WO2006012902A2

WO2006012902A2 - Creation of diversity in polypeptides

Info

Publication number: WO2006012902A2
Application number: PCT/DK2005/000515
Authority: WO
Inventors: Allan Svendsen; Lars Beier
Original assignee: Novozymes A/S
Priority date: 2004-08-02
Filing date: 2005-08-02
Publication date: 2006-02-09
Also published as: US20100136653A1; CA2575878A1; EP1781779A2; US20110287142A1; US8017371B2; US20080003642A1; WO2006012902A3; US8309337B2; AU2005269082A1

Abstract

The inventors realized that the diversity generated by conventional methods may be limited by steric hindrance between amino acid residues in the three-dimensional structures of the resulting polypeptides. The steric hindrance may occur between amino acid residues at widely different positions in the amino acid sequences, e.g. between residues in two different domains of the 3D structure, and resulting polypeptides which include such steric hindrance may never be observed in the conventional recombination methods because they may be expressed in poor yields or may have poor activity or stability. The inventors developed a method to identify and alleviate such steric hindrance in the resulting polypeptides. In an alignment of the three-dimensional structures, steric hindrance is indicated when residues from two different structures are located within a certain distance. Pairs of residues at corresponding positions in the amino acid sequences are not considered, and residues close to the surface (high solvent accessibility) are considered to be less prone to steric hindrance.

Description

CREATION OF DIVERSITY IN POLYPEPTIDES

FIELD OF THE INVENTION

The present invention relates to a method of constructing a hybrid polypeptide from two or more parent polypeptides in order to create diversity. It also relates to hybrid polypep- tides constructed by this method.

BACKGROUND OF THE INVENTION

The prior art describes methods of creating diversity by recombination of DNA se¬ quences encoding two or more polypeptides, followed by transformation of a suitable host or¬ ganism with the recombined DNA sequence and screening of the transformants for enzymatic activity. The recombination may be random or directed. WO 1995022625: US 6368805; J. E. Ness et al., Nature Biotechnology, vol. 20, Dec. 2002, pp. 1251-1255; M. C. Saraf et al., 4142- 4147, PNAS, March 23, 2004, vol. 101, No. 12.

SUMMARY OF THE INVENTION

The inventors realized that the diversity generated by conventional methods may be limited by steric hindrance between amino acid residues in the three-dimensional structures of the resulting polypeptides. The steric hindrance (also referred to as "structural stop codon") may occur between amino acid residues at widely different positions in the amino acid se¬ quences, e.g. between residues in two different domains of the 3D structure, and resulting polypeptides which include such steric hindrance may never be observed in the conventional recombination methods because they may be expressed in poor yields or may have poor activ¬ ity or stability.

The removal of "structural stop codons" can result in improved expression and/or stability of the protein of interest, or in ultimate case expression at all of protein of interest. For example in combining of two or more proteins, i.e. combining multiple hybrids of two or more proteins using various DNA techniques e.g. using shuffling techniques as known in the art (WO9522625, WO9827230 and WO2000482862) the removal of "structural stop codons" from one or more of the included proteins will improve the expression and/or stability of the proteins, and/or create access to a novel diversity not found by other shuffling or hybrid techniques. Combination of protein sequences will often result in accommodation of different sized resi- dues and homologous positions, but not always. Sometimes clashes will occur and especially in the core of the protein. The removal of "structural stop codons" results in novel diversity due to allowance of new region combinations not seen because of presence of "structural stop codons", which otherwise may result in a non functional or non expressed protein. The inventors developed a method to identify and alleviate such steric hindrance in the resulting polypeptides. In an alignment of the three-dimensional structures, steric hin¬ drance is indicated when residues from two different structures are located within a certain dis¬ tance. Pairs of residues at corresponding positions in the amino acid sequences are not taken into consideration since only one of the two residues is expected to be present in the recom- bined polypeptide. Pairs of residues are not taken into consideration if one or both is glycine or if one or both side chains is close to the surface (indicated by a high solvent accessibility) as the residue may be able to reposition to avoid the potential clash.

Accordingly, the invention provides a method of constructing a polypeptide, compris- ing: a) selecting at least two parent polypeptides each having an amino acid sequence and a three-dimensional structure, b) structurally aligning the three-dimensional structures, thereby aligning amino acid residues from different sequences, c) selecting a first amino acid residue from one structure and a second residue from another structure, such that: i) the two residues are not aligned in the superimposition, ii) a non-hydrogen atom of the first residue and a non-hydrogen atom of the second residue are located less than 2.7 A apart, and iii) each of the two residues is not Glycine and has a side chain having less than 30 % solvent accessibility, and d) substituting or deleting the first and/or the second residue such that the substitution is with a smaller residue, and e) recombining the amino acid sequences after the substitution, and f) preparing a DNA-sequence encoding the polypeptide of step e) and expressing the polypeptide in a transformed host organism.

