 1 mM phosphate or 10 mM Mg(I1) have only marginal effects on the final Zn(I1) equilibrium stoichiometry (Fig. 5 C ) . An additional finding that distin- guishes Zn(I1) binding from Cd(I1) binding is its time dependence. The Zn(I1) stoichiometry uersus concentration shown in Fig. 6B is that reached after 72 h of equilibration and the stoichiometry rises slowly over this period to reach this constant value after 72 h. For example, the Zn(I1) stoichiometry at M Zn(1I) is 0.56, 1.16, and 1.6 g atoms/mol of dimer at 2 4 4 8 , and 72 h, respectively. At pH 6.5 Zn(I1) is bound much more tightly to the enzyme in the absence of phosphate than

+ m M P. /'

4 no addit lon

3 t

6 5 70 7.5 80 8.5 9.0

PH

C.

Metal Ion Binding to Alkaline Phosphatase 39 1

is the larger Cd(I1) ion. Binding of phosphate, however, stabilizes the Cd(I1) enzyme-complexes to the point where they appear a t least as stable as those of Zn(I1) (Fig. 5C).

Metal ion binding to the apoenzyme as a function of metal ion concentration cannot be interpreted as a simple reversible equilibrium process. At pH 8, Zn(I1) binding reaches 4 gm atoms/mol of dimer a t lower concentrations of free metal than at pH 6.5, but the apparent stability of the first pair of Zn(I1) ions bound is not appreciably increased by alkaline pH, a somewhat surprising result. Once 4 Zn(I1) ions are bound, if the samples, initially a t lo-' M free metal, are resuspended in metal-free buffer and allowed to re-equilibrate at to M free Zn(II), equilibrium is reached at a much higher stoichiometry of "Zn(II), approximately 4 g atoms/mol of dimer. The stoichiometry does not fall to even 3 g atoms/mol of dimer until -3 x lo-' M free Zn(I1) (Fig. 50). This suggests that the binding of a metal ion to a second site on the monomer must induce conformational changes that stabilize both the fist and second metal ion complexes. Homologous and heterologous metal exchange studies with "Zn(I1) and "'Cd(I1) to be shown below also suggest such cooperative changes, since they indicate large differences in the compar- ative stability of the fully metalated Zn(I1) and Cd(I1) species, differences not suggested by the initial binding stoichiometry as a function of metal ion concentration (Figs. 1 and 5 ) .

Achievement of equilibrium in the redialysis experiments was defined operationally by changing the 50-ml dialysate to "metal-free'' buffer and following the increase of radioactivity in the dialysate and the loss of enzyme-bound radioactivity at successive 24-h periods until no further changes were observed. This took 7-10 days depending on enzyme concentration, and the free metal ion concentration is plotted as determined from the final specific activity of the dialysate. If equilibrium a t a lower free metal ion concentration was desired, the enzyme-containing dialysis bag was resuspended in a second 50 ml of metal-free buffer and the process repeated, as for example the "Zn(I1)-enzyme sample in Fig. 5 0 containing 3 g atoms of Zn(II)/mol and a free [Zn(II)] of 3 X IO-' M. While the above protocol was adequate, individual test experiments were continued for as long as 4 weeks, but did not result in either further increase in the binding when starting with apoenzyme or in loss of metal ion from enzyme that had initially been fully metalated. Thus the failure to reach the same "equilibrium" in the forward and reverse directions does not appear to be due to differences in the methodology, but seems to relate to kinetic barriers to conformational change in the initial partial binding reaction, barriers removed by binding of the full complement of metal ions which results in a more stable species (see "Discussion").

Having observed this "hysteresis" effect on metal-enzyme stability by dialyzing a fully metalated Zn(I1) enzyme, we repeated the experiment on Cd(II),AP and found a similar but not as dramatic increase in stability of the fully metalated enzyme (Fig. 1A). Two g atoms of Cd(I1) remain bound a t 5 X M at pH 6.5, a free Cd(I1) concentration a t which little would bind in the forward arm of this hysteretic binding curve. Phosphate binding remains a relatively more effective stabilizing process for the Cd(I1) ion.

