Production and Use of Non-Native Variants of Domain 2 of Human Placental Bikunin Designed by Directed Molecular Evolution

Background of the invention:

Human placental bikunin, also called hepatocyte growth factor activator inhibitor type 2 (HAI-2) or SPINT2, belongs to the family of Kunitz type inhibitors (Marlor et al., J. Biol. Chem. 272:12202-12208, 1997; Kawaguchi et al., J. Biol. Chem. 272:27558- 27564, 1997; Mϋller-Pillasch et al., Biochim. Biophys. Acta 1395:88-95, 1998). Kunitz domains are polypeptides of about 56-60 amino acids length that inhibit a broad spectrum of serine proteases with different potency. Most of the family members contain three intramolecular disulfide bonds in a conserved spacing. The reversible interaction between the Kunitz type inhibitor and the active site of the serine protease is mediated mainly by a loop of 9 amino acid length located in the N-terminal part of the Kunitz domain (Gebhard et al., Chapter 10 In: Proteinase Inhibitors, Barrett and Salvesen (eds.), Elsevier, Amsterdam, 1986, pp. 375-88; Bode and Huber, Eur. J. Biochem. 204:433-51 , 1992).

Mature placental bikunin contains of 2 Kunitz domains and a single putative transmembrane domain indicating that bikunin is synthesized as a membrane- associated form and might be shedded as a proteolytically truncated form. It has a broad inhibitory spectrum against various serine proteases showing potent inhibitory activities not only to plasmin, trypsin and kallikreins but also to the HGA-activating proteases hepatocyte growth factor activator (HGFA) and hepsin (Delaria et al., J Biol. Chem. 272:12209-12214, 1997; Kawaguchi et al., J. Biol. Chem. 272:27558- 27564, 1997; Kirchhofer et al., FEBS 579:1945-1950, 2005). Thus, although the precise (patho-) physiological function of placental bikunin is not yet defined, it is has been postulated that the protein may have an important role in the regulation of HGF-induced tissue responses and of protease activities related to inflammation, coagulation, fibrinolysis and tumorigenesis.

The prototypical Kunitz type inhibitor is represented by aprotinin which is a 58-amino acid bovine protein (Dembowsky et al., Chapter 10, In: Novel Therapeutic Proteins -

Selected Case Studies, Dembowsky et al. (eds.), WILEY-VCH, Weinheim, 2001 , pp. 225-41). Aprotinin is the active substance in the medicament Trasylol, which is approved for the reduction of perioperative blood loss in CABG patients. Its blood saving properties have been mainly assigned to the antifibrinolytic but anticoagulative activities of aprotinin mediated by the potent inhibition of key serine proteases like plasmin, plasmakallikrein and factor XIa. In addition, its antiinflammatory activity leads to modulation of the systemic inflammatory response (SIRS) following bypass surgery (Mojcik and Levy, Ann. Thorac. Surg. 71 :745-754, 2001). Aprotinin might have further potential for blood saving in major surgery (hip replacement, spinal surgery, liver transplantation) and further indications like trauma (Samama et al., Anesth. Analg. 95:287-293, 2002; Porte et al., Lancet 355:1303- 1309, 2000; Coats et al., Cochrane Database Syst. Rev. 4:CD004896, 2004).

However, as aprotinin is of bovine origin it bears a significant immunogenic potential, which translates into reported cases of hypersensitivity in human patients upon re- exposure (Dietrich et al., Anesthesiology 95:64-71 , 2001 ; Beierlein et al., Ann Thorac. Surg. 79:741-748, 2005). Thus, there ^' s a high medical need for of a less immunogenic functional equivalent to aprotinin.

Placental bikunin due to its human origin is supposed to be non-immunogenic. In addition, functional characterization of a soluble fragment of recombinant bikunin and both of its synthetically prepared Kunitz domains 1 and 2 (KD1 and KD2) revealed that all three proteins are potent inhibitors of kallikreins, plasmin and factor XIa (Delaria et al., J Biol. Chem. 272:12209-12214). Interestingly, a bikunin variant with a mutation in KD2 had almost wt-activity against HGFA, whereas a point mutation in KD1 markedly reduced the activity. From these results it has been concluded that KD1 is mainly responsible for the HGFA inhibitory activity of bikunin (Qin et al., FEBS 436:111-114, 1998). Together these findings led to the conclusion that bikunin or its isolated KDs may be considered as a therapeutic protein for those indications for which aprotinin has shown to be beneficial.

A method for the production of human placental bikunin or its glycosylated domain 1 from mammalian cell culture has been described in US2004/0235715 by Chan et al.. Unfortunately, the analysis of the recombinant domain 2 of bikunin (hBikD2) was

hampered by low expression levels in various expression systems and part of the published biochemical data refers to synthetically derived material (US6583108 by Tamburini et al.; Delaria et al., J Biol. Chem. 272:12209-12214). Thus, an objective of the present invention is to identify derivatives of hBikD2 with improved expression levels and with functional activities at least similar to aprotinin.

Recently, it has been reported that modification of the N-terminus of aprotinin (EP419878 by Ebbers et al.) or hBikD2 (US6583108 by Tamburini et al.) may lead to an improved yield of correctly processed material. A yeast expression system for the preparation of homogeneously processed secreted recombinant aprotinin having the natural N-terminal sequence was disclosed in US5707831 by Apeler et al..

Non-native variants of natural Kunitz-type domains with putative protease inhibitor activity have also been described in US5863893 by Dennis et al. and US5914315 by Sprecher et al..

Variant polypeptides according to the present invention may be designed by directed molecular evolution approaches which in general allow the random generation of a large number of mutants followed by selection for the desired property. Strategies for this type of in vitro protein engineering are based on various techniques of mutagenesis and/or recombination of DNA as has been reviewed in Bloom et al., Curr. Opin. Struct. Biol. 15:447-452, 2005; Kaur and Sharma, Crit. Rev. Biotechnology 26:165-199, 2006.