Further the invention relates to a polypeptide which has at least 80%, 85%, 90%, 95% or 98% or 99% identity to SEQ ID NO: 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25. The invention also relates to a polynucleotide encoding any of the polypeptides.

BRIEF DESCRIPTION OF DRAWINGS

Fig. 1 shows an alignment of various known CGTase sequences. Details are given below. Fig. 2 shows the results of a comparison of two 3D structures. The upper sequence is

1qho for the maltogenic alpha-amylase Novamyl (SEQ ID NO: 17), and for the lower sequence is 1a47 for a CGTase (SEQ ID NO: 5). Details are described in Examples 1 and 2.

Fig. 3 and 4 shows hypothetical sequences with "structural stop codons". Details are described in Examples 6 and 7.

DETAILED DESCRIPTION OF THE INVENTION

Parent polypeptides

According to the invention, two or more parent polypeptides are selected, each having an amino acid sequence and a three-dimensional structure. The parent polypeptides may in particular be selected so as to be structurally similar, e.g. each pair having a amino acid iden¬ tity of at least 50 %, e.g. at least 60 %, 70 % or 80 %. Amino acid identity may be determined as described in US 6162628.

In another preferred embodiment the structurally similar parent polypeptides have a homology of at least 50 %, e.g. at least 60 %, 70 %, 80 %, 90% or 95%. Homology may be de¬ termined as described in WO 2004067737, i.e. by using the GAP routine of the UWGCG pack¬ age version 9.1. The parent polypeptides may be polypeptides having biological activity, structural polypeptides, transport proteins, enzymes, antibodies, carbohydrate binding modules, serum albumin (e.g. human and bovine), insulin, ACTH, glucagon, somatostatin, somatotropin, thy¬ mosin, parathyroid hormone, pituitary hormones, somatomedin, erythropoietin, luteinizing hor¬ mone, interleukin, chorionic gonadotropin, hypothalamic releasing factors, antidiuretic hor- mones, thyroid stimulating hormone, relaxin, interferon, thrombopoeitin (TPO) and prolactin.

The enzyme may have an active site, e.g. a catalytic triad, which may consist of Ser, Asp and His. The parent enzymes may be selected so as to have identical residues in the ac¬ tive site.

Three-dimensional structure Three-dimensional structure is meant to be a known crystal structure or a model struc¬ ture.

The 3D structure of each polypeptide may already be known, or it may be modeled using the known 3D structures of one or more polypeptides with a high sequence homology, using an appropriate modeling program such as Homology, Modeller or Nest. The 3D model may be optimized using molecular dynamics simulation as available, e.g., in Charmm or NAMD. The optimization may particularly be done in a water environment, e.g. a box or sphere.

The Homology, Modeller and Charmm software is available from Accelrys Inc., 9685 Scranton Road, San Diego, CA 92121-3752, USA, http://www.accelrys.com/. The Nest soft- ware is distributed free of charge at http://trantor.bioc.columbia.edu/programs/iackal/index.html. The NAMD software is available at http://www.ks.uiuc.edu/Research/namd/.

Structural alignment of 3D models

5 The 3D models may be structurally aligned by methods known in the art. The struc¬ tural alignment may be done by use of known software. In the structurally aligned models, pairs of residues from different sequences are considered to be aligned when they are located close to each other. The following software may be used:

DALI software, available at http://www.ebi.ac.uk/dali/ 10 CE software available at http://cl.sdsc.edu/

STAMP software available at http://www.compbio.dundee.ac.uk/Software/Stamp/stamp.html Protein 3Dhome at http://www-lecb.ncifcrf.gov/~tsai/ Yale Gernstein Lab - spare parts at http://bioinfo.mbb.yale.edu/align/ - 15 Structural alignment server at http://www.molmovdb.org/aliqn/

In the case of enzymes having an active site, the structural alignment may be a super- imposition of the structures based on the deviations of heavy atoms (i.e. non-hydrogen atoms) in the active sites, e.g. by minimizing the sum of squares of deviations. Alternatively, the super- imposition may be done so as to keep deviations between corresponding atoms below 0.8 A, 20 e.g. below 0.6 A, below 0.4 A,, below 0.3 A or below 0.2 A.

