Probing the role of the trivalent metal in phosphate ester hydrolysis: Preparation and characterization of purple acid phosphatases containing AlIIIZnII and InIIIZnII active sites, including the first example of

Article abstract of DOI:10.1021/ja9837147

Purple acid phosphatases contain a dinuclear Fe³⁺M²⁺ center in their active site (M = Fe²⁺ or Zn²⁺). To resolve the specific role of the ferric ion in catalysis, a series of metal-substituted forms of bovine spl

Full text of DOI:10.1021/ja9837147

J. Am. Chem. Soc. 1999, 121, 6683-6689
6683
Probing the Role of the Trivalent Metal in Phosphate Ester
Hydrolysis: Preparation and Characterization of Purple Acid
Phosphatases Containing AlIIIZnII and InIIIZnII Active Sites, Including
the First Example of an Active Aluminum Enzyme
Maarten Merkx and Bruce A. Averill*
Contribution from the E.C. Slater Institute, Biocentrum Amsterdam, UniVersity of Amsterdam,
Plantage Muidergracht 12, 1018 TV, Amsterdam, The Netherlands
ReceiVed October 20, 1998
Abstract: Purple acid phosphatases contain a dinuclear Fe3+M2+ center in their active site (M ) Fe2+ or
Zn2+). To resolve the specific role of the ferric ion in catalysis, a series of metal-substituted forms of bovine
spleen purple acid phosphatase (BSPAP) of general formula MIIIZnII-BSPAP has been prepared, in which the
trivalent metal ion was systematically varied (MIII ) Al, Fe, Ga, and In). The activity of the AlZn-BSPAP
form was only slightly lower (kcat ≈ 2000 s-1) than that of the previously reported GaZn and FeZn forms (kcat
≈ 3000 s-1). The InZn form was inactive. The kinetics parameters and pH profile of AlZn-BSPAP were
remarkably similar to those of FeZn-BSPAP and GaZn-BSPAP, but AlZn-BSPAP was readily distinguished
from the GaZn and FeZn forms by its 50-70-fold lower inhibition constant for fluoride. These results are not,
at first sight, consistent with intrinsic properties of the trivalent metal ions as they are known from coordination
chemistry. In particular, aluminum has generally been believed to be of little use as a Lewis acid in the active
site of an enzyme because of the slow ligand exchange rates typically observed for aluminum complexes. The
present results are thus in conflict with this general wisdom. The conflict can be resolved either by assuming
that the protein modulates the properties of the aluminum ion such that ligand exchange rates are substantially
enhanced and thus not rate-limiting, or by assuming a catalytic mechanism in which ligand exchange does not
take place at the trivalent metal ion.
Purple acid phosphatases (PAPs) are unique enzymes in that
they use a dinuclear FeIIIFeII or FeIIIZnII center to catalyze the
hydrolysis of phosphate esters.1-3 They are further characterized
by their purple color, which is due to a tyrosinate-to-Fe3+
charge-transfer band, their acidic pH optimum, and their
insensitivity to tartrate, which is an inhibitor of other classes of
acid phosphatases. PAPs have been isolated from mammalian,
plant, and microbial sources.1 The mammalian enzymes, of
which uteroferrin (Uf) and bovine spleen purple acid phos-
phatase (BSPAP) are the best-studied examples, contain two
antiferromagnetically coupled iron ions. The plant purple acid
phosphatase from kidney beans (KBPAP) contains one iron and
one zinc. Despite this difference in metal content, the mam-
malian and plant PAPs display similar enzymatic and spectro-
scopic properties. One of the irons in BSPAP and Uf can be
replaced by zinc without drastic effects on the phosphatase
activity,4-9 while the zinc ion in KBPAP can be replaced by
iron, again without major effects on the catalytic activity.10,11
The kidney bean purple acid phosphatase is the only PAP whose
X-ray structure has been reported.12,13 The PAPs belong to a
much larger group of enzymes that catalyze phosphate ester
hydrolysis and share a sequence motif incorporating most of
the metal ligating amino acids in KBPAP.13-16 X-ray structures
of two Ser/Thr-specific protein phosphatases, protein phos-
phatase 117,18 and protein phosphatase 2B (calcineurin),19,20
(9) Merkx, M.; Averill, B. A. Biochemistry 1998, 37, 11223-11231.
(10) Beck, J. L.; de Jersey, J.; Zerner, B.; Hendrich, M. P.; Debrunner,
P. G. J. Am. Chem. Soc. 1988, 110, 3317-3318.
(11) Beck, J. L.; McArthur, M. J.; de Jersey, J.; Zerner, B. Inorg. Chim.
Acta 1988, 153, 39-44.
(12) Stra¨ter, N.; Klabunde, T.; Tucker, P.; Witzel, H.; Krebs, B. Science
1995, 268, 1489-1492.
(13) Klabunde, T.; Stra¨ter, N.; Fro¨hlich, R.; Witzel, H.; Krebs, B. J. Mol.
Biol. 1996, 259, 737-748.
(14) Koonin, E. V. Protein Sci. 1994, 3, 356-368.
(15) Zhuo, S.; Clemens, J. C.; Stones, R. L.; Dixon, J. E. J. Biol. Chem.
1994, 269, 26234-26238.
* To whom correspondence should be addressed: +31-20-525-5045
(phone); +31-20-525-5124 (fax); BAA@chem.uva.nl (e-mail).
(1) Klabunde, T.; Krebs, B. Struct. Bonding 1997, 89, 177-198.
(2) Wilcox, D. E. Chem. ReV. 1996, 96, 2435-2458.
(3) Stra¨ter, N.; Lipscomb, W. N.; Klabunde, T.; Krebs, B. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 2024-2055.
(16) Barton, G. J.; Cohen, P. T. W.; Barford, D. Eur. J. Biochem. 1994,
220, 225-237.
(17) Goldberg, J.; Huang, H.-B.; Kwon, Y.-G.; Greengard, P.; Nairn,
A. C.; Kuriyan, J. Nature 1995, 376, 745-753.
(18) Egloff, M. P.; Cohen, P. T. W.; Reinemer, P.; Barford, D. J. Mol.
Biol. 1995, 254, 942-959.
(4) Keough, D. T.; Dionysius, D. A.; de Jersey, J.; Zerner, B. Biochem.
Biophys. Res. Commun. 1980, 94, 600-605.