Protein function can be modified and improved in vitro by a variety of methods, including site directed mutagenesis (Alber et al., Nature, 5; 330:41-46, 1987) combinatorial cloning (Huse et al., Science, 246:1275-1281 , 1989; Marks et al., Biotechnology, 10:779-783, 1992) and random mutagenesis combined with appropriate selection systems (Barbas et al., PNAS. USA, 89:4457-4461 , 1992).

The method of random mutagenesis together with selection has been used in a number of cases to improve protein function and two different strategies exist. Firstly, randomisation of the entire gene sequence in combination with the selection of a variant (mutant) protein with desired characteristics, followed by a new round of

random mutagenesis and selection. This method can then be repeated until a protein variant is found which is considered optimal (Schier R. et al., J. MoI. Biol. 1996 263:551-567). Here, the traditional route to introduce mutations is by error prone PCR (Leung et al., Technique, 1 :11-15, 1989) with a mutation rate of approximately 0.7%. Secondly, defined regions of the gene can be mutagenised with degenerate primers, which allows for mutation rates of up to 100% (Griffiths et al., EMBO. J, 13:3245-3260, 1994; Yang et al., J. MoI. Biol. 254:392-403, 1995).

Random mutation has been used extensively in the field of antibody engineering. Antibody genes formed in vivo can be cloned in vitro (Larrick et al., Biochem. Biophys. Res. Commun. 160:1250-1256, 1989) and random combinations of the genes encoding the variable heavy and light genes can be subjected to selection (Marks et al., Biotechnology, 10:779-783, 1992). Functional antibody fragments selected by these methods can be further improved using random mutagenesis and additional rounds of selections (Schier R. et al., J. MoI. Biol. 1996 263:551-567).

Typically, the strategy of random mutagenesis is followed by selection. Variants with interesting characteristics can be selected and the mutated DNA regions from different variants, each with interesting characteristics, combined into one coding sequence (Yang et al., J. MoI. Biol. 254:392-403, 1995).

Combinatorial pairing of genes has also been used to improve protein function, e.g. antibody affinity (Marks et al., Biotechnology, 10:779-783, 1992).

Another known process for in vitro mutation of protein function, which is often referred to as "DNA shuffling", utilises random fragmentation of DNA and assembly of fragments into a functional coding sequence (Stemmer, Nature 370:389-391 , 1994). The DNA shuffling process generates diversity by recombination, combining useful mutations from individual genes. It has been used successfully for artificial evolution of different proteins, e.g. enzymes and cytokines (Chang et al., Nature Biotech. 17:793-797, 1999; Zhang et al. Proc. Natl. Acad. Sci. USA 94:4504-4509, 1997; Christians et al., Nature Biotech. 17:259-264, 1999). The genes are randomly fragmented using DNase I and then reassembled by recombination with each other. The starting material can be either a single gene (first randomly mutated using error-

prone PCR) or naturally occurring homologous sequences (so-called family shuffling).

Fragment-|NkJuced Diversity (FIND™) technology developed by Alligator Bioscience AB is a directed evolution approach involving DNA-shuffling of single stranded DNA fragments as disclosed in WO 98/58080 by Borrebaeck et al., WO 02/48351 by Carlsson et al. and WO 03/97834 by Furebring et al..

Summary of the invention:

It is the purpose of the present invention to provide novel non-native variants of hBikD2 which can be produced as a functional protein at a high level in a recombinant expression system, for example the yeast secretion system, and which show a favorable inhibition profile against serine proteases.

The invention is based on the surprising discovery that by methods of directed molecular evolution chimeras of Kunitz-type proteins like hBikD2 and aprotinin can be generated that exhibit the desired properties.

According to a first aspect of the present invention it was possible to generate a novel variant (SEQ ID NO: 1 in Tab. 1) with the said characteristics containing the core sequence of hBikD2 and flanking sequences of aprotinin by FIND ^® recombination of the hBikD2/aprotinin chimeras 3, 5 and 6 (Example 7).

According to a second aspect of the present invention the expression level of the protein defined by SEQ ID NO: 1 can be further improved by generating novel variants of SEQ ID NO: 1 through random mutagenesis using error prone PCR (Example 8). The mutated proteins (SEQ ID NO: 2 to 37 in Tab.1) are characterized by having at least one additional amino acid exchange and/or amino acid insertion per molecule leading to a further increased expression level in the yeast secretion system while retaining the favorable inhibition profile.

According to a third aspect of the present invention there are provided variants of SEQ ID NO: 1 (SEQ ID NO: 38 to 40) with the said characteristics bearing combinations of the above mentioned mutations (Example 5 and 6).

, Brief description of the figures:

Table 1 : Amino acid Sequences of relevant proteins

Table 2: Expressionlevels of novel hBikD2 variants in yeast secretion system. Range in 2-3 different experiments, values referring to trypsin inhibitory activity of an aprotinin standard

Table 3: Activity of aprotinin and novel hBikD2 variants in various serine protease assays. Mean values of at least 2 experiments.

Detailed description of the invention:

A description of preferred embodiments of the invention follows.

Definitions:

In the present context, the term "non-native variant of hBikD2" is defined as a non- naturally Kunitz domain having disulfide bonds at Cys5-Cys55, Cys14-Cys38, and Cys30-Cys51 and more than 50% amino acid sequence identity to the second domain of human placental bikunin (hBikD2, Y129 to Q186 of NM_021102) but bearing at least one amino acid exchange.

In the present context, the term "improved expression level" is defined as having trypsin inhibitory activities accumulating in the secretions of transformed cells of >10 μg/ml (referring to an aprotinin standard).

In the present context, the term "favorable inhibition profile" is defined as having IC50 values for the inhibition of plasmin, plasmakallikrein and trypsin of below 50 nM.