Selection of amino acid residues

Steric hindrance ("potential clashes") between two amino acid residues is indicated if a heavy atoms (i.e. non-hydrogen) of the two residues are located less than 2.1k, 2.5 A or 2.0 A apart, particularly less than 1.7 A, 1.5 A, 1.2 A, 1.1 A or 1.0 A apart, with the following excep- 25 tions:

Two residues aligned with each other in the structural alignment (pairs of residues at corresponding positions in the amino acid sequences) are not taken into consideration since only one of the two residues is expected to be present in the recombined polypeptide.

Pairs of residues are not taken into consideration if one or both is glycine.

30 Pairs of non-glycine residues are not taken into consideration if one or both side chains has more than 20 %, 25 % or 30 % solvent accessibility as a high solvent accessibility is taken as an indication that the residue may be able to reposition to avoid the potential clash. Solvent accessibility can be calculated by use of the DSSP program, available from Centre for Molecular and Biomolecular Informatics, University of Nijmegen, Toernooiveld 1 , P.O. Box 35 9010, 6500 GL Nijmegen, +31 (0)24-3653391 , http://www.cmbi.kun.nl/qv/dssp/. The DSSP program is disclosed in W. Kabsch and C. Sander, BIOPOLYMERS 22 (1983) pp. 2577-2637. The residue total surface areas of the 20 natural amino acids are tabulated in Thomas E. Creighton, PROTEINS; Structure and Molecular Principles, W.H. Freeman and Company, NY, ISBN: 0-7167-1566-X (1984). To confirm the severity of the potential clash, a local alignment of the two 3D struc¬ tures may then be made by aligning all residues within a distance of 10 A.

The steric hindrance may be identified by a comparison of two complete sequences in order, particularly severe clashes (less than 1.2, 1.1 or 1.0 A apart), to identify potential clashes that may arise no matter how the two sequences are recombined. Alternatively, the comparison may be made between two partial sequences to be combined in a hybrid, and in this case a larger limit may be used for the distance (less than 2.7 A, 2.5 A, 2.0 A, 1.7 A or 1.5 A).

Amino acid substitution

When a potential clash between two residues has been identified, one or both resi- dues is substituted with a smaller residue. In this connection, the residues are ranked as fol¬ lows from smallest to largest: (an equal sign indicates residues with sizes that are practically indistinguishable):

G < A=S=C < V=T < P < L=I=N=D=M < E=Q < K < H < R < F < Y < W The substitution may be such that the two residues after the substitution can form a hydrogen bond, a salt bridge or a cysteine bridge.

Recombination of amino acid sequences

After making amino acid substitutions to alleviate potential clashes, the substituted amino acid sequences are recombined. The recombination may be done by designing hybrids or by gene shuffling. Hybrids may be constructed by switching from one sequence to another between aligned residues. Once constructed, the hybrids can be produced by conventional methods by preparing a DNA sequence encoding it and expressing it in a transformed host organism.

Alternatively, genes can be prepared encoding each substituted amino acid se¬ quence, by shuffling the genes by known methods, transforming a suitable host organism with the shuffled genes. The shuffling can be done, , e.g., as described in WO 1995022625.

In the case of the parent polypeptides being enzymes, the transformants can be screened for enzymatic activity. Enzymes

The parent enzymes may have hydrolase, oxidoreductase or transferase activities, e.g. activities such as protease, lipolytic enzyme, glycosyl hydrolase, laccase, oxidoreductases with oxygen as acceptor (e.g. glucose oxidase, hexose oxidase or galactose oxidase), glycosyl transferase, esterase, cellulase, xylanase, amylase, isoamylase, pullulanase, branching en¬ zyme, pectate hydrolase, cyclodextrin glucanotransferase, or maltogenic alpha-amylase activ¬ ity. One or more of the parent enzymes may have a carbohydrate-binding domain.

The method may particularly be applied to two or more structurally similar enzymes, e.g. belonging to the same family in a structural classification of enzymes. Thus, they may be- long to the same structural family for glycosyl hydrolases and glycosyl transferases as de¬ scribed, e.g., in the following literature. The enzymes may be of family 13 and may particularly include a maltogenic alpha-amylase and a cyclodextrin glucanotransferase.

• Henrissat B., A classification of glycosyl hydrolases based on amino-acid sequence similarities. Biochem. J. 280:309-316 (1991). • Henrissat B., Bairoch A. New families in the classification of glycosyl hydrolases based on amino-acid sequence similarities. Biochem. J. 293:781-788 (1993).

• Henrissat B., Bairoch A. Updating the sequence-based classification of glycosyl hy¬ drolases. Biochem. J. 316:695-696 (1996).

• Davies G., Henrissat B. Structures and mechanisms of glycosyl hydrolases. Struc- ture 3:853-859 (1995).

