Understanding Metallo ß-Lactamases
A major role of zinc in these enzymes is to act as a
Lewis acid rather than providing a nucleophile when hydrolyzing ß-lactam
Michael I. Page
The ability to produce ß-lactamase enzymes is the
major cause of the resistance of bacteria to ß-lactam antibiotics.
These enzymes catalyze hydrolysis of four-membered ß-lactam rings,
converting them into ß-amino acids that, unlike active members
within this class of antibiotics, no longer interfere with bacterial
cell synthesis and growth. The ß-lactamase enzymes are divided
mechanistically into two groups, with those in classes A and C using
serine as an active site residue, and those in class B using a metal
ion, which is invariably zinc.
The class B metallo ß-lactamases were discovered
more that 40 years ago by E. P. Abraham at the University of Oxford in a
strain of Bacillus cereus that also produces two serine ß-lactamases.
By 1985 researchers knew of two metallo ß-lactamases, whereas
currently about 20 different strains, most of them human pathogens, are
recognized as carrying such ß-lactamases. The genes encoding these
enzymes generally are chromosomal but some are on plasmids, which
enables them to spread among microorganisms more readily and thus makes
them a greater cause of public health concern.
The metallo ß-lactamases are broad-spectrum enzymes
that very effectively catalyze the hydrolysis of not only penicillins
and cephalosporins but also other ß-lactam antibiotics, making
pathogens that carry these enzymes a real problem in the clinic. In
particular, they catalyze the hydrolysis of carbapenems, such as
imipenem, that are generally resistant to serine ß-lactamases.
Indeed, when these antibiotics were introduced, no known ß-lactamases
catalyzed their hydrolysis. Currently, although several compounds
effectively inhibit these metallo enzymes in vitro, none is clinically
Metallo ß-Lactamases Being Intensively Studied
The primary sequences of more than 20 metallo ß-lactamases
are now available, along with refined crystal structures for four of
them, each of which appears capable of binding two zinc ions. Although
procedures for purifying and crystallizing these enzymes might distort
this picture in some cases, all these enzymes require at least one zinc
per molecule of enzyme to be catalytically active.
Some investigators divide these enzymes into three
subgroupsdesignated B1, B2, and B3on the basis of the residues
that bind the zinc ions. Among those belonging to the most common B1
type, the zinc ion in the first zinc site is in a four-coordinate
tetrahedral geometry that is bound by three histidine residues, while
the fourth coordination site is occupied by a water molecule. The second
zinc binding site of the B1 proteins contains its metal ion in a
five-coordinate arrangement with trigonal bipyramidal geometry. In this
configuration, three different amino acid residuesaspartate, cysteine,
and histidineserve as ligands, furnishing oxygen, nitrogen, and
sulfur atoms, respectively, to link with zinc. In addition, the zinc
binds two water molecules, one which is not bound elsewhere and another
which bridges with the other zinc ion. The two zinc ions thus are fairly
close to one another, separated by only about 3.5 Å, particularly in
those enzymes that bind these ions very tightly.
These enzymes target the carbon-nitrogen bond and open
the four-membered ring that is characteristic of ß-lactam
antibiotics. In the first step of this hydrolytic reaction, a new bond
forms between the ß-lactam carbonyl carbon and oxygen, which comes
from water, assumed to be the one bound to zinc. In addition to breaking
the carbon-nitrogen bond, various proton transfers occur. Metal ions can
increase the rates of both stepsthe first in which a new
carbon-oxygen bond forms, and the second in which the carbon-nitrogen
bond of the ß-lactam ring breaks. Despite a misplaced emphasis on
that first step, breaking the carbon-nitrogen bond in peptides and,
surprisingly, in ß-lactams is not easy and also requires catalysis.
The zinc cation, with its high positive charge density,
could play a key role in this step, acting in one of several ways. For
one, this charge density reduces the pKa of the water that is bound to
the zinc ion, forming the rough equivalent of a zinc-bound
"hydroxide ion."If this pKa were lowered to 7, then the
acidity of this bound water molecule approaches that of dihydrogen
phosphate ion or MOPS, suggesting that this zinc-bound "hydroxide
ion" would be no more nucleophilic than either of these bases.
Alternatively, the zinc ion might act as a Lewis acid, polarizing the
carbonyl group while stabilizing the oxyanion that is formed following
nucleophilic addition to the carbonyl group of the ß-lactam. Yet
another possibility is that the zinc ion facilitates carbon-nitrogen
bond fission in the ß-lactam ring by directly coordinating to the
nitrogen and stabilizing the nitrogen anion that is expelled. Finally,
the water molecule bound to the zinc could act as a general acid
catalyst, donating a proton to that nitrogen to facilitate
carbon-nitrogen bond fission.
