CA2174525A1 - Customized proteases with altered transacylation activity - Google Patents

Abstract

The invention provides customized proteases (i.e. mutant enzymes), e.g. carboxypeptidase Y, methods of making customized proteases, as well as methods of using customized proteases. The customized proteases of the invention are derived from the known proteases. Altered transacylation reactions include the capability to perform transacylation reactions not substantially catalyzed by the known protease or the capability to perform transacylation reactions with improved yields, or both. The methods of the invention provide for customized proteases through site specific or random mutagenesis of the active site amino acids of the known proteases. The invention also provide for methods of using the customized proteases to prepare a preselected transacylation products. The preselected transacylation products produced can be modified by substitution at the N- or C-terminal with nucleophiles such as L-amino acids, D-amino acids, amino acid amides, and radioactive amino acids.

Description

~ WO9S/20039 PCT~S9S/06682 ~u~loMT~n PROTEASES

Bac~4-~.d of the In~ention Peptides used for pharmaceutical purposes will, in the future, more frequently be produced through the exploitation of genetic engineering. However, genetic engineering has limits to its capabilities. For example, expression of recombinant peptides bearing non-naturally occurring L-amino acids, D-amino acids, radioactive amino acids, and other detectable labels is not possible through recombinant techniques because there is no genetic code which codes for these modifications. In addition, naturally occurring amino acid modifications such as C-terminal amide group substitution, which are routinely performed ln vivo, are difficult to perform ln vitro. These post-translation modifications are important because they often result in the most potent or longest acting form of the peptide and constitute the naturally occurring form,of the peptide often needed for pharmaceutical use.
There are techniques for modification of recombinant peptides. One such technique is C-terminal ~-carboxyl amidation, as described by Bongers et al., Int. J. Pe~tide Protein Res., 40:268 (1992) utilizing an ~ amidating enzyme as described in Henriksen et al., J
Am. Chem. Soc., 114:1876-1877 (1992); and Ohsuye et al., Biochem. BiophYs. Res. Commun., 150:1275-1281 (1988).
However, these techniques are limited to those modifications for which there exists a natural enzyme or chemical method capable of performing the desired modification.
Amidation of peptides has been performed through protease catalyzed replacement reactions (transpeptidation) using an amino acid amide or peptide amide as a nucleophile. Sahina et al., Chem. Pharm.
Bull., 36:4345-4354 (1988); Sahina et al., Chem. Pharm.
3Ull., 37:811-812 (1989); B~eddam et al., J. Pe~tide W095/20039 , 2 1 7 4 5 ~ 5 PCTIu~/OCC~2 ~

~ . . . .

Protein Res., 37:153-160 (1991). Yields using these techniques are typically quite low. However, transpeptidation reactions catalyzed by serine or thiol-proteases, under appropriate reaction conditions, have been carried out in high yields. Breddam et al. (1991) cited su~ra. Although protease catalyzed transpeptidation can be very effective under some circumstances, it is limited to substrates for which a natural protease exists and which exhibits specificity for a peptide bond close to the C-terminus.
Hence, there is a need to provide mutant protease enzymes capable of performing heretofor unknown N- or C-terminal modifications as well as peptide chain elongation with a variety of substrates, especially those substrates that are not reactive with the naturally occurring protease enzyme.

Sl~m~arY of the Invention These and other objectives are provided for by the method of the invention. The invention provides customized proteases (i.e., mutant proteases), methods of making customized proteases, as well as methods of using customized proteases.
Customized proteases are derived from known proteases such as endoproteases, exoproteases, serine proteases and cysteine proteases. A customized protease is a modified version of a known protease designed to provide a protease that is capable of transacylating a preselected substrate with a preselected nucleophile in a transacylation reaction not substantially catalyzed by the known protease. The mutant or customized protease can also exhibit improved or enhanced yields of transacylation. The preferred preselected substrates are peptides having an acidic or basic amino acid at the penultimate position. The preferred preselected nucleophiles are amino acids and amino acid derivatives such as amino acid esters and amino acid amides.

WO95/20039 PCT~S95106682 ~ 74525 The invention also provides methods for O preparing a customized protease. These methods can involve site specific mutagenesis or random mutagenesis.
Site specific mutagenesis can involve replacing a selected amino acid in the active site with a selected amino acid or by replacing the selected active site amino acid with any one of the 20 amino acids randomly.
Random mutagenesis can involve replacing any amino acid of the active site with any of the other 19 amino acids.
One method of the invention involves providing a DNA sequence that encodes the known protease, modifying at least one codon for at least one amino acid in the active site to form a mutant DNA sequence, and transforming a suitable host cell with the mutant DNA to provide for expression of the customized protease. In a preferred version, the codon for the amino acid in the active site is deleted using restriction enzymes and the deleted codon is replaced with an oligonucleotide encoding a different amino acid residue.
Another method of the invention involves modifying a DNA sequence encoding the known protease by inserting stop codons and/or a restriction enzyme recognition site at targeted sites to form a modified DNA sequence encoding an inactive protease. The targeted site preferably includes a codon for an amino acid in the active site which is replaced by the stop codon. A mutant DNA strand is synthesized and amplified by incubating the modified DNA strand in the presence of synthetic enzymes and oligonucleotides and a first degenerate oligonucleotide. The first degenerate ~ oligonucleotide contains a codon for a different amino acid in the targeted site and in place of the amino acid in the active site in the known protease. The mutant DNA strand is then selected and screened by detecting the presence of the customized protease.
The invention also includes a method of using customized proteases to modify a preselected substrate WO95/20039 PCT~S9S/06682 ~
2 1 7~525 by transacylation. The transacylation reaction catalyzed by the customized protease is preferably not substantially transacylated by the known enzyme. A
customized protease ins incubated with a preselected substrate and a preselected nucleophile to form a mixture. The mixture is incubated sufficiently to form a preselected transacylation product, preferably in high yield.
Customized proteases according to the invention can be utilized for post translation modification of recombinant peptides. The transacylation products produced are modified by substitution at the C-terminal or N-terminal end with various nucleophiles (including L-amino acids, D-amino lS acids, amino acid amides, amino acid derivatives, amino acid esters and radioactive amino acids or peptide derivatives including two or more amino acids of which the terminal amino acid is a natural amino acid or an amino acid derivative). It is understood that peptides produced by means other than recombinant technology can be transacylated according to the method of the invention.

Detailed Descri~tion of the Invention The invention provides customized proteases (i.e., mutant enzymes), methods of making the customized proteases, as well as methods of using the customized proteases.
The customized proteases of the invention are derived from known proteases and have transacylation capabilities differing from the known proteases.
Altered transacylation capabilities include the capability to perform transacylation reactions not substantially catalyzed by the known protease (i.e.
yields less than 10~) or the capability to perform transacylation reaction with improved or enhanced yields (i.e., 80 to 100~) or both. The mutant customized ~ WO9S/20039 - ` 2 1 7 4 5 2 5 PCT~S9S/06682 , proteases have been modified so that the protease can perform transacylation reactions with different preselected substrates and/or different preselected nucleophiles than the known protease. The mutant enzymes can also be specifically designed and selected to perform transacylation reactions with a specific preselected substrate and/or nucleophile. A known protease, preferably an exopeptidase, can be customized by replacement of amino acids in the active site so that the customized enzyme can transacylate different preselected substrates and/or nucleophiles. The customized protease can also exhibit enhanced or improved yields of transacylation. Preselected substrates are preferably those that have an acidic or basic penultimate amino acid. Preselected nucleophiles are preferably acidic or basic amino acid amides.
The present disclosure will utilize the following terminology. This terminology is applicable to proteases including both carboxy- and amino peptidases, as well as endopeptidases. Schechter et al., (1967).
The amino acids of the preselected substrates are designated by the letter "P". The amino acids of the substrate on the N-terminal side of the peptide bond to be cleaved by a protease enzyme are designated Pn - ~ P3, P2, Pl, with Pn being the amino acid furthest from the catalytic apparatus of the protease. Amino acids of the preselected substrate on the carboxy side of-the bond to be cleaved by the protease are designated P1', P2', P3 ' . . ~ Pn~ with Pn~ being the amino acid furthest from the catalytic apparatus of the protease. The bond which is to be cleaved by the protease is the P1-Pl' W095/20039 -- PCT/u~ 0~2 ~
2 ~ 74525 bond. Hence, the generic formula for the amino acids of the preselected substrate are as follows:

carboxypeptidase Pn---P3-P2-Pl-P1' amino peptidase Pl-Pl'-P2'-P3'---Pn' endopeptidase Pn~ ~ ~ P3 - P2 - Pl - Pl ~ - P2 ~ - P3 - - - Pn The "active site" of the protease is divided into a number of substrate binding sites and a catalytic apparatus. For example, the catalytic apparatus of serine proteases such as carboxypeptidase Y has a conserved catalytic triad of amino acids including serine, histidine and aspartic acid. The binding sites of the enzymes can include the S1 binding site, the Sl' binding site which includes the C binding site. For carboxypeptidases, the S1 binding site binds the side chain of the penultimate amino acid of the preselected substrate (P1), the S1' binding site binds to the side chain of the carboxy terminal amino acid (Pl'), and the C binding site binds the terminal ~-carboxylate group.
Modification of the active site preferably includes changes to amino acids in one or more of the binding sites.
The terminology for the substrate binding site of a protease is analogous to that for describing the amino acids of the preselected substrate except that the substrate binding sites of the protease are designated by the letter "S". The substrate binding sites for the amino acids on the N-terminal side of the cleaved bond are labelled as Sn . . . S3, S2 ~ Sl . The substrate binding sites for amino acids on the carboxy side of the cleaved bond are designated by "S'". These are labelled as S1~S2~...Sn~. Also, analogous to terminology of the amino acids on the substrate, the catalytic apparatus of the protease is understood to exist between the Sl and ~W095l20039 PCT~S95/06682 2 ~ 74525 .
. 7 S1' substrate binding sites. Hence, a generic formula for describing substrate binding sites of a protease is:

Sn ~ ~ ~ S2 ~ Sl--Sl --S2 ---- Sn The customized proteases of the invention are capable of transacylating substrates. As used herein, transacylation means that the enzyme can catalyze a reaction in which a leaving group is exchanged for a nucleophile. Transacylation reactions include transpeptidation reactions as well as peptide elongation reaction. Transpeptidation as used herein, occurs when single or multiple amino acids or amino acid derivatives act as a leaving group and the nucleophile is a single amino acid or peptide or amino acid derivative. Peptide elongation as used herein involves replacement of a single amino acid with a peptide. Transacylation reactions also include peptide elongation if the leaving group is an alcohol and the nucleophile is a single or multiple amino acid unit.
Customized proteases according to the invention can be utilized for post translation modification of recombinant peptides. The transacylation products produced are modified by substitution at the C-terminal or N-terminal end with various nucleophiles (including L-amino acids, D-amino acids, amino acid amides, amino acid derivatives, amino acid esters and radioactive amino acids or peptide derivatives including two or more amino acids of which the terminal amino acid is a natural amino acid or an ~ amino acid derivative). It is understood that peptides produced by means other than recombinant technology can ~ be transacylated according to the method of the invention.
3S The method of the invention provides for production of customized proteases through the process of site specific and/or random site mutagenesis. The WO9St20039 PCTIU~5,'~C~2 ~

invention further provides for selection and screening of suitably modified customized protease that is capable of catalyzing the preselected transacylation reaction.
Alternatively, the method of the invention provides a means of producing customized protease through the process of traditional mutagenesis. The invention also provides a process for utilizing the customized protease to transacylate a preselected substrate to form modified peptide products.
A. Customized Protease The invention provides for customized proteases. The customized proteases are derived from known proteases and have altered transacylating capabilities differing from the known proteases. As used herein, a customized protease is a modified version of a known protease designed to provide a protease that is capable of transacylating a preselected substrate with a preselected nucleophile in a transacylation reaction not substantially catalyzed by the known protease (i.e., less than 10~). The mutant or customized protease can also exhibit improved or enhanced yields of transacylation. Improved or enhanced yields are yields preferably increased to about 40 to 80~, preferably about 80 to 100~ over the yields of the known enzyme. Known proteases preferably include serine proteases, cysteine proteases and other endo- and exopeptidases. The especially preferred proteases are serine carboxypeptidases. Protease enzymes are capable of hydrolyzing preselected substrates as well as transacylating substrates in whlch the scissile bond is an ester and/or peptide bond.
The general reaction for transacylation is shown below:

~ WO 9S/20039 PCT/US~SIOC6~2 . ~ 2 ~ 7 4 5 2 5 AA2-X + N > AA~-AA2-N + X
le~ t~ d nucleophile transacylation leaving substrate product group Transacylation can occur at the N or C-terminal end of the preselected substrate. While not meant to be a limitation of the invention, it is believed that an enzyme that catalyzes transacylation preferably can bind to or otherwise accommodate the nucleophile. The desired product of the reaction is designated the trahsacylation product.
The general reaction for hydrolysis is:
exo-protease AAl-AA2-X > AAl-AA2-OH + X
substrate H2O hydrolysis leaving product group Yields of a desired transacylation product can depend on the (1) leaving group of the substrate, (2) the 25 nucleophile, and (3) the rates of competing side reactions for the particular substrate and nucleophile.
The competing side reactions can include (l) hydrolysis of the substrate, ( 2 ) hydrolysis of the desired transacylation product, (3) formation of various undesirable side products through transacylation, e.g., addition of nucleophiles to peptides originating from hydrolysis of the substrate, addition of nucleophiles to the desired transacylation product, polymerization of transacylation products, etc. Undesirable side products can include the hydrolysis product of the substrate, the hydrolysis product of the desired transacylation product, and transacylation product of the substrate with one or more of the hydrolysis products. The preferred customized protease of the invention can catalyze the desired transacylation reaction in high yields (i.e., preferably about 40 to 100~ and more W095/20039 PCT/U~95~0CC~2 ~
- - - 2 ~ 74525 preferably 80 to 100~), does not form substantial amounts of undesirable side products, and has a high affinity for the nucleophile. The preferred customized protease also does not substantially form undesirable transacylation side products, especially the transacylation product of a reaction of the hydrolysis product of the substrate with the substrate. As used herein, "substantially" means that the undesirable side products preferably are about 0 to 40~ of the yield and more preferably 0 to 20~ of the yield and most preferably 0 to 5~ of the yield. Some of the undesirable side products can be removed from the reaction mixture by modifying the side product with an antigenic capping agent and removing the capped products with an affinity column. Side products can be modified with antigenic capping agents in a manner similar to that described in U.S. Patent No. ~,049,656 issued September 17, l991, wherein a method for modifying undesirable side products in automated peptide synthesis is described.
A preselected substrate is preferably not substantially transacylated by the known protease. The term substantially as used herein, means the yield of the transacylation reaction with the known protease is preferably about 0 to 40~ and more preferably about 0 to 10~. The preselected substrate can be a naturally occurring peptide, a recombinant peptide, a synthetic , peptide or a peptide in which the C-terminal ~-carboxyl group has been esterified or otherwise modified. The preselected substrate has a core peptide connected to a leaving group at a terminus of the core peptide. The portion of the preselected substrate from which the leaving group is removed and to which the nucleophile is added is the core. Suitable leaving groups are amino acids, small peptides, or alcohols. The preferred leaving groups are small apolar or hydrophillic amino ~ WO95/20039 2 1 7 4 5 25 PCT~S9S,06682 .

acids as well as moieties linked to the peptide core by an ester bond.
The suitability of the preselected substrate is dependent on the substrate specificity of the protease as well as on the nature of the leaving group and the desired final products. While not meant to be a limitation of the invention, it is believed that the suitability of the leaving group is dependent on (1) the desired modification of the preselected substrate; (2) the substrate specificity of the customized protease;
and (3) the manner in which the leaving group binds to the customized protease.
A suitable preselected substrate for transacylation using a customized protease, according to the method of the invention, is of the general formula:
P-A

Wherein P represents the N-terminal or C-terminal core of the preselected substrate and A is the leaving group.
The leaving group (A) can be an amino acid, an amino acid amide, a peptide, a peptide amide, or an alcohol.
If A is an amino acid, amino acid amide, peptide or peptide amide, cleavage of A, from the core (P) is at the peptide bond. If A is an alcohol, cleavage of A
from the core is at the ester bond. Preferably, the preselected substrate is not a substrate that is transacylated by the known protease with high efficiency (i.e., yields of about 0 to 40~). More preferably, the preselected substrate has an acidic or basic penultimate amino acid. Suitable preselected substrates include growth hormone releasing factor (GRF) and derivatives thereof, calcitonin and derivatives thereof, and glucagon-like peptide-l (GLP-l) (SEQ ID NO:l).
Preferred preselected substrates are GRF (1-43)-Ala (SEQ
ID NO:2), GRF (1-24) (SEQ ID NO:3), and GLP-l (SEQ ID
NO:l).