"Zn(I4 t tZn( I4 Exchange at the Active Center ofAlkaline Phosphatase-Exchange of "'Zn(I1) bound at the active center of the enzyme (5 X M) with equimolar stable Zn(I1) in the dialysate at pH 8.0 is plotted in Fig. 6. Note that exchange is always from the 4 Zn(I1) species. Two examples are shown. The first (solid symbols) is "'Zn(II),AP exchanged against stable Zn(I1) in the absence of ligands or Mg(I1). The exchange divides dramatically into two phases. One pair of "'Zn(I1) ions exchanges relatively rapidly with a half-time of -24 h, while

4% \ -\

m M P. + IOmM M g i i _"_ "_ I """"" -. - ., - " .-.-. t I /2 = 840 hr.

-0-

t 1/2 = 24 hr.

0.4 0.3 i \ 0.2

0 80 160 240 320 400 480 560

HOURS FIG. 6. 65Zn(II) ++ Zn(I1) exchange at pH 8.0 in

65Zn(II)2Zn(II)2AP + 1 mM Pi + 10 mM Mg(I1) (0) and in 65Zn(II)4AP (0). Exchange for "'Zn(II),AP breaks down into two first order processes with t 1 / 2 values of 24 and 840 h, respectively (solid lines). Conditions as in Fig. 2.

the second pair exchanges very slowly with a half-time of -840 h. The analogous slow exchange for the Cd(II),AP species even at pH 8.0 is -15 h (Table I). The open symbols in Fig. 6 illustrate the exchange of a species initially labeled by the addition of 2 "Zn(I1) ions in the presence of 10 mM Mg(I1) and 1 mM Pi to insure placement of the Zn(I1) ions at the slowly exchanging sites. This species was then dialyzed against 1 mM P,, 1 X M Zn(II), 10 mM Mg(I1). The half- time of the slowly exchanging Zn(I1) species has now increased to -2500 h. Self-exchange rates of similar magnitude for the Zn(II)4AP species are observed at pH 6.5 (Table I). The binding of the phosphate ligand (still mostly as E - P for the Zn(I1) enzyme at pH 6.5) stabilizes the Zn(I1) at both pairs of sites (Table I). Substantially more than 2 g atoms of "Zn(II)/ dimer remain after 500 h and suggest exchange half-times of greater than 1000 h. These rates should be compared to the Cd(I1) exchange under the same conditions which has a maximum half-time of -70 h for the slowest pair in the phosphoryl enzyme and only -10 h in the unliganded species which remains CdzAP at pH 6.5 (Fig. 1).

In the same manner as Zn(I1) blocks "'Cd(I1) binding (Table 11), Cd(I1) can be shown to block "Zn(I1) binding (Table 11). At 5 X M enzyme and Zn(II), approximately 4 gm atoms of "Zn(I1) are bound to the apoenzyme per mol of dimer after 24-h dialysis in the presence or absence of 1 mM P,. The prior addition of 2 Cd(I1) ions/mol of dimer blocks a pair of these sites. MgfII), 10 mM, in the presence of phosphate will also block a pair for 24 h if the Mg(I1) and Pi are added to the enzyme prior to dialysis. After prolonged dialysis "Zn(I1) exchanges with the Mg(I1). A second pair of Cd(I1) ions also blocks the second pair of potential Zn-binding sites, but even by 24 h of dialysis against "Zn(I1) significant exchange has occurred (Table 11).