(19) Griffith, J. P.; Kim, J. L.; Kim, E. E.; Sintchak, M. D.; Thomson,
J. A.; Fitzgibbon, M. J.; Fleming, M. A.; Caron, P. R.; Hsiao, K.; Navia,
M. A. Cell 1995, 82, 507-522.
(5) Davis, J. C.; Averill, B. A. Proc. Natl. Acad. Sci. U.S.A. 1982, 79,
4623-4627.
(20) Kissinger, C. R.; Parge, H. E.; Knighton, D. R.; Lewis, C. T.;
Pelletier, L. A.; Tempczyk, A.; Kalish, V. J.; Tucker, K. D.; Showalter, R.
E.; Moomaw, E. W.; Gastinel, L. N.; Habuka, N.; Chen, X.; Maldonado,
F.; Barker, J. E.; Bacquet, R.; Villafranca, J. E. Nature 1995, 378, 641-
644.
(6) Beck, J. L.; Keough, D. T.; de Jersey, J.; Zerner, B. Biochim. Biophys.
Acta 1984, 791, 357-363.
(7) David, S. S.; Que, L., Jr. J. Am. Chem. Soc. 1990, 112, 6455-6463.
(8) Merkx, M.; Averill, B. A. Biochemistry 1998, 37, 8490-8497.
10.1021/ja9837147 CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/02/1999

6684 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Merkx and AVerill
which also have this sequence motif, show the presence of a
dinuclear metal center which resembles that of KBPAP, although
the identity of the metals is still the subject of some debate.21-23
enzyme is thus postulated to make use of the strong Lewis
acidity of Fe3+ to generate a hydroxide ion, even at the weakly
acidic pH where it is active (pH optimum ∼6).
In general, trivalent metal ions are stronger Lewis acids than
divalent metal ions. As such, they are widely used as catalysts
in organic synthesis, e.g., the Friedel-Crafts catalyst AlCl3.48
Ligand exchange reactions are, however, generally slower at
trivalent metal ions than at divalent metal ions.49 This fact has
been put forward to explain why the vast majority of metallo-
hydrolases contain divalent metal ions, in particular zinc and
magnesium.50 Despite this apparent disadvantage of trivalent
metal ions, several enzymes are now known to make use of the
strong Lewis acidity of a trivalent metal ion while exhibiting
fairly high turnover numbers: the PAPs and possibly other
phosphatases with the same sequence motif (Fe3+),2,22 nitrile
hydratases (Fe3+ or Co3+),51,52 and intradiol dioxygenases
(Fe3+).53,54
The metal centers in Uf, BSPAP, and KBPAP have been
studied extensively by various spectroscopic techniques, includ-
ing EPR,5,7,9,10,24,25 Mo¨ssbauer,25-30 resonance Raman,7,25,31
NMR,32-34 magnetic susceptibility,5,25,35-37 MCD,38 ENDOR/
ESEEM,39 and EXAFS.40-43 From these studies and from the
X-ray structures of KBPAP and its phosphate and tungstate
complexes,12,13 the structure of the dinuclear metal center in
PAP is now well characterized. Relatively little is known,
however, about the mechanism by which PAPs catalyze the
hydrolysis of phosphate esters. From a chiral substrate experi-
ment with BSPAP, it is known that hydrolysis results in net
inversion of the configuration around phosphorus, which
strongly suggests the direct attack of water/hydroxide on the
phosphate ester without the formation of a phosphorylated
enzyme intermediate.44 Stopped-flow measurements on the
reaction of phosphate with uteroferrin suggest a rapid binding
A classical approach for investigating the role of a metal ion
in enzymatic catalysis is substitution by another metal and
characterization of the resulting perturbation in spectroscopic
and/or enzymatic properties. We have recently reported that both
of the iron ions in BSPAP can be specifically replaced by other
metal ions.8 By preparing GaFe-BSPAP and GaZn-BSPAP, we
showed that the ferric iron can be replaced by Ga3+ without
major effects on kinetics parameters and pH optima. The
coordination chemistry of Fe3+ and Ga3+ is, however, known
to be very similar. Ga3+ and Fe3+ have similar ionic radii, they
show similar ligand exchange rates, and their Lewis acidities
are comparable.55 In the present study, we probe the role of the
trivalent metal ion by the preparation and characterization of
metal-substituted BSPAP forms of the general formula MIII-
ZnII, with two trivalent metal ions that differ significantly from
iron: aluminum and indium. The ionic radius of Al3+ is smaller
than that of Fe3+, while that of In3+ is larger.56 [Al(H2O)6]3+ is
also significantly less acidic than is [Fe(H2O)6]3+ or [Ga-
of phosphate to the Fe2+ 45,46
A mechanism has been proposed
.
in which the phosphate ester also rapidly binds to the Fe2+
,
followed by attack of an Fe3+-bound hydroxide ion.13,44,47 The
(21) Wang, X.; Culotta, V. C.; Klee, C. B. Nature 1996, 383, 434-437.
(22) Rusnak, F.; Yu, L.; Mertz, P. J. Biol. Inorg. Chem. 1996, 1, 388-
396.
(23) Chu, Y.; Lee, E. Y. C.; Schlender, K. K. J. Biol. Chem. 1996, 271,
2574-2577.
(24) Antanaitis, B. C.; Aisen, P.; Lilienthal, H. R.; Roberts, R. M.; Bazer,
F. W. J. Biol. Chem. 1980, 255, 11204-11209.
(25) Averill, B. A.; Davis, J. C.; Burman, S.; Zirino, T.; Sanders-Loehr,
J.; Loehr, T. M.; Sage, J. T.; Debrunner, P. G. J. Am. Chem. Soc. 1987,
109, 3760-3767.
(26) Debrunner, P. G.; Hendrich, M. P.; de Jersey, J.; Keough, D. T.;
Sage, J. T.; Zerner, B. Biochim. Biophys. Acta 1983, 745, 103-106.
(27) Cichutek, K.; Witzel, H.; Parak, F. Hyperfine Interact. 1988, 42,
885-888.
(28) Sage, J. T.; Xia, Y.-M.; Debrunner, P. G.; Keough, D. T.; de Jersey,
J.; Zerner, B. J. Am. Chem. Soc. 1989, 111, 7239-7247.
(29) Suerbaum, H.; Ko¨rner, M.; Witzel, H.; Althaus, E.; Mosel, B.-D.;
Mu¨ller-Warmuth, W. Eur. J. Biochem. 1993, 214, 313-321.