Discovery and Preferred Embodiment:

It has been described that the Kunitz-type inhibitor domain 2 of human bikunin (hBikD2) exhibit similar or improved protease specificities as found for aprotinin, especially with respect to the potency of plasmin and plasmakallikrein inhibition. Moreover, its human origin should allow the repeated use of an hBikD2-based drug in human patients with reduced risk of adverse immune responses. However, despite recent efforts to increase the expression level of recombinant hBikD2, the accessibility of the material remains unsatisfactory. Therefore, it is the purpose of the present invention to provide novel non-native variants of hBikD2 which can be produced as a functional protein at a high level in a yeast secretion system while retaining a favorable inhibition profile against serine proteases.

Expression of hBikD2 using various expression systems, for example the yeast secretion system, do not exceed background levels when assessed by the analysis of the trypsin inhibitory activity accumulating in the secretions of transformed cells (Tab. 2).

In a first optimization step FIND ^® recombination of hBikD2 and aprotinin sequences was employed to create novel non-native hBikD2 variants (Example 7). Due to the limited homology of hBikD2 and aprotinin (50%) recombination was performed with 3 hBikD2/aprotinin chimeras: In the first chimera (chimera # 3) amino acids 1 to 10 and 56 to 58 relate to aprotinin whereas the core sequence from amino acid 11 to 55 is derived from hBikD2. The second chimera (chimera # 5) comprises the N-terminal and core amino acid 1 to 39 from hBikD2 and the C-terminal amino acids 40 to 58 from aprotinin. The third chimera (chimera # 6) consists of amino acids 1 to 39 from aprotinin and amino acid 40 to 58 from hBikD2.

FIND ^® recombination of single stranded DNA-fragments of chimera # 3, # 5 and # 6 led to the generation of a yeast expression library. Surprisingly, screening of about 1000 clones of this library for trypsin inhibitory activity in the conditioned media resulted in the identification of two clones (176E9 and 174H10) with improved expression levels. Sequence analysis revealed that on the amino acid level both clones express the same novel non-native hBikD2 variant which is composed of a hBikD2 core of amino acid 11 to 39 and aprotinin derived flanking residues 1 to 10

and 40 to 58 (SEQ ID NO:1 1 in Tab. 1). However, on a nucleotide level both clones show minor variations in their sequences.

The in depth analysis of the expression level of clone 176E9 with SEQ ID NO: 1 in a 100 ml shaking flask-scale confirmed that the trypsin inhibitory activity recovered from the conditioned medium is about 30-fold above the level of the empty vector control. In terms of the absolute expression level (Tab. 2) values of 34 to 53 μg/ml were reached (referring to an aprotinin standard in the trypsin assay). Moreover, similar to hBikD2 and aprotinin the purified protein of SEQ ID NO: 1 also inhibited plasmin, plasmakallikrein and trypsin with high potencies, i.e. with IC50 < 50 nM (Tab. 3).

Thus, in one embodiment the instant invention provides for non-native variants of hBikD2 generated by directed molecular evolution with more than 50% aminoacid sequence identity to the second domain of human placental bikunin (hBikD2, Y129 to Q186 of NM_021102) but bearing at least one aminoacid exchange. In a preferred embodiment an example of a polypeptide useful in the invention has the amino acid sequence defined by SEQ ID NO: 1.

In a second step the expression of the hBikD2 variant defined by SEQ ID NO: 1 should be further optimized, by generating randomly mutated libraries using error prone PCR (Example 8). Surprisingly, screening of the resulting about 10,000 clones led to the identification of additional non-native hBikD2 variants with further improved expression in the yeast secretion system (Examples 9 and 10). Sequence analysis revealed that these mutated variants are characterized by having at least one amino acid exchange and/or amino acid insertion per molecule (SEQ ID NO: 2 - 37 in Tab. 1). Among the clones with the highest expression levels a variant with a R39G mutation could be identified (clone 176E9-A07, SEQ ID NO: 6) which is characterized by a trypsin inhibitory activity in the secretions of about 70-fold above the level of the empty vector control (63-106 μg/ml, see Tab. 2). As shown in Tab. 3 the purified protein of SEQ ID NO: 6 also inhibited plasmin, plasmakallikrein and trypsin with high potencies, i.e. with IC50 < 50 nM.

Thus, in one embodiment, the instant invention provides for the non-native variants of hBikD2 having the amino acid sequence of SEQ ID NO NO: 2 to 37. In a preferred embodiment the instant invention provides for the non-native variant of hBikD2 having the amino acid sequence of SEQ ID NO: 6.

In a third round of expression optimization mutations identified in the previous step contained in SEQ ID NO: 2 to 37 were combined. Sequences of the resultant variants with further improved expression levels are given in Tab. 1 (SEQ ID NO: 38 to 40).

Thus, in one embodiment, the instant invention provides for the non-native variants of hBikD2 having all kinds of combinations of mutations described by SEQ ID NO NO: 2 to 37. In a preferred embodiment the instant invention provides for the non- native variants of hBikD2 having the amino acid sequence of SEQ ID NO: 38 to 40.

The skilled artisan will recognize that a similar approach of optimization via directed molecular evolution can be applied for increasing the expression levels of other native and non-native Kunitz-type serine protease inhibitors of human and non- human origin.

The non-native variants of hBik-D2 described here can be made synthetically using any standard polypeptide synthesis protocol and equipment. Alternatively, the described variants can be produced by recombinant methods using expression systems based on bacterial, yeast, baculovirus or mammalian expression vectors and the like. A preferred recombinant expression system producing the described variants is the yeast S. cerevisiae.