Several Reasons Why Bacteria Use Zinc in Metallo-Enzymes
Why do bacteria produce a zinc metallo enzyme? One
important requirement is the protein's ability to sequester particular
metal ions from the environment. Ions are selected primarily on the
basis of size, charge, and "hardness," which is a measure of
an ion's relative capacity to combine with oxygen, sulfur, and nitrogen
ligands. One of the interesting things about zinc is that it is on the
borderline in hardness and it therefore combines equally well with
oxygen, sulfur, and nitrogen ligands and, in the case of one of the
classes of ß-lactamases, the metal binds all three of these ligands.
Another important feature of zinc is that although its geometry is
predominantly a four-coordinate tetrahedral, it can change very readily
to five- or six-coordination with little expense of energy. Finally,
zinc ions have a d10 electronic configuration, meaning there is no redox
chemistry and avoiding ligand field stabilization energy and any
complications that would occur by changing the number and nature of
ligands on the zinc.
All these properties make zinc the favored metal for
hydrolytic catalytic activity. To make this enzyme, bacteria have to
sequester zinc from the environment, which is not easy. The hydration
energy of zinc ions is very highat least 500 kilocalories per molemaking
zinc extraction and binding a very energetically expensive process.
Several changes occur when zinc is taken out of the
environment. In aqueous solution, zinc is generally six coordinate with
an octahedral geometry, whereas in the protein it is generally
tetrahedral. Hence, on binding, the ligands change, the geometry of
their binding changes, and the pKa of the water that is bound to the
zinc also changes. Water has a pKa of about 16. However, when it is
bound to zinc, the pKa decreases tremendously to around 9, making the
water much more acidic. When zinc with its bound water molecule is
coordinated to a protein, typically the pKa is reduced even further,
down to about 6 or 8. The main reason for this change in pKa is the
altered coordination chemistry for zinc from an octahedral to a
four-coordinate system, which increases the positive charge density on
the cation and makes the bound water molecule more acidic. This effect
of charge density is important when looking at mutants where the ligands
in the enzyme bound to the metal-ion are changed, because of the effect
that these changes may have on the pKa of the zinc bound water. For
example, if a neutral histidine ligand of zinc is replaced by negatively
charged aspartate or cysteine, then the pKa is increased by about 2
Investigating the Detailed Mechanism of the Metallo
The mechanism of the metallo ß-lactamases may well
not be unique but may vary for different substrates and with the source
of the enzyme. For example, our studies with the Bacillus cereus metallo
ß-lactamase show that, by increasing the zinc ion concentration
100,000-fold, the catalytic activity of the mono form of the zinc enzyme
changes by less than a factor of 2. Thus, although the enzyme can bind
two zinc ions, our kinetic evidence suggests that the second zinc does
not have a significant catalytic effect.
One of the common features of hydrolytic metallo
enzymes, particularly metallo proteases such as carboxypeptidase, is
uncertainty over mechanism. For example, there is often an ionized amino
acidcommonly aspartatein the active site that might act as a
nucleophile to form an anhydride intermediate which, in principle, could
be trapped with methanol to form an ester. However, with the metallo
ß-lactamases, no methyl ester forms and adding methanol does not
affect the rate of hydrolysis. Such experiments, although not ironclad,
indicate that the ß-lactamase mechanism is not nucleophilic.
The pH dependence of the rate of hydrolysis of benzyl
penicillin catalyzed by the metallo enzyme reveals several potentially
important features. First, the catalytic activity is pH independent over
a wide range, with a pH bell profile that is typical of two ionizations
with pKa's of about 6 and 9. Moreover, changing the pH in the acid
region by 1 unit changes catalytic activity 100-fold, suggesting that
two groups are ionizing. The enzyme-catalyzed hydrolysis likely involves
at least three bases and a proton.
The pKa of the zinc-bound water could represent either
of the ionizations indicated at pH 6 or 9 but which can be identified
from the pH dependence of the inhibition constants of some thiol
inhibitors. We made a series of thiol inhibitors that bind to the zinc
ion, and their inhibition constants show a similar pH independence to
that for the hydrolysis reaction, indicating that the water bound to
zinc is ionized at neutral pH and confirming that the pKa of the water
bound to zinc is 5.7. There are two catalytically important groups of
about pKa 6 in this enzyme, one of which is the water bound to zinc and
the other is likely the carboxylate group of the active site aspartate.
An important feature of having two basic groups is that they are both
anions, meaning the transition state is dianionic. This feature may
prove a useful indicator for designing inhibitors of the zinc enzymes.