(30) Rodriguez, J. H.; Ok, H. N.; Xia, Y. M.; Debrunner, P. G.; Hinrichs,
B. E.; Meyer, T.; Packard, N. H. J. Phys. Chem. 1996, 100, 6849-6862.
(31) Gaber, B. P.; Sheridan, J. P.; Bazer, F. W.; Roberts, R. M. J. Biol.
Chem. 1979, 254, 8340-8342.
(32) Lauffer, R. B.; Antanaitis, B. C.; Aisen, P.; Que, L., Jr. J. Biol.
Chem. 1983, 258, 14212-14218.
(33) Scarrow, R. C.; Pyrz, J. W.; Que, L., Jr. J. Am. Chem. Soc. 1990,
112, 657-665.
(H2O)6]3+ 57
. When studying catalytic reactions involving ligand
exchange at a metal, it is important to consider the ligand
exchange rates typical of that metal ion. Thus, ligand exchange
rates are typically slow for Al3+ complexes in comparison to
the corresponding Fe3+ or Ga3+ complexes, while they are
higher for In3+ complexes.49,58-60 As a result, our findings, that
the AlIIIZnII-BSPAP form is ∼60% active as the FeIIIZnII and
GaIIIZnII forms, with a kcat of ∼2000 s-1, while the InIIIZnII
form is totally inactive, are rather surprising. This first
demonstration of an actiVe aluminum-containing enzyme is also
in conflict with the axiom that the relative slow ligand exchange
(34) Wang, Z.; Ming, L.-J.; Que, L., Jr.; Vincent, J. B.; Crowder, M.
W.; Averill, B. A. Biochemistry 1992, 31, 5263-5268.
(35) Sinn, E.; O’Connor, C. J.; de Jersey, J.; Zerner, B. Inorg. Chim.
Acta 1983, 78, L13-L15.
(36) Day, E. P.; David, S. S.; Peterson, J.; Dunham, W. R.; Bonvoison,
J. J.; Sands, R. H.; Que, L., Jr. J. Biol. Chem. 1988, 263, 15561-15567.
(37) Gehring, S.; Fleischhauer, P.; Behlendorf, M.; Hu¨ber, M.; Loro¨sch,
J.; Haase, W.; Dietrich, M.; Witzel, H.; Lo¨cke, R.; Krebs, B. Inorg. Chim.
Acta 1996, 252, 13-17.
(48) March, J. AdVanced organic chemistry: reactions, mechanisms, and
structure; Wiley-Interscience: New York, 1992; pp 534-542.
(49) Wilkins, R. G. Kinetics and mechanism of reactions of transition
metal complexes; VCH: Weinheim, 1991; pp 199-205.
(50) Frau´sto da Silva, J. J. R.; Williams, R. J. P. The biological chemistry
of the elements: the inorganic chemistry of life; Oxford University Press:
Oxford, 1991; pp 101-105.
(38) Yang, Y. S.; McCormick, J. M.; Solomon, E. I. J. Am. Chem. Soc.
1997, 119, 11832-11842.
(39) Doi, K.; McCracken, J.; Peisach, J.; Aisen, P. J. Biol. Chem. 1988,
263, 5757-5763.
(51) Huang, W. J.; Jia, J.; Cummings, J.; Nelson, M.; Schneider, G.;
Lindqvist, Y. Structure 1997, 5, 691-699.
(40) Kauzlarich, S. M.; Teo, B. K.; Zirino, T.; Burman, S.; Davis, J. C.;
Averill, B. A. Inorg. Chem. 1986, 25, 2781-2785.
(41) True, A. E.; Scarrow, R. C.; Randall, C. R.; Holz, R. C.; Que, L.,
Jr. J. Am. Chem. Soc. 1993, 115, 4246-4255.
(52) Payne, M. S.; Wu, S. J.; Fallon, R. D.; Tudor, G.; Stieglitz, B.;
Turner, I. M.; Nelson, M. J. Biochemistry 1997, 36, 5447-5454.
(53) Que, L., Jr.; Ho, R. J. N. Chem. ReV. 1996, 96, 2607-2624.
(54) Frazee, R. W.; Orville, A. M.; Dolbeare, K. B.; Yu, H.; Ohlendorf,
D. H.; Lipscomb, J. D. Biochemistry 1998, 37, 2131-2144.
(55) Martin, R. B. Metal Ions Biol. Syst. 1988, 24, 1-57.
(56) Shannon, R. D. Acta Crystallogr. 1976, A32, 751-767.
(57) Baes, C. F., Jr.; Mesmer, R. E. The hydrolysis of cations; Wiley-
Interscience: New York, 1976.
(58) Crumbliss, A. L.; Garrison, J. M. Comments Inorg. Chem. 1988, 8,
1-26.
(59) Kowall, T.; Caravan, P.; Bourgeois, H.; Helm, L.; Rotzinger, F. P.;
Merbach, A. E. J. Am. Chem. Soc. 1998, 120, 6569-6577.
(60) Kido, H.; Saito, K. J. Am. Chem. Soc. 1988, 110, 3187-3190.
(42) Wang, X.; Randall, C. R.; True, A. E.; Que, L., Jr. Biochemistry
1996, 35, 13946-13954.
(43) Wang, X.; Que, L., Jr. Biochemistry 1998, 37, 7813-7821.
(44) Mueller, E. G.; Crowder, M. W.; Averill, B. A.; Knowles, J. R. J.
Am. Chem. Soc. 1993, 115, 2974-2975.
(45) Aquino, M. A. S.; Lim, J.-S.; Sykes, A. G. J. Chem. Soc., Dalton
Trans. 1992, 2135-2136.
(46) Aquino, M. A. S.; Lim, J.-S.; Sykes, A. G. J. Chem. Soc., Dalton
Trans. 1994, 429-436.
(47) Dietrich, M.; Mu¨nstermann, D.; Suerbaum, H.; Witzel, H. Eur. J.
Biochem. 1991, 199, 105-113.

MIIIZnII Purple Acid Phosphatase
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6685
rates of aluminum render this metal ion useless in the active
sites of enzymes.50,55 Our findings suggest that the rate-limiting
step in the catalytic mechanism of BSPAP may not be a ligand
substitution reaction at the trivalent metal ion. Possible roles
for the trivalent metal ion in phosphate ester hydrolysis are
discussed.
Experimental Section
General. Unless stated otherwise, protein solutions contained 40
mM sodium acetate, 1.6 M KCl, and 20% (v/v) glycerol, pH 5.0. Optical
spectra were measured on an HP8452A diode array spectrophotometer.