The parent enzymes may be lipolytic enzymes belonging to the same homologous family as described at http://www.led.uni-stuttgart.de/families.html. The 3D structures of the lipolytic enzymes may all include a so-called "lid" in open or closed form.

The enzymes may be proteases or peptidases belonging to the same family or sub- family as described by MEROPS in "the Peptidase Database", available at http://merops.sanger.ac.uk/. The proteases may be subtilases, e.g. belonging to the same sub¬ group as described by Siezen RJ and Leunissen JAM, 1997, Protein Science, 6, 501-523; one of these sub-groups is the Subtilisin family.

CGTase The cyclodextrin glucanotransferase (CGTase) may have an amino acid sequence as shown in SEQ ID NOS: 1-16 and may have a three-dimensional structure found under the fol¬ lowing identifier in the Protein Data Bank (www.rcsb.org): S. circulans (1 CDG), alkalophilic Ba¬ cillus (1PAM), S. stearothermophilus (1CYG) or Thermoanaerobacterium thermosulfurigenes (1CIU, 1A47). 3D structures for other CGTases may be constructed as described in Example 1 of WO 9623874. Fig. 1 shows an alignment of the following known CGTase sequences, each identified by accession number in the GeneSeqP database and by source organism. Some sequences include a propeptide, but only the mature peptide is relevant for this invention.

SEQ ID NO: 1. aab71493.gcg S. agaradherens SEQ ID NO: 2. aau76326.gcg Bacillus agaradhaerans

SEQ ID NO: 3. cdg1_paema.gcg Paenibacillus macerans (Bacillus macerans).

SEQ ID NO: 4. cdg2_paema.gcg Paenibacillus macerans {Bacillus macerans).

SEQ ID NO: 5. cdgt_thetu.gcg Thermoanaerobacterthermosulfurogenes (Clos¬ tridium thermosulfurogenes) (SEQ ID NO: 2:) SEQ ID NO: 6. aawO6772.gcg Thermoanaerobacterthermosulphurigenes sp. ATCC

53627 (SEQ ID NO: 3)

SEQ ID NO: 7. cdgt_bacci.gcg Bacillus circulans

SEQ ID NO: 8. cdgt_bacli.gcg Bacillus sp. (strain 38-2)

SEQ ID NO: 9. cdgt_bacsθ.gcg Bacillus sp. (strain 1011 ) SEQ ID NO: 10. cdgt_bacs3.gcg Bacillus sp. (strain 38-2)

SEQ ID NO: 11 cdgu_bacci.gcg Bacillus circulans

SEQ ID NO: 12. cdgt_bacsp.gcg Bacillus sp. (strain 17-1 , WO 2003068976) (SEQ ID NO: 4)

SEQ ID NO: 13. cdgt_bacoh.gcg Bacillus ohbensis SEQ ID NO: 14. cdgt_bacs2.gcg Bacillus sp. (strain 1-1 )

SEQ ID NO: 15. cdgt_bacst.gcg Bacillus stearothermophilus

SEQ ID NO: 16. cdgt_klepn.gcg Klebsiella pneumoniae

To develop variants of a CGTase without a known 3D structure, the sequence may be aligned with a CGTase having a known 3D structure. An alignment for a number of CGTase sequences is shown in Fig. 2. Other sequences may be aligned by conventional methods, e.g. by use the software GAP from UWGCG Version 8.

Maltogenic alpha-amylase

The maltogenic alpha-amylase (EC 3.2.1.133) may have the amino acid sequence shown in SEQ ID NO: 17 (in the following referred to as Novamyl), having the 3D structure de- scribed in US 6162628 and found in the Protein Data Bank with the identifier 1QHO. Alterna¬ tively, the maltogenic alpha-amylase may be a Novamyl variant described in US 6162628. A 3D structure of such a variant may be developed from the Novamyl structure by known meth¬ ods, e.g. as described in T.L. Blundell et al., Nature, vol. 326, p. 347 ff (26 March 1987); J. Greer, Proteins: Structure, Function and Genetics, 7:317-334 (1990); or Example 1 of WO 9623874. Use of hybrid polypeptide

The hybrids may be useful for the same purpose as the parent enzymes.

Thus, a hybrid of a maltogenic alpha-amylase and a cyclodextrin glucanotransferase may form linear oligosaccharides as an initial product by starch hydrolysis and a reduced amount of cyclodextrin and may be useful for anti-staling in baked products.

A hybrid of laccases and/or other enzymes belonging to EC 1.10.3 may be useful for e.g. hair dyeing or reduction of malodor.

