Alteration of enzyme specificity and catalysis by protein engineering Helen M. Wilks and I. John Holbrook University of Bristol, Bristol, UK New substrate specifities can be introduced into existing enzymes for the purpose of making them more suitable for the chemoenzymic synthesis of single compound drugs and other chirai compounds. The most productive route used in the past year has involved the utilization of the catalytic and substrate-binding properties from homologous enzymes found in nature, one example being the broadening of the substrate specificity of yeast alcohol dehydrogenase. Other highlights include the creation of thermostable dehydrogenases that will interconvert NADPH and NADH, and the design of mutant enzymes with improved catalytic rates compared with their wild-type counterparts. Current Opinion in Biotechnology 1991, 2:561-567
Introduction The development of techniques for gene synthesis and manipulation has presented protein scientists with the opportunity of constructing proteins with enzyme properties that are not readily available in nature. This goal can be achieved by either random mutagenesis and screening or by rational design and construction. The latter approach is considered here in a discussion of papers pubfished over the past year. Emphasis is given to papers that describe applications for which enzymes have been redesigned. In theory, existing knowledge of the mechanisms of enzyme catalysis should enable the protein designer to determine the positions in space where amino acid side chains are required for both substrate specificity and catalysis. The next step requires the design of a self-folding polypeptide chain which acts as a scaffold to hold these groups in the desired positions in space. Such design is beyond present day understanding although it can be avoided, at least in part, by using template-assisted synthetic protein folding, where relatively short peptide chains are attached to a stable base plate which serves as a nucleation point from which secondary structure can more easily expand [ 1]. The alternative route is to select a well understood preexisting enzyme framework on which to build new catalysis and specificity. A thermostable protein has the advantage that it will tolerate greater design errors before misfolding occurs. The range of enzyme targets approached is still relatively narrow because of the very detailed knowledge of a protein framework that is required before it can be rationally redesigned. Thus, successes
have all involved proteins that have undergone a long history of detailed mechanistic and structural studies. Consequently, in addition to the papers mentioned here, a much larger body of literature exists in which authors have used protein engineering techniques to raise the understanding of protein frameworks to the point where enzyme redesign can be attempted with some chance of success.
This year's reports have addressed five goals: the creation of proteinases suitable for peptide synthesis; the design of thermostable dehydr0genases which have a broad substrate specificity, making them suitable as catalysts for a wide range of chiral organic syntheses; the creation of thermostable dehydrogenases that will interconvert NADPH and NADH for organic syntheses which require regeneration of the expensive cofactor NADP(H); the narrowing of the substrate specificity of [3-1actamase; and the improvement of the catalytic rate of natural enzymes against natural substrates.
Creation of proteinases suitable for peptide and nucleotide synthesis The serine protease, subtilisin, catalyses the hydrolysis of ester or amide substrates. For the synthesis of peptides, the rate of proteolysis must be reduced while leaving an esterase activity high enough for the synthesis reaction. One way of controlling proteolytic activity is the addition of water-miscible organic solvents such as, dimethylformamide, to the enzyme. In the dimethylformamide-water solvent system the enzyme is more stable, the substrates
Abbreviations ADH--alcohol dehydrogenase; DHFR~ihydrofolate reductase; LDH--lactate dehydrogenase.
(~) Current Biology Ltd ISSN 0958-1669
561
562
Proteinengineering
Fig. 1. Applications of the highly solventstable mutant of subtilisin from Bacillus amyloliquefaciens, subtilisin 8350. (a) The regiospecific esterification of the 5'-hydroxyl of (deoxy)ribonucleosides, giving a 40-100% yield of 5'-ester in 1-2 days. R, H/OH; B, thymidine/uracil/ cytosine/adenine. (b) Production of the analgesic Leu-enkephalinamide, giving a yield of 85% in 45min using 50% dimethylformamide solvent. Boc, tertbutyloxycarbonyl.