W095/20039 .~ PCT/U~3S~C~6~2 ~
2 ~ 74525 A nucleophile, as used herein, is a molecule that donates a pair of electrons to an atomic nucleus to form a covalent bond. A suitable nucleophile can be an amino acid derivative, peptide derivative, ammonia or labelled compound which can be added to the core of the preselected substrate by the customized protease capable of substituting the leaving group for the nucleophile.
A suitable nucleophile can also include agents that can be converted to achieve the desired modification of the transacylation product. For example, photonucleophiles such as those described by Buckardt can be added to the substrate by transacylation with a customized protease and the resulting transacylation product can subsequently be converted to an amide by application of light. A suitable nucleophile can be preselected based upon (1) the desired modification of the final product;
and (2) the ability of the nucleophile to displace the leaving group on the preselected substrate. The preferred nucleophiles include amino acids and amino acid derivatives such as amino acid esters and amino acid amides.
Customized proteases are rendered suitable for a chosen transacylation reaction through modification of the known protease at the "active site." Modifications of the mutant or customized enzyme can be site specific mutations designed to alter the "active site" of the protease so that it can act upon different preselected substrates and/or nucleophiles than the known protease.
Modifications can include substitution, deletion, or insertion of one or more amino acids. The modifications can also be generated by random mutagenesis.
Some of the amino acids in the active sites of proteases are known to those of skill in the art. When the amino acids in the active site of a protease are not known, amino acids equivalent to amino acids in known binding sites of proteases can be identified using standard methods. These methods include identification ~ W095/20039 2 ~ 7 4 5 2 5 PCT/Ub~5~0'Ç~2 of equivalent amino acids by reference to the primary and/or tertiary structure of an enzyme in that class of proteases. For example, a reference enzyme for carboxypeptidases is wheat carboxypeptidase (CPD-WII).
The primary amino acid sequence and the crystal structure of CPD-WII are known (Liao and Remington, J.
Biol. Chem., 265:6528 (1990)) and can serve as reference points to identify equivalent amino acids in other carboxypeptidases.
The amino acid sequence and crystallographic structure of CPD-Y are known as well (Endrizzi et al., BiochemistrY, 33:11106 (1994)) and can similarly be used as reference points to identify equivalent amino acids in other carboxypeptidases.
One method that is applied to identify residues in the active site of a protease with an unknown tertiary structure is comparison of the amino acid sequence of the protease of interest with the amino acid sequence of a homologous protease with a known tertiary structure. Thus, this method can be used to identify amino acids in the protease of interest that are equivalent to amino acids in the active site of the homologous protease. For example, see Olesen et al.
Protein Enqineerinq, 6:409 (1993). Alternatively, the amino acids in the active site can be identified by determination of the tertiary structure using X-ray crystallography or NMR techniques.
In a preferred version, carboxypeptidase Y is modified by substitution of amino acids in the active site. These amino acids are preferably found in the S
or Sl' binding sites. Preferred amino acids of the S
binding site include Tyrl47, Leul78, Tyrl85, Tyrl88, Asn241, Leu245, Trp312, Ile340 and Cys341. Preferred amino acids of the S1' binding site include Trp49, Asn51, Gly52, Cys56, Thr60, Phe64, Glu65, Glu145, Tyr256, Tyr269, Leu272, Ser297, Cys298 and Met398. Amino acids W095/20039 PCT/U~5~G~Q2 ~
.., , ~ , . ~ . . . .
14 ~ 745Z5 equivalent to these residues of the binding sites of carboxypeptidase Y are also preferred.
Amino acid substitutions in the Sl binding site can preferably result in a mutant protease capable of transacylating a preselected substrate with a basic or acidic penultimate amino acid (Pl). Amino acid substitutions in the Sl' binding site can preferably result in a mutant protease capable of performing transacylation reaction on preselected substrates with large apolar amino acid leaving groups and/or nucleophiles such as large apolar amino acids, proline and proline amide. The especially preferred enzyme is a customized carboxypeptidase that has different amino acid residues in a position equivalent to amino acid residue 178 or 398 of carboxypeptidase Y. A preferred substituent amino acid is serine at position 178.
The preferred customized protease is a carboxypeptidase that is capable of transacylating a preselected substrate having an acidic or basic amino acid such as growth hormone releasing factor with a C-terminal alanine (GRF (1-43)-Ala) (SEQ ID NO:2) and arginine as the penultimate amino acid. The especially preferred mutant carboxypeptidase catalyzes formation of growth hormone releasing factor with a leucine amide.
Selection can also involve choosing the different amino acid that will be substituted into the active site. While not meant to be a limitation of the invention, one way the amino acid to be inserted into the active site can be selected is by predicting the effect on the binding interaction between the preselected substrates and the customized protease.
These substitutions can be conservative amino acid substitutions.
In general, the binding interaction between the substrate binding sites of the proteases and preselected substrate and/or nucleophile can directly affect the transacylation process. Although not ~ W095/20039 ,~ ~, t ~ r r~ 2 1 7 4 5 2 5 PCT~S95/06682 intended to be a limitation of the invention, it is believed that the interactions between the amino acids of the binding sites and the amino acids of the substrate effect formation of the enzyme substrate complex (ES) which is a first step to transacylation.
For serine or thiol proteases, enzyme substrate interactions lead to the formation of a tetrahedral transition state intermediate and subsequently to the cleavage of the peptide (ester) bond between the substrate core and leaving group. The resultant acyl-or thio-acyl species then undergoes nucleophilic attack (aminolysis) to form the transacylated product.
The binding interaction involved in formation of the ES complex include three major types:
electrostatic interactions, hydrogen bonds, and van der Waal forces. Amino acid substitutions within the substrate binding sites can cause changes in the transacylation capabilities of the protease through changes in the binding affinity of an enzyme for a substrate (i.e. formation of ES complex), through modification of the interaction of the enzyme with the transition state as well as through interactions affecting the rate of competing aminolysis and hydrolysis reactions. Amino acid substitutions that affect each of these stages can be predicted based upon the preselected substrate with leaving group and the preselected nucleophile. The preferred substitutions include the substitution of Asn51 with glutamine and Leul78 with serine in carboxypeptidase Y.
= 30 The mutant or customized protease can also be a protease that exhibits enhanced transacylating capabilities. Enhanced transacylating capability can be determined by determining an increase in the yield of the transacylation product. Preferably, the increase in 35 the yield is about 40 to 100~ and more preferably about 80 to 100~ increase over the yield catalyzed by the known protease. The preselected substrate can either be Wogs/20039 ~ PCT~S9S/06682 ~
2 t 74525 a substrate that can be transacylated by the known protease but at low yields (i.e., about 10-40~) or a substrate not substantially cleaved by the known protease (i.e., less than 10~).

B. Methods for Preparinq Customized Protease The method of the invention provides for preparing a customized protease derived from a known protease and that has a modified active site and that functions to alter the transacylation capability of the known protease. As used herein, the customized enzymes can be modified to transacylate a new substrate by mutating one of more amino acids in one or more of the substrate binding sites of a known protease. It is possible that mutation of as few as one amino acid in one substrate binding site can provide for transacylation of a substrate that was not a suitable substrate for the known protease. The active site can also be modified to provide an enzyme having enhanced transacylation capability, i.e., higher yields of transacylation.
Mutation of amino acids of the substrate binding sites can alter one or more functionalities which affect transacylation of a preselected substrate by a customized protease such as: (1) affinity of the customized protease for the core peptide portion of the substrate; (2) affinity of the customized protease for the leaving group or nucleophile; and (3) preference of catalysis of aminolysis over the competing hydrolysis reaction.

a. Providin~ a DNA se ence of known proteases to be mutated According to a method of the invention, mutation of a known protease to produce a customized protease can be accomplished througn site specific mutagenesis, random site mutagenesis and traditional mutagenesis. The first two methods require knowledge of WO95l20039 f ~ 7 4 5 2 5 PCT~S9S/06682 the DNA sequence of known proteases and the location of codons which code for the substrate binding site amino acids. Amino acids in the active site of proteases are either known to those of skill in the art or can be identified by analogy to known proteases as described herein. The corresponding DNA sequence encoding the known protease and the location of codons for amino acids in the active sites are either known to those of skill in the art or can be derived from the amino acid sequence. For example, the DNA sequence and restriction map of the PRC1 gene which encodes carboxypeptidase Y (CPD-Y) and a source of the DNA
sequence is described in Valls et al., Cell, 48:887-889 (1987). DNA sequences encoding known proteases can be obtained from an electronic database such as SwissProt, GeneBank, and EMBL. Once these sequences are identified, publications identifying vectors containing the DNA sequence can be located and used by those of skill in the art to prepare a customized protease.
If the DNA sequence and location of substrate binding sites is not available for a known enzyme, the putative DNA sequence can be derived from the amino acid sequence of the known protease. The amino acid sequence of the known protease can be used to prepare synthetic oligonucleotide probes. The probes can be used to identify DNA sequences encoding the known protease in suitable organisms by standard methods as described by Maniatis et al, A Guide to Molecular Cloninq (1989).
Once a DNA sequence encoding a known protease is isolated, codons corresponding to amino acids in the active site can be identified as described herein.
Active site amino acids can be identified by comparing the primary or tertiary structure to other known active sites of other proteases or by X-ray crystallography as described herein.

W095/20039 PCT~S9S/066~2 ~
.., ~ ,., ~ .,. ~ .

b. ModifYin~ amino acids in the active site Amino acids in the active site can be modified by modifying the codon encoding the amino acids in the DNA sequence encoding the known protease. Amino acids in the active site and the location of codons encoding these amino acids are either known to those of skill in the art or can be determined using standard methods.
Amino acids in the active site preferably include those found in the Sl, Sl' or C binding sites. The codon or codons encoding amino acids of the active site of the protease are included in a targeted site on the DNA
sequence. The targeted site includes the DNA sequence that is going to be mutated. One or more than one codon can be changed in the targeted site. Optionally, the targeted site can also include the DNA sequence surrounding the codon or codons for an amino acid in the active site. The DNA sequence of the targeted site surrounding the codon preferably includes about 3 to 9 nucleotides on either side of the codon or codons for amino acids in the active site. Modifications of codons include substitution, insertion or deletion of the codon.
The codons encoding amino acids of the active sites are preferably modified to encode a different amino acid than that of the known protease. In site specific modification, a selected amino acid in the active site can be changed either randomly or by substitution with a codon for a specific amino acid. In random site modification any number of codons can be modified by substitution with any number of amino acids.
The modification of the codon results in a customized protease with altered transacylation capabilities.
The preferred codons for modification are those that encode amino acids in the Sl or Sl' binding sites of carboxypeptidases. While not meant to be a limitation of the invention, it is believed that modification of amino acids in the Sl site can result in ~ W095/20039 ~ S i ~ 2 1 74525 PCTIU~95/OC~Y2 ; ~ .

a mutant protease capable of transacylating a preselected substrate with a basic or acidic penultimate amino acid (Pl). Likewise, it is believed that a modification of amino acids in the Sl' binding site can result in a mutant protease capable of transacylating a preselected substrate with an acidic or basic leaving group and/or amino acid amides as nucleophiles.
Choice of a specific amino acid substitution at a random or specific location can be based on the known or inferred mechanism of interaction of the binding site amino acid and the substrate. From this a rational inference is made, using knowledge of the properties of amino acids, of what amino acid substitution will provide the appropriate interaction to effect transacylation of the preselected substrate.
Known amino acid properties which may be considered when selecting specific amino acid substitutions include electronic and steric factors. For example, specific amino acid substitution selection may be based on pKa values (of ~-carboxyl and side chain hydrogens), amino acid side chain length, and amino acid side chain polarity at various pH. While not meant to be a limitation of the invention, the effect of the amino acid substitution can be predicted based upon the interactions involved in binding and catalysis as described herein for carboxypeptidase. Alternatively, if no data is available from which to make a rational inference of specific amino acid substitutions, random selection of amino acid substitutions can be made both with respect to the amino acids of the active site to be changed and the amino acid substitutions to be made.
According to the method of the invention, site specific and random site mutagenesis are used to mutate the known protease and can be accomplished through incorporation of an oligonucleotide containing a mutated or modified codon at the chosen or targeted codon location. Other methods of random and site specific W095/20039 re PCT/U~5S/OG6~2 ~
~ ~.. . . .
2 ~ 74525 mutagenesis can be employed as described by Maniatis, cited su~ra. Preferred methods of incorporation of the oligonucleotide into the DNA sequence encoding the known protease to produce a modiied DNA sequence include polymerase chain reaction (PCR) and standard cloning technology.
Oligonucleotides containing a mutated or modified codon can be obtained by standard methods.
These standard methods include synthesis by automated methods. Methods for automated DNA synthesis are known to those of skill in the art. The synthetic oligonucleotides are comprised of a variable and a constant region and preferably are about 20 to 60 nucleotides long. The length of the oligonucleotide is dependent on two main factors; (l) the number of variable regions the oligonucleotide is coding for; and (2) the length of the constant regions.
The variable region of the oligonucleotide contains the nucleic acid codons which code for the mutated amino acids of the substrate binding sites. The codons for amino acids are known to those of skill in the art. The variable region of oligonucleotide can be designed to include a codon for a specific amino acid or any number of random amino acids. Therefore, the minimum number of codons in the variable region is three, which represents the codon for a single amino acid. The codons of the variable region correspond to the location of the codons to be mutated in the known protease. The variable region is flanked by the constant region of the oligonucleotide. If an oligonucleotide contains more than one variable region, there are constant regions between variable regions.
The constant regions are necessary to incorporate the oligonucleotide into the customized protease gene and include codons corresponding to those of the known protease at that location (i.e., that are not mutated~.
The length o the constant region can depend on the 2 1 7 4 5 2 5 PCT~S9S/06682 r means by which the oligonucleotide will be incorporated into the customized gene and the number of amino acid modifications included in the variable region.
Preferably, the constant region includes about 3 to 50 nucleotides on either side of the variable region, and more preferably about 3 to 30 nucleotides on either side of the variable regions.
Once formed, the synthetic oligonucleotides are incorporated into the DNA sequence for the known protease in frame and at the targeted location. One way this insertion can occur is by cleavage with at least one appropriate restriction endonuclease so that the targeted site is deleted, followed by ligation of the synthetic oligonucleotide into the site that was deleted. Appropriate restriction endonucleases can be determined by ex~mln'ng the nucleotide sequence around the targeted site and by the size of the synthetic oligonucleotide to be inserted at the site. The recognition sequences of restriction enzymes are known to those of skill in the art, and an appropriate combination of enzymes can be readily selected by one of skill in the art.
In a preferred version, the codon for Asn51 of carboxypeptidase Y is modified to encode glutamine 51.
The PCR1 gene encodes carboxypeptidase Y and can be obtained from plasmid pTSY3 which has been deposited with the American Type Culture Collection in Rockville, MD on October 26, 1993 and given Accession No. 75580.
An oligonucleotide including a codon for glutamine at a site corresponding to the codon for Asn51 such as:
GGATCCGGTCATCCTTTGGTTG_AAGGGGGT (SEQ ID NO:4) (oligo - N5lQ) (underline indicates changes to the codon at the targeted site) can be synthesized by automated DNA synthesis. The BamHI fragment of PRCl includes the codon for amino acid -WO9S/20039 -- PCT~S9S/06682 ~
2 1 ~525 51. The DNA sequence of carboxypeptidase Y surrounding the codon for Asn51 can be deleted from the BamHI
fragment of PRCl with restriction endonucleases such as BstXI and SmaI. The synthetic oligonucleotide can then be introduced in place of the deleted DNA sequence at the SmaI-BstXI site of the BamHI fragment. The modified BamHI fragment is then inserted back into the entire DNA
coding sequence for carboxypeptidase Y using known methods to form a modified DNA sequence. The sequence of the modified DNA sequence can be confirmed using dideoxy sequencing methodology.
Once the modified DNA sequence is obtained, it can be introduced into a suitable host cell, selected and expressed to yield the customized protease with the modified active site and that functions to alter the transacylation activity of the known protease.
The modified DNA sequence is preferably incorporated into a vector to provide for selection and expression. Suitable vectors include the yeast bacterial shuttle vectors YEp24, pRA21~BAM, pYSP1, pTSY3, p~A21, and pYSP32. The modified or mutated DNA
sequence can be incorporated into the vectors by standard methods as described by Maniatis et al., cited su~ra, and Nielsen et al., A~pl. Microbiol. Biotechnol., 33:307 (1990).
Once combined with a vector, the vector is introduced into a suitable host cell for selection and expression. Suitable host sells include bacteria such as E. Coli and yeast such as S cerevisiae. Preferred host cells include S. cerevisiae strains having isogenic vpl mutations, delta-prcl mutations and ura3 mutations.
Especially preferred hosts are S cerevisiae strains that have vpl mutations resulting secretion of active CPD-Y as described in Nielsen et al., cited supra.
The preferred vector is a plasmid pTSY3 which is the yeast bacterial shuttle vector YEp24 with a 3.2 kb DNA insert containing the PRCl gene under the ~ W095/20039 PCT~S9S/06682 .. r v- -~ r ~ 1 74525 control of the PRC1 promoter. This plasmid has been deposited with the American Type Culture Collection, Rockville, MD, and given Accession No. 75580.
Suitable host cells are transformed by standard methods including transformation of calcium phosphate, calcium chloride or lithium acetate competent cells, microinjection, electroporation, and the like.
Transformed cells can be selected~based upon the presence of antibiotic resistance genes on the vector in the case of E. coli and based upon the presence of URA3 in the case of yeast. Transformed yeast cells can be screened for the production of mutant protease activity.
Transformed yeast cells producing mutant proteases can be screened by detecting the ability of the transformed cells to hydrolyze a peptide substrate using standard methods as described by Nielsen et al., cited suPra.
Mutant proteases that can perform transacylation reactions with a preselected substrate and/or nucleophile can be further selected by assaying for transacylation activity by standard methods including those described in Examples 2 and 3.
Once the transformed cells are selected and amplified, mutant proteases can be purified using standard methods such as high performance liquid chromatography and affinity chromatography as described in Example 1.
In an alternative version, a novel method of the invention involves mutagenizing a known protease to form a customized protease having altered transacylating capability. The basic technique of the method involves in vitro DNA synthesis primed by mutagenic degenerate synthetic oligonucleotides. The method provides efficient screening of a large population of mutant transformants which eliminates the wild type background due to unmutagenized plasmids in the subsequent functional screens.

W095/20039 PCT~S9S/06682 2 ~ 7 4 5 25 The steps of the method include providing a DNA sequence encoding a known protease. DNA sequences encoding known proteases are either publicly available or can be obtained by standard methods as described herein.
Once the DNA sequence encoding the known protease is obtained, one or more targeted sites are selected. A targeted site of the DNA preferably includes at least one codon for an amino acid of the active site to be modified as described herein. Once targeted sites are identified, the DNA sequence is modified at each targeted site by insertion of stop codons and optionally restriction endonuclease sites at the location to be mutated. Codons for stop codons are designated amber, ochre and opal, and the sequences of the stop codons are known to those of skill in the art.
The DNA sequences recognized by a restriction endonuclease are known to those of skill in the art.
The DNA sequence including the restriction endonuclease site can be adjacent to the stop codon or it can overlap with the stop codon. The oligonucleotide sequence inserted at the target site can be prepared by standard methods including automated DNA synthesis.
The inserted oligonucleotide sequence is preferably about 3 to 60 nucleotides long and can be inserted into one or more targeted site using standard methods such as in vitro DNA synthesis as described by Maniatis et al, cited suDra. Once the oligonucleotide having a stop codon and optionally restriction endonuclease site is inserted into a targeted site of the DNA encoding the known protease, a modified DNA
sequence encoding an inactive known protease is formed.
The presence of stop codons results in the expression of truncated forms of the known protease lacking activity.
The DNA sequence encoding an inactive known protease is introduced into a vector, preferably a phagemid vector. The vector is transformed into ~ WO95/20039 PCT~S~S/06682 suitable host cells such as E. coli for amplification.
Once amplified, the vector is isolated and single stranded DNA can be prepared. Optionally, the DNA
- sequence can be introduced into a vector carrying an inactive antibiotic resistance gene such as a gene - encoding ampicillin resistance that has a frameshift mutation. A preferred phagemid vector is the pYSP1.
A mutant DNA strand encoding the customized protease can be synthesized by incubating the single stranded DNA with one or more first degenerate oligonucleotides in the presence of DNA polymerase and DNA ligase. A first degenerate oligonucleotide has variable and constant regions as described previously herein. The variable region of a degenerate oligonucleotide includes at least one mutated codon for an amino acid in the active site of the known protease and that has been targeted. The mutated codon is found at the same location with respect to the surrounding DNA
as the codon for the amino acid in the known protease.
An oligonucleotide is degenerate if the mutated codons are randomly changed to encode any one of the 20 amino acids. A degenerate oligonucleotide at the mutated codon has the sequence of NNN wherein N corresponds to any one of the four nucleotides. A degenerate oligonucleotide preferably includes about 10 to 50 nucleotides on both sides of the mutated codon. The codons of the constant region correspond to the codons of the known protease at the targeted location in the known protease. Degenerate oligonucleotides can be for any of the 20 amino acids and are randomly generated using known methods and as described in Olesen et al., cited supra.
When the DNA sequence encoding the known protease contains more than one targeted site that has been modified by stop codons and/or restriction endonuclease sites, it may be des-rable to synthesize a mutant DNA strand that is modifiea at only one or some W095l20039 PCT~SsS/06682 ~
, . . ,. . ,. , ,~ . .
- ~1 74525 of the targeted sites. When mutation is desired only at some of the targeted sites having stop codons, the other targeted sites in the mutant enzyme should have the sequence of the known enzyme at these other targeted sites. One or more second oligonucleotides can be included in the DNA synthesis mixture that function to ensure that the other targeted sites that are not to be mutated in the mutant DNA strand have the sequence of the known enzyme as described in Olesen et al., cited suDra. The second oligonucleotides include codons corresponding to those in the known protease that have been replaced by stop codons and optionally restriction endonuclease sites at targeted sites. Each second oligonucleotide has the same sequence at a specific targeted site as the known protease A second oligonucleotide preferably has about 20 to 60 nucleotides.
The synthesis mixture can also optionally include an oligonucleotide that provides for repair of an antibiotic resistance gene. When the DNA sequence encoding the known protease is inserted into a vector, that vector preferably includes a mutated antibiotic resistance gene such as ampicillin resistance.
Including one or more oligonucleotides that can repair the mutation of the antibiotic resistance gene in the DNA synthesis mixture results in a mutant DNA strand that has a functional antibiotic resistance gene. An oligonucleotide that can repair the antibiotic resistance gene includes codons that provide for the correct DNA sequence at the relevant mutation in the resistance gene. These oligonucleotides are known to those of skill in the art or can be prepared by standard methods as described in herein.
Once synthesis of the mutant DNA is complete, desirable mutant enzymes encoded by the DNA can be selected and screened for the ability to act on a preselected substrate for alteraticn of transacylation ~ WO9S/20039 2 1 7 4 5 2 5 PCT~5~CB2 .

capability. The mutant DNA can be selected and amplified first in a suitable host cell such as E coli followed by introduction into a suitable host cell that can secrete proteases such as S. cerevisiae strains.
Selection methods for transformed cells include selecting for antibiotic resistance based upon the presence of an antibiotic resistance gene on a vector.
Methods for selection in yeast using selectable marker genes are known to those of skill in the art.
Transformed cells can be screened to identify cells having the customized protease with the desired functional activity. The desired customized protease can include a protease that can hydrolyze a preselected substrate having an acidic or basic penultimate amino acid, a protease that can catalyze transacylation reaction in which a preselected substrate is modified by a preselected nucleophile, and a protease with enhanced transacylation capabilities. The preferred customized protease is a carboxypeptidase that can modify a preselected substrate with a basic penultimate amino acid with a terminal leucine amide. One or more screening and selecting methods can be employed.
For example, mutant enzymes capable of hydrolyzing a preselected substrate are first selected and screened for hydrolysis activity and then those selected mutants are screened for transacylation activity. Substrate hydrolysis is used as a first level selection to ensure that the mutant is capable of acting on the preselected substrate. The mutant enzymes, capable of hydrolyzing the preselected substrate are then further screened for transacylating capability.
Selection of transformants expressing hydrolysis activity of the preselected substrate can be performed using the plate activity method or the color overlay method. For detection of low level customized proteases, the color overlay method is preferred.