EXAMPLES

Example 1 : Comparison of complete sequences

Superimposition of parent enzymes

Two glycosyl hydrolases of family 13 were selected. One was a maltogenic amylase (Novamyl) having the amino acid sequence shown in SEQ ID NO: 17 and having a 3D struc¬ ture published under number 1 QHO. The other was a CGTase having the amino acid se¬ quence shown in SEQ ID NO: 5 and the 3D structure 1A47, and this was also taken to repre- sent the structure of the highly homologous CGTase having the sequence SEQ ID NO: 6. The two 3D structures were superimposed so as to align the active sites, and the alignment of resi¬ dues of the two sequences is shown in Fig. 2 Aligned residues shown vertically above each other, with gaps inserted to separate non-aligned residues.

Identification of potential clashes The two structures were analyzed, and the following unaligned residues were identi¬ fied as having a side chain with less than 30 % solvent accessibility and with a heavy atom less than 1.5 A (or less than 1.0 A) apart from a heavy atom of a residue in the other structure. The following pairs of residues were found to come within 1.0 A. The potential clashes are shown as CGTase residue and atom, Novamyl residue and atom, and distance in A:

D209 OD2 A676 CB 0.89

L261 CD1 K270 NZ 0.93

D267 CG N266 O 0.94

D267 OD1 N266 O 0.48

M307 CE L286 CD1 0.77

H503 CD2 K7 NZ 0.97

T509 OG1 Y574 CZ 0.65

V626 CB Y181 CZ 0.41

V626 CG1 Y181 OH 0.99 V626 CG2 Y181 CD2 0.76

K651 NZ P592 CG 0.35

The above residues are marked by asterisks in Fig. 2.

Example 2: Comparison of complementary sequences

To design hypothetical hybrids, residues in a partial sequence of Novamyl (SEQ ID NO: 17) were compared with residues in the complementary part of the CGTase sequence (SEQ ID NO: 6), and residues with heavy atoms located less than 1.7 A apart were identified. The potential clashes are shown as in Example 1. The identified residues are marked with as¬ terisks in Fig. 2.

Novamyl 1-494 + CGTase 495-683

H503 CD2 K7 NZ 0.97

N575 O Y317 OH 1.68

V626 CB Y181 CZ 0.41

CGTase 1-494 + Novamvl 495-686

D3 C R545 NH2 1.36

D209 OD2 A676 CB 0.89

Novamvl 1-499 + CGTase 500-683

H503 CD2 K7 NZ 0.97

N575 O Y317 OH 1.68

V626 CB Y181 CZ 0.41

CGTase 1-499 + Novamvl 500-686

D3 C R545 NH2 1.36

D209 OD2 A676 CB 0.89

Novamvl 1-410 + CGTase 410-683

H503 CD2 K7 NZ 0.97

N575 O Y317 OH 1.68

V626 CB Y181 CZ 0.41

Novamvl 1-378 + CGTase 378-683

N409 OE1 R354 N 1.63

H503 CD2 K7 NZ 0.97

N575 O Y317 OH 1.68 V626 CB Y181 CZ 0.41

Novamvl residues 1-204 + CGTase residues 207-683

W219 CZ2 L75 CD2 1.66

H503 CD2 , K7 NZ 0.97

V626 CB Y181 CZ 0.41

CGTase residues 1-139 and 207-683 + Novamvl residues 131-204

V626 CB Y181 CZ 0.41

Example 3: Construction of hybrids

Hybrids were constructed with the following combinations of Novamyl residues and CGTase residues (SEQ ID NO: 6) and with substitutions of Novamyl residues as indicated to alleviate potential clashes. For comparison, similar variants were constructed without substitu¬ tions.

The first eight of the above hybrids are found in SEQ ID NO: 18 to SEQ ID NO: 25.

Example 4: Screening of hybrids for amylase activity

Four hybrids of the previous example were produced by preparing a DNA-sequence encoding the hybrid and expressing the hybrid in a transformed organism cultivating a trans- formant, and the amylase activity was assayed by letting the culture broth act on Phadebas (dye-labelled substrate, available from Pharmacia) and measuring the absorbance at 650 nm. The amylase assay was made at pH 5.5 at two different temperatures: 5O⁰C and 60⁰C. Refer¬ ence hybrids without substitutions were included for comparison.

Example 5: Baking with hybrids. Further two hybrids were produced by cultivating a transformant and tested for baking.

The two hybrids are:

BaHyI : CGTase (SEQ ID NO: 6) residue 1-139 + Novamyl (SEQ ID NO: 17) residue 131-204 + CGTase (SEQ ID NO: 6) residue 207-683; and

BaHy2: Novamyl (SEQ ID NO: 17) residue 1-577 + CGTase (SEQ ID NO: 6) residue 580-683 + Y181A mutation in Novamyl.

