Concerning two-metal cooperativity in model phosphate hydrolysis

Article abstract of DOI:10.1006/bioo.2001.1221

Three series of bimetallic ligands were tested for cooperativity in the hydrolysis of phosphate esters. It was shown that rate enhancements were in part contributed by binding to the hydrophobic linkers when the substrates were also hydrophobic, and two metal cooperativity was not found to be present. Kinetic order tests were performed and shown to be superior to previous methods for analyzing cooperativity.

Full text of DOI:10.1006/bioo.2001.1221

Bioorganic Chemistry 29, 345–356 (2001)
doi:10.1006/bioo.2001.1221, available online at http://www.idealibrary.com on
Concerning Two-Metal Cooperativity in Model
Phosphate Hydrolysis
Martin Leivers and Ronald Breslow1
Department of Chemistry, Columbia University, New York, New York 10027
Received April 19, 2001
Three series of bimetallic ligands were tested for cooperativity in the hydrolysis of phosphate
esters. It was shown that rate enhancements were in part contributed by binding to the hydropho-
bic linkers when the substrates were also hydrophobic, and two metal cooperativity was not
found to be present. Kinetic order tests were performed and shown to be superior to previous
methods for analyzing cooperativity.
᭧ 2001 Elsevier Science (USA)
INTRODUCTION
In nature, some hydrolytic enzymes, including phosphatases, utilize two metals
and substrate binding pockets in a bifunctional catalytic mechanism (1). A variety of
metals are used in the natural enzymes (Mg+2, Co+2, Zn+2, Fe+2,+3) and the exact
function of the two metal centers and their respective characteristics have been the
subject of speculation and model study. To date, work in our group (2) and in others
(3,4) has focused on ligand systems with increased complexity and preorganization,
with the goal of greater rate enhancements and bifunctionality.
In our prior work, we demonstrated that increasing catalyst rigidity and incorporat-
ing a bimetallic ligand system (L11 and L12, Fig. 7) led to improved catalysis. With
hopes of further increasing catalytic power, we have now studied a new set of
macrocyclic ligands. We see effects that at first sight could be interpreted to indicate
bifunctional bis-metal ion catalysis. However, by varying metal ion concentrations
and ligand concentrations we have shown that the observed effects reflect hydrophobic
binding rather than bifunctional catalysis. Thus we have also reexamined our previous
system (2) to see whether its results also can be explained by hydrophobic, rather
than bifunctional, effects.
EXPERIMENTAL
General. Proton and carbon NMR spectra were obtained from Bruker 300- or
400-MHz spectrometers. Mass spectra were obtained from electron impact (ESI).
Chromatography was performed using Merck silica gel 60.
1 To whom correspondence should be addressed.
345
0045-2068/01 $35.00
᭧ 2001 Elsevier Science (USA)
All rights reserved.