Utility:

The agents described herein are useful in the treatment of the following diseases: blood loss during operations with increased risk of bleeding; therapeutic intervention into thromboembolism (e.g. after operations, accidents); bleeding from postoperative surgery; shock; polytrauma; sepsis; disseminated intravascular coagulation (DIC); multiorgan failure (MOF); unstabile angina; myocardial infarction; stroke; embolism;

deep venous thrombosis (DVT); inflammatory diseases (e.g. asthma, rheumatism); invasive tumor growth and metastasis; therapeutic intervention into pain or edema (edema of brain of spinal marrow); prevention of activation of hemostasis during dialysis; treatment of symptoms of skin aging (elastosis, atrophy, wrinkling, vascularly changes, pigmentary changes, actinic keratoderma, blackheads, cysts); wound healing; melanoma; treatment of symptoms of melanoma (actinic keratoderma, basal cell carcinoma, invasive squamous-cell carcinoma, malignant melanoma); multiple sclerosis; fibrosis; cerebral haemorrhage; inflammation of brain or spinal cord; infections of brain; tendinopathy.

Examples:

The present invention is further illustrated in the following examples which are not intended to be in any way limiting to the scope of the invention as claimed.

Example 1 :

Cloning of domain 2 from human placental bikunin gene (hBikD2) Routine cloning tasks were carried out according Sambrook et al. (Molecular Cloning Cold Spring Harbor, 1989). For the isolation of plasmid DNA from E.coli (mini- and midipreps) Qiagen-tips (Qiagen) were used. The host organism employed for transformation was E.coli strain DH5α (invitrogen). The extraction of DNA fragments from agarose gels was carried out with the aid of Qiagen gel extraction kit according to the manufacturers protocol (Qiagen). Oligonucleotides for PCR and sequencing reactions were purchased from Operon, synthetic genes (optimized for S.cerevisiae coden-usage) from Geneart.

For PCR experiments kits from Qiagen (Hot Star Mastermix), Stratagene (PfuUltra Hotstart DNA Polymerase) or Novagen (KOD HiFi, Hot Start and XL DNA Polymerases) were used according to each manufacturers protocol. All vector constructs were confirmed by cycle-DNA-sequencing using fluorescence- labelled terminators (Big Dye Terminator, Version 1.1 , Firma Applied Biosystems ) on an 3100 Avant Genetic Analyzer (Applied Biosystem).

The 58 amino acids coding sequence comprising Kunitz-type domain 2 (Y129 - Q186) from the human placental bikunin gene (hBikD2, NM_021102) was purchased

as synthetic gene from Geneart (optimized for S.cerevisiae codon-usage). Additional oligonucleotides 5' and 3' to the coding sequence include restriction enzyme recognition sites, which were used for in-frame subcloning of hBikD2 into the yeast secretion vector plU10.10.W (Apeler, Chapter 12, In: J. Knablein (ed.), Wiley-VCH, Modern Biopharmaceuticals, 1021-1032, 2005). The synthetic gene coding for hBikD2 was cloned into vector pPCR-Script, also purchased from Geneart.

The DNA sequence of the synthetic gene coding for hBikD2 is as following:

GGGCGAATTGGGTΆCCGATTCCCATCTATTTTTACTGCTGTTTTGTTTGCTGCTTC TTCT GCTTTGGCTTATGAAGAGTATTGTACTGCTAATGCTGTTACTGGTCCATGTAGAGCTTCT TTTCCAAGATGGTATTTTGATGTTGAGAGAAATTCTTGTAACAACTTCATCTATGGTGGT TGTAGAGGTAACAAAAATTCTTATAGATCTGAAGAGGCTTGCATGTTGAGATGTTTTAGA CAATAATAACTCGAGGAGCTCCAGCTTTTGTTCCC

Restriction enzyme recognition sites (5' Kpnl: GGTACC, BsaBI: GATnnnnATC; 3' Xhol: CTCGAG, Sad: GAGCTC) are underlined.

The deduced amino acid sequence of hBikD2 is as following :

YEEYCTANAVTGPCRASFPRWYFDVERNSCNNFIYGGCRGNKNSYRSEEACMLRCFR Q hBikD2

Example 2:

Cloning of chimera # 5

Chimera # 5 comprises 58 amino acids (aa), consisting of 39 aa of the human placental bikunin gene (hBikD2, aa 1 - 39 ) and of 19 aa of the bovine aprotinin gene (BPTI, bovine pancreatic trypsin inhibitor, aa 40 - 58, NM_001001554 ). Chimera # 5 was purchased as synthetic gene from Geneart (optimized for S.cerevisiae codon- usage). Additional oligonucleotides 5' and 3' to the coding sequence include restriction enzyme recognition sites, which were used for in-frame subcloning of chimera # 5 into the yeast secretion vector pi U 10.10. W (Apeler, 2005). The synthetic gene coding for chimera # 5 was cloned into vector pPCR-Script, also purchased from Geneart.

The synthetic gene coding for chimera # 5 had the following sequence :

GGGCGAATTGGGTACCGATTCCCATCTATTTTTACTGCTGTTTTGTTTGCTGCTTCT TCT GCTTTGGCTTATGAAGAATATTGTACTGCTAATGCTGTTACTGGTCCTTGTAGAGCTTCT TTTCCAAGATGGTATTTTGATGTTGAAAGAAATTCTTGTAATAACTTCATATATGGTGGT TGTAGAGCTAAAAGAAATAACTTCAAATCTGCTGAAGATTGTATGAGAACTTGTGGTGGT GCTTAATGACTCGAGGGAGCTCCAGCTTTTGTTCCCTT

Restriction enzyme recognition sites (5' Kpnl: GGTACC, BsaBI: GATnnnnATC; 3' Xhol: CTCGAG, Sad: GAGCTC) are underlined.