are more soluble and synthesis reactions are thermodynamically favoured over hydrolysis. A highly solventstable mutant of subtilisin from Bacillus amyloliquefaciens has been obtained by six site-specific mutations (Met50Phe, Gly169Ala, Asn76Asp, Gln206Cys, Tyr217Lys and Asn218Ser). This mutant is 100 times more stable than the wild-type enzyme in aqueous solution at room temperature and 50 times more stable in anhydrous dimethylformamide [2--]. The mutations were chosen as a result of a combination of screening following random mutagenesis, and comparison with the amino acid sequences of other serine proteases. Overall, the new enzyme (subtilisin 8350) not only retains the hydrolytic properties of the wild-type enzyme but also has slightly improved catalytic ettlciency. With the use of activated esters, such as enol esters, the enzyme has activity that is useful for regiospecific esterification of the 5'-hydroxyl of (deoxy)ribonucleosides (Fig. l a). The low peptidase and high esterase activity of subtilisin 8350 provides quick, direct chemosynthetic routes to useful peptides such as
the analgesic Leu-enkephalinamide, giving 85% yield in 45 min (Fig. lb). Another approach to improving subtilisin from B. a m y loliquefaciens for peptide synthesis is to alter the substrate specificity of the enzyme such that it can differentiate between amide and ester substrates. Replacement of Met222 with Phe reduces the volume of the Sl' binding pocket, which results in a reduction of amide hydrolysis without affecting catalysis of esters. Mutations in the Sl binding pocket can also be made in order to alter the specificity properties of the enzyme. This construct was. not tested in preparative peptide syntheses [3"].
Creation of thermostable dehydrogenases with broad substrate specificity Lactate dehydrogenases (LDHs) and horse liver alcohol dehydrogenases (ADHs) have proved to be useful cata-
Fig. 2. Synthesis of chiral carbinols catalysed by la~:tate dehydrogenase (LDH) and horse liver alcohol dehydrogenase (ADH). The source of reducing equivalents is formate via NADH, which is generated cyclically from NAD with formate dehydrogenase. R1 = substrate side chain in both cases; R2 = H with ADH and C O O - with LDH.
Alteration of enzyme specificity and catalysis by protein engineering Wilks and Holbrook lysts for the synthesis of chiral carbinols. Such reactions usually take place in enzyme reactors. Preparative scale chiral reductions of ketones require the regeneration of NADH by a cheap reductant, such as formic acid (see Fig. 2).
,
Introduction The development of techniques for gene synthesis and manipulation has presented protein scientists with the opportunity of constructing proteins with enzyme properties that are not readily available in nature. This goal can be achieved by either random mutagenesis and screening or by rational design and construction. The latter approach is considered here in a discussion of papers pubfished over the past year. Emphasis is given to papers that describe applications for which enzymes have been redesigned. In theory, existing knowledge of the mechanisms of enzyme catalysis should enable the protein designer to determine the positions in space where amino acid side chains are required for both substrate specificity and catalysis. The next step requires the design of a self-folding polypeptide chain which acts as a scaffold to hold these groups in the desired positions in space. Such design is beyond present day understanding although it can be avoided, at least in part, by using template-assisted synthetic protein folding, where relatively short peptide chains are attached to a stable base plate which serves as a nucleation point from which secondary structure can more easily expand [ 1]. The alternative route is to select a well understood preexisting enzyme framework on which to build new catalysis and specificity. A thermostable protein has the advantage that it will tolerate greater design errors before misfolding occurs. The range of enzyme targets approached is still relatively narrow because of the very detailed knowledge of a protein framework that is required before it can be rationally redesigned. Thus, successes
have all involved proteins that have undergone a long history of detailed mechanistic and structural studies. Consequently, in addition to the papers mentioned here, a much larger body of literature exists in which authors have used protein engineering techniques to raise the understanding of protein frameworks to the point where enzyme redesign can be attempted with some chance of success.