Bovine spleen purple acid phosphatase was isolated as previously
described.8,61 Preparations had A280 nm/A536
ratios of ∼14. Apo-
nm
BSPAP, FeZn-BSPAP, and GaZn-BSPAP were prepared as previously
described.8 Apo-BSPAP typically contained 0.05 mol of Fe per mole
of protein and had a residual phosphatase activity of ∼0.1%. Protein
concentrations were determined by measuring the absorption of the
tyrosinate-to-Fe3+ charge-transfer band at 536 nm (ꢀ ) 4080 M-1 cm-1
)
Figure 1. Time dependence of the formation of phosphatase activity
after incubation of apo-BSPAP (4.2 µM) with 200 µM ZnCl2 and 200
µM GaCl3, 202 µM FeCl3, 202 µM Al(NO3)3, 201 µM InCl3, and no
trivalent metal ion added. Incubations were performed at 33 °C in a
solution containing 35 mM sodium acetate, 1.4 M KCl, and 17% (v/v)
glycerol, pH 5.0. Activity assays were performed at 22 °C in 100 mM
Na-MES, 200 mM KCl, pH 6.00, with 10 mM p-NPP.
for FeFe-BSPAP and FeZn-BSPAP and the absorbance at 280 nm (ꢀ
) 60 000 M-1 cm-1) for the apo-enzyme and the other metal-substituted
forms of BSPAP. For all experiments involving apo-BSPAP, plastic
disposable labware was boiled in 5% HCl and rinsed with Millipore
water, while glassware was stored in 10% HCl and rinsed with Millipore
water just before use. Sephadex G-25 columns (Pharmacia) were
washed with 2-3 volumes of 2 mM 1,10-phenanthroline, followed by
several volumes of Millipore water and buffer. Buffers were treated
with Chelex-100 (BioRad) to remove metal impurities.
Enzyme Kinetics. Activity assays, determination of kinetics pa-
rameters KM and kcat, and determination of pH profiles were done as
previously described using p-nitrophenyl phosphate (p-NPP) as substrate
at pH 6.0.8 Fluoride inhibition was studied in 100 mM NaOAc, 200
mM KCl, pH 5.0, by measuring the formation of phenol from the
hydrolysis of phenyl phosphate at 278 nm (ꢀ ) 870 M-1 cm-1) for
nine substrate concentrations and four or five different fluoride
concentrations. The entire data set was fit to the Michaelis-Menten
equation for mixed-competitive inhibition using the program MacCur-
veFit 1.4 (Kevin Raner Software).
Time Dependence of the Formation of MIIIZnII-BSPAP. A 4.2
µM apo-BSPAP solution was incubated with 200 µM ZnCl2 and 200
µM of the trivalent metal (GaCl3, FeCl3, Al(NO3) 3, InCl3, or no trivalent
metal) at 33 °C. At several times after metal addition, an aliquot was
taken and assayed for activity using the continuous assay in a buffer
containing 100 mM Na-MES, 200 mM KCl, and 10 mM p-NPP, pH
6.00, at 22 °C. An effective ꢀ410 nm of 1266 M-1 cm-1 was used for
p-nitrophenol at this pH.
(INAA) at the Interfacultair Reactor Instituut, Delft, The Netherlands.
Samples for INAA were desalted by repeated concentration/dilution
with Millipore water using a Centricon-30 concentrator and lyophilized
by freeze-drying in polyethylene INAA vials.
Results
In a previous study, it was shown that FeIIIZnII-BSPAP and
GaIIIZnII-BSPAP can be obtained simply by the simultaneous
addition of the trivalent metal and Zn2+ to apo-BSPAP, followed
by removal of the excess metal ions.8 These results showed that
the apo-enzyme contains two clearly different metal sites: a
“ferric” site with a high affinity for trivalent metals and a
“ferrous” site with a high affinity for divalent metals. With the
goal of resolving the role of the ferric ion in phosphate ester
hydrolysis, it was decided to prepare a series of metal-substituted
BSPAP forms of the general formula MIIIZnII-BSPAP. Zn2+ was
chosen as the invariant divalent metal instead of the native Fe2+
,
since metal-substituted forms having Fe2+ at the divalent metal
site may, in principle, disproportionate, resulting in the re-
formation of some of the native FeIIIFeII-BSPAP form. Such a
disproportionation has been observed for GaFe-BSPAP.8
Apo-BSPAP was incubated with a ∼40-fold excess of ZnCl2
and a ∼40-fold excess of either FeCl3, GaCl3, Al(NO3)3, InCl3,
or no trivalent metal at 33 °C, and the formation of phosphatase
activity was monitored as a function of time after metal addition
(Figure 1). Under these conditions, the addition of gallium
results in an almost immediate formation of GaZn-BSPAP
activity (within 3 min). The addition of Fe3+ also results in the
formation of phosphatase activity, but the formation of FeZn-
BSPAP is much slower. Surprisingly, the addition of Al3+ also
resulted in phosphatase activity. The formation of this putative
AlZn-BSPAP form is even slower than that of FeZn-BSPAP.
The addition of either In3+ and Zn2+, or Zn2+ in the absence of
any trivalent metal, gave rise to a low level of phosphatase
activity (∼4% of GaZn-BSPAP activity), which did not increase
further with time. Furthermore, this phosphatase activity disap-
peared almost completely after incubation for 10 min in the
presence of the strong reductant sodium dithionite, which
strongly suggests that it is due to FeZn-BSPAP. This is
consistent with the amount of residual iron that was determined
Titration of Indium Binding to Apo-BSPAP. Apo-BSPAP (12.8
µM) was incubated with 98 µM ZnCl2 and various amounts of InCl3
for 5.5 h at 22 °C. Then, 95 µM GaCl3 was added. After incubation of
the solution for at least 20 min at 22 °C, the activity of the solution
was measured using the continuous assay in a buffer containing 100
mM Na-MES, 200 mM KCl, and 10 mM p-NPP, pH 6.00, at 22 °C.
Preparation of AlZn-BSPAP. Al(NO3)3 (∼200 µM) and ZnCl
2
(∼200 µM) were added to a solution of apo-BSPAP (∼30 µM) and
incubated at 37 °C for 8 h. Precipitated protein was removed by
centrifugation (15000g). Excess metal ions were removed by repeated
concentration/dilution with metal-free buffer using a Centricon-30
concentrator (Amicon).