346
LEIVERS AND BRESLOW
Materials. Tetrahydrofuran was distilled from Na/benzophenone, pyridine was dis-
tilled from CaH2. All other materials were used as received unless indicated.
Phosphate hydrolysis. IR monitoring of the release of p-nitrophenoxide anion and
HPLC conditions for UpU hydrolysis were reported earlier.2 Hydrolysis studies with-
out error bars were conducted in duplicate, those with error bars were conducted 3–8
times per data point and the standard deviation is shown.
N-Tosyl cyclen (L2). To a solution of cyclen (43 mg, 0.25 mmol) in pyridine (15
ml) was added dropwise a solution of tosyl chloride (47.6 mg, 1 eq) and the mix
stirred for 1.5 h. The solution was made basic (pH 13, 10% KOH aq.) and extracted
with CHCl3 (3 ϫ 20 ml) and then dried (Na2SO4). Chromatography (SiO2, 5–10%
1
MeOH (w/30% NH4OH)/CHCl3) yielded the product as a white solid. H NMR
(CDCl3) ␦: 7.72 (2H, d, ArH, J ϭ 8Hz), 7.33 (2H, d, ArH, J ϭ 8Hz), 3.24 (4H, m
CH2NSO2-), 2.80 (8H, m, CH2N), 2.69 (4H, m, CH2N), 2.42 (3H, s, CH3). ms (ESI):
343 (M ϩ H, 100).
Bis cyclen 4,4Ј biphenyl di sulfonamide (L1). Performed in the same manner as
1
described for N-tosyl cyclen yielding an off-white solid. H NMR (CDCl3) ␦: 7.95
(4H, d, ArH, J ϭ 9Hz), 7.72 (4H, d, ArH, J ϭ 9Hz), 3.30 (8H, m CH2NSO2-), 2.80
(16H, m, CH2N), 2.69 (8H,m, CH2N).
N-mesyl cyclen (L3). Performed in the same manner as described for N-tosyl cyclen
yielding a white solid. 1H NMR (CDCl3) ␦: .3.41 (4H, m CH2NSO2CH3), 2.91 (3H, s,
CH2NSO2CH3), 2.85 (8H,m, CH2N), 2.70 (4H,m, CH2N).
4,4Љ p-terphenyl dialdehyde. To a solution of 4-bromobenzaldehyde dimethylacetal
(25 ml, 34.5 g, 150 mmol) in THF (150 ml) at Ϫ78ЊC, was added n-BuLi (1.2 eq,
112 ml) over 30 min. A white precipitate appeared over the next 60 min. Then a
solution of benzoquinone (0.3 eq, 43 mmol, 4.6 g) in THF (40 ml) was added over
30 min. The mixture was then allowed to warm to 10ЊC and quenched with NH4Cl
(sat. aq. ca. 40 ml). The organics were separated and the water layer extracted with
CH2Cl2 (3 ϫ 50 ml). The crude was quickly purified through silica chromatography
(50–60% EtOAc/CHCl3). The intermediate compound decomposes at room tempera-
ture and was used immediately.
To the dienol (17.3 g) in THF (300 mL) at Ϫ20ЊC., HI (47% aq., 50 ml) was added
dropwise over 15 min. The mixture was slowly allowed to warm to room temperature
over 2 h. and then quenched with NaOH (3N, ca 100 ml) and the solids removed.
These were washed with H2O and dried under vacuum to yield the pure dialdehyde
1
as a white solid (4.1g, 33%). H NMR (CDCl3) ␦: 10.1 (2H, s CHO), 8.0 (4H, d,
ArH, J ϭ 8 Hz), 7.83 (4H, d, ArH, J ϭ 8 Hz), 7.80 (4H, s, ArH).
General procedure for reductive amination (15). To a solution of dien (2 ϫ 10Ϫ2
M) in MeCN is added a solution of the aldehyde (0.3 M) in MeCN and the mixture
is stirred for 1–5 h. During this time a precipitate may be noted (especially in case
of the macrocycles). Then a suspension of NaBH4 (4 eq) in MeOH is added and the
mixture stirred at room temperature overnight. The resulting suspension is quenched
with NH4Cl (sat. aq.) and then extracted with CH2Cl2 (3x). The compound is then
purified on silica (CHCl3:MeOH:NH4OH (25:5:1).
1
Bisdibenzofuran methyl dien macrocycle- L5. H NMR (CDCl3) ␦ 8.07 (4H, s,
ArH), 7.51 (4H, d, ArH, J ϭ 8.5 Hz), 7.30 (4H, d, ArH, J ϭ 8.5 Hz)), 4.05 (8H, s,
ArCH2), 2.90 (16H,m, CH2N).

TWO-METAL COOPERATIVITY IN PHOSPHATE HYDROLYSIS
1
347
N,N dibenzyl dien- L9. H NMR (CDCl3) ␦ 7.3 (10H, m, ArH), 3.78 (4H, s,
ArCH2), 2.72 (8H,m, CH2N). ms (ESI): 284.3 (M ϩ H, 100).
N,N bisbiphenyl methyl dien L10. Prepared in the above manner. The biphenyl 4-
1
aldehyde was added as a solution in CHCl3. H NMR (CDCl3) ␦: 7.31-7.25 (18H,
m, ArH), 3.78 (4H, s, ArCH2), 2.72 (8H,m, CH2N) ms (ESI): 436.2 (M ϩ H, 100).
p-Terphenyl methyl dien macrocycle L7. Prepared in the above manner. The dialde-
hyde was dissolved in CHCl3 (1.2 ϫ 10Ϫ2 M) and allowed to condense with dien
1
over 15 h. H NMR (CDCl3) ␦: 7.53 (8H, s, ArH), 7.51 (8H, d, ArH, J ϭ 8 Hz),
7.37 (8H, d, ArH, J ϭ 8 Hz), 3.82 (8H, s, ArCH2), 2.90 (16H,m, CH2N). ms (ESI):
715.4 (M ϩ H, 100).
2,5,7 triazanonane L8. To a solution of diethyl iminodiacetate (6 g, 31.7 mmol)
in MeOH:MeCN (1:1, 50 ml) was added KCN (ca. 50 mg) and the solution was
saturated with MeNH2. The solution was stirred at room temperature for 20 h and
the solvents were removed.
The crude diamide was then treated with BH3 (1 M in THF, 4.2 eq, 130 ml) and
refluxed for 20 h. The mixture was then cooled and MeOH (10 ml) was added
dropwise until no more H2 was evolved. The mixture was acidified with HCl (3M,
FIG. 1. Phosphate ester substrates used in the kinetic studies and synthesis of the cyclen-based series
of ligands.