This year's reports have addressed five goals: the creation of proteinases suitable for peptide synthesis; the design of thermostable dehydr0genases which have a broad substrate specificity, making them suitable as catalysts for a wide range of chiral organic syntheses; the creation of thermostable dehydrogenases that will interconvert NADPH and NADH for organic syntheses which require regeneration of the expensive cofactor NADP(H); the narrowing of the substrate specificity of [3-1actamase; and the improvement of the catalytic rate of natural enzymes against natural substrates.
Creation of proteinases suitable for peptide and nucleotide synthesis The serine protease, subtilisin, catalyses the hydrolysis of ester or amide substrates. For the synthesis of peptides, the rate of proteolysis must be reduced while leaving an esterase activity high enough for the synthesis reaction. One way of controlling proteolytic activity is the addition of water-miscible organic solvents such as, dimethylformamide, to the enzyme. In the dimethylformamide-water solvent system the enzyme is more stable, the substrates
Abbreviations ADH--alcohol dehydrogenase; DHFR~ihydrofolate reductase; LDH--lactate dehydrogenase.
(~) Current Biology Ltd ISSN 0958-1669
561
562
Proteinengineering
Fig. 1. Applications of the highly solventstable mutant of subtilisin from Bacillus amyloliquefaciens, subtilisin 8350. (a) The regiospecific esterification of the 5'-hydroxyl of (deoxy)ribonucleosides, giving a 40-100% yield of 5'-ester in 1-2 days. R, H/OH; B, thymidine/uracil/ cytosine/adenine. (b) Production of the analgesic Leu-enkephalinamide, giving a yield of 85% in 45min using 50% dimethylformamide solvent. Boc, tertbutyloxycarbonyl.
are more soluble and synthesis reactions are thermodynamically favoured over hydrolysis. A highly solventstable mutant of subtilisin from Bacillus amyloliquefaciens has been obtained by six site-specific mutations (Met50Phe, Gly169Ala, Asn76Asp, Gln206Cys, Tyr217Lys and Asn218Ser). This mutant is 100 times more stable than the wild-type enzyme in aqueous solution at room temperature and 50 times more stable in anhydrous dimethylformamide [2--]. The mutations were chosen as a result of a combination of screening following random mutagenesis, and comparison with the amino acid sequences of other serine proteases. Overall, the new enzyme (subtilisin 8350) not only retains the hydrolytic properties of the wild-type enzyme but also has slightly improved catalytic ettlciency. With the use of activated esters, such as enol esters, the enzyme has activity that is useful for regiospecific esterification of the 5'-hydroxyl of (deoxy)ribonucleosides (Fig. l a). The low peptidase and high esterase activity of subtilisin 8350 provides quick, direct chemosynthetic routes to useful peptides such as
the analgesic Leu-enkephalinamide, giving 85% yield in 45 min (Fig. lb). Another approach to improving subtilisin from B. a m y loliquefaciens for peptide synthesis is to alter the substrate specificity of the enzyme such that it can differentiate between amide and ester substrates. Replacement of Met222 with Phe reduces the volume of the Sl' binding pocket, which results in a reduction of amide hydrolysis without affecting catalysis of esters. Mutations in the Sl binding pocket can also be made in order to alter the specificity properties of the enzyme. This construct was. not tested in preparative peptide syntheses [3"].
Creation of thermostable dehydrogenases with broad substrate specificity Lactate dehydrogenases (LDHs) and horse liver alcohol dehydrogenases (ADHs) have proved to be useful cata-
Fig. 2. Synthesis of chiral carbinols catalysed by la~:tate dehydrogenase (LDH) and horse liver alcohol dehydrogenase (ADH). The source of reducing equivalents is formate via NADH, which is generated cyclically from NAD with formate dehydrogenase. R1 = substrate side chain in both cases; R2 = H with ADH and C O O - with LDH.
Alteration of enzyme specificity and catalysis by protein engineering Wilks and Holbrook lysts for the synthesis of chiral carbinols. Such reactions usually take place in enzyme reactors. Preparative scale chiral reductions of ketones require the regeneration of NADH by a cheap reductant, such as formic acid (see Fig. 2).
,