Preparation of InZn-BSPAP. InCl3 (∼200 µM) and ZnCl2 (∼200
µM) were added to a solution of apo-BSPAP (∼40 µM) and incubated
at 22 °C for 3 h. Precipitated protein was removed by centrifugation
(15000g). Excess metal ions were removed by repeated concentration/
dilution with metal-free buffer using a Centricon-30 concentrator.
Metal Analyses. Metal analyses for Fe, Al, and Zn were performed
on a Hitachi180-80 polarized Zeeman atomic absorption spectrometer
equipped with a graphite furnace. Adventitious metal ions were removed
from buffer by passage through a Chelex-100 column. Analysis of
indium was performed by instrumental neutron activation analysis
(61) Vincent, J. B.; Crowder, M. W.; Averill, B. A. Biochemistry 1991,
30, 3025-3034.

6686 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Merkx and AVerill
Table 1. Kinetics Parameters for Various Metal-Substituted Forms
of Bovine Spleen Purple Acid Phosphatasea
enzyme
kcat × 10-3 (s-1
)
KM (mM)
Ki (F-) (µM)e
AlIIIZnII
1.90 (0.04)b
4.2 (0.1)b
5.5 (0.4)b,d
nd
3 (0.2)
140 (70)
nd
200 (30)
nd
GaIIIZnII
3.09 (0.44)b,d
InIIIZnII
<0.02b
FeIIIZnII
2.84 (0.18)b,d
1.76 (0.09)c,d
3.3 (0.4)b,d
1.2 (0.2)c,d
FeIIIFeII (native)
a Numbers in parentheses are standard deviation values. b Assays
were performed at 22 °C and pH 6.00 in a buffer containing 100 mM
Na-MES and 200 mM KCl with p-NPP as substrate. c Assays were
performed at 22 °C and pH 6.0 in a buffer containing 100 mM Na-
MES, 200 mM KCl, 15 mM sodium ascorbate, and 0.2 or 2 mM
Fe(NH4)2(SO4)2 with p-NPP as substrate. d Data from ref 8. e Assays
were performed at 22 °C and pH 5.0 in a buffer containing 100 mM
NaOAc and 200 mM KCl with phenyl phosphate as substrate.
for this apo-BSPAP preparation, ∼0.05 Fe per protein. Thus,
the addition of either Zn2+ alone or In3+ in combination with
Zn2+ does not result in any phosphatase activity other than that
attributable to a small impurity of FeZn-BSPAP. It should be
noted that apo-BSPAP is not fully stable under the conditions
used in this experiment (33 °C). On a time scale of hours,
substantial denaturation of apo-BSPAP occurs, so that dena-
turation of apo-protein competes with formation of FeZn-BSPAP
and AlZn-BSPAP. The final activity levels observed in this
experiment (Figure 1), therefore, do not reflect the relative
specific activities of the purified GaZn-BSPAP, FeZn-BSPAP,
and AlZn-BSPAP derivatives.
Figure 2. Titration of indium binding to apo-BSPAP. Apo-BSPAP
(12.8 µM) was incubated with 98 µM ZnCl2 and various amounts of
InCl3 at 22 °C for 5.5 h in a solution containing 35 mM sodium acetate,
1.4 M KCl, and 17% (v/v) glycerol, pH 5.0. Next, 95 µM GaCl3 was
added. Phosphatase activity was determined ∼20 min after gallium
addition. Activity assays were performed at 22 °C in 100 mM Na-
MES, 200 mM KCl, pH 6.00, with 10 mM p-NPP.
p-NPP at 22 °C and pH 6.00. Overall, the kinetics parameters
of AlZn-BSPAP are remarkably similar to those of GaZn-
BSPAP and FeZn-BSPAP. The KM value is close to that of
FeZn-BSPAP, while kcat is decreased 1.5-fold compared to those
of FeZn-BSPAP and GaZn-BSPAP. It should be noted, however,
that the aluminum-substituted enzyme is still a very efficient
phosphatase, having a turnover rate of nearly 2000 s-1. Fluoride,
which is an uncompetitive inhibitor at pH 5.0,62 binds much
more strongly to AlZn-BSPAP than to GaZn-BSPAP or FeZn-
BSPAP. The pH dependence of the phosphatase activity of
AlZn-BSPAP was determined for the hydrolysis of p-NPP at a
constant p-NPP concentration of 50 mM (not shown). The pH
profile is bell-shaped with apparent pKa values of 5.2 and 7.1,
and is thus very similar to that of FeZn-BSPAP under identical
conditions (pKas of 5.3 and 7.1).8
InZn-BSPAP. The absence of any activity after incubation
of apo-BSPAP with indium and zinc means either that apo-
BSPAP is not able to bind indium or that InIIIZnII BSPAP is
formed but is inactive. If In3+ is able to bind at the trivalent
metal site, it should prevent the binding of gallium and,
therefore, the development of phosphatase activity resulting from
GaZn-BSPAP. Figure 2 shows a titration experiment in which
12.8 µM apo-BSPAP was incubated with excess Zn2+ and
various amounts of In3+ for 5.5 h at 22 °C. Subsequently, 95
µM Ga3+ was added, and the phosphatase activity was deter-
mined after another 10-20 min incubation time. Figure 2 shows
that the binding of one In3+ ion per apo-BSPAP precluded the
formation of active GaZn-BSPAP. Independent evidence for the
formation of BSPAP with a dinuclear InIIIZnII center came from
metal analysis of InZn-BSPAP that had been prepared by the
addition of excess In3+ and Zn2+. Metal analysis showed that,
even after repeated concentration/dilution in metal-free buffer,
the BSPAP samples still contained equimolar amounts of indium
(1.2 mol/mol of protein) and zinc (1.1 mol/mol of protein) and
only very little iron (0.05 mol/mol of protein).
AlZn-BSPAP. The slow formation of AlZn-BSPAP requires
either very long incubation times or the use of higher temper-
atures. AlZn-BSPAP was prepared by the incubation of apo-
BSPAP with 200 µM Al(NO3)3 and 200 µM ZnCl2 at 37 °C.