Zinc Ions in ß-Lactamases Likely Act in Dual
Role when Hydrolyzing Antibiotics
We believe that the zinc ions in metallo enzymes act in
a dual role during hydrolysis of ß-lactam antibioticsproviding,
first, a Lewis acid to polarize the carbonyl group and, second, a
"hydroxide ion" bound to zinc to act as a nucleophile to
attack the carbonyl group to form a tetrahedral intermediate. Because
the zinc-bound hydroxide is only a weak nucleophile, this step is
reversible. The role of the aspartate is to deprotonate the hydroxyl
group in the intermediate to form a dianion. Because the aspartate then
exists in its neutral acidic form, it can act as a proton donor to the
ß-lactam nitrogen to facilitate carbon nitrogen bond fission to
form the end products.
In the ß-lactamase enzymes that contain two zinc
ions, the bridging ion might be acting as the immediate nucleophile
during the hydrolytic step. In such enzymes, the zinc-bound
"hydroxide ion" is an extremely weak nucleophile, meaning the
first addition step to the ß-lactam would almost certainly be
reversible. The other water that is coordinated to the second zinc can
act as a general acid, a proton donor to facilitate carbon nitrogen bond
fission to form the products. There is an intermediate formed in these
reactions, at least with nitrocefin as a substrate, and these findings
suggest that the zinc ion coordinates directly to the ß-lactam
nitrogen and thereby stabilizes nitrogen anion expulsion, which is
subsequently protonated, according to Steve Benkovic at Pennsylvania
State University in University Park, Pa., and his collaborators.
Site-directed mutagenesis of the mono-zinc form of the
ß-lactamase from B. cereus indicates that the pKa of the
zinc-bound water can be increased by making the zinc effectively less
positive due to a change in ligands around the metal ion. A higher pKa
reduces the concentration of the zinc-bound hydroxide ion at pH values
below the pKa but, at the same time, makes it a better nucleophile.
However, because the zinc has less positive charge density, these
changes will decrease the ability of zinc to stabilize the oxygen anion
and polarize the ß-lactam carbonyl.
If the zinc is acting mainly as a Lewis acid stabilizing
the tetrahedral intermediate, then one would expect the mutant to show
reduced activity, whereas if its main role is to provide a water-bound
nucleophile, then activity would be expected to increase. Replacing one
of the zinc-bound histidines by serine, in fact, reduces the activity of
the enzyme. The pH dependence of the rate of hydrolysis of ß-lactams
catalyzed by the mutant now shows two separate ionizations on the acid
limb. With serine replacing histidine, the positive charge density on
the zinc is decreased, thereby increasing the pKa of the zinc-bound
water to 7.3, separating it from the aspartate. The decreased activity
of this particular mutant demonstrates that one of the major roles of
zinc is to act as a Lewis acid rather than to provide the nucleophile.
By decreasing the positive charge density on zinc and increasing the pKa
of the zinc-bound water, the ability of the zinc to stabilize the
developing negative charge density on the ß-lactam carbonyl oxygen
Because one of the important functions of zinc is to
polarize the ß-lactam carbonyl group, we made a series of cysteinyl
dipeptide inhibitors of these enzymes in each of which the thiol anion
can coordinate to the zinc ion. Hydrophobic residues increase the
ability of the zinc enzyme to bind these inhibitors. Among a series of
diastereoisomers, the DD diastereoisomer proved a better inhibitor than
the others. In penicillins and cephalosporins, the acyl amino side chain
has an l stereochemistry at C6/7 and a D stereochemistry at the C3/4
bearing carboxylate, respectively. If the cysteinyl dipeptides are bound
to the zinc, the stereochemistry has to be inverted to enable the best
molecular recognition of the other substituents.
Bounaga, S., A. P. Laws, M. Galleni, and
M. I. Page. 1998. The mechanism of catalysis
and the inhibition of Bacillus cereus zinc-dependent ß-lactamase
Biochem. J. 331:703-711.
Bounaga, S., M. Galleni, A. P. Laws, and
M. I. Page. 2001. Cysteinyl peptide inhibitors
of B. cereus zinc ß-lactamase. Bioorg. Med. Chem. 9:503-510.
de Seny, D., C. Prosperi-Meys, C. Bebrone,
G. M. Rossolini, M. I. Page, P. Noel, J.-M. Frere, and M. Galleni.
Mutational analysis of the two zinc binding sites of the Bacillus
cereus 569/J/9 metallo-ß-lactamase. Biochem. J., in press.
Page, M. I., and A. P. Laws.
1998. The mechanism of catalysis and the inhibition of ß-lactamases.
Chem. Commun. 1998:1609-1617.
Paul-Soto, R, R. Bauer, J.-M. Frere, M.
Galleni, W. Meyer-Klaucke, H. Nolting, G. M. Rossolini, D. de Seny, M.
Hernandez-Valladares, M. Zeppezauer, and H.-W. Adolph. 1999.
Mono- and binuclear Zn2+- ß-lactamase J. Biol. Chem. 274:13242-13249.