W095/20039 ~ ~ PCT~S95/06682 According to the invention, transformed cells expressing customized proteases can be selected for preselected substrate hydrolysis activity using a plate activity method. This method of the invention utilizes a bacterial or yeast host cell which requires an amino acid for growth. The amino acid required for growth is provided to the transformed host cells as a C-terminal amino acid of a peptide. By culturing the transformants on a media deficient in the leaving group amino acid, only those transformants capable of releasing the leaving group from the peptide substrate can grow. This method is described in Olesen et al., Protein Enq., cited supra. For example, a vps strain of S. cerevisiae which requires Leu for growth is used to select for transformants expressing a mutant CPD-Y gene capable of cleaving a terminal Leu from a preselected substrate such as N-blocked-X-Leu-OH dipeptides (X = Glu, Gly, Ser, His, Pro, Trp, or Lys). Only cells which express a protease which can release the terminal Leu from these substrates can grow on these plates.
A preferred method for selection of transformants which express a customized protease capable of hydrolyzing a preselected substrate is the color overlay method. According to this method, transformant colonies are overlaid by agar containing a chromogenic substrate which reveals customized protease activity. The chromogenic substrate turns color upon reaction with the product formed by catalytic action of the customized protease. In a preferred version, the transformed cells are incubated with a preselected peptide or amino acid substrate. The preselected peptide or substrate such as N-acetyl-L-alanine ~-naphthyl ester is acted upon by transformed cells expressing a customized protease and the reaction product is detected by simultaneously overlaying the transformed cells with a chromogenic agent such as WO95/20039 PCT~S9S/06682 ~ 1 7 4 5 2 5 .

Garnet Red that changes colors upon exposure to the reaction products.
Customized proteases which have been mutated and screened for the capacity to act on a preselected substrate can be further screened for transacylation capability. One method that can be employed to screen for transacylation capability is that described in Examples 2 and 3. Customized proteases are purified from the transformed cells by methods known in the art and as described in Examples 2 and 3. The purified customized protease is mixed with the preselected substrate under conditions favorable for transacylation such as pH of about 7 to 9.5 and in the presence of a suitable nucleophile. Transformants which produce customized proteases capable of catalyzing transacylation of the preselected substrate can be identified by following the appearance and/or amount of the desired reaction product by standard methods.
In a preferred version, the PRC1 gene encoding carboxypeptidase Y is modified by insertion of DNA
se~uences at target sites as follows:
TAA GCT TCC (SEQ ID NO:5) at Tyrl47 ochre stop codon HindIII
TGA ATT CCT (SEQ ID NO:6) at Leul78 opal stop codon EcoRI
TAG CCC GGG TGT (SEQ ID NO:7) at Glu214, Arg216 amber stop codon SmaI
TAA ATC GAT ACC (SEQ ID NO:8) at Ile340, Cys341 ochre stop codon ClaI

The target sites in carboxypeptidase were selected based on homology to carboxypeptidase-WII. The modified DNA
sequence encodes an inactive carboxypeptidase Y. The DNA sequence as modified is amplified as phagemid vector pYSP32 in E. coli. The preferred vector also contains an inactive frame shifted ampicillin resistance gene.

WO95/20039 PCT/U~7S~0C~2 2 1 7452~

Single stranded DNA of pYSP32 can be generated by standard methods.
The single stranded DNA can be incubated with up to four first degenerate oligonucleotides, each 20 to 60 nucleotides long and containing degenerated codons at the center. A preferred first degenerate oligonucleotide for position 178 includes the sequence:
GGC NNN ACT
wherein NNN is situated at the center of the oligonucleotide and corresponds to the location of the codon for amino acid 178. NNN is a codon for any of the 20 amino acids. Degenerate oligonucleotides can be synthesized by automated synthesis.
One or more second oligonucleotides can be included in the preferred synthetic mixture and in which case they each replace one of the first oligonucleotides. The second oligonucleotide includes codons for amino acids at positions 147, 178, 215 and 216, and 340 and 341 of carboxypeptidase Y, as follows:
TCC TAC GCC (SEQ ID NO:9) Tyrl47 TTG GAA CGT TGT (SEQ ID NO:10) Glu215, Arg216 TTC ATC TGT ACC (SEQ ID NO:11) Ile340, Cys341 GGC CTC ACT (SEQ ID NO:12) Leul78 The presence of these second nucleotides ensures the mutant DNA strand is synthesized without stop codons at other targeted sites. One preferred mutagenesis mixture contains a first degenerate oligonucleotide for position 178 and a second oligonucleotide for each position of Tyrl47, Glu215, Arg216, Ile340 and Cys341.
Optionally, the preferred DNA synthesis mixture also contains one or more third oligonucleotides, ampicillin repair oligonucleotides.
The repair oligonucleotides provide for synthesis of a mutant DNA sequence having a functional ampicillin ~ WO95/20039 ~ 2 1 7 4 5 2 5 PCT~S9SI06682 ,. ,~ . . . ..

resistance gene operably linked to the DNA sequence encoding the mutant or customized protease.
The mixture of single stranded DNA and oligonucleotides is incubated in the presence of precursor nucleotides, DNA polymerase and DNA ligase.
Mutant DNA sequences encoding the customized protease are formed and can optionally be linked to a functional ampicillin resistance gene. The mutant DNA is introduced into a suitable mismatch repair deficient bacteria for amplification and selection and subsequently introduced into yeast for screening. The preferred suitable yeast cell is a S. cerevisiae strain that has a vps mutation, and/or requires at least one amino acid for growth. Especially preferred S.
cerevisiae strains include W2579, K2579LLR and JHRY20-2C~3.
The transformed E. coli cells can be selected first for the ability to grow in the presence of ampicillin. Subsequently, transformants are further selected for the ability to grow on a medium deficient in the amino acid required for growth and in the presence of the peptide substrate containing the amino acid for growth at the C-terminus. For example, a host cell that cannot normally grow without leucine is transformed with mutant DNA and plated onto leucine deficient medium supplement with a preselected peptide having the following formula:
PnPl-A
where A is leucine and P1 is an amino acid not sufficiently acted on by known protease such as any one of Glu, Gly, Ser, His, Pro, Trp or Lys. Transformants that can grow on leucine deficient medium supplement with a preselected polypeptide such as Pn-Lys-Leu can act on the preselected polypeptide to release leucine, thereby providing a source of leucine for growth. These expressed proteases are then screened for the ability to transacylate a preselected polypeptide such as Pn-Arg-Ala W095/20039 ~ ~- ` 2 1 7 4 5 2 5 PCTIu~5S~OCC~2 where Pn is growth hormone releasing factor and leucine amide.
Customized proteases which favor aminolysis over hydrolysis can be produced using traditional mutagenesis. According to this method of the invention, mutation of the amino acid composition of a known protease is accomplished by subjecting DNA or cells containing a vector encoding a DNA sequence for known protease to a mutagenic agent such as W light, nitrosoguanidine, ethylmethyl sulfonate, bisulfite, dimethyl sulfate, formic acid, hydrazine, hydroxylamine, methoxylamine, nitrous acid, potassium sulfate, and others. Methods of traditional mutagenesis are known in the art and are described, for example: for chemical in 15 vitro mutagenesis: Myers et al., Science, 229:242-247 (1985); Hayatsu, Methods EnzYmol., 45:568-587 (1976);
Shortle et al., Methods EnzYmol., 100:457-568 (1983);
Kadonga et al., Nucl. Acids. Res., 13:1733-1745 (1985);
Busby et al., J. Mol. Biol., 154:197-209 (1982); and 20 Loeb, Cell, 40:483-484 (1985); for nucleotide misincorporation: Kornberg, ~NA Re~lication, Freeman, San Francisco (1980) at page 724; and Kunkel et al., J.
Biol. Chem., 254:5718-5725 (1979); for inco~poration of nucleotide analogs: Flavel et al., J. Mol. Biol., 25 89:255-272 (1974); Dierks et al., Cell, 22:659-706 (1983)1 Dodson et al., PNAS, 79:7440-7444 (1982); Eadie et al., Nature, 308:201-203 (1984); Grossberger et al., PNAS, 78:5689-587 (1981); and Mott et al., Nucl. Acids Res., 12:4139-4152 (1984).
The mutated vectors are tnen incorporated into a suitable expression system and the expressed customized enzymes are selected ana screened. The methods for selecting and screening customized protease produced by traditional mutagenesis can be performed as described above for transformants produced by site specific and random site mutagenesis. Plate activity and color overlay selection can be utilized to select ~ W095/20039 / .~.~ 2 1 74525 PCT/u~ 2 for those transformants which express a customized protease capable of acting on a preselected substrate.
Those customized proteases capable of acting on a preselected substrate are purified and mixed with a preselected substrate and nucleophile under conditions favorable for transacylation to screen for enzymes capable of catalyzing the transacylation of the preselected substrate.

c. Method of transacylating substrates usinq customized ~roteases.
The invention also provides a method for using the customized enzyme to transacylate a preselected substrate with a preselected nucleophile. This method is useful to add nucleophiles such as D-amino acids, modified amino acids and radiolabelled amino acids to the termini of recombinantly produced peptides to form transacylation products. This method can also be applied to naturally occurring or synthetic peptides to form useful analogs or derivatives.
The customized protease of the invention is designed to either enhance transacylation capabilities (i.e., yields) or act on a preselected substrate and/or nucleophile poorly accepted by the known proteases. The customized protease can be prepared and selected by the methods described herein. The preselected substrate is selected depending on the desired transacylation product. The preselected substrate is preferably not substantially transacylated by the known proteases.
"Substantially" in this context means that the yields of transacylation with the preselected substrate and with a particular nucleophile are preferably about 0 to lO~
- under standard conditions. The preselected substrate preferably has a basic or acidic amino acid as the penultimate amino acid. The preferred amount of the preselected substrate depends on the substrate specificity of the enzyme for the substrate as well as the solubility of the substrate in the chosen solvent WO9S/20039 PCT~S9S/06682 ~
~ t ~5~5 and is about 0.2 to 10 mM. A preselected substrate can be a naturally occurring peptide, a synthetic peptide or a recombinantly produced peptide.
The preselected nucleophile is preferably not an effective nucleophile with the known protease. The nucleophiles are preferably amino acids, radioactively labelled amino acids, and amino acid amides.
Nucleophiles can be prepared by standard synthetic methods known to those of skill in the art such as described in Breddam et al., Int. J. Peptide Res., 37:153-160 (1991). A preferred amount of a nucleophile also depends on the affinity of the enzyme and solubility of the nucleophile in the chosen solvent and is about 10 mM to 2M.
Reaction conditions resulting in high yields of the desired product can vary with a given enzyme substrate system. Reaction conditions can be altered to minimize degradation and polymerization of the products.
Such side reactions may, when using ester substrates together with a serine carboxypeptidase, be avoided by increasing the pH above 8.0 when aqueous solvents are employed. Alternatively, side reactions can be avoided by conducting the reaction in an organic solvent.
Transacylation can be performed in aqueous buffer solution. Preferred buffer solutions include 50mM HEPES and 5mM EDTA, pH 7.5 or 50mM CHES and 5mm EDTA, pH 9.5. It is of importance that the chosen buffer is unable to act as a nucleophile in the transacylation reaction. The preferred pH for transacylation using an alcohol leaving group is preferably about pH 6.5 to 9.5 and more preferably pH
7.5 to 8.5. The preferred pH for transacylation using an amino acid or peptide derivative leaving group is preferably about pH 5.5 to 8.5, and more preferably about preferably pH 6.5 to 7.5.
The production of the trGnsacylation product is monitored by HPLC or other appropriate analytical ~ W095/20039 2 ~ 7 4 5 2 5 PCT~S5~/0'6~2 technique. The reaction can be stopped by additlon of an acidic solution to bring the pH of the reaction mixture down to about pH 1 to 3. Alternatively, the reaction can be stopped by addition of an enzyme inhibitor such as phenyl methane sulfonyl fluoride (PMSF), or diisopropyl phosphoryl fluoridate (DFP). The transacylation product can be separated from the reaction mixture by reverse phase chromatography, hydrophobic interaction chromatography, ion exchange chromatography, or HPLC.
Alternatively, the transacylation reaction can be performed in organic solvents for those enzymes substrate systems capable of functioning in organic solvents. Suitable organic solvents for the transacylation reaction include dimethyl sulfoxide (DMSO), N,N'-dimethylacetamide and other similar solvents. The methodology for transacylation in organic solvents is described in Bongers et al., Int. J. PePtide Protein Res., 40:268 (1992).
In a preferred example for transpeptidation using an amino acid leaving group in aqueous solution, the preselected peptide substrate, GRF (1-43)-Ala (SEQ
ID NO:2~, is dissolved in a 5~ solution of acetic acid.
The nucleophile, leucine amide, is dissolved in 50 mM
HEPES, 5 mM EDTA to a final concentration of 500 mM.
25 ~l of a 40 mM solution of GRF (1-43)-Ala (SEQ ID
NO:2) is added pr. 950 ~l of nucleophile solution and the pH is add to 7.5 at 20C. The customized protease is added to the mixture in 25 ~l of water pr. ml 30 solution, resulting in an enzyme concentration of about 0.002 to 0.07 mg/ml. The reaction is followed by HPLC
and is stopped when no additional product is formed by the addition of one volume of 2.5~ trifluoracetic acid.

W095/20039~~ - . PCT/u~9~/OG6B2 ~
2 ~ 74525 Site Specific Mutagenesis of CarboxYPePtidase Y Active Site 5The three-dimensional structure of carboxypeptidase Y suggests that the side c~; n.~ of Trp49, Asn51, GlyS2, Cys56, Thr60, Phe64, Glu65, Glu145, Tyr256, Tyr269, Leu272, Ser297, Cys298 and Met398 are important in the active site of the enzyme. These amino acid residues were mutated by site specific mutagenesis to form enzymes with single, double, or triple mutations.
The following mutants of carboxypeptidase Y
have been constructed using standard methods:
Asn51 has been replaced with: Alanine (N51A) Cysteine (N51C) Glycine (N51G) Valine (N51V) Aspartic Acid (N5lD) Glutamic Acid (N51E) Glutamine (N5lQ) Serine (N5lS) Threonine (N5lT) Glu 65 has been replaced by: Alanine (E65A) Glutamine (E65Q) Glu145 has been replaced by: Alanine (E145A) Glutamine (E145Q) Serine (E145S) Asparagine (E145N) Aspartic Acid (E145D) Trp49 has been replaced by: Phenylalanine (W49F) In addition, enzymes having more than the mutation as shown below have been constructed:
N51A + E145A
E65A + E145A
N51A + E65A + E145A
E65Q + E145Q
The general methodology used to construct the mutant enzymes is by site specific mutagenesis using the polymerase chain reaction.

W095/20039 , 21 74525 PCT~S9S/06682 For example, a plasmid pUC-~30 was constructed by inserting a 1112 bp BamHI fragment of the PRCl gene from pYTS3 containing the coding region for all amino acid residues involved in the formation of the active site into the unique BamHI site in the polylinker of pUCl9. Yanisch-Perron et al., Gene, 33:103 (1985);
Stevens et al. Yeast Cell Bioloqy, Ed. J. Hicks, New York, Alan R. Liss, at pages 519-536 (19 _). The orientation of the fragment was opposite of the lacZ
gene. pUC-~30 contains unique BstXI, EcoRI, NaeI and SmaI restriction sites which can be used in cloning and mutagenesis procedures.
The mutations W49F and N51A were made by the polymerase chain reaction (PCR) (Innes et al., 1990) in a Perkin Elmer Cetus DNA Thermal Cycler using a Gene Amp kit (Perkin Elmer Cetus) on pUC-~30 with GTTTCTGTCCTTGTGAGACAAAATTTCAGA (SEQ ID NO:13) (oligo wtll) and with either GGATCCGGTCATCCTTTTCTTGAACGGG (SEQ
ID NO:14) (oligo W49F) or GCAAAGGA~CCGGTCATCCTTTGGTTGGCAGGGGGCCA (SEQ ID NO:15) (oligo N51A). Nucleotides underlined are different from wild-type. Cleavage with BstXI allowed insertion of the PCR fragment into a SmaI-BstXI vector fragment of pUC-~30.
The mutation E145A was made by PCR with GCAAGGCGATTAAGTTGGGT (SEQ ID NO:16) (oligo pUCl9 spl) and GGCGTAGGAAGCCCCAGCGAT (SEQ ID NO:17) (oligo E145A) on pUC-~30. Cleavage of the PCR fragment with EcoRI
allowed introduction into a NaeI-EcoRI vector fragment of pUC-~30.
The mutations E65A and N51A+E65A were produced by fusion of two overlapping PCR fragments using either pUC-~30 or pUC-~30-N51A as template. Fragment 1 was generated with CTGTTCTTTGCATTAGGACCC (SEQ ID NO:18) (oligo E65A) and (oligo wtl) and fragment 2 with (oligo pUC19sql) and (oligo E145A). An additional PCR reaction was performed on the fused fragment with oligo pUC19sql W095/20039 ~ 4 5 2 ~ PCT/U~S~GC82 ~

and oligo wtl. The resultant fragment was cut with EcoRI and BstXI, thus, removing the unwanted mutation on position 145, and ligated into a pUC-~30 vector fragment cut with the same restriction enzymes.
N51A+E145A, E65A+El45A, and N51A+E65A+E145A
were made by proper combination of the above listed mutations exploiting the EcoRI site in the polylinker and exploiting that BstXI cleaves between position 65 and 145.
The mutated sequences were introduced into the PRCl gene by transferring the mutated 1112 bp BamHI
fragment into the vector pRA21~Bam. The fragment inserted into pRA21~Bam was controlled for the absence of any non-silent secondary mutation by sequencing using the Taq Dye-Dideoxy~ terminator cycle sequencing kit and the model 373A DNA-sequencing system from Applied Biosystems, USA.
Site directed mutagenesis on position 51 and 145 was performed using polymerase chain reaction and restriction endonuclease cleavage as described herein.
The following oligonucleotides were used in the mutagenesis reactions:

GCAAAGGATCCGGTCATCCTTTGGTTGGACGGGGGCCA (SEQ ID NO:19 (oligo N51D), GGATCCGGTCATCCTTTGGTTG_AAGGGGGCCA (SEQ ID NO:20) (oligo N51E), GGATCCGGTCATCCTTTGGTTGCAGGGGGT (SEQ ID NO:21) (oligo N51Q), GGATCCGGTCATCCTTTGGTTGAGTGGGGGT (SEQ ID NO:22) (oligo N51S), GGATCCGGTCATCCTTTGGTTGACTGGGGGT (SEQ ID NO:23) (oligo N5lT), GGCGTAGGAATCCCCAGCGAT (SEQ ID NO:24) (oligo E145D), GGCGTAGGAATTCCCAGCGAT (SEQ ID NO:25) (oligo E145D) GGCGTAGGATTGCCCAGCGAT (SEQ ID NO:26) (oligo E145Q), GGCGTAGGATGACCCAGCGAT (SEQ ID NO:27) (oligo E145S).