The deduced amino acid sequence of chimera # 5 (aa from BPTI are uderlined) is as following :

YEEYCTANAVTGPCRASFPRWYFDVERNSCNNFIYGGCRAKRNNFKSAEDCMRTCGG A chimera # 5

Example 3:

Cloning of chimera # 6

Chimera # 6 comprises 58 amino acids (aa), consisting of 39 aa the bovine aprotinin gene (BPTI, bovine pancreatic trypsin inhibitor, aa 1 - 39, NM 001001554) and of 19 aa of the human placental bikunin gene (hBikD2, aa 40 - 58). Chimera # 6 was purchased as synthetic gene from Geneart (optimized for S.cerevisiae codon-usage). Additional oligonucleotides 5' and 3' to the coding sequence include restriction enzyme recognition sites, which were used for in-frame subcloning of chimera # 6 into the yeast secretion vector plU10.10.W (Apeler, Chapter 12, In: J. Knablein (ed.), Wiley-VCH, Modern Biopharmaceuticals, 1021-1032, 2005). The synthetic gene coding for chimera # 6 was cloned into vector pPCR-Script, also purchased from Geneart.

The synthetic gene coding for chimera # 6 had the following sequence :

GGGCGAATTGGGTACCGATTCCCATCTATTTTTACTGCTGTTTTGTTTGCTGCTTCT TCT GCTTTGGCTAGACCAGATTTTTGTTTGGAACCACCATATACTGGTCCATGTAAAGCTAGA ATTATTAGATACTTCTATAATGCTAAAGCTGGTTTGTGTCAAACTTTTGTTTATGGTGGT TGTAGAGGTAACAAAAATTCTTATAGATCTGAAGAGGCTTGTATGTTGCGTTGTTTTAGA CAATAATGACTCGAGGGAGCTCCAGCTTTTGTTCCCTT

Restriction enzyme recognition sites (5' Kpnl: GGTACC, BsaBI: GATnnnnATC; 3' Xhol: CTCGAG, Sad: GAGCTC) are underlined.

The deduced amino acid sequence of chimera # 6 (aa from BPTI are uderlined) is as following :

RPDFCLEPPYTGPCKARI IRYFYNAKAGLCQTFVYGGCRGNKNSYRSEEACMLRCFRQ Chimera # 6

Example 4:

Cloning of chimera # 3

Using a PCR reaction, N- and C-terminal amino acids of the synthetic gene mentioned in example 1 were exchanged to the corresponding amino acids in the

BPTI protein sequence.

PCR-primers used in the PCR-reaction were deduced from BPTI, corresponding to aa 1-17 (primer A) and aa 56-58 (primer B). In addition, primer A at the 5' end exhibits a recognition site for the restriction enzyme BsaBI, primer B a recognition site for Xhol.

The primers A and B used had the following sequences: Primer A:

5' -caccgattcccatctattttcactgctgtcttqttcqctqcttcttctqctttqqctAG ACCAGATTTCTGCtTqGAGCCA CCATATACTGGTCCATGTAGAGCTTCT-B'

Primer B:

5 ' -ttactcqaqctaTTAAGCACCACCACATCTCAACATGCAAGCCTCTTCA-3 '

BPTI-specific nucleotides are printed in capital letters, lower case letters are for flanking sequences, restriction enzyme recognition sites (BsaBI: gatnnnnatc; Xhoi: ctcgag) are underlined.

The PCR mixture contained 10ng hBikD2 Plasmid-DNA, 10 pMol Primer A, 10 pMol Primer B, 1 mM dNTPs, IxPCR reaction buffer (Novagen), 1 mM MgSO ₄ , 1 U KOD Hot Start DNA Polymerase (Novagen) in a total volume of 50μl. The 'cycle'- conditions were 2 min. at 94°C, 25 cycles of, in each case, 1 min. at 94°C, 1 min. at 5O ⁰ C, 1.5 min. at 72°C and a subsequent 10 min incubation at 68°C.

The PCR product coding for chimera # 3 had the following sequence:

caccqattcccatctattttcactgctgtcttqttcqctqcttcttctgctttggct aqa ccagatttctgcttggagccaccatatactggtccatgtagagcttcttttccaagatgg tattttgatgttgagagaaattcttgtaacaacttcatctatggtggttgtagaggtaac aaaa attcttatagatctgaagaggcttgcatgttgagatgtggtggtgcttaatagct cqaqtaa

Restriction enzyme recognition sites (5' BsaBI: GATnnnnATC; 3' Xhol: CTCGAG) are underlined.

The deduced amino acid sequence of chimera # 3 (aa from BPTI are underlined) is as following:

RPDFCLEPPYTGPCRASFPRWYFDVERNSCNNFIYGGCRGNKNSYRSEEACMLRCGG A chimera 3

Example 5:

Cloning of variants -mut10. -mut11.-mut12 (SEQ ID NO: 38 - 40)

The variants 176E9-mut10,-mut11 , and -mut12 comprised 58 aa each (SEQ ID NO:

38 - 40) and were purchased as synthetic genes from Geneart (optimized for

S.cerevisiae codon-usage). Additional oligonucleotides 5' and 3' to the coding sequences include restriction enzyme recognition sites, which were used for in-frame subcloning of the variants into the yeast secretion vector plU10.10.W (Apeler,

Chapter 12, In: J. Knablein (ed.), Wiley-VCH, Modern Biopharmaceuticals, 1021-

1032, 2005).

The synthetic genes coding for variants 176E9-mut10, -mutH , and -mut12 (SEQ ID

NO: 38 - 40) were cloned into vector pPCR-Script, also purchased from Geneart.