When no further increase of phosphatase activity was measured
(after ∼8 h), denatured protein was removed by centrifugation,
and the supernatant was concentrated/diluted several times with
metal-free buffer to remove excess aluminum and zinc. Metal
analysis showed the presence of 1.0 mol of aluminum, 0.7 mol
of zinc, and 0.1 mol of iron per mole of protein. Like GaFe-
BSPAP and GaZn-BSPAP8, AlZn-BSPAP lacks the character-
istic purple color of the FeFe- and FeZn-BSPAP forms and is
colorless (Figure S1, Supporting Information), indicating the
absence of the Fe(III) at the trivalent metal site. The very weak
residual band at ∼550 nm is probably due to a small amount
(∼10%) of FeZn-BSPAP present in this preparation. This iron
probably also gives rise to the small EPR signal at g ) 4.3
which is present in the EPR spectrum of AlZn-BSPAP. More
importantly, the typical EPR spectrum of the native, antiferro-
magnetically coupled FeIIIFeII cluster with g-values of 1.87, 1.74,
and 1.58 is completely absent in the spectrum of AlZn-BSPAP
(Figure S2, Supporting Information). The specific activity of
AlZn-BSPAP is comparable to that of the native FeFe-BSPAP
and ∼60% that of FeZn-BSPAP. The addition of dithionite to
FeFe- and FeZn-BSPAP has been shown to result in the
reduction of the ferric iron and inactivation of these BSPAP
forms,8 but AlZn-BSPAP retained ∼90% of its phosphatase
activity in the presence of dithionite, which is consistent with
the replacement of Fe3+ by Al3+. The 10% decrease of
phosphatase activity in the presence of dithionite is consistent
with the residual 0.1 Fe detected by AAS and the residual
features observed in the optical and EPR spectra. Like those of
FeZn- and GaZn-BSPAP,8 the phosphatase activity of AlZn-
BSPAP was insensitive to oxidation by hydrogen peroxide,
Discussion
Preparation and Characterization of AlZn-BSPAP and
InZn-BSPAP. In this study we demonstrate that the “ferric”
indicating that the native Fe2+ has been replaced by Zn2+
.
Table 1 compares the kinetics parameters of the various MIII-
ZnII forms and the native FeIIIFeII enzyme for the hydrolysis of
(62) Pinkse, M. W. H.; Merkx, M.; Averill, B. A. Biochemistry, accepted
for publication.

MIIIZnII Purple Acid Phosphatase
Table 2. Comparison of the Catalytic Activity of MIIIZnII-BSPAP and Some Properties of the Trivalent Metal
kexchange (s-1
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6687
)
cat (s-1
)
KM (mM)
ionic radius (Å)a
pKa [M(H2O)6]3+ b
[M(H2O)6]3+
[M(H2O)5OH]2+
M(acac)3
c
M(III)
k
Al
Fe
Ga
In
1.9 × 103
2.8 × 103
3.1 × 103
4.2
3.3
5.5
nd
0.54
0.65
0.62
0.80
5.5
2.7
3.0
4.4
1.3c
9.1 × 10-5
3.3 × 10-3
1.6 × 10-3
2
1.6 × 102 c
4.0 × 102 c
g107 d
1.2 × 105 e
(0.6-2.0) × 105 f
<20
a From ref 56. b All values converted from ref 57 to 0.16 M ionic strength. c From ref 60 and references therein. d From ref 59. e From ref 49.
f From ref 72.
metal site of BSPAP can accommodate a variety of trivalent
metal ions. Gallium, iron, and aluminum all yield active
enzymes. Indium does bind to apo-BSPAP, but InIIIZnII BSPAP
is inactive. The only proteins for which the interaction with
trivalent metal ions have been studied extensively are the Fe3+
transport/binding proteins transferrin, lactoferrin, and ovotrans-
ferrin.63 In these proteins, the trivalent metal is coordinated by
two tyrosinates, one histidine, one carboxylate, and a nonprotein
ligand, carbonate,64,65 while the Fe3+ in the purple acid
phosphatases is coordinated by one tyrosinate, one histidine,
one terminal carboxylate, one bridging carboxylate, probably a
bridging hydroxide, and either water or hydroxide.12,13 Alumi-
num, gallium, and indium all bind quite strongly to transferrin,
and a free energy relationship has been observed between the
stability constants for hydroxide and phenolate binding and their
binding constants to transferrin, suggesting that electrostatic
interactions rather than, e.g., ionic radii determine the binding
strength.66 The rate of formation of the MIIIZnII BSPAP strongly
depends on the trivalent metal, with rates decreasing in the order
Ga3+ . Fe3+ > Al3+. The rapid formation of GaZn-BSPAP
was used to advantage in studying the formation of InZn-
BSPAP, because it allowed the amount of free trivalent metal
sites to be determined by their reaction with gallium to form
the active GaZn-BSPAP species. For transferrin, the relative
order of the rates of complex formation differs from that found
for apo-BSPAP (Ga > Al > Fe > In).67 For both BSPAP and
transferrin, these rates of complex formation do not follow the
trends found for ligand exchange rates for trivalent metal ions,
indicating that some process other than a simple ligand exchange
is determining these rates.
Our previous work on the gallium-substituted BSPAP forms
showed that the kinetics parameters and the pH profiles of FeFe-,
GaFe-, FeZn-, and GaZn-BSPAP are similar but not identical.8
The small differences that were detected were consistent. Thus,
the same effects on kcat and KM and the pH profiles were
observed for the Fe3+-to-Ga3+ substitution when going from
FeFe-BSPAP to GaFe-BSPAP and when going from FeZn-
BSPAP to GaZn-BSPAP. The same is true for the Fe2+-to-Zn2+
substitution. The differences between the various MIIIZnII
BSPAP forms may, therefore, be ascribed with some confidence
to chemical differences that are due to the identity of the trivalent
metal, and not to some artifact introduced by their preparation.
The value of kcat is calculated on the basis of the amount of
protein. Since our AlZn-BSPAP preparations contain 0.7 Zn
per protein, it may be argued that one should calculate kcat on
the basis of the amount of zinc. Doing this would increase the
value of kcat to 2700 s-1, which would mean that the activity of
AlZn-BSPAP is the same as those of FeZn-BSPAP and GaZn-
BSPAP.
Relating Enzymatic Properties to General Properties of
the Trivalent Metal. The trivalent metals used in this study
are all spherically symmetrical, having either fully filled (d10:
Al3+, Ga3+, and In3+) or half-filled, high-spin d-shells (d5:
Fe3+). The absence of specific ligand field effects facilitates
the comparison of the enzymatic properties of these metal-
substituted enzymes to the general properties of the trivalent
metal. In Table 2, some of these properties (ionic radius in
octahedral complexes, ligand exchange rates for [M(H2O)6]3+
,
[M(H2O)5OH]2+
, and [M(acac)3], and pKa values for
[M(H2O)6]3+) are compared with the enzymatic properties kcat
and KM for the MIIIZnII BSPAP form containing the same
trivalent metal.