~ WO95l20039 ~ , PCT~S9S/06682 2~ 74525 Underlined nucleotides are different from wild-type.
All fragments generated by the PCR reaction were ligated into pUC-~30 after cleavage with the appropriate restriction enzymes such as EcoRI (E145) or BstXI
(Asn51). The absence of any non-silent secondary mutation was confirmed by sequencing using the Taq Dye-Dideoxy~ terminator cycle sequencing kit and the model 373A DNA-sequencing system from Applied Biosystems, USA.
The mutant enzymes containing cysteine (N51C), valine (N51V), or glycine (N51G), glutamine (E65Q), and asparagine (E145N) were prepared in a similar manner.
Once the DNA sequence encoding the PCR enzyme was mutated as described herein, it was transferred into the vector pRA21~Bam. This vector was then introduced into S. cerevisiae strain (W2579(prc~LEU2 ura 3-51, pep 4-3 ) as described by Ito et al. J. Bacteriol., 153:163-168 (1983) using lithium acetate. The stability of plasmids in stationary-phase yeast cells was determined by plating culture samples on YPD medium (1 yeast extract [Difco], 2~ peptone [Difco] and 2~
glucose) and subsequent replica-plating of about 100 colonies to SC medium lacking uracil. Yeast cultures were grown in MU/pro medium plus 50 mM potassium phosphate.
Mutant enzymes were purified from a one liter culture grown under the conditions previously described.
(Nielsen et al., 1990). Growth media containing secreted mutant enzyme was adjusted to pH 4.4 with concentrated acetic acid and then directly applied to the GYBS-Sepharose affinity column according to the procedure of Johansen et al. (1976). N51A, which did not bind effectively to GYBS-Sepharose, was subjected to diafiltration against 10 mM NaH2PO4, pH 7.0, using a Pellicon system (Millipore) and then purified by ion exchange chromatography on a DEAE Fractogel 650 column (2.6 x 6 cm). The column was washed until A280 was below W095/20039 PCT/u~5S~0(~8~ ~
. ~
2~ 74525 0.01 and elution was accomplished with a linear salt gradient from 0 to 0.5 M NaCl in 10 mM NaH2PO~, pH 7Ø
The elute was concentrated-using an Amicon cell and applied to a Sephacryl-S300 column (1 cm x 100 cm) equilibrated with 50 mM NaH2PO4, pH 7Ø Fractions with constant specific activity were pooled, concentrated and dialyzed against water. All enzyme preparations were stored frozen in water at -18C.
The purity of the mutant enzymes was ascertained by SDS-PAGE on 12.5~ homogeneous gels using the PhastSystem from Pharmacia. The concentration of CPD-Y mutants was determined spectrophotometrically using A2~0 (1 mg/ml = 1.48) (Johansen et al., (1976) cited supra).
Once prepared, the mutant or customized enzymes can be evaluated for a change in the transpeptidation reaction.

~XAMPLE 2 Carboxypeptidase Y Mutants With Improved Characteri~tics in Transacylation Reac~ion~
Usin~ Amino Acids as NucleoPhiles The binding site in carboxypeptidase Y (CPD-Y) for the negatively charged C-terminal carboxylate group of peptide substrates has been identified using site directed mutagenesis as described herein. While not meant to be a limitation of the invention, it is believed that the carboxylate group of the peptide substrate binds to the side ch~inq of Asn51 and Glu145 in the Sl' binding pocket. Both side chains can act as hydrogen bond donors. The side chains of Asn51 and Glu145 appear to be oriented by hydrogen bonds with Glu65 and Trp49 which, therefore, have an indirect function in the binding of the carboxylate group of peptide substrates.
Serine carboxypeptidases also catalyze the hydrolysis of peptide esters and this activity increases with pH and remains constant in the pH range 7 to 9.5.

~ W095/20039 ;~ 2 1 7 4 5 25 PCT~S9S/06682 Thus, at basic pH, the esterase activity is high and the peptidase activity is low. These unique properties, combined with an ability to catalyze transacylation reactions with amino acids or amino acid derivatives as nucleophiles, suggest that serine carboxypeptidases can be useful in peptide synthesis. However, many substrates and nucleophiles cannot be catalyzed by known enzymes. With amino acid methyl esters as nucleophiles, medium yields (40-80~) are obtained but this method is complicated by the risk of further elongation of the product (oligomerization). It would be valuable to use amino acids rather than amino acid amides as nucleophiles.
It has been demonstrated that mutant carboxypeptidase enzymes can bind to and catalyze peptide elongation with amino acids as nucleophiles in higher yields. With amino acids as nucleophiles, yields exceeding 60~ are obtained in a few cases but yields of 10-40~ are much more common and H-Pro-OH, H-Glu-OH and H-Asp-OH are not accepted as nucleophiles. Thus, the yields obtained with amino acids as nucleophiles are rarely satisfactory. The low yields with amino acids as nucleophiles are not due to degradation of the product since the reaction is carried out at basic pH where the peptidase activity is very low (see above), thus, securing accumulation of the peptide product in the reaction mixture.
Mutants of carboxypeptidase Y were examined for the capacity to transacylate certain substrates 30 using amino acids as nucleophiles. Some amino acid substitutions in the active site of mutant - carboxypeptidase Y enzymes were also made knowing that they were not likely to improve the yields of transacylation reactions, but rather to investigate the mechanism of action of the binding and catalysis with certain types of nucleophiles.

W095/20039 ~ i~ 452~ PCT~s~slorc~2 ~

CPD-Y was obtained from Carlbiotech, Copenhagen, Denmark. Amino acids and buffers were from Bachem, Switzerland or Sigma, USA. The mutants E65A, E65Q, E145A, E145Q, E61A+E145A, E65Q+E145Q, N51S, N51Q, N51A, and N51A+N145A were prepared as described in Example 1. The purity of the enzymes was ascertained by SDS polyacrylamide gel electrophoresis.
Aminolysis reactions were carried out in the following way. The nucleophile was dissolved in 50 mM
Hepes, 5 mM EDTA and pH was adjusted to 7.5. Five (5) ~l of substrate (8 mM FA-Ala-OBzl in methanol) was added to 190 ~l nucleophile solution followed by 5 ~l enzyme, resulting in a substrate concentration of 0.2 mM. For reactions carried out at pH 9.5, Hepes was replaced with Ches. During the reaction, 20 ~1 aliquots were removed from the reaction mixture and added to 50 ~
trifluoroacetic acid to quench the reaction. The reactant composition was determined by HPLC using a Waters HPLC equipped with a C-18 Waters Novapac 4 reverse phase column and various gradients of acetonitrile in 0.1~ trifluoroacetic acid. The separation was monitored at 302 nm allowing the direct quantification of the products from the integrated peak areas. The composition of the reaction mixture was determined at least twice during the reaction, the first time when 20-50~ (preferably 35~) of the ester substrate had been consumed in the reaction and the second time when 50-90~ (preferably 80~) of the substrate had been consumed. The products were collected and identified by amino acid analysis after acid hydrolysis using a Pharmacia Alpha Plus analyzer.
Further identification was obtained by co-chromatography of authentic standard compounds. The fraction of aminolysis (fa) was expressed as the ratio between the formed aminolysis product and the sum of all products being formed, i.e., unconsumed substrate was disregarded in the calculations. The KN(app~, representing W095/20039 ~-4~- t~ 's" _ 2 1 7 4 5 2 5 PCT~S9S/06682 the nucleophile concentration at which fa is half the maximum value (a measure for the dissociation constant of the nucleophile), and fa~x (the highest possible fa obtained at saturation of the enzyme with nucleophile) were determined by fitting the values of fa obtained at a minimum of seven concentrations of nucleophile to a saturation. The value of fa obtained at the highest possible nucleophile concentration is designated fa~at.
The ability of CPD-Y to catalyze transacylation reactions with amino acids or amino acid derivatives acting as nucleophiles in competition with water can be studied. A study of the beneficial effects of structural alterations within the Sl' binding site on the productive binding of amino acids to effect aminolysis is preferably performed with an ester substrate. Transacylation reactions should preferably be performed at slightly basic pH to m~;m; ze the esterase activity and m;n;~;ze the peptidase activity.
When amino acids are used as nucleophiles the product peptide is very slowly degraded by the enzyme and, as a consequence, it accumulates in the reaction mixture, FA-Ala-OBzl is hydrolyzed at very high kC~t/~
by CPD-Y and the prepared mutant enzymes allowing the use of low concentrations of enzyme (0.5 ~g/ml). The fact that the peptide products are hydrolyzed at much lower kCat/~ prevents degradation of the aminolysis product. At pH 7.5, with H-Val-OH as added nucleophile, two products were formed: FA-Ala-OH (hydrolysis) and FA-Ala-Val-OH (aminolysis). At each concentration of nucleophile, the fraction undergoing aminolysis reaction was constant with time and independent of the concentration of substrate remaining in the reaction mixture. This indicates that the product FA-Ala-Val-OH
is not hydrolyzed by the enzyme under the employed reaction conditions and, thus, the ratio of the two products reflect the relative rates of the two competing reactions. With increasing concentrations of H-Val-OH, WO95l20039 ~ 2 ~ 7~25 PCT~S95/06682 ~

fa increased but it did not exceed 0.32. The correlation between the fraction of aminolysis and concentration of H-Val-OH was unaffected by the presence of 0.5 M NaCl in the reaction medium. Hence, the ratio of rates of the hydrolysis and aminolysis reaction is independent of the ionic strength of the reaction medlum .
The observation that CPD-Y becomes saturated with H-Val-OH is consistent with the previous demonstrations that nucleophiles bind to the acyl-enzyme intermediate of serine carboxypeptidases prior to the deacylation reaction. The correlation between fa and the concentration of nucleophile is consistent with KN~app~
and famax values of 13 mM and 0.32, respectively. The reaction was also carried out at pH 9.5 and the values for KN(app~ and fa~x were 19 mM and 0.31, respectively, suggesting that the ratio of the two forms of the nucleophile (amino/ammonium) has little influence on the synthesis parameters.
The influence of the side chain of the amino acid nucleophile on famax and KN(app~ was investigated.
KN(app) decreased in the following order:

H-Gly-OH ~ H-Ser-OH ~ H-Lys-OH > H-Gln-OH
H-His-OH ~ H-Ala-OH ~ H-Val-OH > H-Leu-OH
H-Phe-OH.
(See Table I). with H-Pro-PH and H-Glu-OH, no aminolysis product was obtained. Thus, hydrophilic amino acids bind much less efficiently than hydrophobic ones. The highest fa~x values were obtained with the amino acids that bind the least efficient: with H-Gly-OH, H-Ala-OH, H-Ser-OH, and H-Arg-OH, the famax ~ values exceeded 0.8, whereas with the hydrophobic and efficiently binding H-Val-OH, H-Leu-OH and H-Phe-OH, values below 0.4 were obtained.

~ W095/20039 ~ ,. 2 l 7 4 5 2 5 PCT~S9S/06682 TABLE I
CPD-Y Catalyzed Aminolysis of an Ester Substrate Using A~; n~ Acids as Nucleophiles The Influence of Amino Acid Structure on the Parameters fo~ the ~m~ n~ly8i8 Reaction Nucleophile fa8~t fam~ KN(~PP) H-Gly-OH 0.600.92 + 0.01 1500 + 50 H-Ser-OH 0.780.87 i 0.01 310 + 20 H-Lys-OH 0.790.81 i 0.03 190 i 30 H-His-oH 0.200.27 i 0.01 87 ~ 9 H-Gln-OH 0.640.90 i 0.02 150 i 8 H-Glu-OH 0 n.d. n.d.
H-Ala-OH 0.850.91 i 0.02 83 i 7 H-Val-OH 0.300.32 i 0.01 13 + 1 H-Leu-OH 0.320.35 i 0.01 10 i 1 H-Phe-OH 0.140.15 i 0.01 9.5 + 1 H-Pro-OH 0 n.d. n.d.

All reactions were performed at pH 7.5 as de~cribed herein. The values for KN~PP) are in mM. The concentrations at which fa5a~ was obtained were:
H-Gly-OH = 2.9 M, H-Ser-OH - 3.2 M, H-Lys-OH = 2.7 M, H-His-OH ~ 0.25 M, H-Gln-OH ~ 0.37 M, H-Glu-OH = l.l M, H-Ala-OH = 1.4 M, H-Val-OH - 0.48 M, H-Leu-OH = 0.14 M, H-Phe-OH = 0.14 M and H-Pro-OH = N.D.

With amino acid amides as nucleophiles, famax consistently exceeds 0.85 and is essentially independent of the hydrophobicity of the side chain.
While not meant to be a limitation o~ the invention, the low values observed with some amino acids might be explained by the interaction between enzyme and the ~-carboxylate group of the amino acid when it is combined with tight binding of its side chain to the W095/20039 ~ PCT~S9S/06682 enzyme. The interaction between enzyme and ~-carboxylate group of some amino acid nucleophiles apparently had an adverse effect on fa~ suggesting that alteration of this interaction could have a beneficial effect. Amino acid nucleophiles could occupy a position similar to that of the Pl' amino acid residue of peptide substrates. If this is the case, then the amino acids in CPD-Y involved in the binding of the C-terminal carboxylate group of peptide substrates would also be important for the binding of the ~-carboxylate group of amino acid nucleophiles. The binding of the C-terminal carboxylate group of peptide substrates is dependent on hydrogen bonds from the side chain of Asn51 and Glu145, the latter with the carboxylic acid group in its protonated form, and the optimal position of these two side chains is secured by hydrogen bonds from Glu65 and Trp49. However, this is the situation at low pH where the enzyme efficiently binds peptide substrates: at pH
7.5 where the aminolysis reactions were carried out, Glu145 is in its deprotonted negatively charged form which cannot participate in the binding of the ~-carboxylate group. In fact, this negative charge may substantially influence the binding mode of the amino acid nucleophile. Hence, the transacylation reactions might be influenced by mutational replacements of Asn51 and Glu145 and possibly also, due to indirect effects, by replacement of Glu65.
CPD-Y mutants with replacements at positions 51, 65 and 145 were investigated for their ability to catalyze transacylation reactions using H-Val-OH and H-Leu-OH as nucleophiles (Table II).

~ W095t20039 ' ! ~ `. 2 1 7 4 5 2 5 PCT~S9S/06682 TABLE II
Carboxypeptidase Mutants with Improved Characteristics of Acylation .

H-Val OH H-Leu-OH
Mutant fa~ K~,pp, fa~ K~(~pp\

Wild-type0.32 + 0.0113 ~ 10.35 + 0.01 10 ~ 0.8 N51A+El45A 0' n.d. 0' n.d.
NSlA 0.02' n.d. 0' n.d.
N51S 0.05~ n.d. 0.03' n.d.
N51Q 0.97 + 0.0261 + 3 0.96 ~ 0.01 42 + 1 E145A 0.48 + 0.01 79 + 6 0.30 + 0.01 40 + 2 E145Q 0.68 + 0.02 41 i 4 0.66 + 0.06 29 + 8 E65A 0.91 + 0.01 69 + 2 0.96 + 0.01 42 + l E65Q 0.91 + 0.02 75 + 4 0.97 + 0.01 56 + 1 E65A+E1~5A > 0.8> 500 > 0.8 > 0.8 E65A+E145Q 1.00 + 0.03 240 + 15 > 0.8 > 0.89 All reactions were performed at pH 7.5 as described herein.
The values for K~(,pp~ are in mM.
A The low values prevented the demonstration of saturation of the enzyme and consequently they must be regarded as fa~,~ values.

The complete removal of the hydrogen bond donating capacity of the side ch~i ns at positions 51 and 145 by incorporation of Ala at both positions, i.e. N51A+E145A, abolished the capacity of the enzyme to accept amino acids as nucleophiles. The same result was obtained with the single mutant N51A suggesting that Asn51 is important for the binding of amino acid nucleophiles prior to the attack on the acyl-enzyme intermediate.
- However, with the other single mutant E145A, significant aminolysis was obtained and this is consistent with the notion that Glu145 at pH 7.5 plays a minor role in the binding of amino acid nucleophiles. Nevertheless, with H-Val-OH and H-Leu-OH as nucleophiles, KNt~PP~ was 4-5 W095/20039 .~ PCT~S9~10C6~ ~
` 2 ~ ~45~

times higher using E145A as compared with the wild-type enzyme.
It is common to all the positions 65 and 145 mutants, i.e. E65A, E65Q, E145A, E145Q, E65A+El45A and E65Q+E145Q, that KN~app~ with H-Val-OH and H-Leu-OH
increased by a factor of 5 to 10. Simultaneously, an increase in famax was observed except with E145A. With E65A and E65Q, the fa~x values exceeded 0.90 and, as a result, it is possible with these enzymes to obtain much higher yields when hydrophobic amino acids are used as nucleophiles. Although not meant to be a limitation of the invention, Glu65 and Glu145 do not appear to be directly involved in the binding of the ~-carboxylate group of amino acid nucleophiles at the pH where the reaction is carried out, they seem to exert an indirect influence such that their replacement affect the binding mode, as indicated by the elevated famax values. When amino acid amides are used as nucleophiles, the presence of the negatively charged Glu145 does not appear to exert a negative effect since the famax values consistently are high.
Asn51 was replaced with other amino acid residues and these enzymes were tested. Replacing Asn51 with Ser or Gln affected famax in different directions.
With N51S and H-Val-OH and H-Leu-OH as nucleophiles, fam~x was 0.05 and 0.03, respectively (Table II). With N51Q, they were 0.97 and 0.96, respectively, and with the wild-type enzyme, famax was 0.35 with both nucleophiles.
With FA-Phe-OMe as substrate and H-Val-OH as nucleophile, the same influence of the group at position 51 on famax was observed (data not shown). The kinetic parameters for the hydrolysis of FA-Phe-Val-OH by these mutants are: kCat = 1500 min~l, KM = O . 085 mM, kCae/KM = 18000 min-l mM~l (N51Q); kCa~ = 8100 min~l, ~ =
O.17 mM, kCat/KM = 48000 min~l mM-l (N51S) as compared with kCat = 6500 min~l, KM - 0.047 mM, k~t/KM = 140000 min~l mM-l _ W095/20039 . - PCT~S9Sl06682 ~ ~ 74525 ., ,.~

(wild-type). Normally, in serine protease catalyzed peptide bond hydrolysis, the acylation step is rate-limiting. However, this is not the case in CPD-Y
catalyzed hydrolysis of FA-Phe-X-OH substrates (X =
amino acid residue) since the kCae values show only little dependence on the structure of X. Thus, the rate of the deacylation step exerts an influence on KCae. The kCae values for the hydrolysis of FA-Phe-Val-OH increase in the order N51Q ~ wild-type < N51S and this correlates inversely with fam~x values with H-Val-OH as nucleophile which decrease in the order N51Q > wild type ~ N51S.
The results in Table II show that, among the mutants tested, the highest fa~x values were obtained with N51Q and, as a consequence, this enzyme was deemed the most suitable supplement to the wild-type enzyme as a catalyst in this particular type of peptide synthesis.
This was further investigated by determination of the KN~app~ and fam~ values for a number of different amino acid nucleophiles using 51Q (Table III) and the results may be compared to those obtained with the wild-type enzyme. For H-His-OH and the hydrophobic amino acids, the fa~x values were significantly higher with N51Q as compared with the wild-type enzyme. However, with some amino acids, the Asn51 Gln had no or only small effect on fa~x, i.e., those where famax with the wild-type enzyme was zero or exceeded 0.9. With all nucleophiles, KN(app~
was significantly higher with N51Q as compared with the wild-type enzyme and this had the consequence that famAx with some amino acids could not be reached due to limited solubility of the nucleophile. The value of fa obtained at the highest possible concentration of nucleophile, i.e. fa8ae, has been listed in Table III and this value represents the highest possible yield that can be obtained in aqueous solution at pH 7.5. A
comparison of the fa8at values obtained with the wild-type enzyme leads to the conclusion that N51Q is WO9S/20039 ~ r , PCT~S~5~ 2 ~- r ~ ~ 7 7 4 ;) ~

superior to the wild-type enzyme with the hydrophobic amino acids as nucleophiles.