The synthetic genes had the following sequences (restriction enzyme recognition sites - 5' BsaBI: GATnnnnATC; 3' Xhol: CTCGAG, Sad: GAGCTC - are underlined):

variant - mut10 (coding for protein defined by SEQ ID NO: 38)

GGGCGAATTGGGTACCGATTCCCATCTΆTTTTTACTGCTGTTTTGTTTGCTGCTTC TTCT GCTTTGGCΆAGACCAGATTTTTGTTCTGAATCTCCATATACAGGTCCTTGTAGAGCTTC T TTTCCAAGATGGTATTTCGACGTTGAAAGAAATTCTTGCAACAATTTCATTTATGGTGGT TGTGGTGCTAAAGGTAACAATTTCGAATCTGCCGAAGATTGTATGAGAACTTGTGGTGGT GCTTAATAACTCGAGGAGCTCCAGCTTTTGTTCCC

variant - mut11 (coding for protein defined by SEQ ID NO: 39)

GGGCGAATTGGGTACCGATTCCCΆTCTATTTTTACTGCTGTTTTGTTTGCTGCTTC TTCT GCTTTGGCAAGACCAGATTTTTGTTTGGAACCACCATATACAGGTCCTTGTAGAGCTTCT TTTCCAAGATGGTATTACGACGTTGAAAGAAATTCTTGCAACAATTTCATTTATGGTGGT TGTGGTGCTAAAGGTAACAATTTTAAATCTGCCGAAGATTGTATGAGAACTTGTGGTGGT GCTTAATAΆCTCGAGGAGCTCCAGCTTTTGTTCCC

variant - mut12 (coding for protein defined by SEQ ID NO: 40)

GGGCGAATTGGGTACCGATTCCCATCTATTTTTACTGCTGTTTTGTTTGCTGCTTCT TCT GCTTTGGCAAGACCAGATTTTTGTTTGGAATCTCCATATACAGGTCCTTGCAGAGCTTCT GTTCCAAGATGGTATTTTGACGTCGAATCAAATTCTTGTAACAATTTCATTTATGGTGGT TGTGGTGCTAAGAGAAACAATTTTAAATCTGCCGAAGATTGTATGAGAATTTGCGGTGGT GCTTAATGACTCGAGGAGCTCCAGCTTTTGTTCCC

The deduced amino acid sequences of variants 176E9-mut10, -mut11 , and -mut12 are as following:

RPDFCSESPYTGPCRASFPRWYFDVERNSCNNFIYGGCGAKGNNFESAEDCMRTCGG A -mut 10 / SEQ ID NO: 38 RPDFCLEPPYTGPCRASFPRWYYDVERNSCNNFIYGGCGAKGNNFKSAEDCMRTCGGA -mut 11 /SEQ ID NO: 39 RPDFCLESPYTGPCRASVPRWYFDVESNSCNNFIYGGCGAKRNNFKSAEDCMRICGGA -mut 12 / SEQ ID NO: 40

Example 6:

Cloning of chimera # 3. 5 und 6 and variants -mut10 (SEQ ID NO: 38). -mut11 (SEQ ID NO: 38). and -mut12 (SEQ ID NO: 38) into the yeast secretion vektor plU10.10.W The E.coli / yeast shuttle vector pYES2 (invitrogen) was modified and used for the construction of a yeast secretion vector plU10.10.W (Apeler, Chapter 12, In: J. Knablein (ed.), Wiley-VCH, Modern Biopharmaceuticals, 1021-1032, 2005). The PCR reaction mixture of example 4 was purified with a purification kit (Qiagen), cleaved with the restriction enzymes BsaBI and Xhol and ligated into the yeast secretion vector plU10.10.W, which was cleaved with the same restriction enzymes accordingly. Chimera # 5 and # 6 and each of the variants -mut10, -mut11 , and - mut12 cloned in pPCRscript were cleaved with the restriction enzymes BsaBI and Xhol, the corresponding inserts isolated and ligated each into the yeast secretion vector plU10.10.W as described above. Plasmid DNA from each transformation

event was isolated, cleaved with the restriction enzymes BsaBI and Xhol and positive clones sequenced.

The deduced aa sequences of chimera # 3, # 5, # 6 and variants -mut10 (SEQ ID NO: 38), -mut11 (SEQ ID NO: 39), and -mut12 (SEQ ID NO: 40) are as following (MFα-pre-sequences are italicized, aa derived from BPTI and mutated aa in variants are underlined):

chimera 3 chimera 5 MfJFPSIFTAyLf ¹ AaSSALARPDFCLEPPYTGPCKARIIRYFYNAKAGLCQTFVYGGCRGNKNSYRSEEACML RCFRQ chimera 6

SEQ ID NO: 38 SEQ ID NO: 39 SEQ ID NO: 40

Example 7:

Generation of FIND ^® libraries

350 base pair DNA fragments coding for hBik-D2/aprotinin chimeras were amplified by polymerase chain reaction (PCR) from the chimeric constructs chimera # 5, chimera # 6 and chimera # 3 described in Example 2 to 4, using the forward primer δ'-CTGGAATTCCACTGCTGTTTTGTTTGCTGC and the reverse primer 5'-

CTCAAGCTTGACTTCAGGTTGTCTAACTCCTTCC. All primers were purchased from MWG, Ebersberg, Germany.

The PCR reactions contained 1 ng of either chimeric construct, 0,5 μM of each primer, 200 μM of each dNTP (New England Biolabs, cat # N0440S, N0441S, N0442S, N0443S), 1x Dynazyme reaction buffer, 1 U Dynazymell DNA Polymerase (Finnzymes, cat # F-501 L ) and 0,02 U Phusion DNA polymerase (Finnzymes, cat # F-530-S) in a total volume of 50 μl. The PCR program comprises 25 cycles of 94°C for 15 sec, 57°C for 30 sec min and 72°C for 30 sec and finally elongation at 72°C for 7 minutes.

Single-stranded DNA (ssDNA), representing sense and antisense strands, was prepared as follows: The PCR products were divided into two batches and digested separately with EcoRI or Hindlll (New England Biolabs, cat # R0101 L and R0104S,

respectively), creating a 5'phosphorylated end on either the sense or the antisense strand. The 5'phosphorylated strand was digested using a Strandase™ ssDNA Preparation Kit (Novagen, Merck Biosciences, cat # 69199-3), leaving the unphosphorylated strand of DNA intact. In both cases, the obtained ssDNA was analyzed by agarose gel (Cambrex, cat # 50080) electrophoresis, purified using Recochip (TaKaRa, cat # 9039) according to manufacturer's recommendations, and finally ethanol precipitatated.