'%d(II) tt Zn(II) Exchange at the Active Center of Alka- line Phosphatase-Ordinarily the exchange rates would be expected to reflect the kinetic metal dissociation constants and Zn(I1) would be expected to replace Cd(I1) at least as fast as Cd(I1) replaces itself and probably faster. Such is not the case as shown in Fig. 7. "5Cd(II)rAP at pH 6.5 exchanges with equimolar Zn(I1) with a half-time of -16 h, the same order of magnitude as observed for Cd(1I) self-exchange. In contrast, if "'Cd(I1)~AP-P is used (induced by adding 1 mM Pi at pH

392 Metal Ion Binding to Alkaline Phosphatase -

'I5Cd -Zn

0.3 1 0 2

0 80 160 240 320 400 480 560 640 720

HOURS

FIG. 7. "'Cd(I1) t) Zn(I1) exchange at pH 6.5 in 1L'Cd(II)4AP (0); "'Cd(II)AF' + 1 mM Pi + 10 mM Mg(I1) (0) and "'Cd(II),AP (0) and 115Cd(II),Cd(II),AP + 1 mM Pi + 10 m M Mg(I1) (0). Conditions: 5 X M enzyme, 5 X M free Zn(II), 0.01 M Tris-0.1 M NaCl, pH 6.5, 4 "C.

6.5) then both pairs of "'Cd(I1) ions of the Cd(II)4 protein are extremely resistant to Zn(I1) exchange with half-times of 400 h or greater (Fig. 7). The maximum half-time for Cd(I1) self- exchange of this species is 70 h (Fig. 2, Table I). The large difference in Cd(I1) c) Cd(I1) versus Cd(I1) Zn(I1) exchange (a 17-fold longer heterogeneous exchange) cannot be explained by simple equilibrium processes. This suggests that the conformational state of the Cd(II)4AP or Cd2(II)Mga(II)AP must be significantly different from their Zn(I1) counterparts, such that Zn(I1) exchange at one of the three sites as an isolated event does not achieve an immediate stability, unlike the homogeneous Zn(I1) species. Addition of 10 mM Mg(I1) has a marginal effect on the Cd(I1)-Zn(I1) exchange of the "'Cd(II), species, actually accelerating exchange slightly. This may relate to competition of Mg(I1) for one of the pairs of sites.

The precise role of Mg(I1) cannot be assessed well in these exchange systems, since competition between the two stable ion species cannot be followed by this method. Competition of Mg(I1) with single radioactive species will be illustrated below. Addition of 1 mM Pi and 10 mM Mg(I1) to the "'Cd(II),AP species has a complex effect in that "'Cd(I1)- Zn(I1) exchange starts out rapidly and then slows dramatically (Fig. 7). This probably relates to the stabilization of the second pair of binding sites in the presence of phosphate and the dramatic slowing may relate to the formation of relatively stable " 'C~AZ~RAP-P hybrids.

""Co(Ig Binding to Apoalkaline Phosphatase-At pH 8.0 -2 g atoms of Co(I1) are bound per enzyme of dimer at a free Co(I1) concentration of M (Fig. 8). As the free Co(I1) concentration is raised to lo-" M, the stoichiometry of enzyme- bound Co(I1) rises to -10 g atoms/mol of of dimer. This large stoichiometry of nonspecific metal ion binding at alkaline pH is not unique to Co(I1). At pH 8.0, if the Cd(I1) concentration is raised to M, the stoichiometry is also -10 rather than 6 as observed at pH 6.5 (Fig. 1). Enzyme-bound Co(I1) above 4 g atoms/mol of dimer is fairly loosely bound, since it is readily removed by dialysis against metal-free buffer and the stoichiometry falls to -4 g atoms of Co(II)/mol of enzyme dimer. In contrast to the earlier figures for Zn(I1) and Cd(II), the points representing stoichiometry after dialysis are plotted on the x axis at the position corresponding to the original labeling concentration to illustrate the fraction of Co(1I) di-

alyzed from each sample. Final free Co(I1) concentration in each was M. Of the 4 nondialyzable bound Co(I1) ions, two are removed by dialysis if 10 mM Mg(I1) is added to the dialysate.