~WO95/20039 z .. `~ ; :' 2 1 74525 PCT/US9S/06682 ~ o o o a. m N ~a~ 0 r ,1 ~ ~
Z
+l +l +l +l +l ~ +l +l ~ +l o o o r o a ~ ~7 o u~
o ~ ~ 0 In 0 ~1 ~1 ~1 ~( t`') ~ N N ~1 ~1 ~1 O O O O O O O O O
~U X . . . . . . . .
O O O O O O O O O
~ +, +~ +, +~ +~ +~ +, +, +, V O ~ N r ~I r o ~I N IS) 111 V~j ~, ~ ~ 0 0 N a~ a~ ~ ~ ~

~ ~1 ~ 3 ~
V o ` O 0 a~ o d' LS ) O (~ ~
~ ~ a ~O r r N ~ 0 ~
a ~ ~ o o o o o o o o o o o H ~ Z ~ o d~ O O O
H m~ O ~ ~ cr~

I-- ~ G o o o o o O r~l (~
~ N O ~D N ~ 151 ~D ~r t~
ao~, ~ A ~1 u~ U~ N
.,1 ~ O

o o o o o o o o 40J ~ ~ o o o o o o o o n ~ ~ o +l +l +l +l ~ ~ +l +
L p ~x A r r ~ O o~ r ~ a~
n~ E CD C~ U') O, a ~V w O O O ~ O O O O
~i ~
o .a ~ ~D r ~ ~ ) a:) r r ,~ . . . . . . . . .
O o o o o o o o o o o ~ O O O O O O O O O O O
G
a) s s~

WO95/20039 ~ ~ -- PCT~S9510~2 Alteration of Leaving Group Preference for Mutant Carboxypeptida~e Y
Catalyzed Transpeptidation Reactions With ester substrates, serine carboxypeptidases exhibit no dependence on the nature of the leaving group. However, with peptide substrates where an amino acid acts as leaving group, this is not always the case. With carboxypeptidase Y, the most commonly employed serine carboxypeptidase, high yield of transpeptidation is only achieved when the leaving group is a hydrophilic amino acid. However, since carboxypeptidase Y in hydrolysis reactions exhibits a preference for hydrophobic amino acid leaving groups (Pl' amino acids), it would be beneficial, due to higher rate and specificity of the reaction, if such leaving groups were permissible in transpeptidation reactions as well.
This would also permit modification of peptides and proteins, as isolated from natural sources, which presently are excluded due to hydrophobic C-terminal amino acid residues. The low yields due to the described leaving group dependence has prevented further development of a method for incorporation of labelled amino acid residues into peptides and proteins with the purpose of facilitating amino acid sequence determinations.
Mutants of carboxypeptidase Y were examined for the capacity to enhance yields of transacylation with substrates having leaving groups that are not hydrophilic. Some amino acid substitutions made in the active site of mutant carboxypeptidase Y enzymes were also made knowing they were not likely to improve the yields of transacylation reactions, but rather to investigate the mechanism of action of the leaving group dependence.
Carboxypeptidase Y was obtained from Carlbiotech, Copenhagen, Denmark. All amino acids/amino acid derivatives were purchased from Bachem, Switzerland. The mutations in the structural gene for WO95/20039 PCT~S9S/06682 ~ ¢ ; 21 74525 carboxypeptidase Y were carried out as described in Example I. The mutants N51S, N51Q, E65A, and E145A were isolated as described in Example I and their purity was ascertained by SDS polyacrylamide gel electrophoresis performed on a Pharmacia Phast-system. The substrates FA-Ala-Gln-OH, FA-Ala-Arg-OH, FA-Ala-Lys-OH, Hippuryl-L-Phe-OH, and Hippuryl-L-~-Penyllactic acid were purchased from Bachem, Switzerland. The preparation of the following substrates were carried as previously described: FA-Ala-OBzl, FA-Ala-Ala-OH, FA-Ala-Val-OH, FA-Ala-Leu-OH, and FA-Ala-Phe-OH.
Breddam et al., Carlsberq Res. Comm., 49:535 (1984).
Aminolysis reactions were carried out in the following way. The nucleophile was dissolved in 50 mM
HEPES, 5 mM EDTA and pH was adjusted to 7.5, 5 ~l substrate (8 mM FA-Ala-OBzl or FA-Ala-Xaa-OH in methanol) was added to 190 ~l nucleophile solution followed by 5 ~l enzyme diluted in buffer to the appropriate concentration, resulting in a substrate concentration of 0.2 mM. During the reaction, 20 ~l aliquots were removed from the reaction mixture and added to 50 ~ trifluoroacetic acid to quench the reaction. The reactant composition was determined by HPLC using a Waters HPLC equipped with a C-18 Waters Novapac 4~ reverse phase column and various gradients of acetonitrile in 0.1~ trifluoroacetic acid. The separation was monitored at 302 nm allowing the direct quantification of the products from the integrated peak areas. The products were collected and identified by amino acid analysis after acid hydrolysis (Pharmacia Alpha Plus analyzer) and by co-chromatography of authentic standards. The fraction of aminolysis (fa) was expressed as the ratio between the formed aminolysis product and the sum of all products being formed, i.e.
unconsumed substrate was disregarded in the calculations. The KNapp (a measure for the dissociation constant of the nucleophile) and fa~ax (the highest W095/20039 , PCT~S95/06682 . . . ~ , t ~
2 ~ 7~525 possible fa obtained at saturation with nucleophile were determined by fitting the obtained fa values to the equatiOn fa = famax/ (l+KN,app/N) The enzymatic hydrolysis of FA-Ala-OBzl and the peptide substrates was followed on a Perkin Elmer A9 spectrophotometer. The substrates were dissolved in methanol at a concentration of 0.5 to 8 mM. A total of 25 ~l of substrate solution was added to 955 ~l of 50 mM
HEPES, 5 mM EDTA, pH 7.5, to give initial substrate concentrations in the cuvette, s0 of 1.25 to 200 ~M.
The cleavage of the substrate was followed with time by monitoring the decrease in absorbance at 337 nm and from this the initial rates of hydrolysis v0 was determined.
For FA-Ala-OBzl, the kCat/KM value was determined by fitting to the following form of the Michaelis-Menten equation: v0 = eO (kcat/~)/(l/KM + 1/so). For the peptide substrates, the kCat~ values were determined by fitting to the equation v0 = (kcat/KM)eOs0~ which is valid when so~KM. All fits of the parameters kCat/~, famax and KN app were performed using the Grafit program (Leatherbarrow, R.J., Grafit version 3.0, Erithacus Software Ltd., Staines, U.K., 1990).
While not meant to be a limitation of the invention, it is believed that in CPD-Y catalyzed hydrolysis reactions, a substrate acylates the essential serine residue which subsequently is deacylated by water, hence completing the hydrolysis reaction. When an amine nucleophile, e.g., an amino acid or amino acid amide, is added to the reaction mixture, the acyl-enzyme will be partitioned between water and the added amino component, in the latter case forming a new peptide bond (transpeptidation).
The saturation curves can be described by equation (1).
famax fa =
KN,app (1) 1 +
N

~ WO 95/20039 , , , 2 1 7 4 5 2 5 PCT/US5S~
r ~

The upper limit of the fa, termed fa~x, is reached when N~KN,~pp. fa~x can reach a maximum value of 1. This is, however, rarely obtained in practice. The concentration at which fa~x/2 is reached, termed KN app/
describes the affinity of the nucleophile for the enzyme and the dissociation of the aminolysis product.
Peptide esters may function as substrates and, in this case, reaction with the amine component causes elongation of the peptide. The ratio of the hydrolysis to aminolysis reaction is not influenced by the nature of the alcohol leaving group, but that there is a pronounced effect of the nucleophile on fa~x. Hence, fa~x values ranging from 1.00 with H-Gly-NH2 to 0.15 with H-Phe-OH as nucleophile are observed.
When a peptide functions as substrate, in which case the C-terminal amino acid residue is exchanged with the added amino component, the hydrophobicity of the amino acid leaving group exerts a pronounced influence on the hydrolysis/aminolysis ratio.
20 With H-Val-NHz or H-Gly-NH2 a nucleophile and FA-Ala-XAA-OH (XAA = H-Arg-OH, H-Lys-OH, H-Gln-OH, H-Ala-OH, H-Val-OH, H-Leu-OH, and H-Phe-OH) as substrates, the observed fa~x values reached 1.00, 0.99, 0.93, 0.94, 0 . 22, 0 .16, and 0.056, respectively (see Table IV). The 25 differences in fa~x suggests that the rate of dissociation of the amino acid leaving group is comparable to that of hydrolysis of the acyl-enzyme with the leaving group bound. Furthermore, they coincide with the observation that the fa~x values obtained with the ester substrate and H-Arg-OH, H-Lys-OH, H-Gln-OH, H-Ala-OH, H-Val-OH, H-Leu-OH, and H-Phe-OH as nucleophiles are 0.87, 0.81, 0.90, 0.91, 0.32, 0.35, and 0.15, respectively, and thus similar to the fa~x values obtained when these amino acids act as leaving groups (see Table IV).

WO95/20039 .~ PCT~S9S/06682 ~
2 ~ 7452~

TABLE IV
Influence of the ~;no Acid Nucleophile/Leaving Group On the fa~ Values Obtained with Wild-Type CPD-Y

FA-Ala-OBzl + FA-Ala-Xaa-OH
H-XA-OH H-Xaa-OH H-Val/Gly-NH2 H-Gln-OH 0.90 + 0.03 0.93 + 0.01' H-Lys-OH 0.81 + 0.03 0.99 + 0.01 H-Arg-OH 0.87 + 0.01 1.00 i 0.02b . H-Ala-OH 0.91 i O.02 0.94 + 0.01b H-Val-OH 0.32 i 0.01 0.22 + o.01b H-Leu-OH 0 35 i 0.01 0.16 + o.01b H-Phe-OH 0.15 i 0.01 0.06 + 0.01b H-Gly-NH2 was used as nucleophile and b H-Val-NH2 was used as nucleophile.
Using FA-A-OBzl as substrate, the observed fa~ was 1.00i0.01 with H-Gly-NHz and 0.98~0.01 with H-Val-NH2.

It has previously been suggested that low yields in aminolysis reactions were due to degradation of the products, a theory based on the fact that any product of an aminolysis reaction will be a substrate for CPD-Y. However, in the reactions studied here with FA-Ala-OBzl as substrate, the kCle/K~ for the hydrolysis of the products, under the conditions used in the aminolysis reactions, are 50-100 fold lower than that of the substrate (data not shown). Consequently, no degradation of the aminolysis procucts is detected with FA-Ala-OBzl as substrate and FA-Ala-Xaa-OH (Xaa =
Ala-OH, Val-OH, Leu-OH, Val-NH2) as product.
In order to be able to alter yields obtainable with a specific amino acid acting as leaving group/nucleophile, it is helpful to know what contributes to low famax values. Wrile not meant to W095/20039 PCT~S~5/OGC~2 ~ 2 1 74525 limit the invention, it is believed that the wide range of results with amino acids as leaving groups indicates that the magnitude of the rate constants which enter the expression for fa~x are associated with the nature of the amino acid side-chain. It appears that the more hydrophobic the leaving group, the lower the famax. A
significant increase in famax can be achieved by structural alterations within the binding pocket for the side-chain of the Pl' amino acid residue. Alternatively, modification of the interaction between the ~-carboxylate group of the leaving group or nucleophile and the binding site for the C-terminal carboxylate group (Asn51, Glu65 and Glu145) will influence famax.
Alterations within this region result in CPD-Y
derivatives with changed fa~x values in transacylation reactions with FA-Ala-OBzl as substrate and various amino acid nucleophiles.
Aminolysis reactions were carried out with the substrates Hippuryl-L-Phe-OH (peptide bond, the N-terminal of the leaving group is a H2N- group) and Hippuryl-L-~-Phenyllactic acid (ester bond, the N-terminal of the leaving group is a HO- group), using H-Gly-NH2 as nucleophile. These two reactions gave us a possibility to evaluate whether the nature of the Phe side-chain carboxyl-terminal or the amino group determines the low famax value observed with Phe. The aminolytic parameters with phenylalanine or L-~-Phenyllactic acid, respectively, as the leaving group and H-Gly-NH2 as nucleophile were determined: famax were found to be 0.33+0.01 and 0.96iO.01, respectively, and KN app to be 0.63+0.06 and 0.47~0.05, respectively. Thus, these reactions suggest that the nature and positioning of the ~-amino group contributes to the low yield obtained with some amino acids.
While not meant to be a limitation of the invention, it appears that the conformation in which an amino acid binds within the ~l' site may facilitate or WO95/20039 PCT~S5Sl0CC8~ _ 2 ~ 7~52S

restrain the access of water to the acyl component covalently attached to the essential serine residue (Serl46) due to the positioning of the amino group of the leaving group/nucleophile. It should, thus, be possible to achieve increased fa~x values if the binding mode of a specific amino acid within the Sl' binding site is altered to prevent nucleophilic attack of water on the acyl-enzyme.
The reaction FA-Ala-Xaa OH + H-Val-NH2 ~
FA-Ala-Val-NH2 + H-Xaa-OH was studied with these mutants.
With the mutant N51Q, the fa~x values were 0.98, 0.91, and 0.80 with Xaa = Ala, Val, Leu, respectively.
Similar results were obtained in the reaction FA-Ala-OBzl + H-Xaa-OH (Xaa = Ala, Val, Leu, and Phe) ~
FA-Ala-Xaa-OH + HOBzl (see Table V), where the fa~x values with Xaa = Val and Leu are 4-5 fold higher than those obtained with the wild-type enzyme. The observation that N51Q exhibits increased fa~x values in transpeptidation reactions is probably due to changes in the binding of amino acids within Sl'. The mutation results in increased reaction with amine relative to water. This result would be consistent with a shorter distance between the acyl-enzyme and the ~-amino group of the nucleophile/leaving group.

~ W095/20039 , 2 1 7 4 5 2 5 PCTlu~5S;~6~2 TABLE V
Influence of the Amino Acid Nucleophile on fa~x and RN,~PP Values Using FA-Ala-OBzl as Substrate and CPD-Y and Various Mut~nts Enzyme Nucleophile fa~xK~,,pp K~,~pp/fa~x (mM) (mM) Wild type H-Ala-OH 0.91 + 0.0283 + 7 91 H-Val-OH 0.32 + 0.0113 i 1 43 H-Leu-OH 0.35 i 0.0110 + 1 29 H-Phe-OH 0.15 i 0.0110 + 1 N51S H-Ala-OH 0.56 i 0.05140 + 4 250 H-Val-OH n,d,bn,d,b n.d.
H-Leu-OH n.d.bn.d.b n.d.
H-Phe-OH n.d.bn.d.b n.d.

N51Q H-Ala-OH 0.99 + 0.01250 + 15 250 H-Val-OH 0.97 + 0.0261 + 2 63 H-Leu-OH 0.96 ~ 0.0142 i 1 44 H-Phe-OH 0.89 + 0.0329 + 1 33 E65A H-Ala-OH 0.98 + 0 02280 + 20 290 H-Val-OH 0.91 ~ 0.0169 + 2 76 H-Leu-OH 0.96 ~ 0.0145 + 2 47 H-Phe-OH 0.89 + 0.0331 i 1 35 E145A H-Ala-OH n.d.' n.d.' 7100 H-Val-OH 0.48 + 0.0179 + 6 160 H-Leu-OH 0.30 + 0.0140 ~ 2 130 H-Phe-OH 0.31 + 0.0121 + 1 68 n.d. Not determ.ined due to:
b low fa values in the ~mined concentration range or K~pp values much larger than N.

In contrast to this, introduction of a serine at position 51 results in fa~x values that are reduced to approximately 50~ of the wild-type value (Tables V
and VI).

W0 95/20039 ~ 2 1 7 4 5 2 5 PCT/ub9s;o C~

TABLE VI
Influence of the Sub~trate Leaving Group on fa~ and R~,~p Values With H-Val-NH2 as Nucleophile and CPD-Y and Various Mutants EnzymeNucleophile fa~ K~,pp R~,pp/fa~
(-mM) (m Wild type FA-Ala-OBzl0.98 + 0.011.7 _ 0.1 1.7 FA-Ala-Ala-OH 0.94 _ 0.011.8 _ 0.1 1.9 FA-Ala-Val-OH 0.22 _ 0.011.5 _ 0.1 6.6 FA-Ala-Leu-OH 0.16 + 0.011.6 + 0.1 9.9 FA-Ala-Phe-OH 0.06 _ 0.011.1 _ 0.1 19.6 N51S FA-Ala-OBzl 0.97 + 0.01 1.6 _ 0.1 1.7 FA-Ala-Ala-OH 0.88 _ 0.011.9 _ 0.1 2.2 FA-Ala-Val-OH 0.13 + 0.011.8 _ 0.1 13.9 FA-Ala-Leu-OH n.d.' n.d.' n.d FA-Ala-Phe-OH n.d.' n.d.~ n.d N51Q FA-Ala-OBzl 0.96 + 0.01 2.1 _ 0.4 2.1 FA-Ala-Ala-OH 0.98 _ 0.013.1 _ 0.1 3.2 FA-Ala-Val-OH 0.91 _ 0.012.4 _ 0.1 2.6 FA-Ala-Leu-OH 0.80 _ 0.012.4 _ 0.1 3.0 FA-Ala-Phe-OH 0.77 _ 0.013.1 + 0.1 ??

E6sA FA-Ala-OBzl 0.97 _ 0.01 4.2 _ 0.1 4.4 FA-Ala-Ala-OH 0.96 _ 0.025.2 _ 0.5 5.5 FA-Ala-Val-OH 0.67 + 0.015.7 _ 0.3 8.5 FA-Ala-Leu-OH 0.71 + 0.015.2 _ 0.1 7.4 FA-Ala-Phe-OH 0.66 + 0.014.7 _ 0.1 7.1 E145A FA-Ala-OBzl 0.98 + 0.013.4 _ 0.04 3.5 FA-Ala-Ala-OH n.d. b n.d. b n.d.
FA-Ala-Val-OH 0.80 + 0.014.2 + 0.2 5.2 FA-Ala-Leu-OH 0.65 _ 0.014.1 _ 0.1 6.3 FA-Ala-Phe-OH 0.49 _ 0.014.1 _ 0.1 8.4 n.d. Not determined due to:
~ low fa values in the ~mi n~ concentration range or b considerable degradation of the product.