The effect of the two hybrids in straight dough was compared to that of CGTase with respect to a number of parameters: Softness of breadcrumb, elasticity, and mobility of free water. Approximately 1 mg/kg of flour was dosed.

The two hybrids improve the softness of breadcrumb as compared to CGTase. The two hybrids improve the elasticity as compared to CGTase.

BaHy2 improves the mobility of free water as compared to CGTase, whereas BaHyI has the same effect as CGTase.

Example 6: Structural stop codons - impact on diversity.

This example illustrates the possible outcome of a hybridization between two proteins having the sequences SeqA and SeqB (figure 3):

If combination sites (marked with |) comprises a "structural stop codon" (marked with X), the resulting protein not be expressed properly or maybe even not at all. Segment 14 in SeqA and segment 7 in SeqB indicates such potential clashes due to the presence of "structural stop codons". The result will be a lowering of the diversity, as combinations containing these two segments most likely not will be able to accommodate the clashes and therefore not be pre¬ sent in the diversity of protein molecules.

If X in SeqA and /or SeqB is made smaller the accommodation might result in a func¬ tional protein. Accommodation may also be obtained by changing the shape or charge of the residue e.g. I to L and D to N. The "structural stop codon" can also be removed by inserting the proper match of residues by mutating the particular residues and/ or mutating the surrounding residues around the clashing residues thus creating accommodation. Smaller residues can be found in the list; G < A=S=C < V=T < P < L=I=N=D=M < E=Q < K < H < R < F < Y < W.

If the "structural stop codon" gives 100% non-functional protein - the lowering of diver¬ sity is 25% for one "structural stop codon" residue pair - compared to the situation without any "structural stop codons". That is the diversity for the segments are 2²⁰ = 1048576 and for the clashes it is 2¹⁸ = 262144.

Example 7: Structural stop codons - impact on diversity when combining more than two proteins.

In this example we have three proteins illustrated by SEQ1 , SEQ2 and SEQ3 (figure 4). SEQ1 has a "structural stop codon" with SEQ2 called X. SEQ1 has a "structural stop codon" with SEQ3 called Z, and SEQ 2 has a structural stop codon" with SEQ3 called Y. The diversity will hereby be lowered dramatically as exemplified above. We will have the common equation for the number of nonstructural stop codon" containing proteins termed D for diver¬ sity in the cases where the "structural stop codons" pairs are found in separate segments not containing other "structural stop codons" and the number of segments are higher or equal to the number of pairs:

Equation I: D = N^κ - P*N^(K"2) where D is diversity without "structural stop codons", N the number of proteins, K the number segments, and P the number of pairs (ie. X, Y and Z).

For other situations e.g. with "structural stop codons" in the same segment or other situations other equations can be derived. Using equation I we get D to be 2/3 for the numbers shown in present example and for the numbers in shown in the above example we get 0.75. Consequently the diversity may be in¬ creased significantly by removing "structural stop codons". Example 8: Structural stop codons - impact on extending combination possibilities for proteins with low homology to a better result.

One important aspect is the possibility of combining more distant related proteins by hybridisation or shuffling techniques and not only closely related proteins. The combination by hybridisation or shuffling techniques may go below the 90, or the 80, or the 70, or the 60, or the 50 percent homology level. At the upper level of homology, around 70-90 percent homol¬ ogy, the amount of diversity - meaning the number of active clones coming out of a hybridisa¬ tion or shuffling experiment - or at the lower level around 50-80 percent homology creation of active clones at all might be the outcome.

Example 9: Example on finding "structural stop codons" for combining proteins e.g. shuffling or hybrid formation.

The set of parent sequences are analyzed using the 3D structures. The 3D structures can be based on existing known structures or obtained by X-ray crystallography, NMR meth- ods or modeled using appropriate modeling programs like NEST, MODELLER or HOMOL¬ OGY. The two structures are superimposed by optimizing the RMSD of the C-alpha atom dis¬ tances using a appropriate program as listed in the description. The superimposed structures are analyzed for possible clashes between residues. For each type of atoms (a,b), where atom a is in structure A and atom b is in structure B the distance d(a,b) between the atoms is calcu- lated as the standard Euclidian distance. All atom pairs with distance smaller than a given pre¬ defined threshold are potentially structural clashes. A set of rules is imposed to filter out atom pairs with distance smaller than the threshold which are not to be considered as clashes. The rules are: i. Atom pairs that form part of the residue that are aligned in the alignment based on the superimposition are filtered out. ii. Atom pairs that form part of residues that are adjacent to aligned residue are filtered out. iii. Atom pairs where both atoms are backbone atoms are filtered out. iv Atom pairs that form part of residues that are both surface exposed are filtered out. Surface exposed can be computed based on the "solvent exposed surface area" computed by the DSSP-program by division by the standard accessibilities in the following list; A=62, C=92, D=69, E=156, F=123, G=50, H=130, l=84, K=174, L=97, M=103, N=85, P=67, Q=127, R=211 , S=64, T=80, V=81 , W=126 and Y=104. The threshold fro interatomic distances can be 3A, or 2.7A, or 2.5A or 2.3A, or 2.1 A or 2A. The minimal relative surface exposed area for filtering out an atom pair is 20% or preferably 30% for each residue. The found clashes are visualized and inspected in a graphic display program. Example 10: "Structural stop codons" for combining Protease - Subtilisin S8A