Structural and Functional Effects of Mg(I4"In the preparation of samples of both the Zn(II)2AP and Cd(II)zAP for NMR studies a t enzyme concentrations M (>loo mg/ ml), if 10 mM Mg(I1) is added and the enzymes then dialyzed, the Mg(I1) content rapidly falls to -2 g atoms/mol. Mg(I1) can be removed by extensive dialysis against metal-free buffer. The precise role of Mg(I1) in the structure and function of this enzyme with two or more pairs of potential metal binding sites is more difficult to assess than that of the IIB metal ions, since it requires the prior presence of a IIB metal ion at the most stable metal binding site before Mg(I1) effects are de- monstrable. Mg(I1) appears to bind considerably less tightly than either Zn(I1) or Cd(I1) even to the pair of sites it preferentially occupies and for which all three metal ions compete. If the Mg(I1) concentration is raised to 10 mM it can successfully compete for this pair of sites as illustrated in Figs. 1, 5, and 8. On an equimolar basis, however, it is a poor competitor, hence stoichiometric titrations which achieve maximal Mg(I1) effects are difficult or impossible.

A functional effect of Mg(I1) on the phosphorylation of cadmium alkaline phosphatase is readily demonstrated. At -4 mM concentrations of H"P04=, Cd(II),AP phosphorylates relatively slowly (i.e. hours) a t pH 6.5 as measured by the formation of E-'"P (Fig. 9). Final equilibrium stoichiometry in such reactions in our hands has always been -1 mol of phosphate/mol of dimer for the Cd(II),AP (Fig. 9). The rate, but not the final stoichiometry, of phosphorylation is dramatically enhanced by the addition of Mg(I1). The half-time of phosphorylation progressively decreases from 20-30 min at 0.1 mM Mg(I1) (not significantly different from that in the absence of Mg(I1)) to -1 min a t 5 and 10 mM Mg(I1) (Fig. 9).

In contrast to Mg(II), the stoichiometric addition of a second pair of Cd(I1) ions brings about the identical rate enhancement of the phosphorylation step, but also raises the final stoichiometry of phosphorylation to between 1.5 and 2.0 mol of E-P/mol of dimer (Fig. 9). The increase in the stoichiometry of formation of E-P by the Cd(II),AP was first detected in the NMR samples by integrating the '"P-NMR signal (15), an integration that consistently shows -2 mol of E-P/ mol of enzyme dimer. While in some samples, especially in the presence of excess Cd(II), determination of E-P stoichiometry by acid precipitation has indicated -2 mol/mol enzyme dimer (15), the acid precipitable :12P is usually below 2

/ After MF 1

FIG. 8. 6oCo(II) binding to apoalkaline phosphatase uersus [Co(II)] at pH 8.0. Enzyme-bound "'Co(II)/dimer after 76 h of labeling (0). Same samples after 48 h of dialysis against metal-free buffer (0). Same samples after 48 h of dialysis against metal-free buffer containing 10 mM Mg(I1) (0). Conditions: 1 X M enzyme from to M co(II) , 5 X M enzyme from to 10 -~ M Co(II), and 2 X M enzyme from to lo-" M Co(I1); 0.01 M Tris HCI-0.1 M NaCl, pH 8.0, 4 "C.


0 10 20 0 I O 20 30 40 50 60

M I N HOURS

FIG. 9. Effect of Mg(II) on the phosphorylation (E-32P/E) of Cd(I1) alkaline phosphatases. Conditions: 0.01 M Tris HC1,O.Ol M NaOAc, 0.1 M NaCl, pH 6.5, 5.26 X M alkaline phosphatase, 3.93 x M H3'P04'-. Metal ion stoichiometry: Cd(II)/alkaline phosphatase = 4.07 (e); Cd(II)/alkaline phosphatase = 2.03, Mg(II)/ alkaline phosphatase = 2.04 (0); Cd(II)/alkaline phosphatase = 2.0, [Mg(II)] = 5 m~ (V); Cd(II)/alkaline phosphatase = 2.0, [Mg(II)] = 10 mM (0).