Stability constants for hydroxide and other oxygen and
nitrogen ligands are similar for iron(III) and gallium(III) but
lower for the corresponding aluminum complexes.58 The greater
acidity of [Fe(H2O)6]3+ relative to that of [Al(H2O)6]3+ has been
correlated with the greater electronegativity of Fe(III) (øFe(III)
) 1.96; øAl(III) ) 1.61), which means that the Fe-O bond has
some covalent character, while the Al-O bond is purely
electrostatic. Whether p-NPP binds to the trivalent metal ion
remains to be established, but the very similar substrate affinities
found for AlZn-, FeZn-, and GaZn-BSPAP suggest that such
an MIII-O bond does not contribute significantly to substrate
binding. The finding that AlZn-BSPAP is inhibited much more
strongly by fluoride than GaZn- and FeZn-BSPAP indicates that
the noncompetitive inhibition is due to fluoride binding at the
trivalent metal ion, where it is likely to replace a coordinated
water/hydroxide. The stronger binding of fluoride to aluminum
is in agreement with the hard-soft acid-base theory: fluoride
is a hard base and Al3+ is a harder acid than both Fe3+ and
Ga3+
.
The acidic limb of the pH optimum of BSPAP has been
attributed to a M(III)-coordinated water ligand with a pKa around
5 in the enzyme-substrate complex47. The similar pH profiles
for FeZn-BSPAP and AlZn-BSPAP seem to be in conflict with
this proposal, since the pKa of the M(III)-coordinated water
would be expected to be higher for AlZn-BSPAP than for FeZn-
BSPAP. The pH profile may, therefore, be the result of the (de)-
protonation of amino acid side-chain groups or substrate. As
will be discussed below for ligand exchange rates, some care
should be taken when comparing the thermodynamic properties
of, e.g., the hexaaqua complexes with those of the metal-protein
complexes. It may be that, in complexes with several anionic
ligands, these trends in pKa values differ from the trend in the
pKa values of the first deprotonation step in the hexaaqua
complexes, as shown in Table 2.
(63) Baker, E. N.; Lindley, P. F. J. Inorg. Biochem. 1992, 47, 147-
160.
(64) Anderson, B. F.; Baker, H. M.; Dodson, E. J.; Norris, G. E.; Rumball,
S. V.; Waters, J. M.; Baker, E. N. Proc. Natl. Acad. Sci. U.S.A. 1987, 84,
1769-1773.
(65) Bailey, S.; Evans, R. W.; Garratt, R. C.; Gorinsky, B.; Hasnain, S.;
Horsburgh, C.; Jhoti, H.; Lindley, P. F.; Mydin, A.; Sarra, R.; Watson, J.
L. Biochemistry 1988, 27, 5804-5812.
(66) Li, H.; Sadler, P. J.; Sun, H. Eur. J. Biochem. 1996, 242, 387-
393.
In most discussions of the relationship between ligand
exchange rates and the nature of the metal ion, the water
(67) Harris, W. R.; Chen, Y.; Wein, K. Inorg. Chem. 1994, 33, 4991-
4998.

6688 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Merkx and AVerill
exchange rates of the hexaaqua complexes of the metal ions
are compared.50,55,68 The water exchange rates for the trivalent
metal ions studied in this work decrease in the order In3+
.
Ga3+, Fe3+ > Al3+, which also reflects the increase in charge/
ionic radius ratio for these metal ions.58 The difference between
the kcat for AlZn-BSPAP (1900 s-1) and the water exchange
rate for Al(H2O)6 (∼1 s-1) is remarkably large. At first glance,
these facts seem incompatible with a catalytic mechanism that
includes a ligand exchange reaction at the trivalent metal. It is
dangerous, however, to compare ligand exchange rates for two
complexes with the same metal ion but with a different ligand
environments. In purple acid phosphatases, the ligand will not
experience the full 3+ charge of the trivalent metal as in the
hexaaqua complexes, because the trivalent metal is coordinated
by several negatively charged ligands: one tyrosinate, one end-
on bound carboxylate, a bridging carboxylate, and probably also
a bridging hydroxide. The effect of such a negatively charged
ligand can be clearly seen by comparing the water exchange
rates for [M(H2O)6]3+ and the corresponding [M(H2O)5OH]2+
complexes (Table 2). It has also been reported that FeIII-EDTA
and FeIII-porphyrins have much faster water exchange rates of
Figure 3. Three mechanistic proposals for the role of the PAP metal
center in the hydrolysis of phosphate monoesters. Starting at the top
center of the figure with the free enzyme, all mechanisms have the
initial coordination of the phosphate ester to the divalent metal ion in
common. In mechanism 1, the phosphate ester is subsequently attacked
by the hydroxide that is coordinated monodentately to the ferric iron.
In mechanism 2, the phosphate ester binds in a bridging mode to both
metal ions and is then attacked by the bridging hydroxide. In mechanism
3, an Fe3+-coordinated hydroxide acts as a general base to deprotonate
a water in the second coordination sphere of the metal, which then
attacks the phosphate ester coordinated to the divalent metal ion.
∼105 s-1 69
. On the other hand, in the all mechanisms proposed
thus far for the PAPs (vide infra), a negatively charged
phosphate rather than a neutral water ligand needs to be
displaced by a water molecule, which might be expected to slow
the exchange reaction.13,42
Since so many factors may contribute to the rate of ligand
exchange, the absolute rates of ligand exchange reactions in
model complexes cannot be directly correlated with kcat values.
If one compares ligand exchange rates for various trivalent
metals with the same ligand environment, however, the relative
rates between corresponding complexes of the various metals
seem fairly constant.60 If a ligand exchange reaction at the
trivalent metal ion is the rate-determining step in PAP catalysis,
kcat would be predicted to be similar for FeZn-BSPAP and GaZn-
BSPAP but at least a factor of 10 lower for AlZn-BSPAP. The
relatively high activity of AlZn-BSPAP is, therefore, inconsistent
with a rate-limiting ligand exchange reaction at the trivalent
metal ion, although it cannot be excluded that ligand exchange
is partially rate-limiting for the aluminum-containing enzyme.