After the superimposition of the two X-ray structures of BPN' (1 SBT - also disclosing the amino acid sequence) and Savinase (1SVN - also disclosing the amino acid sequence) using a suitable display software like INSIGHT Il from Accellrys inc. a "structural stop codon" can be found i.e. a clash between to residues with distance lower than a certain threshold here 2.5A. The residues giving a clash can be seen are located in the core of the two proteins and having the following residues below 2.5A apart to 1198 from Savinase structure 1SVN and I268 BPN' 3D structure 1 SBT. Mutation of either 1SVN to 1198V or A or G or T, or the SBT se¬ quence to I268V or A or G or T will remove the interaction. So for example making the hybrid construction 1SVN sequence A1-G219 and 1 SBT sequence N218-Q275 should include the mutations suggested above to obtain the best result regarding expression.

Example 11 : "Structural stop codons" for combining protease TY145 and Savinase

After the superimposition of the two X-ray structures of TY145 (see patent application WO2004067737 A3, also disclosing the amino acid sequence (SEQ ID NO: 1)) and Savinase (1SVN - also disclosing the amino acid sequence) using a suitable display software like IN¬ SIGHT Il from Accellrys inc. a "structural stop codon" can be found i.e. a clash between to residues with distance lower than a certain threshold here 2.1 A: TY145 P308 clashes with Savinase 1198

TY145 W101 clashes with Savinase M119 TY145 103 clashes with Savinase W113 Savinase Y263 clashes with TY145 Mainchain

Example 12: "Structural stop codons" for combining lipases:

Two hybrid enzymes consisting of the N-terminal from Thermomyces lanuginosus li¬ pase (TLL, SEQ ID NO: 26) and the C-terminal from Fusarium sp. lipase (KVL, SEQ ID NO: 27) have been constructed (Construct 1 and Construct 2). The point of crossover resides within conserved regions within the two enzymes. A study of the three-dimensional structure of Thermomyces lanuginosus lipase 1GT6 and a model of the Fusarium sp. lipase build based on the 1GT6 structure reveals two places of residue clashes when making the two hybrid con¬ structs. In general the following "structural stop codons" can be found: TLL F142 clashes with KVL F136 TLL T64 clashes with KVL F24 TLL 1222 clashes with KVL Y226 TLL F80 clashes with KVL 160 TLL F55 clashes with KVL A62

The structural problem has been alleviated by introduction of the following mutations T64G and T64G/I222L into the two hybrid enzymes Construct 1 and Construct 2, respectively.

The constructs for two specific hybrids are (the numbers are taken for KVL and TLL protein sequences):

Construct 1. KVL 1-28 and TLL 29-269 Construct 2. KVL 1-28 and TLL 29-227 and KVL 225-267 Construct 3. KVL 1-28 and TLL 29-269 and TLL T64G

Construct 4. KVL 1-28 and TLL 29-227 and KVL 225-267 and TLL T64G and TLL I222L

The 3D structures of the KVL lipase was build using the Accelrys software HOMOLOGY pro¬ gram - other suitable software like NEST could also be used.

Example 13: "Structural stop codons" for combining laccases: Analyzing the three dimensional structure of the Coprinus cinerius laccase (CLL, SEQ ID NO:28) and the three dimensional structure model of Myceliophthora thermophila laccase (MTL, SEQ ID NO: 29) build using the NEST software based on the Melanocarpus albomyces laccase structure (1GW0 - also disclosing the amino acid sequence), it can be found that several "structural stop codons" can be found. Focusing on the core "structural stop codons" the following residues can found to be important to mutate. There are the following important "structural stop codons" that has to be removed before attempting shuffling of the two laccases of CCL and MTL:

MTL M301A and/or CCL F124L CCL E239A or D CCL E453A

MTL W464L MTL W420F

There are besides the mentioned changes other important issues concerning the cystin bridges MTL C301/C267 and CCL C135/C222. Securing of no overlaps in theses re- gions are of great importance. To avoid the problems the following are a plausible way to go further::

MTL C379S/C345S and CCL C135G/C222V Alternatively "transfer" CCL cystinbridge to MTL: MTL G193C/V281C.