100 - w 5 N >

w z w > t- Z

-

-

- - a

5 0 -

I I I I I I 0 1 2 3 4 5 6

gm AT Zn or Co / mole DIMER FIG. 10. Reactivation of apoAP (5 X low5 M) by Zn(I1) and

Co(I1) as a function of moles of Me(L1) added/mol of enzyme dimer in 0.01 M Tris HC1-0.1 M NaCl (0, Zn(I1); 0 Co(II)), and in the additional presence of 10 mM Mg(1I) (e, Zn(I1); ., Co(11)). Following 1 h of incubation activity was determined at pH 8.0 as described under "Materials and Methods." Activity is expressed as the per cent recovered relative to the activity of the native enzymes from which the apophosphatase samples were prepared. Activity of individual preparations of the native enzyme varied from 2OoO to 2500 @mol of substrate hydrolvzed/h/mg of enzyme.

mol by 0.2 to 0.5 mol. The frequently observed stoichiometry of 1.5 may be accounted for by the disproportionation of Cd(II),AP into a mixture of Cd(I1)zAP-P and Cd(II)6AP-P2 which tends to occur on the phosphorylation of the Cd(1I)JP species (19).

A Mg(1I) effect is also prominent in reactivation titrations of the apoenzyme with Zn(I1) or Co(I1) (Fig. 10). Two Zn(1I) or Co(I1) ions invariably restore complete activity to the apoenzyme if 10 mM Mg(I1) is present. In the absence of Mg(II), however, up to 6 mol of Zn(I1) or Co(I1) are required,

although this number is somewhat variable depending on conditions.

DISCUSSION

Alkaline phosphatase can potentially bind three pairs of metal ions over a free metal ion concentration range from 1 O P M to IO-" M (Figs. 1 and 5 ) . The magnitudes of the binding constants of the successively more weakly bound pairs are not sufficiently different under many conditions to resolve the binding curves for each of the three pairs of metal ions. Because of this one might raise the question of whether metal ion binding to the apoenzyme produces a strictly homogeneous species (each type of site fully occupied) until all 6 metal ions are bound. While this is clearly a potential problem, several conditions of metal ion binding can be found which enable one to distinguish the properties of the different pairs during less than maximal metal ion binding. These distin- guishing features will be used to designate the pairs of sites as A, B, and C in decreasing order of binding affinity. From the direct binding curves for ""Cd(II), apoalkaline phosphatase binds approximately two Cd(I1) ions at 5 X M Cd(I1) and pH 6.5 (Fig. 1). Since the binding of the first two Cd(1I) ions in the absence of phosphate is part of a smooth binding curve achieving a stoichiometry of 6 Cd(I1) ions/dimer at lo-:' M Cd (Fig. l A ) , this data alone cannot identify the first two as a unique pair. The following additional data has to be used to argue that these represent homogeneous A-site Cd(I1). A second pair of Cd(I1) ions is induced to bind by phosphate, hence one of the pairs is not binding in the M Cd(I1) range. This latter pair involves binding to sites whose affinity is markedly increased by formation of the phosphoseryl intermediate (Fig. 1) and will be designated the B sites. The enzyme at pH 6.5 with only two Cd(I1) ions bound per dimer shows a single ":@d(II) NMR signal corresponding to both ""Cd(I1) ions as will be detailed in the following paper (19). Thus, a unique pair of sites is occupied, which will be defined as the A sites.

Not unexpectedly both A- and B-site Cd(I1) complexes are made more stable by raising the pH (Fig. l ) , but only the less stable pair is successfully competed for by Mg(I1) (Fig. 1C). Competition with Mg(I1) is thus another operational defini- tion of the B-site Cd(I1) ions. Once the stable homogeneous 4 Cd(I1) species is formed it is easy to demonstrate by the time dependence of the "'Cd(I1) ++ Cd(I1) self-exchange that there are two types of site present characterized by differential exchange constants (Figs. 2-4, Table I). While both sites are stabilized by the presence of the phosphate ligand (either as E-P, pH 6.5 or E-P, pH 9.0), they are still distinguished by different exchange rates in the phosphate complexes and the sites can be differentially labeled by the order of addition of pairs of "'Cd(I1) and stable Cd(I1) ions (Fig. 3 ) .