W095/20039 .~:~ ` PCT~S5S/OC~'d2 This is consistent with the fact that no products are observed with FA-Ala-OBzl as substrate and valine, leucine or phenylalanine as nucleophile and with alanine fam~x was reduced from 0.91 to 0.56 (Table V). Thus, with N51S the decreased famax values observed with FA-Ala-OBzl are also found with FA-Ala-Xaa-OH as substrate, hence repeating the pattern found with N51Q, i.e., that similar changes in famax are found with either leaving group. While not meant to be a limitation of the invention, this might be due to the serine in the mutant being shorter than the asparagine and subsequent steric changes in the active site that result in more rapid hydrolysis. The results from the aminolysis reactions are in good agreement with those obtained from steady state kinetics, which suggest N51 interacts with the C-terminal ~-carboxylate group of peptide substrates.
The characteristics of the mutant E61A are quite identical to those of N51Q. Hence, it exhibits an increase in famax in the reaction: FA-Ala-OBzl + H-Xaa-OH
(Xaa = Val, Leu and Phe) ~ FA-Ala-Xaa-OH + HOBzl to approximately 0.9 (see Table V) as well as in the reactions where these amino acids act as leaving groups.
In the latter reactions, famax reaches 0.67, 0.71 and 0.66 with -Val-OH, -Leu-OH and -Phe-OH, respectively.
Again, we observe similar changes in famax with the same amino acid as leaving group/nucleophile, the same pattern as previously was found with the mutants on position N51.
With E145A, famax was increased from 0.22 to 0.890 with valine as leaving group and from 0.16 to 0.65 with leucine. But with valine as nucleophile, famax is - only increased from 0.35 to 0.48, and with leucine reduced from 0.35 to 0.30. Thus, E145A does not show a similar increase in famax with the same amino acids as nucleophiles. In this way the characteristics of E145A
deviate from those of N51S, N51Q and E65A. The fact W095/20039 PCT~S95/06682 ~ ~n 2 ~ 7 4 5 2 5 that with E145A the fa~x values in reactions with FA-Ala-OBzl and FA-Ala-Xaa-OH deviate is consistent with to E145 not being involved in the binding of the carboxyl group of an amino acid nucleophile.
From these mutants it has been demonstrated that it is possible to alter the leaving group preference of CPD-Y, hence increasing its potential use in various transpeptidation reactions.

Alteration o~ Nucleophile Specificity By Mutant CarboxYPePtidase The characterization of a series of mutationally altered derivatives of this carboxypeptidase Y (CPD-Y) has delineated the nature of the interaction between the C-terminal carboxylate group of the substrate and the enzyme. While not meant to be a limitation of the invention, it is believed that hydrogen bonds from the side c~i n.~ of Asn51 and Glu145 appear to be responsible for the binding of the C-terminal carboxylate group of peptide substrates. The peptidase activity of CPD-Y is optimal at acidic pH.
CPD-Y also catalyzes the release of amino acid amides from peptide amides but this activity is optimal at basic pH. It is likely that at the basic pH range Asn51 interacts with the carbonyl oxygen of the C-terminal carboxyamide group while Glu145 in its deprotonated (carboxylate) form interacts with the -NH2 group of the substrate. Glu65 is hydrogen bonded to Asn51 and Glu 145 thereby orienting the two side ~h~i n~ involved in C-terminal recognition. When these amino acid residues are exchanged by site directed mutagenesis with amino acid residues without the capacity to function as hydrogen bond donors, the kCat/KM for the release of amino acids and amino acid amides from the C-terminus of peptides and peptide amides is drastically reduced.
However, when other amino acid residues with such a WO95/20039 ~ - PCT~9S/OGC~2 ~ 2 1 74525 capacity are incorporated at these positions, much more activity is retained.
In a number of cases, higher yields with nucleophiles containing a blocked ~-carboxylate group, in particular amino acid amides, would be highly desirable. This problem can be approached by protein engineering. In particular, one could imagine that such changes might improve the synthetic capacity of the enzyme by (a) permitting the use of nucleophiles which are poorly accepted by the wild-type enzyme, (b) securing better binding of nucleophiles, (c) providing higher yields, and (d) reducing the rate of degradation of products. We here report the effects of such mutations on transacylation reactions with amino acid derivatives as nucleophiles. Some amino acid substitutions made in mutant proteases were made knowing that they were not likely to enhance yields of transacylation, but rather to study the mechanism of action of binding and catalysis with nucleophilies containing a blocked ~-carboxylate group.
Carboxypeptidase-Y was obtained from Carlbiotech, Copenhagen, Denmark. H-Val-OPr, H-Val-OBu, H-Val-NHCH3 and H-Val-NHC2Hs were from Peptech, Sydney, Australia. All other amino acids and amino acid derivatives were from Bachem, Switzerland. The mutants Asn 51 , Gly, Cys and Val in the structural gene for CPD-Y and subsequent expression and purification of the mutants N51G and N51C were carried as described in Example 1. The purity of the enzymes was ascertained by SDS polyacrylamide gel electrophoresis. The preparation of the mutants N51A, N51D, N51T, N51Q, N51S, E145A and E145D has previously been described. The mutations Asn51 ~ Cys, Asn51 ~ Gly and Asn51 ~ Val were constructed. However, N51V was not expressed and, thus, only N51C and N51G were isolated. FA-Ala-OBzl, FA-Phe-Ala-OH and FA-Phe-Leu-OH were synthesized by standard methods.

W095/20039 . ~ ~ ~ 7~52~ PCT~S9S,06682 ~

All enzymatic activities toward FA-substrates were determined spectrophotometrically at 329-337 nm using a Perkin Elmer lambda 7 or lambda 9 spectrophotometer thermostated at 25C. The hydrolysis was carried out in 0.05 M Mes, 1 mM EDTA, pH 6.5 for peptide substrates and in 0.05 M Hepes, 1 mM EDTA, 2.5 (v/v), pH 7.5 for ester substrates, kCat and Km values were determined using the Enzfitter program.
Among the previously described CPD-Y
derivatives, mutationally altered at positions 51, 65 and/or 145, a number were chosen to be investigated for their applicability in peptide synthesis. For the study of aminolysis reactions, the excellent CPD-Y substrate FA-Ala-OBzl was selected and initially the ability of all the CPD-Y derivatives to hydrolyze this substrate was investigated (data not shown). With most mutants the hydrolysis of FA-Ala-OBzl was only slightly affected by replacement of Asn51. In fact, with a number of mutants, elevated kC~t/KM values were observed. With E65A, N51G and, in particular, E145Q, reduced kCat/KM were observed, possibly due to the effects of minor conformational changes. However, apart from this result, it appears that the interaction between enzyme and the -OBzl leaving group, in spite of its bulkiness, is not adversely affected by mutational replacements at positions 51, 65 and 145. Thus, provided that these enzymes may bind nucleophiles in a productive mode, they should be able to catalyze aminolysis reactions.
The reactions with the valine amides were investigated. With H-Val-NH2, the fraction undergoing aminolysis reaction increased with increasing concentration of nucleophile until it reached 1.00. The results were analyzed according to a model which assumes that nucleophile binds to the acyl-enzyme intermediate prior to the deacylation reaction. The correlation between fa and the concentrations of nucleophile are consistent with K~(app, and famax values of 16 mM and 1.00, ~ W095/20039 ~ 2 1 7 4 5 2 5 PCT~S5~5/0G6~2 respectively. Thus, the hydrolysis reaction is completely excluded by high concentrations of nucleophile. With the corresponding free amino acid, H-Val-OH, the values for KN(app, and fa~x, as determined under identical conditions, were 13 mM and 0.32, respectively. Thus, compared to H-Val-NH2, the binding of the free amino acid is much looser and, when it is bound, the binding mode evidently is much less favorable for the aminolysis reaction. The beneficial effect of blocking the carboxylate group is presumably due to elimination of the adverse effects of charge repulsion.
The contribution of Asn51 and Glu145 to the productive binding of amino acid amide nucleophiles was investigated by determination of KN(~PP) and fa~x with H-Val-NH2 using CPD-Y derivatives mutationally altered at positions 51 and 145 (see Table VII).

W095/20039 ~ 2 ~ 7~5 PcT/u~sl~6~82 ~

TABLE VII
Use of CPD-Y Mutants for Aminolysis of FA-Ala-OBzl using H-Val-NHl and N-al~yl Derivatives Hereof as Nucleophiles Nucleophile Enzyme H-val-NH2 H-Val-NH-CH3 Wild-tyPe KN(~PP) 1 . 6 i 0.1 290 + 50 famax 0.97 _ 0.010.80 _ 0.05 N51G KN(app~ 12 _ 2360 _ 20 famax 0.89 i 0.020.76 + 0.07 N51A KN(app) 34 i 1~ 1200 famax 1.00 i 0.01 ~ 0.8 N51S KN(app) 3 9 + 0.1360 i 50 famax 1.00 + 0.010.79 _ 0.05 N51C KN(~PP~ 11 _ 21200 _ fa~x 0 95 + 0.010.72 + 0.04 NSlD KN(aPP) 11 + 1 510 _ 40 famax 0.95 i 0.010.70 i 0.03 N51Q KN~app) 1.5 + 0.1 f amax O ~ 9 7 i 0.01 E145A KN(~PP) 3.1 0.1 140 i 13 famax 0.98 _ 0.01 0.97 i 0.03 E145Q KN(app) 8.4 1.0 famax 0.08 _ 0.01 E145D KN(app) 2.3 i 0.1580 i 30 fam~x 0.98 + 0.010.59 _ 0.02 E65A KN(app) 2.9 _ 0.3 famax 0.96 + 0.01 N51A+E145A KN(app1 100 i 7 640 + 40 fa~x 0-95 _ 0.020.98 + 0.03 E65A+E145A KN(app) 110 + 3 47 4 famax 0.76 + 0.010.33 + 0.01 ~W095/20039 ~ PCT~S9Sl06682 -- `: 21 74525 With N51A and N51G, the KN(app~ values were increased 8-20 fold and the fam~x values remained high. While not meant to be a limitation of the invention, it is believed these results demonstrate that the capacity of the group to donate a hydrogen bond is not required to properly orient the amino acid amide prior to the nucleophilic attack on the acyl-enzyme intermediate but it is apparently important for the binding of the nucleophile.
These results are consistent with Asn51 functioning as hydrogen bond donor with the carbonyl oxygen of the carboxyamide group of the nucleophile as acceptor.
Asn51 can be replaced with other hydrogen bond donors without impairing this wild-type enzyme. However, the Asn51 Cys mutation caused a significant increase in KN(app~ and this is consistent with hydrogen bonds involving Cys being rather weak. With N51D, the value for KN(app, was also significantly elevated but this result is difficult to interpret since it is possible that Asp51 may interact with -NH2 group of the nucleophile.
With E145A and E65A, the values for KN(app~ were only doubled and the famax value remained unchanged.
Thus, compared to Asn51, Glu145 and Glu65 appear to be much less important for the interaction with amino acid amide nucleophiles. However, the possibility existed that the remaining glutamic acid, which is negatively charged at pH 7.5, could function in the binding of amino acid amides. To investigate this, the double mutant E65A-E145A was tested and it was found that KN(app~
was drastically increased while famax was somewhat reduced. Thus, the absence of both glutamic acids is detrimental to the binding of amino acid amides but the presence of one of them is sufficient to secure tight binding. In the wild-type enzyme, Glul45 is negatively charged while Glu65 is uncharged and, accordingly, Glu145 is the one interacting with amino acid amides.
The very low famax and significantly elevated KN(app~
obtained with E145Q confirm the significance of Glu145 WO9S/20039 PCT~S9S/06682 s- 2 ~ 74525 ~

in the interaction with amino acid amides. Most likely, the C-terminus of peptide amides interact with the enzyme in an analogous way with the negatively charged Glu145 acting as a hydrogen bond acceptor. Since both Glu65 and Glu145 in the single mutants may function in this capacity there apparently may be some latitude concerning the length of the hydrogen bond. This is suggested by the fact that shifting the carboxylate group at position 145 one carbon atom away, i.e. E145D, has very small effects on both parameters. The high KN(~PP) obtained with the double mutant N51A+E145A and H-Val-NH2 as nucleophile show that the effects of the Asn51 ~ Ala and Glu145 Ala mutations are approximately additive but the famax remained high.
Valine N-methyl amide (H-Val-NH-CH3) binds much less efficiently to CPD-Y than the unblocked H-Val-NH2.
However, the fa~x is almost as high (0.80).
Substitution of Asn51 for Gly, Ala, Ser, Asp or Gln and &lu145 for Ala, Asp or Gln only affected the values for fa~x and KN(~PP1 moderately, suggesting that neither Asn51 nor Glu145 is involved in the interaction with H-Val-NH-CH3. The binding mode of this nucleophile, therefore, remains unclear.

Altering Substrate Preference of Carboxypeptidase Y by a No~el Strategy of Mutagenesis Selected targets of PRCl were mutagenized randomly and subse~uently screened for mutants expressing carboxypeptidase Y (CPD-Y) with increased activity toward poor Pl substrates. From an alignment of the primary structures of CPD-Y and a wheat carboxypeptidase (CPD-W) (Breddam et al., Carlsberq Res.
Commun., 52:55-63i 65-71 and 297-311 (1987)) as well as the crystal structure of CPD-W (Liao et al., J. Biol.
Chem., 265:6528-6531 (1990), it was predicted which amino acid residues might constitute the surface of the ~W095/20039 ^~ 2 1 7 4 5 2 5 PCT~S9SIOCC~2 CPD-Y Sl binding pocket. The corresponding codons of PRC1 were chosen as targets for saturation mutagenesis.
The basic technique of the mutagenesis was in - vitro DNA synthesis primed by mutagenic (degenerate) synthetic oligonucleotides using single-stranded phagemid DNA as template, followed by transfection of Escherichia coli. To facilitate efficient screening of a large population of mutant transformants, a new and generally applicable mutagenesis strategy was developed which eliminates the wild-type background due to unmutagenized plasmids in the subsequent functional screens. Although several existing procedures are highly efficient with respect to the frequency of mutagenesis, a residual portion of the plasmids produced will still be unmutated. If the mutation frequency is, say, 90~, and if lOs transformants are produced, then 104 will be wild type. After transformation of yeast, the wild-type transformants can make it very difficult to select the desired mutants in the subsequent screens.
To overcome this problem, combinations of stop codons and restriction sites were introduced at each position in the PRC1 gene that was to be targeted by a degenerate oligonucleotide. The stop codons ensure that the e.g.
10~ unmutated plasmids will encode inactive CPD-Y, while the introduced restriction sites allow us to monitor the mutation frequency. The mutagenesis itself is performed according to the protocol of Lewis et al. (Nucleic Acids Res., 18:3439 (1990)) on single-stranded phagemid DNA
carrying a frameshift mutated ampicillin resistance gene and the nonsense mutated prcl gene. Mutant strand synthesis is primed in a single reaction with a number of degenerate oligonucleotides to produce mutations in the PRC1 gene and an oligonucleotide that repairs the ampicillin resistance gene. The resulting DNA is used to transform a mismatch-repair deficient E. coli strain to ampicillin resistance. Propagation in ampicillin-W095/20039 ~ 2 1 ~ ~ 525 PCT~S9S/06682 ~

containing medium selects for progeny of the mutantstrand.
Two prcl deletion strains of yeast were transformed with the plasmid population. In one strain, CPD-Y enters the vacuole, while in the other strain CPD-Y is missorted, secreted and activated extracellularly due to a vps mutation as described by Nielsen et al., Appl. Microbiol. Biotech., 33:307 (1990).
The vps strain, which requires leucine for growth, was used to search for desired mutants in a single direct screen. The transformed cells were plated on synthetic medium lacking leucine but containing one of various N-blocked-X-Leu-OH dipeptides. Only cells which express a protease that can release the terminal leucine can grown on these plates.
The Vps~ strain was used to screen for desired mutants in two steps. First, transformant colonies were overlaid by agar containing a chromogenic substrate which reveals CPD-Y activity. In the second step, the CPD-Y activity from the positive colonies was estimated towards different substrates in a chromogenic microtiter dish assay.
By this procedure, a total of >105 independent mutants was produced, some of which exhibit increased activities (k~a~/~) toward certain substrates by a factor of up to 150. A number of these mutant plasmids have been sequenced and the encoded enzymes have been purified and characterized kinetically.

Reaqents:
CBZ-X-Leu-OH peptides and N-acetyl-L-alanine ~-naphthyl ester (AANE) were from Bachem; horseradish peroxidase type I, Crotalus atrox L-amino acid oxidase type VI, o-dianisidine and Fast Garnet Red GBC salt were from Sigma. Oligonucleotides were synthesized on an Applied Biosystems 394 DNA-RNA Synthesizer. LB, 2xYT
and SOC medium were prepared according to Sambrook et ~ 095/20039 ~ 2 1 7 4 5 2 5 PCT~S9S/06682 al., Molecular Cloning. A Laboratory Manual, Cold Spring Harbor, NY (1989). SC and YPD medium were prepared according to Sherman, Methods EnzYmol., 194:3-21 ~1991), with slight modifications (Tullin et al., Yeast, 7:933-94i (1991)).

Strains:
E. coli BMH71-18mutS (thi supE ~(lac-proAB) [mutS::TnlO] F'[proAB~ laqIq lacZAM15]) (Kramer et al.
Cell, 38:879 (1984); Zell et al. EMBO J., 6:1809 (1987)); E. coli JM109 (recAl supE44 endAl hsdR17 gyrA96 relAl thi ~(lac-proAB) F'[traD36 proAB~ laqIq lacZ~M15]) (Yanisch-Perron et al., Gene, 33:103 (1985));
S. cerevisiae JHRY20-2C~3 (MATa ~prcl leu2-3 leu2-112 ura3-52 his3-~200 prcl-A3::HIS3) (Blachly-Dyson et al., J. Cell Biol., 104:1183 (1987)); S. cerevisiae W2579 (MATa Aprcl leu2-3 leu2-112 ura3-52 vpll-l) (Nielsen et al. (1990) cited su~ra.). The vpll gene has been renamed vpsl (Robinson et al., Mol. Cell. Biol., 8:4936 (1988)). K2579LLR was isolated in the present study as a spontaneous mutant of W2579 that requires less leucine for growth.

Plasmids:
The 2.2 kb EcoRI 2~ fragment taken from YEp24 (Botstein et al., Gene, 8:17-24 (1979)) was blunted and inserted into the blunted ClaI site of pSELECT (Lewis et al., 1990 cited suPra.) in the orientation that places the HINDIII site close to lacZ. The 1.1 kb HindIII URA3 fragment taken from YEp24 was blunted and inserted clockwise into the blunted StyI site of the resulting plasmid to produce pYSEL. The SalI-PvuII fragment of pWI3 (Winther et al., Eur. J. Biochem., 179:681 (1991)) with the PRCl gene under control of its own promoter was blunted and inserted clockwise into the blunted SalI-BamHI fragment of pYSEL to produce the phagemid shuttle vector pYSPl. A combined stop codon/restriction site W095/20039 2 7 7 ~ 5 25 PCT~S95/06682 ~

was introduced into pYSP1 at each of the four selected targets for mutagenesis of the PRC1 gene (Table VIII).
The resulting plasmid, pYSP32, carries ori, tet~, bla-, the fl ssDNA replication origin, a functional fragment of 2~, URA3 and prcl. Single-stranded DNA produced from this phagemid is complementary to the sequence indicated in Table VIII.