Example 14: "Structural stop codons" for combining xylanases:

Analysing the three dimensional structure of the Bacillus agaradherens xylanase (BAX), having the X-ray structure 1QH7 (also disclosing the amino acid sequence), and the three dimen¬ sional structure of Bacillus halodurans xylanase (BHX) having the X-ray structure 1XNB (also disclosing the amino acid sequence), it can be found that several "structural stop codons" can be found. Focusing on the core "structural stop codons" the following residues can found to be important to mutate: BAX R49 clashes with BHX Y165 BAX K53 clashes with BHX Y5 BAX K136 + E56 clashes with BHX R73 BAX F163 clashes with BHX F145 BAX L199 clashes with BHX W42 BAX M28 clashes with BHX W6

Analysing the three dimensional structure of the Bacillus agaradherens xylanase (BAX), having the X-ray structure 1QH7, and the three dimensional structure model of Paenibacillus sp. xy¬ lanase (PSX) having the X-ray structure 1 BW (also disclosing the amino acid sequence), it can be found that several "structural stop codons" can be found. Focusing mostly on the core "structural stop codons" the following residues can found to be important to mutate: BAX R49 clashes with PSX Y166 + Q7 BAX K53 clashes with PSX Y5 BAX L199 clashes with PSX W42 BAX F163 clashes with PSX F146 BAX M28 clashes with PSX W6 BAX Y195 clashes with PSX N54.

Claims

1. A method of constructing a polypeptide, comprising: a) selecting at least two parent polypeptides each having an amino acid sequence and a three-dimensional structure, b) structurally aligning the three-dimensional structures, thereby aligning amino acid residues from different sequences, c) selecting a first amino acid residue from one structure and a second residue from another structure, such that: i) the two residues are not aligned with each other in the structural alignment, ii) a non-hydrogen atom of the first residue and a non-hydrogen atom of the second residue are located less than 2.7 A apart, and iii) each of the two residues is not Glycine and has a side chain having less than 30 % solvent accessibility, and d) substituting or deleting the first and/or the second residue with a smaller residue, and e) recombining the amino acid sequences after the substitution, and f) preparing a DNA-sequence encoding the polypeptide of step e) and expressing the polypeptide in a transformed host organism.

2. The method of claim 1 , wherein each pair of parent polypeptides has an amino acid identity of at least 50 %,

3. The method of claim 1 , wherein each pair of parent polypeptides has a homology of at least 50%.

4. The method of any preceding claim, further comprising a) superimposing the structures so as to align each non-hydrogen atom located < 10 A of an atom in the first or the second residue, and b) selecting two residues that are less than 1.5 A apart in the new superimposition.

5. The method of any preceding claim wherein the two selected residues after the substitution can form a hydrogen bond, a salt bridge or a cysteine bridge.

6. The method of any preceding claim wherein a non-hydrogen atom of the first residue and a non-hydrogen atom of the second residue are located less than 1.7 A apart, particularly less than 1.5 A, 1.2 A, 1.1 A or 1.0 A apart.

7. The method of claim 1 wherein each parent polypeptide is an enzyme having an active site, and the structural alignment is done so as to align each non-hydrogen atom of the amino acid residues of the active sites

8. The method of the preceding claim wherein the enzymes belong to glycosyl hydrolase fam- 5 ily 13, particularly comprising a cyclodextrin glucanotransferase and a maltogenic alpha- amylase.

9. The method of claim 6 or 7 which further comprises producing a polypeptide having the recombined amino acid sequence, testing the polypeptide for an enzymatic activity and select¬ ing an enzymatically active polypeptide.

10 10. A polypeptide which has an amino acid identity of at least 80% to SEQ ID NO: 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25.

11. A polynucleotide encoding any polypeptide of the above claim.

12. A polypeptide which: a) has an amino acid sequence which is a hybrid of a maltogenic alpha-amylase and 15 a cyclodextrin glucanotransferase, b) has a smaller residue at a position corresponding to: i) D209, L261 , D267, M307, H503, T509, V626, K651 of SEQ ID NO: 6 or ii) K7, Y181 , N266, K270, L286, Y574, P592, S676 of SEQ ID NO: 17, and c) has hydrolytic activity on starch.

20 13. A dough comprising the polypeptide of the preceding claim.