While the "on" constants for the formation of these enzyme- metal complexes may be smaller than the diffusion limit, it still seems reasonable that the slow step, and the one account- ing for the major difference in relative stabilities of a series of metalloalkaline phosphatases or their phosphoenzyme forms, is a relatively small "off" constant. The kinetic constants characterizing the self-exchange reactions can be used to distinguish the Cd(I1) complexes of the A and B sites. At pH 9.0 where both A and B sites are occupied in the unliganded enzyme the tight Cd(I1) site has a he, of -4 x s", while the second site has a he, of -8 X 10"' s" (Table I ) . While formation of E-P (pH 6.5) or E . P (pH 9.0) stabilizes both A and B sites (Table I), the relatively large dissociation constant for the B site results in induction of binding to the B site by the phosphate ligand at low pH (Fig. 1A ). There is a functional consequence of this, since binding of Cd(I1) or Mg(I1) to the


B sites accelerates the phosphorylation rate dramatically (Fig. 9). In view of the dramatic stabilization of both A and €3-site Cd(1i) by the phosphate ligand, it is not certain which site is the most stable to exchange in the phosphoderivatives. There is dat.a from both ""Cd and "'P NMR to suggest that once the enzyme is phosphorylated the A site becomes most labile to exchange (19,30).

The two pairs of sites found for Cd(I1) also represent two pairs of Zn(1I) binding sites, since Cd(I1) binding is almost completely blocked by 4 Zn ions, a block which is made more effective by the phosphate ligand (Table 11). The second pair of sites also appears to be shared by Mg(II), since addition of 10 mM Mg(I1) and phosphate to a 2 Zn(I1) enzyme partially blocks the binding of Il5Cd(II) to the second pair. There is apparently a special stability of the mixed Mn(II)2Mg(II)2 phosphoenzyme. While the Mn(I1) enzyme has been exten- sively characterized by ESR and 2 Mn(I1) ions shown to occupy unique low symmetry sites (27), 2 Mn(I1) ions alone are not very effective in blocking "'Cd(I1) binding (Table 11). In contrast the combination of Mn(I1) and Mg(i1) in the phosphoenzyme is especially effective (Table 11). This suggests that the stability of metal complexes a t both sites is critically influenced by occupancy of the second site and by the presence of the phosphate ligand and perhaps by the properties of the metal ion as well, e.g. ionic radius. The ready formation and characterization of stoichiometric 4 metal enzymes with both Zn(I1) and Cd(I1) tends to emphasize the roles of these ions in the B sites. Yet under physiological conditions the concentrations of free metal ion in cell-free fluids (-0.01 mM for Zn(I1) and 5-10 mM for Mg(I1) (25, 26)) may favor Mg(Ii) as the B site metal ion.

X-ray diffraction studies of crystals of the Cd(I1) phosphoryl enzyme in the presence of 50 mM Cd(I1) have recently identified three closely spaced Cd(I1) binding sites on each monomer, the second and third separated from the catalytic or A site by -4 and -5 A, respectively (28).:' The catalytic or A site contains histidyl ligands and appears to correspond to that identified by Cu(I1) ESR to contain three equivalent nitrogens as ligands to the Cu(I1) ion (29). A second pair 3.9 A away is close to seryl 102 and would appear to be the site to which Cd(I1) binding is induced by phosphate at low pH.

The dialysis of "Zn(I1) and "'Cd(I1) into samples of the apoenzyme indicate rather large dissociation constants for even the more tightly bound metal ions, i.e. not less than 2 X 10" M for Zn(1I) (Fig. 5 ) . This led us to consider the possibility that binding of the second pair of metal ions might result in conformational changes which further stabilized the first pair. Such a conformational change appears to take place, since if one re-establishes equilibrium by redialysis of the "Zn(II)4AP species against metal-free buffer and measures the achievement of equilibrium by following radioactivity in the dialysate, a much smaller apparent dissociation constant is indicated, i.e. -IO-' M even for the second pair (Fig. 5 D ) .