The FIND ^® experiments were initiated by fragmenting sense and antisense ssDNA, respectively, with Exonuclease I (Exo I) (100 U/g DNA, (New England Biolabs, cat # 0293L)), Exonuclease V (Exo V) (50 U/g DNA (USB, cat # 70040Y)) and Exonuclease VII (Exo VII) (5 U/g DNA (USB, cat # 70082Y)) in separate tubes under conditions recommended by the manufacturers.

The ssDNA fragments resulting from the exonuclease digestions were recombined in a first PCR reaction (PCR 1) essentially as described above with the exceptions that the DNA polymerase used was Phusion (1 U) and that the PCR mix contained 1x Phusion HF reaction buffer, no added primers and 60ng ssDNA from each strand, in a total volume of 50 μl. The material from PCR 1 was then amplified in a second PCR reaction (PCR 2) also as described above with 5μl PCR 1 product, 0,5μM primers (same as used for the initial amplification of DNA fragments), 1x AmpliTaq reaction buffer and 1 ,25U AmpliTaq® DNA polymerase (Applied Biosystems, cat # N8080171) in a 50 μl total volume. The reassembled full-length genes were subsequently cloned into pGEM®-T Vector System I (Promega, cat # A3610) and sequenced.

In the first FIND ^® reaction, sense ssDNA from chimera # 5 was recombined with antisense ssDNA from chimera # 3 and this resulting library was in a second FIND ^® round used to prepare sense ssDNA that was recombined with antisense ssDNA from chimera # 6.

Example 8:

Generation of randomly mutated library

A randomly mutated KPI library using hBikD2/aprotinin chimeras 176E9 (SEQ ID NO: 1) and 174H10 (SEQ ID NO: 2) as starting material was generated by subjecting a 50:50 mixture of the genes for these clones to GeneMorphll PCR Mutagenesis Kit (Stratagene, cat # 200550) according to the manufacturer's instructions. To obtain a sufficient number of mutations (1-2 amino acid substitutions per sequence), three consecutive rounds of GeneMorph Il was performed. In each round, 0.1 ng template was used. Primers used in the GeneMorph Il reactions were forward: 5 ^' - GCTGCTAGCTCTGCTTTGGCTAGACCAGATTTC and reverse: 5 ^' -

GATGCATGCTCGAGCTATTAAGCACCACC.

Example 9:

Cloning and expression of variants

A pYES2-vector containing MF pre-pro-leader sequence and hBikD2 was modified to remove a Nhel-site after the URA-3 using Quikchange multisitedirected mutagenesis kit (Stratagene #200513) according to the manufacturer's instructions. The primer used in the reaction was 5 ^' -

GATGAATTGAAAAGCTAACTTATCGATGATAAGCTGTC. The resulting vector was further modified to introduce an Nhel-site in the leader sequence using Quikchange Il mutagenesis kit (Stratagene, #200524) according to the manufacturer's instructions using the primers 5 ^' -

CTGCTGTTTTGTTTGCTGCTAGCTCTGCTTTGGCTTATGAAGAG and 5 - CTCTTCATAAGCCAAAGCAGAGCTAGCAGCAAACAAAACAGCAG. From the resulting vector, the hBikD2-sequence was removed using Nhel (New England Biolabs #R0131) and Sphl (New England Biolabs #R0182) and replaced by a stuffer- fragment encoding a stop-codon created by annealing of the two phosphorylated oligos forward: 5'P-CTAGCTCTGCTTTGGCTTAACTCGAGCATG and reverse 5 ¹ P- CTCGAGTTAAGCCAAAGCAGAG.

The libraries ware cloned into the modified pYES2 expression vector, generated as described above, using Nhel (New England Biolabs #R0131) and Sphl (New England Biolabs #R0182) and transformed into the E. coli strain ElectroTen Blue (Stratagene #200159) using electroporation according to the manufacturer's instructions. The transformation was plated out on Q-trays (Genetix #X6023) with LB

agar (Miller) (Merck Cat# 1.10283.0500) with 50μg/ml Ampicillin (Calbiochem) and incubated over night in 37°C. The resulting colonies were scraped off the plate and used for extracting the plasmid DNA using HiSpeed Plasmid Midi Kit (Qiagen #12643).

For expression of the library (ligated into pYES2 prepared from ElectroTen Blue as described above), 1 μg of supercoiled plasmid was transformed into 10 ⁸ cells of S. cerevisiae according to Gietz high-efficiency transformation protocol (Methods Enzymol. 2002; 350:87-96). Thus, the S. cerevisiae strain Hansen teleomorph (ATCC 201149) was grown in 2xYPD (100g/l YPD-broth, Sigma #Y-1500, 40g/l Glucose, Sigma #G7528) supplemented with 25μg/ml Chloramphenicol (Calbiochem, #220551) at 30 ⁰ C and 200rpm for 16-18h. This o/n-culture was used to inoculate 50ml 2xYPD with 25μg/ml Chloramphenicol at a starting titer of 5-10 x 10 ⁶ cells/ml. This culture was grown at 30 ⁰ C, 200rpm until the titer was at least 2x10 ⁷ cells/ml. To 10 ⁸ of these cells, a transformation mix consisting of 240μl 50% PEG 3500 (Sigma, #P-3640), 36μl 1.0M Litium Acetate (Sigma #L-4185), 50μl 2mg/ml salmon sperm DNA (Sigma, #D1626) and 1 μg of DNA to be transformed was added. Cells and transformation mix were vortexed vigorously and incubated at 42°C for 60min. The cells were pelleted and 1 ml of water was added and the cells were plated on Q-trays with SC-Ura agar (20g/l agar (Saveen, #B1000-1), 20g/l glucose (Sigma, #G7528), 1.,92g/l yeast synthetic drop-out media (Sigma, #Y-1501), 6.7g/l Yeast nitrogen base without amino acids (Fluka, #51484) and incubated at 30 ⁰ C over night.