348
LEIVERS AND BRESLOW
5 ml) and the solvents removed. Aqueous KOH (10%, to pH 10) was added and the
mixture was extracted with CH2Cl2 (4ϫ 20 ml). The aqueous layer was lyophilized
and extracted with CHCl3. The extracts were combined and distilled (water aspirator,
bp 100–110ЊC) to yield the pure product as a colorless oil (0.4 g). The triamine must
1
be kept refrigerated and away from CO2 and moisture. H NMR (CDCl3) ␦: 3.73
(6H, s, NCH3), 3.48 (4H,s, CH2N).
RESULTS AND DISCUSSION
Cyclen ligands. The ligands (L1–3) were formed through condensation of cyclen
(1, 4, 7, 10-tetraazacyclododecane) with the corresponding sulphonyl chloride (Fig.
1). In L1 two metal ligands are separated by a 4,4Ј-biphenyl-bis-sulfonyl spacer,
while L2 is an analogous monotopic ligand carrying a toluenesulfonyl group, and L3
is also a simple ligand carrying a methansulphonyl group. Both copper and zinc
complexes were examined with L1 and L2; L3 was tested only as the zinc complex.
As a final control, cyclen itself was examined as the zinc and copper complex. As
in previous work, we examined catalysis of the cyclization of the RNA mimic HPNPP
(2-hydroxypropyl p-nitrophenyl phosphate) by metal complexes of our ligands. The
results are shown in Table 1.
From these results, it can be seen that bimetallic cooperativity is not the dominant
factor in the catalytic ability. Comparing the dimeric L1 with its monomeric analog
L2 does not decrease the rate enhancements for copper, but the opposite is true for
the zinc case. These differences are small, however, and thus it seems likely that
another effect is dominating the catalysis. Comparing these results with those for the
control ligand (L3) leads one to suspect that the hydrophobic effect is the dominant
factor in catalytic ability for this ligand system. Without elaborating this series of
ligands, another ligand set was investigated with the hopes of finding bimetallic
cooperativity within a macrocyclic scaffold.
Dien-based ligands. The imine cyclization of Martell and coworkers provides a
facile route into macrocyclic bimetallic ligands (4). In this method, a diamine and a
TABLE 1
Kinetic Data for the Transesterification of RNA Model HPNPP with the Cyclen-Based Ligands with
Cu+2 and Zn+2
b
Ligand
Spacer
Substrate
105 ϫ kobs sϪ1a
kobs/kuncat
L1
L1
L2
L2
L3
Cyclen
Cyclen
4,4Ј BP (Cu+2)
4,4Ј BP (Zn+2)
Mono-Ts (Cu+2)
Mono-Ts (Zn+2)
Mono-Mes (Zn+2)
(Cu+2)
HPNPPc
HPNPPc
HPNPPc
HPNPPc
HPNPPc
HPNPPc
HPNPPc
3.81
4.41
6.07
3.15
0.029
0.23
0.90
224
245
337
175
1.6
13
(Zn+2)
50
a Run in 30% DMSO and 1 ϫ 10Ϫ2 M Hepes (pH 8.00) at 30ЊC.
b Observed first-order rate coefficients with 2.5 ϫ 10Ϫ4 M dimer and 5 ϫ 10Ϫ4 M monomer and one
eq. of metal per binding site.
c kuncat HPNPP ϭ 1.8 ϫ 10Ϫ7 sϪ1
.