Such data imply that binding of metal ions at one of the three sites in each monomer influences the dissociation constants of the metal ions at the other two sites. The phosphate

"Analysis of the electron density map obtained from the Cd(I1) phosphoryl-enzyme using anomalous dispersion shows that the metal binding site originally designated A in the crystal structure (28) consists of two separate Cd(I1) sites 3.9 A apart (A and A,) (J. M. Sowadski, M. Handschumacher, K. Murthy, and H. W. Wyckoff (1982) Nature, submitted for publication). The site originally labeled B in the crystal structure appears to consist of carboxyl ligands exclusively and probably corresponds to the site defined as C from the ""Cd NMR in the following paper (19). The new site Ar, much closer to the catalytic or A site and close to the phosphoseryl residue is likely to correspond to the site defined here as B and to which Cd(I1) binding is induced by phosphorylation.

ligand has an even more striking ability to stabilize these metal-enzyme complexes (Figs. 1, 2, and 6 ) , a fact confirmed by the effect of phosphate binding on the ".'Cd NMR signals from these sites as shown in the following paper (19). In view of the close spacing of the three metal binding sites, less than 10 8, apart,:' and the direct coordination of the phosphate of E -P t o t he A site Cd(I1) ion as shown in the following two papers (19, 30), it is not surprising that there is positive cooperativity between the three metal ion sites, particularly the first two (A and B), and the phosphate ligand. While behavior of this kind may not be unexpected for a multinuclear chelate complex undergoing formation of a mixed complex with phosphate, the cooperativity of metal binding to the alkaline phosphatase monomer can also be shown by the '"'Cd and '"P NMR methods (19,30) to involve interactions between the subunits of the dimer. For example, the binding of phosphate to Cd(II),AP i s shown to be associated with migration of Cd(I1) from the A site of one monomer to the B site of the phosphorylated active cent.er of the opposite monomer, a migration which requires the presence of subunit-subunit interactions to produce the relative destablization of the A site of one of the monomers (19).

The metal exchange reactions of the fully metalated phosphorylated dimer are extremely slow (Figs. 2 and 6, Table I) and can account for the slow dissociation of metal ions in dialysis experiments involving fully metalated enzyme. The failure in many instances, however, to bind a full complement of metal ions to the apoenzyme except in the presence of phosphate or at higher metal ion concentrations than expected from the exchange experiments, may relate to the existence of several conformational states of alkaline phosphatase which are of similar stability, but interconvert very slowly. The kinetic barriers which separate these states can be overcome by additional binding of pairs of metal ions or the phosphate ligand. Once established, the final conformation does not appear to be readily reversed. On the other hand, if the apoenzyme is binding less than a full complement of metal ions two conformational states of the protein may exist si- multaneously and not interconvert except on a time scale of days. While the radioactive metal binding cannot prove the existence of several conformational states, the chemical shifts, coupling, and exchange broadening of '"'Cd(11) NMR signals (and in many instances the "'P NMR signals as well) can easily discriminate between different conformational states of the protein and show that several can exist in the same sample. These species interconvert very slowly if at all despite the fact that a pathway for interconversion must exist (19).

The catalytic roles of two closely spaced metal ions at the active center of alkaline phosphatase remain speculative at the present time. One of the roles of Zn(I1) is clearly to produce conformational changes which result in phosphate binding as has been documented previously (16). Another possible role would be to generate the nucleophile, either coordinated -OH or HzO, required to hydrolyze the phosphoseryl intermediate. In addition, it seems highly probable, in common with phosphoryl transfer enzymes in general, that charge neutralization at the negative phosphate dianion is one of the functions of the metal ion in order to facilitate a nucleophilic attack of the seryl hydroxyl on the incoming phosphorus atom. Spatial relationships alone could account for a mechanistic advantage to separating these two functions by incorporating two metal ions. Indeed, binding of CdUI) or Mg(I1) at the B site accelerates phosphorylation, and as will be shown in the following paper (19) the phosphate of E - P coordinates one, but not both Me(I1) ions at the active center. The charge/radius ratio of the particular pair of metal ions occupying the A and B sites may significantly influence the


topology and stability as well as the catalytic efficiency of the Bid . Chem. 252, 7053-7061 resultant metalloenzymes and may account for the slow turn- 14. Otvos. J. D., Alger, J. R.. Coleman, J . E., and Armitage, 1. M. over of the Cd(I1) enzyme, the alteration in apparent pK, of phosphorylation-dephosphorylation (see Ref. 19), and the dif- 15. Otvos, J. D., Armitage, I. M., Chlebowski, J. F., and Coleman. J.