The InZn-BSPAP is the first example of a purple acid
phosphatase for which substitution of the ferric iron by another
trivalent metal results in inactivation of the enzyme. In3+ is a
weaker Lewis acid then Ga3+, Fe3+, and Al3+, which may be
the reason for its inactivity. Since the ionic radius of In3+ is
also larger, we cannot exclude the possibility that the binding
of In3+ causes structural changes in the active site that affect
the catalytic activity.
Figure 3 shows three different mechanistic proposals for the
role of the two metal ions in hydrolysis. All mechanisms have
as a common feature the initial coordination of the phosphate
ester to the divalent metal ion. In mechanism 1, proposed by
Witzel and co-workers,13,47 the phosphate ester is subsequently
attacked by the hydroxide, which is bound to the ferric iron in
a monodentate fashion. The final step is then release of
phosphate. Mechanism 2 has been proposed for protein phos-
phatase 1 and calcineurin, on the basis of the X-ray structures
of their oxoanion complexes,18,19 and recently also for uterof-
errin.42 In this mechanism, the substrate binds in a bridging mode
to both metal ions. The µ-OH bridge is then activated to act as
the nucleophile that attacks the phosphate ester. Both mecha-
nisms 1 and 2 involve ligand substitution reactions at the
trivalent metal ion in which phosphate is replaced by a water/
hydroxide ligand. The similar kcat values for AlZn-, FeZn-, and
GaZn-BSPAP suggest that this release of phosphate is not the
rate-determining step in catalysis, which leaves the hydrolysis
reaction itself as the most likely rate-limiting step for both
mechanisms 1 and 2.
A third mechanistic possibility is that ligand substitution at
the trivalent metal ion does not occur during catalysis (mech-
anism 3). In this mechanism, the hydroxide that is coordinated
to the trivalent metal (either end-on or bridging the two metals)
acts as a general base to deprotonate a second water molecule,
which then attacks the phosphate ester. In the final step,
phosphate is released from the divalent metal ion. A similar
role for the trivalent metal ion has recently been proposed for
the Fe(III) or Co(III) ions in the enzyme nitrile hydratase.51 Two
classes of nitrile hydratases exist, one containing a low-spin
Fe(III) metal center and one containing a low-spin Co(III) metal
center. Both classes share sequence homology, and the coor-
dination environments of the trivalent metals are believed to
be similar. Typical values for kcat are 1000-2000 s-1 for both
enzyme classes. Since Co(III) complexes are known to be
kinetically inert, it was proposed that the chemistry is taking
place in the second coordination sphere of the trivalent metal,
in a way similar to that depicted in mechanism 3.
Trivalent Metals as Lewis Acids in Enzyme Catalysis. In
general, trivalent metal ions are better Lewis acids then divalent
metal ions, and consequently one might expect the former to
be widely distributed in enzymes. In fact, however, only a few
well-characterized examples exist: the purple acid phosphatases
and possibly other phosphatases with the same sequence motif,
nitrile hydratase, and intradiol dioxygenases. In all these
enzymes, the trivalent metal ions in these enzymes are bound
via several anionic ligands, which will decrease their Lewis
acidity compared to that of their hexaaqua complexes but at
the same time increase their ligand exchange rates. Two options,
(68) Lippard, S. J.; Berg, J. M. Principles of bioinorganic chemistry;
University Science Books: Mill Valley, CA, 1994; pp 28-31.
(69) Kolski, G. B.; Plane, R. A. J. Am. Chem. Soc. 1972, 94, 3740-
3744.

MIIIZnII Purple Acid Phosphatase
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6689
therefore, seem open to metallohydrolases: using a divalent
metal ion such as zinc with high ligand exchange rates in a
coordination sphere that makes the metal ion as acidic as
possible (e.g., all neutral histidine ligands, as in carbonic
anhydrase), or using a trivalent metal ion such as Fe(III), which
is intrinsically a much better Lewis acid, in an anionic ligand
environment, which will increase the rate of ligand substitution
at the trivalent metal ion.
Aluminum has no known biological role, despite the fact that
it is the most abundant metal on earth and a good Lewis acid.
Its slow ligand exchange rates are generally considered to be
an important reason for its absence in biology; for example, in
a review on the bioinorganic chemistry of aluminum, it was
concluded that, “The slow ligand exchange rate renders Al3+
useless as a metal ion at the active sites of enzymes.” 55 In
another recent review on aluminum in biological systems,
Williams also suggested that, “The Al3+ ion is not active in
kinases in all probability because of its movement with a
restricted rate constant of seconds. Here lies the problem of high
valent metal ions.” 70 The present results, however, demonstrate
that an aluminum enzyme can be prepared that is an efficient
catalyst of phosphate ester hydrolysis. The absence of aluminum
in biology should, therefore, be attributed to other factors, such
as its problematic transport and storage or its interference with
Mg2+ at Mg2+ binding sites for phosphate esters (DNA/RNA,
ATP, etc.).71
Conclusion. The phosphatase activity of AlZn-BSPAP is
almost as great as that of the FeZn- and GaZn-BSPAP forms
(kcat ∼2000 vs ∼3000 s-1), while InIIIZnII-BSPAP is inactive.
This first demonstration of an active aluminum-containing
enzyme is in conflict with the axiom that the relative slow ligand
exchange rates of aluminum render this metal ion useless in
the active sites of enzymes.55,70 Our findings suggest that the
rate-limiting step in the catalytic mechanism of BSPAP may
not be a ligand substitution reaction at the trivalent metal ion.
Acknowledgment. We thank Femke Mensonides and Bart
van Ulden for doing some of the initial kinetics experiments,
Martijn Pinkse for measuring the fluoride inhibition constants,
Thea van Meerten of the Interfacultair Reactor Instituut Delft
for the indium analysis by INAA, and Ingeborg Kooter and Rico
Funhoff for critically reading the manuscript.
Supporting Information Available: Optical spectra of
FeFe-BSPAP and AlZn-BSPAP (Figure S1) and EPR spectra
of FeFe-BSPAP, FeZn-BSPAP, and AlZn-BSPAP (Figure S2)
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA9837147
(71) Macdonald, T. L.; Martin, R. B. Trends Biochem. Sci. 1988, 13,
15-19.
(72) Hugi-Cleary, D.; Helm, L.; Merbach, A. E. J. Am. Chem. Soc. 1987,
109, 4444-4450.
(70) Williams, R. J. P. Coord. Chem. ReV. 1996, 149, 1-9.