TWO-METAL COOPERATIVITY IN PHOSPHATE HYDROLYSIS
349
dialdehyde are condensed and reduced, forming a 2 ϩ 2 macrocycle. With this
technique we prepared ligands L4–L7, each carrying two diethylenetriamine units
separated by two spacers of increasing length. Control ligands L9 and L10 (Fig. 2)
were formed through simple reductive amination of diethylene triamine and the
corresponding monoaldehyde. Ligand L8 was formed by borane reduction of the
corresponding diamide (5). The biphenyl and dibenzofuran dialdehydes needed for
ligands L6 and L5 were synthesized from the corresponding dibromides through
halogen-metal exchange and condensation with DMF. p-Terphenyl dialdehyde for the
synthesis of L7 was synthesized by adaptation of a published procedure (6).
In previous work (2) with bimetallic enzyme mimics, rigid spacers (aryl) performed
better than less rigid spacers (alkyl). This was thought to arise from preorganization
of the ligand into an active geometry, thus reducing the entropic cost of reorganization.
In our current study, poly-aryl spacers were chosen and incorporated into a rigid
macrocyclic framework. These would be able to constrain the geometry yet provide
flexibility in the synthesis in order to test other factors in catalysis.
One of these factors is the dependence of rate enhancement on metal–metal distance.
It was hoped that an optimal M-M distance could be found through a systematic
˚
study. In natural bimetallic phosphatases this distance varies from 3–5 A (1). In this
series, ligands were synthesized and tested, which varied the M-M distance from
˚
3–12 A.
Instead of there being an ideal length, all ligands over a certain size produced large
rate enhancements in the catalyzed hydrolysis of bis-p-nitrophenyl phosphate (BNPP)
and in the cyclization of HPNPP. Indeed, the largest ligand tested (L7) produced the
highest rates of catalysis for HPNPP, and essentially also for BNPP (Table 2). Even
FIG. 2. Structures of the diethylene triamine (dien)-based macrocycles and butterfly series of ligands.

350
LEIVERS AND BRESLOW
TABLE 2
Kinetic Data for Phosphate Hydrolysis and Transesterification with Dien-Based Ligands and
Zn+2 (from Fig. 2)a
c
Ligand
Spacer
1,4 Ph
DBF
4,4Ј BP
4,4Љ TP
Mono-Me
Mono-Bn
Mono-4 BP
1,4 Ph
DBF
4,4Ј BP
4,4Љ TP
Mono-Me
Mono-Bn
Mono-4 BP
1,4 Ph
4,4Ј BP
Mono-Me
Substrate
kobs ϫ 105 sϪ1b
kobs/kuncat
L4
L5
L6
L7
L8
L9
L10
L4
L5
L6
L7
L8
L9
L10
L4
L6
L8
HPNPPd
HPNPPd
HPNPPd
HPNPPd
HPNPPd
HPNPPd
HPNPPd
BNPPe
BNPPe
BNPPe
BNPPe
BNPPe
BNPPe
BNPPe
UpUf
2.5
12.1
11.9
23.8
2.7
140
674
661
1320
150
210
927
330
1410
2240
1980
none
28
3.9
16.7
0.11
0.47
0.74
0.654
none
0.0092
0.0858
0.22
0.12
0.07
260
230
120
74
UpUf
UpUf
a Run in 30% DMSO and 1 ϫ 10Ϫ2 M Hepes (pH 8.00).
b Observed first-order rate coefficients with 2.5 ϫ 10Ϫ4 M dimer and 5 ϫ 10Ϫ4 monomer and one eq.
of metal per binding site.
c kuncat HPNPP ϭ 1.8 ϫ 10Ϫ7
d Run at 30ЊC.
s s .
Ϫ1, kuncat BNPP ϭ 3.3 ϫ 10Ϫ9 Ϫ1, kuncat UpU ϭ 9.8 ϫ 10Ϫ9 sϪ1
e Run at 55ЊC.
f Run at 41ЊC with 20% DMSO and 5 ϫ 10Ϫ2 M Tris buffer (pH 8.4).
invoking outer sphere effects, the distance between the metals precludes cooperativity
and suggests that another effect is operating.
In addition to providing rigidity to the ligand system, the spacer groups may also
provide hydrophobic binding surfaces for the model substrate. Two of the substrates
in this and many other studies are activated phosphates carrying phenyl groups.
Cleavage of these model substrates is monitored by the release of nitrophenoxide
anion. The catalysts (comprised of metal binding domains and various hydrophobic
linkers) may form, by our hypothesis, hydrophobic regions capable of binding the p-
NO2Ph-moiety. We and other groups have used binding cavities in phosphatase mimics
(calixarenes2h, ␤-cyclodextrin7) in order to increase catalysis rates through sub-
strate binding.
Nature normally uses two metals to perform the hydrolysis of phosphates. Rate
enhancements of 1017 are necessary to cleave DNA on the seconds time scale (8). It
is logical to assume that nature found it necessary to use two metals cooperatively
in order to achieve these high rate enhancements. Enzymes that model this reaction
should show the same dependence for two metals over one. Thus, various methods
have been devised to investigate cooperativity.