ferences in stoichiometry of phosphate binding 16. Applebury, M. L., Johnson, B. P., and Coleman, J . E. (1970) J . cooperativity) between the homogeneous IIB metal enzymes Biol. Chem. 245,4968-4976 and the Mg(I1) hybrids. 17. Malamy, M. H., and Horecker, B. L. (1964) Biochemistry 3, 1893-

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

E. (1979) J. Biol. Chem. 254,4707-4713

1.

2.

3.

4.

5.

6. 7.

8.

9.

10.

11.

12.

13.

REFERENCES Harris, M. I., and Coleman, J. E. (1968) J. Biol. Chem. 243,5063-

Applebury, M. L., and Coleman, J. E. (1969) J . Biol. Chem. 244,

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

Plocke, D. J., Levinthal, C., and Vallee, B. L. (1962) Biochemistry

Bosron, W. F., Kennedy, F. S., and Vallee, B. L. (1975) Biochem-

Reid, T. W., and Wilson, I. B. (1971) The Enzymes 4, 373-415 Simpson, R. T., and Vallee, B. L. (1968) Biochemistry 7, 4343-

Reynolds, J. A. and Schlesinger, M. J. (1969) Biochemistry 8,

Petitclerc, C., Lazdunski, C., Chappelet, D., Moulin, A,, and

Trotman, C. N. A,, and Greenwood, C. (1971) Biochem. J. 124,

Csopak, H., and Szajn, H. (1973) Arch. Biochem. Biophys. 157,

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

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

5073

308-3 18

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

1, 373-378

istry 14, 2275-2282

4350

588-593

Lazdunski, M. (1970) Eur. J. Biochem. 14, 301-308

25-30

374-379

chemistry 15, 3710-3716

1897 18. Bradshaw, R. A., Cancedda, F., Ericsson, L. H., Neumann, 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

19. Gettins, P., and Coleman, J. E. (1982) J. Biol. Chem. 257,396-407 20. Applebury, M. L., and Coleman, J. E. (1969) J . Biol. Chem. 244,

21. Csopak, H., Falk, K.-E., and Szajn, H. (1972) Biochim. Biophys.

22. Coleman, J. E., and Vallee, B. L. (1960) J. Biol. Chem. 235,390-

23. Hughes, T. R., and Klotz, I. M. (1956) Methods Biochem. Anal.

24. Stranks, D. R., and Wilkins, R. G. (1957) Chem. Rev. 57, 743-866 25. Zinc (Medical and Biologic Effects of Environmental Pollutants)

Subcommittee on Zinc, National Research Council (1979) Uni- versity Park Press, Baltimore

26. Wacker, W. E. C., and Vallee, B. L. (1958) New Engl. J. Med.

27. Weiner, R. E., Chlebowski, J. F., Haffner, P. H., and Coleman, J .

28. Sowadski, J. M., Foster, B. A,, and Wyckoff, H. W. (1981) J . Mol.

29. Taylor, J. S., and Coleman, J. E. (1972) Proc. Natl. Acad. Sci. U.

30. Gettins, P., and Coleman, J. E. (1982) J. Biol. Chem. 257,408-416

709-718

Acta 258,466-472

395

3,265-299

259,431-438,475-482

E. (1979) J. Biol. Chem. 254,9739-9746

Biol. 150, 245-272

S. A. 69,859-862

Zn(II), 115mCd(II), 6oCo(II), and Mg(I1) Binding to Alkaline ...

Documents

Transcript of Zn(II), 115mCd(II), 6oCo(II), and Mg(I1) Binding to Alkaline ...