temperatures and pH, and at high concentrations of reactants and organic
(co)-solvents, which is crucial for the construction of process-stable proteins
for biotechnological applications. In addition to improved stability, enzymes also
can be engineered for enhanced (stereo)selectivities, which represents an equivalent to ligand tuning of homogeneous catalysts. Only some key issues are
discussed below since this area requires special expertise in molecular biology –
not necessarily a playground of synthetic organic chemists. For a deeper understanding, excellent introductory chapters can be found in recent books and
reviews [325–332].
There are two distinct philosophies to enzyme engineering:
1. Rational protein design requires detailed knowledge of the three-dimensional
structure of an enzyme, preferably from its high-resolution crystal structure or
NMR measurements [333]. Alternatively, a computer-generated homology
model may help, if the sequence identity is high enough. A sequence identity
of ~70% translates into a reasonably well-defined model showing a root mean
square deviation of 1–2 Å, which drops to a low value of 2–4 Å for proteins
having only ~25% identity. In a first step, docking of the substrate to the active
site allows to identify amino acid residues, which appear to interact closely with
the structural features of the substrate during binding. Steric incompatibilities,
such as collisional interference between residues, insufficient substrate binding
in large pockets, or nonmatching polarities between hydrogen bonds or salt
bridges can be identified and proposals for the replacement of (usually only
few) amino acids can be made. The corresponding mutants are generated and
tested for their catalytic properties. Sometimes, this rational approach yields
impressive results, but quite often mutant enzymes tell us that the rational
analysis of the substrate binding based on a static (crystal) structure is insufficient to explain the dynamic process of protein (re)folding upon formation of the
enzyme–substrate complex, which is a prerequisite to support the dynamics of
protein catalysis [334]. In addition, the tempting notion that mutations close to
the active site are always better than distant ones is only a single aspect of a more
complex story [335].
2. Directed evolution requires the availability of the gene(s) encoding the enzyme
of interest, a suitable (microbial) expression system, a method to create mutant
libraries, and an effective selection system – while structural information is
irrelevant here.
9 Traditionally, mutant libraries are created by error-prone polymerase chain reaction (epPCR) with low mutation frequencies (1–3 mutations
per 1000 base pairs). Since the possible number of mutants generated from a
given protein exponentially increases by the number of mutations,
10 the crucial
9 The principle of directed evolution was first described by M. Eigen, see [336].
10 The possible number of mutants generated from a protein possessing 200 amino acids are 3800
variants for a single mutation, 7,183,900 for two mutations, and 8,429,807,368,950 for only four
mutations. Complete ramdomization would result in 20
200 enzyme variants, which is more than the
mass of the universe, even if only one molecule of each enzyme were to be produced.
2.1 Hydrolytic Reactions
77
(co)-solvents, which is crucial for the construction of process-stable proteins
for biotechnological applications. In addition to improved stability, enzymes also
can be engineered for enhanced (stereo)selectivities, which represents an equivalent to ligand tuning of homogeneous catalysts. Only some key issues are
discussed below since this area requires special expertise in molecular biology –
not necessarily a playground of synthetic organic chemists. For a deeper understanding, excellent introductory chapters can be found in recent books and
reviews [325–332].
There are two distinct philosophies to enzyme engineering:
1. Rational protein design requires detailed knowledge of the three-dimensional
structure of an enzyme, preferably from its high-resolution crystal structure or
NMR measurements [333]. Alternatively, a computer-generated homology
model may help, if the sequence identity is high enough. A sequence identity
of ~70% translates into a reasonably well-defined model showing a root mean
square deviation of 1–2 Å, which drops to a low value of 2–4 Å for proteins
having only ~25% identity. In a first step, docking of the substrate to the active
site allows to identify amino acid residues, which appear to interact closely with
the structural features of the substrate during binding. Steric incompatibilities,
such as collisional interference between residues, insufficient substrate binding
in large pockets, or nonmatching polarities between hydrogen bonds or salt
bridges can be identified and proposals for the replacement of (usually only
few) amino acids can be made. The corresponding mutants are generated and
tested for their catalytic properties. Sometimes, this rational approach yields
impressive results, but quite often mutant enzymes tell us that the rational
analysis of the substrate binding based on a static (crystal) structure is insufficient to explain the dynamic process of protein (re)folding upon formation of the
enzyme–substrate complex, which is a prerequisite to support the dynamics of
protein catalysis [334]. In addition, the tempting notion that mutations close to
the active site are always better than distant ones is only a single aspect of a more
complex story [335].
2. Directed evolution requires the availability of the gene(s) encoding the enzyme
of interest, a suitable (microbial) expression system, a method to create mutant
libraries, and an effective selection system – while structural information is
irrelevant here.
9 Traditionally, mutant libraries are created by error-prone polymerase chain reaction (epPCR) with low mutation frequencies (1–3 mutations
per 1000 base pairs). Since the possible number of mutants generated from a
given protein exponentially increases by the number of mutations,
10 the crucial
9 The principle of directed evolution was first described by M. Eigen, see [336].
10 The possible number of mutants generated from a protein possessing 200 amino acids are 3800
variants for a single mutation, 7,183,900 for two mutations, and 8,429,807,368,950 for only four
mutations. Complete ramdomization would result in 20
200 enzyme variants, which is more than the
mass of the universe, even if only one molecule of each enzyme were to be produced.
2.1 Hydrolytic Reactions
77