TWO-METAL COOPERATIVITY IN PHOSPHATE HYDROLYSIS
351
One that has been widely used is to synthesize separate control compounds con-
taining only one metal binding site. The relative rate enhancements between these
monometallic control ligands and the bimetallic ligands usually are an indication of
the amount of cooperativity involved. Unfortunately, it is sometimes difficult to
construct a good control compound that simply removes one of the metals, without
changing other aspects of the complex. In practice, it is seen that rate enhancements
are not greatly changed through the addition of another metal in these model systems.
The relative hydrolysis of monomeric vs bimetallic ligands is usually a small fraction
of the total rate enhancement. In one case (4), Martell finds that the active catalyst
is the monometallic complex, and the bimetallic complex has reduced catalytic power.
In some of the ligands tested in the present study, no cooperativity is possible due
to the large separation between the metals. It is probable that the hydrophobic effect
is governing the rate enhancements. Even though the hydrophobic effect is operating
in this set of ligands, do some exhibit cooperativity? In our study, monomeric ligands
were also synthesized in order to test cooperativity. As is seen in Table 2, even with
one metal, high rate enhancements can be obtained given enough hydrophobic area.
A biphenyl unit is necessary to achieve a rate enhancement comparable to the bimetallic
ligands. The relatively nonhydrophobic L9 (N-methyl) and the smaller L8 (N-benzyl)
groups have diminished catalysis.
From this data it does not seem that incorporation of two metals is necessary for
large rate enhancements. The enhancements appear to be due to reduction of the
activation barrier through hydrophobic binding and preorganization of the substrate
in the ligand. However, there still may be cooperation between the metals in the
macrocyclic ligands tested, but the amount of cooperation appears to be low and the
hydrophobic effect is the dominant component. In addition, these compounds also
illustrate the care needed in constructing an appropriate monomeric control ligand.
Simply bifurcating the active ligand L6 (with a biphenyl spacer) and comparing it to
the monomeric ligand L8 (with a benzyl control) does not yield an appropriate metric
for estimating the level of cooperativity involved in catalysis.
If a comparison of bimetallic ligands with monometallic control compounds is not
the ideal way to measure the extent of cooperativity, what tests are appropriate?
Looking only at the ligand in question and varying the amount of coordinated metal
(from 0 to 2 per ligand) is an ideal control experiment. This control experiment was
performed in two ways: (1) at a constant ligand concentration, varying the metal
concentration; (2) at a constant metal concentration, varying the ligand concentration.
Considering case 1: if we treat the two metal binding sites as independent, simple
statistics gives the probability of finding the three possible states of the ligand (LM0,
LM1, LM2) with respect to metal concentration. (9) It can be seen from plots of these
speciations that functions dominated by LM1 have a maximum and then decrease as
metal concentration is increased, and those dominated by LM2 have a sigmoidal
shape. Assuming that both LM1 and LM2 have some catalytic function, an equation
relating catalysis vs. metal concentration can be made (10).
This series of experiments was performed for two ligands (L6, L9) with the RNA
model substrate HPNPP. The results are shown in Figs. 3 and 4. The graph of the
bimetallic ligand (L6—Fig. 4) shows that the mono- and bimetallic species have
equivalent catalysis. Thus, cooperativity is not involved in the catalysis of phosphate

352
LEIVERS AND BRESLOW
FIG. 3. Transesterification of RNA model (HPNPP) at constant monomeric ligand (L9) concentration
and varying Zn.
transesterification of HPNPP with L6. The same experiment was performed with the
monometallic ligand L9. This of course did not show a maximum, because only one
metal can be bound per ligand (11).
The second method of analyzing cooperativity (setting [Zn] constant and varying
[ligand]) was also performed. Assuming that there is no cooperativity in the binding
of the two metals (as is assumed in the above treatment as well), changing the ligand
concentration should give a measurement of cooperativity. This approach has been
reported previously as evidence for cooperativity but it has not been adopted as a
standard test for cooperativity (12). If free zinc is not an effective catalyst and the
ligand is a strong chelator, a graph of rate vs ligand concentration should indicate if
the more potent species is the 2:1 or 1:1 (metal:ligand) complex. The results of this
experiment are shown with L6 and HPNPP in Fig. 6. This is in accord with the first
method that shows that there is no evidence of cooperativity.
FIG. 4. Transesterification of RNA model (HPNPP) at constant dimeric ligand (L6) concentration
and varying Zn. Curve fit with equation in Ref. 10, with k1/k2 ϭ 0.9