~Tvo gs/20039 2 1 7 4 5 2 5 PCT/US95/06682 dP O ^ O O O O ~( O O O ~ o o o U~ ~ ~ U.C
h .U
~ V C) V V C~ V V V V U 3 r V U VVUVVVVV' U
E-( Z ~ E

u V tl5 E~ Z E~ Z Z E~ ~ Z ~, 4~
o C~ 0 V V V Z V X Z V V Z a --z E~ æ z E~ ~ z ~c V ~ ~ ~ X Z ~ ~ Z
V ~ V V V ~) V V V V C
o ~ 4 a~ . - - - .......... a C! C
V V VVVVVVVV C~

v r~
_ Z ~ R-' ~ V v V Z Z V Z V Z V (~ ~ r~ O
H 3 U V ~/ V V V V V V V V V 3 4---H, ,~ O ~ ~ E~ E~ E~ ~ S
t;7 . E . . . . . . . . 'c 3 .-V~ ~ ~uuJ~
a o + + + + + + + + ~ ~c ~ c U V 1~ VVVVVVVV S' L~ '¢ V C ~¢ .¢ .~ ,~ .cl ,¢ ,¢ ,¢ S .,~
E~ c , V Z; Z Z V Z V v V C.) V ~1 _ c a~
C~ V ~C _ VVVVVVVV ~C~_ '7C ~ V ~ ) r, V V V V V V V V 4_1 ~. V
L ~ V 0~ E~ ' V V V V V V V V ~ ? ~
a, ~ + + + + + + + + ~ s o ..... ._ V V ~J V V V V V V V V --~ -~ 3 ~_) V -~ V ~,) V V V V V U
:~ V E~ H ~: V V V V V V V V ~ .
CL~ H ~I a~
V~` V S~ V ~z; V V V V V V ~
V ~ ~ ~: Z ~ ¢ C
v cnv ~ ~c v v v v v v v v ~ ,~ ~ vvvvvvvv ~c V C~ O ........ (~ ,~
c~ O - - - - a a 0 ~ ~ ~; a~
V .. -......... ~ ~ V
~, ~ v ~ ~ ~ ~ ~ ~ r ~ . r -r ~ /V
J_) ~I C 0~ U1 ~O lQ U~ U~ U~ tq , _ u. 2 -, ~v a) ~v a) a) a~ ~v ~v ~ 1 h (V a) aJ ~V ~V a) O ~V
V 3 ~ V cn cn cn cn cn cn cn cn Z

WO95/20039 ^ ~ 2 ~ ~ 4 5 2 5 PCT~S9S/06682 ~

Transformation:
Transformation of E. coli was performed with a Bio-Rad Gene Pulser set at 25 ~F, 200 Q and 2.5 kV in 2 mm cuvettes according to Dower et al., Nucleic Acid Res., 16:6127 (1988). Transformation of yeast was performed according to Schiestl and Gietz, Curr. Genet., 16:339 tl992).

Preparation of sinqle-stranded phaqemid DNA:
E. coli JM109 transformed with pYSP32 was grown to an OD600 of 0.5 in 2 x YT + 50 mg/l tetracycline. One milliliter of this culture was superinfected with 20 ~l of a >109 p.f.u./ml M13K07 helper phage stock in a 500 ml Erlenmeyer bottle. After incubation for 1 hour at 37C, 200 ml 2 x YT + 50 mg/l tetracycline + 50 mg/l kanamycin was added. After incubation with agitation overnight at 37C, ssDNA was purified by standard procedures (Sambrook et al., _ Guide to Molecular Clonina (1989)). Kanamycin selects for cells superinfected with helper phage M13K07.

Mutaqenesis:
Mutagenesis was performed according to the protocol of Lewis et al. (1990) (cited su~ra) with slight modifications. Single-stranded pYSP32 (0.2 pmol) was mixed with 0.5 pmol ampicillin-repair oligonucleotide (Promega, Altered Sites Kit) and 2 pmol of each mutagenic degenerate oligonucleotide in 80 ~l 2 x annealing buffer (20 mM Tris-HCl pH 7.5, 10 mM MgClz, 50 mM NaCl), heated to 70C for 5 minutes and allowed to cool slowly ( 1 hour) to room temperature. After the annealing reaction 10 ~l 10 x synthesis buffer (100 mM
Tris-HCl pH 7.5, 5 mM of each of the four dNTPs, 10 mM
ATP, 20 mM DTT), 10 Weiss units T4 ligase (New England Biolabs), 20 units T4 DNA polymerase (Promega) and H2O to 100 ~l were added. The polymerase/ ligation reaction was incubated at 37C for 2 hours. Subsequently the reaction mixture was phenol extracted, ethanol precipitated, resuspended and used to transform electro-O9S/20039 : PCT~S9SI06682 ~ 2174~25 , competent E. coli BMH71-18mutS cells. After incubating the transformed cells for 45 min at 37C in SOC, an aliquot was plated on LB plates with 60 mg/l ampicillin to determine the number of transformants, while the rest of the cells were grown overnight in 50 ml LB with 60 mg/l ampicillin. A plasmid preparation from this primary culture will contain up to about 50~ ampicillin sensitive plasmids originating from the unmutated parent strand. To enrich ~or the mutants, 1 ~g miniprep DNA
from this culture was used for a secondary transformation of JM109. One microgram of miniprep DNA
from the JM 109 transformant culture was used to transform the yeast strains K2579LLR and JHRY20-2C~3.

Se encin~:
DNA sequencing was performed by the Applied Biosystems dsDNA Taq DyeDeoxy~ terminator procedure for use with the Applied Biosystems Model 373A DNA
sequencing system.
Plate activitY screen:
Yeast strain K2579LLR was transformed with the mutated population of pYSP32 and plated on SC-ura-leu supplemented with 1.5 mM of a poor CBZ-X-Leu-OH peptide substrate as an enzyme-dependent leucine source. P1 of the substrate was either Glu, Gly, Ser, ~is, Pro, Trp or Lys. The cells can grow on this medium only if they express proteases capable of releasing the C-terminal leucine. However, CPD-Y is not the only protease secreted by the cells that can catalyze this cleavage.
Thus, a minimum level of CPD-Y activity is necessary not only to release sufficient leucine for growth, but also in order to discriminate CPD-Y dependent growth from the background growth. Based on experiments with characterized CPD-Y mutants, it is estimated that the kCa~/~ must be above -10,000-25,000 min~lmM~l, dependent on the substrate.

, WO95/20039 2 1 7 4 ~ 25 PCT~S9S/06682 ~

Color overlaY screen:
Yeast strain JHRY20-2C~3 was transformed with the mutated population of pYSP32 and plated on SC-ura plates, which were incubated at 30C until colonies reached a diameter of 2-3 mm. Then each plate was overlaid with a fresh mixture of 3 ml 0.6~ agar in water at 50C and 2 ml dimethylformamide containing 0.25~ AANE
at room temperature. After incubation for 5 min at room temperature, 5 ml 0.4~ Fast Garnet Red GBC salt in 10 mM
sodium phosphate pH 7.0 buffer was poured on top; after 5 min incubation the plates were then rinsed in tap water. Colonies expressing active CPD-Y appear red, while those lacking vacuolar CPD-Y activity appear white (modified from Jones, 1977). Colonies expressing active CPD-Y were isolated by streaking onto SC-ura plates.

Cou~led assaY screen:
After incubation for 3-4 days on SC-ura plates of JHRY20-2C~3 transformants expressing active mutant CPD-Y, ~1 mm3 of cells was picked with a toothpick and suspended in 200 ~l water. After aliquoting 20 ~l cell suspension into each of eight microtiter wells, 100 ~l substrate mixture (50 mM MES pH 6.5, 1 mM MnCl2, 50 ~g/ml peroxidase, 100 ~g/ml L-amino acid oxidase, 100 ~g/ml o-dianisidine and 1 mM CBZ-X-Leu-OH) were added. Pl o~ the substrate (X) was either Phe, Glu, Gly, Ser, His, Pro, Trp or Lys. The plates were incubated at room temperature, and OD 560 was measured at times 0, 1 and 16 hour on a Perkin Elmer Lambda Microplate Reader (modified from Lewis and Harris, 1967, cited supra).
The activity of each mutant was normalized to that of the wide type as follows. First, all absorption values were corrected by subtracting the background (absorption at same time point in wells without cells).
The difference in correction absorption between two time points (corresponding to the amount of hydrolysis) was then normalized to account ~or variations in cell ~ 095/20039 ~ .. PCT~S9Sl06682 number, by dividing by the OD of the cells (OD of well with cells at time point 0 minus OD of well without cells at time point 0). Finally, the obtained activity estimate was divided by the corresponding estimate for the wild type. Time point 0 and 1 hour were used to calculate the relative activity towards CBZ-Phe-Leu-OH, while time points 0 and 16 hour were used to calculate the relative activity towards all other substrates.

Isolation and kinetic characterization of mutant forms of CPD-Y:
Over-expression of selected mutant enzymes was performed according to Nielsen et al. (1990) cited supra. The mutant BamHI-BamHI PRCl fragments of pYSP32 were inserted into the GAL expression vector pRA21 and introduced into yeast strain K2579LLR. The plasmid pRA21 is derived from p72UG (Nielsen et al. (1990) cited suPra) by replacing the 918 bp BglII-SalI fragment with the 638 bp BglII-PvuII fragment of pWI3, thereby deleting the BamHI site downstream of the PRCl gene.
Mutant forms of CPD-~ were purified from culture supernatants by affinity chromatography as described in Example 1. Kinetic parameters were determined by measuring rates of hydrolysis at 25C of CBZ-X-Leu-OH
substrates spectrophotometrically at 224 nm in 10 mM
sodium phosphate buffer pH 6.5. The concentrations of substrates ranged from 0.01 to 0.5 mM. Accurate measurements were not possible with substrate concentrations exceeding 0.5 mM due to the high absorption of the substrate.
By studying the crystal structure of the wheat carboxypeptidase CPD-W (Liao et al (1990) cited suPra), the amino acid positions Pro60, Tyrl56, Leul87, Phe224, Ile225, Val334 and Val335 were found to constitute the bottom, end and sides of the Sl binding pocket of carboxypeptidase W. These positions of CPD-W correspond to amino acid positions Pro54, Tyrl47, Leul78, Glu215, Arg216, Ile340 and Cys341 of CPI~-Y according to an WO95/20039 2 i 7 ~ 5 25 PCT~S~S~C~2 ~
,., . ` :

alignment of the primary structures of the two enzymes (Breddam et al. (1987) cited su~ra; Sorensen et al., Carlsberq Res. Commun., 52:285 (1987)). We hypothesized that these amino acid residues constitute the Sl binding poc~et of CPD-Y. As Pro60 contributes very little to the surface of Sl of CPD-W, Pro54 of CPD-Y was not included in any of the mutagenesis experiments, and as Tyrl56 is next to the active Serl55 of CPD-W and its contribution to the surface of Sl is also minor, Tyrl47 of CPD-Y was only included in one mutagenesis experiment.
The crystal structure of yeast carboxypeptidase Y (Endrizzi et al., Biochemistry, 33:11106 (1994)) confirms that this model is correct in many of its features. The model correctly predicts 4 of 9 residues in the Sl binding site. Pro54 of carboxypeptidase Y is predicted to be at most a marginal part of the Sl binding site; it is in fact absent from the Sl site. The model that used the crystal structure of wheat carboxypeptidase-W and the sequence of yeast carboxypeptidase Y yielded accurate guidance for construction of mutants of carboxypeptidase Y.
Eight different series of mutagenesis of pYSP32 were carried out, randomly mutating various numbers of codons, from 1 to 6, simultaneously (Table VIII): series 1, all codons reverted to wild type;
series 2, positions 147, 178, 215, 216, 340 and 341 degenerated; series 3, positions 178, 215 and 216 degenerated; series 4, positions 178, 340 and 341 degenerated; series 5, positions 215, 216, 340 and 341 degenerated; series 6, position 178 degenerated; and series 8, positions 340 and 341 degenerated. In all series, all mutant codons present in pYSP32 that we not subjected to random mutagenesis were reverted to wild type (Table VIII). All degenerate mutagenic oligonucleotides were designed relatively long (33 bases) to minimize biased annealing of oligonucleotides WO95/20039 ~ ~ PCT~S95/06682 . .
2 1 7452~

complementary to the introduced stop codons and restriction sites.
A sensitive two-step screening strategy was employed. First, to differentiate transformants expressing active CPD-Y , we have used a modification of the overlay procedure of Jones (1977) for staining CPD-Y
containing yeast colonies. Instead of using APNE
(N-acetyl-DL-phenylalanine ~-naphthyl ester) as a CPD-Y
substrate, we have used AANE, since we find that the background staining of inactive CPD-Y transformants is much lower with AANE, thus making it possible to identify transformants with low CPD-Y activity.
Furthermore, it was anticipated that most active CPD-Y S
mutants would accommodate the smaller Pl side chain of AANE better than the larger side chain of APNE.
Colonies of yeast strain JHRY20-2C~3 transformed with DNA from all mutagenesis series were screened by this assay. In the wild type control mutagenesis, series 1, half of the transformant colonies express active CPD-Y (Table VIII). If the mutagenesis event at each oligonucleotide target is independent of that of the other three targets, this number corresponds to a mutation frequency of 84~ at each target. Of lOs tested transformants from mutagenesis series 2, none had detectable CPD-Y activity. In this series, six codons were mutated simultaneously. Fewer codons were mutated in series 3-6 and 8, which yielded between 0.4 and 10~
positive transformants, expressing a wide range of CDP-Y
activities as indicated by the color intensities in the overlay assay. Mutagenesis series 7 yielded 50~
transformants expressing active CPD-Y and the level of activity of all transformants was indistinguishable from that of the wild type transformants, suggesting that positions 215 and 216 can be varied with little effect on activity towards AANE.
In the second step, we estimated the CPD-Y
activity of positive mutants toward eight N-blocked WO95/20039 2 1 7 4 5 2 5 PCT/U~5S/OGG82 ~

dipeptides (listed in Table IX), by a modi~ication of a chromogenic microtiter dish assay, involving L-amino acid oxidase, peroxidase and o-dianisidine, previously used to monitor carboxypeptidase S activity (Lewis and Harris, 1967; Wolf and Weiser, 1977).

~WO9S/20039'' ~ PCT~S9S/06682 ~ ` 2 1 74525 TABLE IX
Estimated CPD-Y Activities of Selected Mutants r - Relative to Wild Type Transformants, Towards an Initial Concentration of lmM of Eiqht CBZ-X-Leu-OH Substrates P1 side chain of substrate Phe Glu Gly Ser Pro Trp His Lys Wide-type CPD-Y
Mutations:
178Ser 0.7 1.3 1.1 0.6 1.8 1.21.1 9.2 178Cys 1.5 l.l 0.9 0.5 0.3 1.01.7 6.9 178Cys/215Thr/2i6Trp 1.0 1.1 0.8 0.41.6 0.9 1.1 5.3 178His/215Pro/216Thr 0.9 l.0 0.8 1.11.2 1.5 1.1 3.6 l7sAla/2lsser/2l6Arg 1.2 1.4 1.l 1.30.6 ~ .o 4.0 178Ser/215Ala/216Pro 1.5 2.0 0.6 0.72.8 1.9 0.9 8.0 178His/215Thr/216Phe 0.9 1.2 0.8 1.01.2 1.7 1.1 3.1 178Asn/21sAla/216Gly 2.8 2.0 0.7 0.70.9 2.0 1.3 4.4 178Gly/340Ile/341Cys 0.8 0.8 0.3 0.30.5 0.5 0.8 4.6 Relative activities >2.0 are emphasized in bold print W095/20039 2 ~ 74525 PCT~S95/06682 ~

Transformants of yeast JHRY20-2C~3 expressing active CPD-Y from mutagenesis series 3-8 were tested in this assay using an initial substrate concentration of 1 mM. Compared with the wild type, transformants from mutagenesis series 7 all exhibited activity levels around 100~ towards all eight tested CBZ-X-Leu-OH
substrates. Neither very low nor very high activity levels were observed, suggesting that positions 21S and 216 have little influence on Pl preference.
Transformants from mutagenesis series 4, 5 and 8 generally exhibited low and varied activity levels.
Most of these exhibited between 10 and 100~ CPD-Y
activity and none exhibited >200~ activity. We conclude that Ile340 and/or Cys341 are important for ~,kCa~ or production of the enzyme. Transformants from mutagenesis series 3 and 6 showed a broad range of activities with an average around 100~. Several mutants were fund with very low activity levels, just as several were found with very high activity levels towards one of the substrates, CBZ-Lys-Leu-OH. In one case the activity was 9 times the activity of the wild type. To determine which amino acids had been substituted in a number of the most interesting mutants, plasmid DNA was recovered and sequenced (Table IX). The mutant exhibiting the highest increase in activity towards CBZ-Lys-Leu-OH was found to have Leul78 substituted with Ser.
To determine the kinetic parameters of the mutants exhibiting the highest increase in activity, CPD-Y was purified from the mutants 178Ser and 178Ser/215Ala/216Pro by affinity chromatography and analyzed kinetically (Table X).

~ W095/20039 ~ 2 1 74525 PCT~S~S/0'6~2 TABLE X
Kinetic Parameter~ of Two Selected Mutant CPD-Y Forms Compared with the Wild Type SubstrateCPD-Y kc~t K~ k~t/K~
(mM/min/mM)(mM) (min-~nM~
l) CBZ-Phe-Leu-OH Wild type13600 0.036 380,000 178S 12000 0.060 200,000 178S/215A/216P 5800 0.053 110, 000 CBZ-Lys-Leu-OH Wild typend nd 178S nd nd 3,700 178S/215A/216P nd nd 2,200 CBZ-Ser-Leu-OH Wild type800 0.070 12,000 178S 420 0.070 6,000 178S/215A/216P 240 0.080 3,000 CBZ-Ala-Leu-OH Wild type-25000 -0.7 36,000 178S nd nd 4,800 178S/215A/216P nd nd 2,400 CBZ-Glu-Leu-OH Wild typend nd 178S nd nd 178S/215A/216P nd nd nd: Values could not be determined as K~ is >0.5 mM.

W09~20039 ~ 2 1 7 4 5 2 5 PCT~S95/OCC~2 ~

The kCa~/~ values of these enzymes towards~CBZ-Lys-Leu-OH
are 3700 and 2200 min~lmM~~ respectively, corresponding to a 155- and 91-fold increase relative to the wild type value of 24 min~lmM~l. These results confirm the importance of the 178Ser mutation and the smaller effect of changes in the side ch~; nq at positions 215 and 216.
By introducing stop codons at all targets for random mutagenesis, the wild type background normally present due to unmutagenized plasmids was eliminated, and the wild type PRC1 gene can now only occur by rare mutational reversion. The frequency of such revertants will depend on the mutagenesis efficiency, on how many codons are mutagenized simultaneously and on the number of codons synonymous to the mutagenized codons. With an overall mutagenesis efficiency of 50~, theoretical reversion frequencies of 6 x 10-9, 1 x 10-4, 7 X 10-5, 2 x 10-6, 5 X 10-2, 1 X 10-3 and 7 x 10-l would be expected for mutagenesis series 1-8 respectively. These low frequencies of wild types enabled selection for rare mutants with increased activity. Without the elimination of the wild type background, it would have been difficult to select the desired mutants from a pool of perhaps 0.5~ active mutants and 50~ unmutagenized wild type transformants. By this strategy the binding site of any enzyme that has a functional screen can be dissected, amino acid by amino acid, to determine which positions are productive mutagenesis targets.
Furthermore, the introduction of a rare restriction site at the oligonucleotide target prior to the mutagenesis simplifies the task of confirming the mutation and reduces sequencing to a minimum.
The results suggest that Leul78, Ile340 and Cys3~1 contribute to Pl specificity, conceivably by forming part of the surface of the Sl binding pocket.
All CPD-Y mutants with increased activity towards Lys in Pl were found to have mutations at position 178.
Surprisingly, substituting Leul78 by Ser, Cys, Ala, Gly, ~ WO 95/20039 ~ r i,' ~ ~( ` . 2 1 7 4 5 2 5 PCT~S~SJ0G6Q2 Asn or His, all give the same general effect - more activity towards substrates with Lys in Pl (Table X). An explanation for the increased activity towards Lys in P
of CPD-Y-178 mutants might be that the introduced mutation causes a conformational change of whichever residues constitute the end of S1 in CPD-Y, corresponding to 224 and 225 of CPD-W.
CPD-Y can catalyze the transpeptidation of proline insulin (INS-Pro-Lys-Ala-OH) to produce human insulin amide (INS-Pro-Lys-Thr-NH2), and it has previously been shown that the Cys341 Hg2~ modified enzyme, which has 15 times more activity towards Lys in Pl, produces 26~ human insulin amide product in a transpeptidation reaction (Breddam et al. 1984), whereas the wild type enzyme only produced 4.2~. Thus, the 178Ser mutant, with its 155-fold increase in activity toward Lys in Pl, might give even higher yields in such a transpeptidation reaction.
Mutants selected for the ability to hydrolyze a poor CBZ-X-Leu-OH peptide substrate as described herein can also be screened for the ability to transacylate preselected polypeptide substrates and nucleophiles as described in Examples 2 and 3.
All publications and patent applications in this specification are indicative of the level of ordinary skill in the art to which this invention pertains. All publications and patent applications are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated by reference.
It will be apparent to one of ordinary skill in the art that many changes and modifications can be made in the invention without departing from the spirit or scope of the appended claims.