TWO-METAL COOPERATIVITY IN PHOSPHATE HYDROLYSIS
353
FIG. 5. Transesterification of RNA model (HPNPP) at constant zinc concentration (0.5 mM) and
varying dimeric ligand (L6)
Another common group of substrates for phosphatase mimics are RNA dimers.
The most common of these is 3Ј → 5Ј uridylyl uridine (UpU). We wondered if this
substrate behaved differently with our catalysts than the nitrophenyl phosphates
(BNPP, HPNPP). The UpU is less hydrophobic and should provide data that does
not create a bias toward the hydrophobic linkers. Three of the ligands were tested
(L4, L6, L8) and the results are given in Table 2. In contrast to the previous two
nitrophenyl based substrates, the most powerful ligand was not the most hydrophobic,
L6. The nonhydrophobic monomeric ligand L8 performed close to the hydrophobic
bimetallic ligand L6 with L4 moderately superior.
The hypothesis is that hydrophobic effects in binding and catalysis are diminished
with UpU relative to the nitrophenoxide based substrates (HPNPP, BNPP). With
this in mind cooperativity was investigated in a series of experiments keeping zinc
FIG. 6. Transesterification of UpU at constant zinc concentration (3 mM) and varying dimeric ligand
(L4) concentration.

354
LEIVERS AND BRESLOW
FIG. 7. Ligands previously synthesized in Ref 2.
concentration constant and varying the ligand concentration. It was hoped that a
maximum would be found for the 2:1 metal:ligand complex, which would be indicative
of cooperativity. The results are shown in Fig. 6. It can be seen that there is little
variance with respect to the ratio between metal and ligand. Put in another way, the
zinc catalyses the hydrolysis equally well in a monometallic complex or a bimetal-
lic complex.
1,4,7-Triazacyclododecane ligands. Finally, two ligands (L11 and 12) (2) that were
previously reported by our group were resynthesised by a slight variation on the
published procedure (Fig. 7) (13). Initially they had been thought to be cooperative
based upon comparison with monomeric controls. In particular, ditopic ligand L11
with a biphenyl spacer was seen to be more effective than L12 with a phenylene
spacer in the zinc catalyzed cleavage of HPNPP and BNPP, while the shorter ditopic
ligand L12 was superior for the hydrolysis of p-nitrophenyl phosphate with zinc.
However, we now had to be concerned about the possibility that the superiority of
L11 over L12 was a reflection of its greater hydrophobicity rather than its particular
metal ion spacing. Thus we examined a series of metal:ligand ratios in the cyclization
of HPNPP in order to assess the amount of cooperativity present in the system. The
results are shown in Figs. 8 and 9. Neither figure conforms to that predicted for
cooperativity. Thus, it is concluded that cooperativity is not, in fact, present for this
ligand as originally proposed (14).
FIG. 8. Transesterification of RNA model (HPNPP) at constant dimeric ligand (L6) concentration
and varying Zn. Curve fit with equation in Ref. 10, with k1/k2 ϭ 1.1

TWO-METAL COOPERATIVITY IN PHOSPHATE HYDROLYSIS
355
FIG. 9. Transesterification of RNA model (HPNPP) at constant dimeric ligand (L12) concentration
and varying Zn. Curve fit with equation in Ref. 10, with k1/k2 ϭ 1.1
We have shown that in the case of some model phosphatase substrates, care must
be taken when constructing control compounds. In these cases the hydrophobic effect
can take precedence over cooperativity and spurious results are obtained. Two related
methods were examined that provide a better test for cooperativity than construction
of control compounds, and they gave equivalent results. It is suggested that bifurcated
versions of purported cooperative bimetallic ligands not be used as evidence for
cooperativity. Studies varying metal to ligand ratios are superior and should be used
as a metric for cooperativity. An RNA dimer was also tested in this manner, and it
was found that this substrate does not have as large a dependence on the hydrophobic
effect as the two model substrates based on p-nitrophenyl phosphates, but cooperativity
was still not seen in the ligands tested.
REFERENCES
1. (a) Stra˚ter, N., Lipscomb, W. N., Kablunde, T., Krebs, B. (1996) Angew. Chem. Int. Ed. Engl. 35,
2024. (b) Wilcox, D. E. (1996) Chem. Rev. 96, 2435.
2. Chapman, W., and Breslow, R. (1995) J. Am. Chem. Soc. 117, 5462.
3. Vance, D., and Czarnik, A. (1993) J. Am. Chem. Soc. 115, 12165. McCue K., and Morrow, J. (1999)
Inorg. Chem. 38, 6136. Valtancoli, B., et al. (1999) Inorg. Chem. 38, 4115. Shanghao, L., and
Hamilton, A. (1999) Chem. Comm. 587. Shiro, M., et al. (1996) J. Am. Chem. Soc. 118, 3091. Chuan,
H., and Lippard, S. J. (2000) J. Am. Chem. Soc. 122, 184. Reinhoudt, D., et al. (1999) J. Org. Chem.
64, 6337. Chin, J., et al. (1998) J. Am. Chem. Soc. 120, 8079.
4. Jurek, P., and Martell, A. E. (1999) Inorg. Chem. Acta. 287, 47.
5. Winans, C. F., and Adkins, H. (1933) J. Am. Chem. Soc. 55, 4167.
6. Cristofaro, M., and Chamberlin, R. (1994) J. Am. Chem. Soc. 116, 5089.
7. Breslow, R., and Dong, S. (1998) Chem. Rev. 98, 1997.
8. Williams, N., Takasaki, B., Wall, M., and Chin, J. (1999) J. Acc. Chem. Res. 32, 485.
9. Define ⌰ as the fraction of site A that is filled at a given metal concentration, thus ⌰ ϭ ([metal] и
kd)/(1 ϩ ([metal] и kd)). In our case there are two equal sites, thus the population of finding the
ligand with no metal: LM0 ϭ (1 Ϫ ⌰)2; and LM1 ϭ 2 и ⌰ и (1 Ϫ ⌰)2; LM2 ϭ (⌰)2
10. Rate ϭ k1[LM1] ϩ k2[LM2]. If k1/k2 ϾϾ 1 then there is anticooperativity, if k1/k2 ϭ 1 then there is
neither cooperativity nor anticooperativity, if k1/k2 ϽϽ then there is cooperativity. Hess, V. L., and
Szabo, A. (1979) J. Chem. Ed. 56, 289.

356
LEIVERS AND BRESLOW
11. In both graphs, however, it was noted that a slight sigmoidal curve appeared at low [Zn]. This would
normally indicate positive cooperativity, but is inconsistent with the maximum seen at a specific
[Zn] for L6 (Figure 5). Because this sigmoidal shape is present in both the bimetallic and monomeric
ligands, another explanation was sought. As prepared, the substrate (HPNPP) is a salt with Ba+2. At
low concentrations the zinc has to compete with the binding of the barium (at 1 equivalent per
ligand) and thus a depression in the rate is seen at low concentrations of zinc. As the [Zn] is raised,
it is able to successfully compete with the barium and the line follows the predicted shape. In order
to further illuminate this effect, additional Ba+2 was added at a constant ligand and constant (low)
zinc concentration. It is seen that the added Ba+2 does indeed function by suppressing the rate
of hydrolysis.
12. This test has been used to show cooperativity through a sigmoidal M:L plot. Yashiro, M., Ishikubo,
A., Komiyama, M. (1995) J. Chem. Soc. Chem. Commun. 1793.
13. From t-butyldimethylsilyl-1,5-dichloro-3-pentanol. Reese, C., and Thompson, E. A. (1988) J. Chem.
Soc. Perkin Trans. 1, 2881.
14. The two metal binding sites in all three series of ligands are assumed to be independent when binding
metals. If however there is cooperativity involved in binding these analyses are not valid. The
metal:ligand studies still show that cooperativity is not involved. Assuming that metal binding is
absolutely cooperative, then in a 1:1 ratio of metal:ligand there is 50% population of LM2 and 50%
LM0. The relative rate should therefore be 50% of the maximum. If the rate is greater than that then
cooperativity is not shown.
15. Ligands L4 and L6 were prepared previously: Ragunathan, K. G., and Schneider, H.-J. (1996) J.
Chem. Soc. Perkin 2, 2597.