W 095l20039 - - PCT~US9SN6682 2 l 7452~ --SEQUENC_ LISTING

(1) GENERAL INFORMATION:
(i) APPLICANT:
(A) NAME: Carlsberg A/S
(B) STREET: Vasterfaelledve; 100 (C) CITY: Copenhagen V
(E) COUNTRY: Denmark (F) POSTAL CODE (ZIP): DK-1799 tA) NAME: Breddam, Klaus (B) STREET: Sondervangsve; 283 (C) CITY: Glostrup (E) COUNTRY: Denmark (F) POSTAL CODE (ZIP): DK-2600 (A) NAME: Kielland-Brandt, ~orten C.
'B) STREET: Lundbyesgade 3 C) CITY: Copenhagen V
E) COUNTRY: Denmark ~F) POSTAL CODE (ZIP): DK-1771-(A) NAME: Mortensen, Uf~e Hasbo B) STREET: Nyrnberggade 6, 3TV
C) CITY: Copenhagen S
E) COUNTRY: Denmark ~F) POSTAL CODE (ZIP): DK-2300 (A) NAME: Olesen, Kjeld Ove (B) STREET: Nitiveg 9, 3th ~C) CITY: Frederiksberg (E) COUNTRY: Denmark (F) POSTAL CODE (zIe): DK-2000 (A) NAME: Stennicke, Henning Ralf B) STREET: Sindshvilevej 3A, 4TV
~C) CITY: Frederiksberg E) COUNTRY: Denmark ~F) POSTAL CODE (ZIP): D~-2000 (A) NAME: Wagner, Fred W.
(B) STREET: Route 1, Box 77B
(C) CITY: Walton (D) STATE: Nebraska (E) COUNTRY: USA
(F) POSTAL CODE (ZIP): 68461 (ii) TITLE OF INVENTION: Customized Proteases (iii) NUMBER OF SEQUENCES: 31 (iv) COMPUTER READABLE FORM:
(A) MEDIUM TYPE: Floppy disk (B) COMPUTER: IBM PC compatible (C) OPERATING SYSTEM: PC-DOS/MS-DOS
(D) SOFTWARE: PatentIn Release #1.0, Version #1.25 (EPO) (vi) PRIOR APPLICATION DATA:
(A) APPLICATION NU~BER: US (Unknown) (B) FILING DATE: 27-OCT-1994 (vi) PRIOR APPLICATION DATA:
(A) APPLICATION NUMBER: US 08/144,704 (B) FILING DATE: 28-OCT-1993 ~W 095/20039 PCTrUS9S/06682 ~ , .

(2) INFORMATION FOR SEQ ID NO~
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 36 amino acids (B) TYPE: amino acid (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (vii) IMMEDIATE SOURCE:
(B) CLONE: GLP-1 (glucagon-like peptide-1) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:1:
His Asp Glu Phe Glu Arg His Ala Glu Gly Thr Phe Thr Ser Asp Val Ser Ser Tyr Leu Glu Gly Gln Ala Ala Lys Glu Phe Ile Ala Trp Leu Val Lys Gly Arg (2) INFORMATION FOR SEQ ID NO:2:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 44 amino acids (B) TYPE: amino acid (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (vii) IMMEDIATE SOURCE:
(B) CLONE: GRF (1-43)-Ala (Growth hormone releasing factor) (xi) SEQUENCE DESCRIPTION: SEQ ID W0:2:
Tyr Ala Asp Ala Ile Phe Thr Asn Ser Tyr Arg Lys Val Leu Gly Gln Leu Ser Ala Arg Lys Leu Leu Gln Asp Ile Met Ser Arg Gln Gln Gly Glu Ser Asn Gln Glu Arg Gly Ala Arg Ala Arg Ala W 095/20039 PCTnU5~5/CG6~2 2~ 74525 ,. . ... ~, .

(2) INFORMATION FOR S-Q ID NO:3:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 24 amino acids (B) TYPE: amino acid (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (vii) IMMEDIATE SOURCE:
(B) CLONE: GRF (1-24) (Growth hormone releasing ~ctor) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:3:
Tyr Ala Asp Ala Ile Phe Thr Asn Ser Tyr Ary Lys Val Leu Gly Gln Leu Ser Ala Arg Lys Leu Leu Gln (2) INFORMATION FOR SEQ ID NO:4:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 31 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo N51Q

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:4:

(2) INFORMATION FOR SEQ ID NO:5:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 9 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:5:

WO 9S/20039 ~ ~ 2 1 7 4 5 2 5 PCT/US95,'C 6~82 . :. .; .
' ^ 89 (2) INFORMATION F02 SEQ ID NO:6:
(i~ SEQUENCE CHARACTERISTICS:
(A) LENGTH: 9 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:6:

(2) IN~ORMATION FOR SEQ ID NO:7:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 12 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:7:

(2) INFORMATION FOR SEQ ID NO:8:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 12 base pair (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:8:

(2) INFORMATION FOR SEQ ID NO:9:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 9 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:9:

W 095/20039 PCTrUS55/0'f~2 ~
2 ~ 7~525 (2) INFORMATION FOR SEQ ID NO:10:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 12 base pairs (8) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) ~kQDkN~k DESCRIPTION: SEQ ID NO:10:

(2) INFORMATION FOR SEQ ID NO:11:
(i) ~kQukN~k CHARACTERISTICS:
(A) LENGTH: 12 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (yenomic) (xi) kyukN~k DESCRIPTION: SEQ ID NO:11:

(2) INFORMATION FOR SEQ ID NO:12:
(i) SkQDkN~: CHARACTERISTICS:
(A) LENGTH: 9 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:12:

(2) INFORMATION FOR SEQ ID NO:13:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 30 base pairs (B) TYPE: nuclBic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) I~EDIATE SOURCE:
(B~ CLONE: Oligo wtll (xi) SEQUENCE DESCRIPTION: SEQ ID NO:13:

~ W O95/20039 2 1 7 4 5 2 5 PCT/USgs/Oc~Q2 ~ s ~

(2) INFORMATION FOR SEQ ID NO:14:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 28 base pairs tB) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear tii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo W49F

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:14:
GGATCCGGTC ATC~ Cl TGAACGGG 28 (2) INFORMATION FOR SEQ ID NO:15:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 38 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo N51A

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:15:
'CAAAGGATC CGGTCATCCT ll~l.GGCA GGGGGCCA 38 (2) INFORMATION FOR SEQ ID NO:16:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 20 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo pUC19 8pl (xi) SEQUENCE DESCRIPTION: SEQ ID NO:16:

; PCTrUS9S/06682 21 7452~

~2) INFORMATIOW FOR SEQ ID NO:17:
(i) S~Qu~N~. CHARACTERISTICS:
(A) LENGTH: 21 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo E145A

(Xi) SkQU~:N~ DESCRIPTION: SEQ ID NO:17:

(2) INFORMATION FOR SEQ ID NO:18:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 21 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo E65A and oligo wtl (xi) SEQUENCE DESCRIPTION: SEQ ID NO:18:

(2) INFORMATION FOR SEQ ID NO:19:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 38 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: lin~ar (ii) ~OLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo N51D

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:19:

-~ W O 95/20039 J ~, , r 2 1 7 4 5 2 5 PCTrUS95/06682 ~ .~, , , ~

.. 93 t2) INFORMATION FOR SEQ ID NO:20:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 33 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo NSlE

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:20:

(2) INFORMATION FOR SEQ ID NO:21:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 30 base pairs (B) TYPE: nucleic acid (C) STRA~D~DN~S: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo N51Q

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:21:
,GATCCGGTC ATCCTTTGGT TGCAGGGGGT 30 (2) INFORMATION FOR SEQ ID NO:22:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 31 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo N515 (xi) SEQUENCE DESCRIPTION: SEQ ID NO:22:

W O95t20039 ~ ; PCTrUS95106682 ~
~ 74525 (2) IWFORMATION FOR SEQ ID NO:23:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 31 b~se pairs tB) TYPE: nucleic acid tC) STRANDEDNESS: single tD) TOPOLOGY: linear tii) MOLECULE TYPE: DNA tgenomic) tvii) IMMEDIATE SOURCE:
(B) CLONE: Oligo NSlT

(Xi) g~yU~N~: DESCRIPTION: SEQ ID NO:23:
GGATCCGGTC ATC~L.LG~ TGACTGGGGG T 31 t2) INFORMATION FOR SEQ ID NO:2~:
(i) ~yU~N~ CHARACTERISTICS:
(A) LENGTH: 21 base pairs (B) TYPE: nucleic acid tC) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
(B) CLONE: Oligo E145D

txi) SEQUENCE DESCRIPTION: SEQ ID NO:24:
~GCGTAGGAA TCCCCAGCGA T 21 t2) INFORMATION FOR SEQ ID NO:25:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 21 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:25:

_WO 95/20039 /~ 2 1 7 4 5 2 5 PCT/US9S/06682 . ~ . , ~ . .
~ ~ t r ~

t2) INFORMATION FOR SEQ ID NO:26:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 21 ~ase pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (vii) IMMEDIATE SOURCE:
. (B) CLONE: Oligo E145Q

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:26:

(2) INFORMATION FOR SEQ ID NO:27:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 21 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA ~genomic) tvii) IMMEDIATE SOURCE:
(B) CLONE: Oligo E145S

(xi) ~:Q~NCk DESCRIPTION: SEQ ID NO:27:

(2) INFORMATION FOR SEQ ID NO:28:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTY.: 9 base pairs (B) TYPE: nucl~ic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:28:

wo g5/20039 . ~ .......... . .; PCT/US~510J~~ ~
2 ~ 74525 (2) INFORMATION FOR SEQ ID NO:29:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 12 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:29:

(2) INFORMATION FOR SEQ ID NO:30:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 12 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: lincar (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:30:

(2) INFORMATION FOR SEQ ID NO:31:
(i) S~QU~NC~ CHARACTERISTICS:
(A) LENGTH: 12 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:31:
TT~ ~NN~N~ CC 12

Claims (44)

WHAT IS CLAIMED IS:

1. A customized protease derived from a known protease, wherein the customized protease has a modified active site and is capable of catalyzing a transacylation reaction in which a preselected substrate is modified with a preselected nucleophile, and wherein the same transacylation reaction is not substantially catalyzed by the known protease.

2. The customized protease of claim 1, wherein the customized protease is capable of transacylating at the C-terminal end of the preselected substrate.

3. The customized protease of claim 1, wherein the customized protease is capable of transacylating at the N-terminal end of the preselected substrate.

4. The customized protease of claim 1, wherein the preselected nucleophile is a single or multiple amino acid unit.

5. The customized protease of claim 1, wherein the customized protease is derived from a known exopeptidase.

6. The customized protease of claim 1, wherein the customized protease is derived from a known serine or cysteine protease.

7. The customized protease of claim 1, wherein the customized protease is derived from a known carboxypeptidase.

8. The customized protease of claim 1, wherein the customized protease is derived from a known endoprotease.

9. The customized protease of claim 1, wherein the preselected substrate has a basic or acidic penultimate amino acid.

10. The customized protease of claim 1, wherein the preselected nucleophile is selected from the group consisting of an acidic amino acid amide, a basic amino acid amide, a peptide amide, an amino acid and an amino acid ester.

11. The customized protease of claim 1, wherein the preselected substrate corresponds to GRF (1-43)-Ala and the preselected nucleophile is a leucine amide.

12. A customized carboxypeptidase derived from a known carboxypeptidase, wherein the customized carboxypeptidase has a modified active site and is capable of catalyzing a transacylation reaction in which a preselected substrate is modified by a preselected nucleophile, wherein the preselected substrate has an acidic or basic amino acid at the penultimate position and the preselected nucleophile is an amino acid amide.

13. The customized carboxypeptidase of claim 12, wherein the preselected substrate has an arginine at the penultimate position.

14. The customized carboxypeptidase of claim 13, wherein the preselected substrate is GRF (1-43)-Ala and the preselected nucleophile is leucine amide.

15. A customized protease of claim 12, wherein the modified active site has a different amino acid at the S1 binding site than the known carboxypeptidase.

16. A customized protease according to claim 12, wherein the modified active site has a different amino acid at the S1' site than the known carboxypeptidase.

17. A customized carboxypeptidase derived from a known carboxypeptidase, wherein the customized carboxypeptidase has a different amino acid residue at a position equivalent in the known carboxypeptidase to position 178 or 398 of carboxypeptidase Y.

18. A customized protease exhibiting enhanced transacylation capability which is derived from a known protease, wherein the active site of the known protease has been modified so that the transacylating activity of the known protease has been enhanced.

19. A customized protease according to claim 18, wherein the known protease is a carboxypeptidase and the active site has been modified by substituting the amino acid residue at a position equivalent to 178 in carboxypeptidase Y with serine.

20. A method for preparing a customized protease derived from a known protease and that has a modified active site and that functions to alter the transacylating capability of the known protease comprising the steps of:
(a) providing a DNA sequence that encodes the known protease;

(b) modifying at least one codon for at least one amino acid residue in the active site to form a mutant DNA sequence, wherein the modified codon encodes a different amino acid residue and wherein the different amino acid residue provides a customized protease with an altered transacylation capability; and (c) transforming a suitable host cell with the mutant DNA sequence to provide for expression of the mutant DNA sequence to yield the customized protease.

21. A method according to claim 20, wherein the DNA
sequence corresponds to a DNA sequence for carboxypeptidase Y.

22. A method according to claim 20, wherein the DNA
sequence encodes an endoprotease.

23. A method according to claim 20, wherein the amino acid is an amino acid at the S1 binding site of the active site of carboxypeptidase Y.

24. A method according to claim 23, wherein the amino acid is selected from the group consisting of Tyr147, Leu178, Tyr185, Tyr188, Asn241, Leu245, Trp312, Ile340, Cys341 and mixtures thereof.

25. A method according to claim 20, wherein the amino acid is an amino acid at the S1' binding site of the active site of carboxypeptidase Y.

26. A method according to claim 25, wherein the amino acid is an amino acid selected from the group consisting of Trp49, Asn51, Gly52, Cys56, Thr60, Phe64, Glu65, Glu145, Tyr256, Tyr269, Leu272, Ser297, Cys298, Met398 and mixtures thereof.

27. A method according to claim 26, wherein the amino acid is methionine 398.

28. The method according to claim 20, wherein the step of modifying further comprises:
(a) deleting the codon for the at least one amino acid residue in the active site; and (b) replacing the deleted codon with an oligonucleotide, wherein the oligonucleotide encodes the different amino acid residue.

29. A method according to claim 20, wherein the step of transforming a suitable host cell further comprises:
(a) transforming a strain of S. cerevisiae that is capable of secreting proteases;
(b) selecting the transformed cells for expression of the mutant protease; and (c) screening the expressed mutant protease for the ability to catalyze transacylation reactions with a preselected substrate and/or preselected nucleophile.

30. A method for mutagenizing a known protease to form a customized protease having an altered transacylation capability comprising the steps:
(a) providing a DNA sequence for a known protease;
(b) modifying the DNA sequence of at least one targeted position by inserting at the targeted position a DNA sequence having a stop codon and/or a DNA sequence recognized by a restriction enzyme to form a modified DNA sequence encoding an inactive form of the known protease, wherein the targeted position includes a codon for an amino acid in the active site of the known protease to be mutated and the stop codon is inserted in place by the codon for the amino acid in the active site;
(c) generating single stranded modified DNA;
(d) synthesizing and amplifying a mutant DNA
sequence encoding the customized protease by incubating the single stranded modified DNA with one or more first degenerate oligonucleotides that contain a codon for a different amino acid than the amino acid present in the active site and in at the targeted site of the known protease; and (e) selecting and screening for the mutant DNA
sequence encoding the customized protease.

31. A method according to claim 30, wherein the step of synthesizing and amplifying includes one or more second oligonucleotides, wherein the second oligonucleotide includes the codons of the known protease at a targeted site and replaces one or more of the first degenerate oligonucleotides.

32. A method according to claim 30, wherein the DNA
sequence for the known protease substantially corresponds to the DNA sequence of carboxypeptidase Y.

33. A method according to claim 32, wherein the targeted site encodes an amino acid selected from the group consisting of Tyr147, Leu178, Tyr185, Tyr188, Glu215, Arg216, Asn241, Leu245, Trp312, Ile340, Cys341 and mixtures thereof.

34. A method according to claim 33, wherein the targeted site encodes amino acid 178.

35. A method according to claim 30, wherein the modified DNA sequence has a stop codon selected from the group consisting of amber, ochre and opal.

36. A method according to claim 35, wherein the modified DNA sequence has an opal stop codon and restriction endonuclease recognition site for EcoRI
at the targeted site.

37. A method according to claim 30, wherein the step of generating an effective amount of single stranded modified DNA comprises the steps:
(a) inserting the modified DNA sequence in a phagemid shuttle vector, wherein the phagemid shuttle vector has a mutated ampicillin resistance gene;
(b) amplifying the shuttle vector in a suitable host cell; and (c) generating single stranded modified DNA.

38. A method according to claim 37, wherein the step of synthesizing and amplifying further comprises one or more oligonucleotides that repair the ampicillin resistance gene.

39. A method according to claim 38, wherein the step of selecting comprises:
(a) transforming a suitable host cell, wherein the host cell requires at least one amino acid to grow;
(b) selecting the transformed cells that can grow on medium deficient in the amino acid required for growth supplemented with a preselected peptide containing the required amino acid at the C-terminus; and (c) identifying the cells with the DNA
sequence encoding the customized protease by screening for mutant proteases that are capable of transacylating a preselected substrate that is not substantially acted upon by the known protease.

40. A process for modifying a preselected substrate by transacylation using a customized protease derived from a known protease comprising:
(a) incubating a customized protease with a preselected substrate and a preselected nucleophile to form a mixture, wherein the customized protease has a modified active site and wherein the preselected substrate and the preselected nucleophile are not substantially transacylated by the known protease; and (b) incubating the mixture sufficiently to form a preselected transacylation product.

41. A method according to claim 40, wherein the preselected substrate is growth hormone releasing factor with a C-terminal alanine.

42 A method according to claim 40, wherein the preselected nucleophile is a leucine amide.

43. A method according to claim 40, wherein the customized protease is a carboxypeptidase Y having a modified active site, wherein the active site has a serine at amino acid 178.

44. A method according to claim 40, wherein the preselected transacylation product is growth hormone releasing factor having a leucine amide at the C-terminal.