For expression of protein from individual mutant clones, colonies resulting from transformation of the library into S. cerevisiae were picked using a Genetix QPixll to 96-well plates (Greiner #655180) with 150μl SC5 (20g/l Bacto Yeast extract (BD #212750), 6.7g/l KH ₂ PO ₄ , (Merck #1.04873), 1g/l MgSO ₄ x 7H ₂ O (Merck, #1.05886), 2g/l (NH ₄ ) ₂ SO ₄ (Sigma, #A2939), 2% glucose (Merck, #1.08337) and 1 ml/l trace element solution (5g/l Tritriplex 3 (Merck 1.08418), 2g/l FeSO ₄ x 7H ₂ O (Sigma #F8633), 0.1g/l ZnSO ₄ x 7 H ₂ O (Sigma #Z4750), 0.03g/l MnCI ₂ x 4H ₂ O (Sigma #M3634), 0.3g/l H ₃ BO ₃ (Merck, #1.00165), 0.2g/l CoCI ₂ x 6H ₂ O (Sigma #C8661), 0.01g/l CuCI ₂ x 2H ₂ O (Aldrich #22.178-3), 0.02g/l NiCI ₂ x 6H ₂ O (Sigma #N6136), 0.03g/l Na ₂ MoO ₄ x 2H ₂ O (Merck, 1.06521), pH3.5) supplemented with 25μg/ml

Chloramphenicol. Plates were sealed with Nunc sealing tape (#236366) and cultures were incubated at 30 ⁰ C for 48 hours and sample supernatants were harvested.

Example 10: Trypsin screening assay hBikD2/aprotinin clones with enhanced expression levels were identified using a trypsin assay in which the enzyme trypsin catalyses the cleavage of the substrate N _α - Benzoyl-L-arginine 4-nitroanilide hydrochloride (L-BAPA), which can be measured as absorbance at 405 nm. The trypsin assay was performed in clear, flat-bottom 384- well plates (Greiner, cat # 781186) in a Beckman Coulter robot system consisting of Beckman Multimek 96, Multidrop 384 Titertek, BMG FluoStar reader and Cytomat Incubator as follows. Trypsin from bovine pancreas (Sigma, cat. # T4665, 10200 U/ml) at a concentration of 1 mg/ml in 50 mM Tris, pH 8.4, was diluted in buffer (0.2 M Triethanolamine hydrochloride, 0.02 M NaCI, 10 ml 5% Tween-80, pH to 7.8) to a final concentration of 50 U/ml. Trypsin solution was added to each well, followed by the addition of sample supernatants, at a final dilution of 7.5 or 15 times, and the plate was preincubated at 23 or 28 ⁰ C for 10 minutes. The substrate L-BAPA (Sigma, cat #. B-3279) (dissolved to 2 mg/ml in a buffer consisting of 4% DMSO and 0.001 % Tween-80) was then added to the plate at a final concentration of 0.77 mM. Following 10 minutes incubation a first absorbance reading at 405 nm was performed. A final absorbance reading was performed after yet another incubation step, either 45 minutes at 23°C, or 75 minutes at 28 ⁰ C and change in absorbance was calculated.

Example 11 :

Fermentation of yeast cells expressing SEQ ID NO:1 in medium to large scale

The following media were used for growth and fermentation of yeast cells:

Trace element solution:

Titriplex III (Merck 8418) 5 g/l

FeSO ₄ ^• 7H ₂ O (Merck 3965) 2 g/l

ZnSO ₄ ^• 7H ₂ O (Merck 8883) 0,1 g/l

MnCI ₂ ^• 4H ₂ O (Merck 5927) 30mg/l

H ₃ BO ₃ (Merck 165) 0,3 g/l

CoCI ₂ ^• 6H ₂ O (Merck 2533) 0,2 g/l

CuCI ₂ ^• 2H ₂ O (Merck 2733) 10mg/l

NiCI ₂ ^• 6H ₂ O (Merck 6717) 20mg/l

Na ₂ MoO ₄ • 2H ₂ O (Merck 6521) 30mg/l

Stock cultures of yeast transformants were prepared by mixing of 1-ml aliquots of a seed culture with 1 ml of a glycerol-solution (80%) in polypropylene vials and storage at -140 ⁰ C.

Yeast fermentation was performed either in medium scale (100 ml) or large scale (10,000 to 25,000 ml) format. For the seed culture production for medium scale fermentation a 50-ml shake flask filled with 10 ml of SD2 medium was inoculated with a stock culture and incubated on a shaker (240 rpm) for 2-3 days at 28-30 ⁰ C. 3 ml of seed culture were used to inoculate 100 ml of SC5 medium in a 1-L shake

flask. Subsequent incubation was perfomed for 4 day at 240 rpm and 28-30 ⁰ C. pH- values of the cultures were adjusted to 5-6 once a day with 5 N NaOH and cultures were fed on day 1 to 3 with 1 ml of 50%-yeast extract solution and 4 ml of a 4 M glucose-solution.

For large scale fermentation the Bioreactor system (Wave Biotech, Tagelswangen, CH) was employed. Bioreactor bags were inoculated with 300 (for 10,000 ml SD5 medium) to 750 ml (for 25,000 ml SD5 medium) seed culture and incubation was performed for 4 days at 30 ⁰ C and a wave frequency of 32/min (angel: 10°, aeration: 0.25 l/min) in a fed batch mode continuously adding a 100- to 250-fold volume of the above mentioned feeding solutions per day. pH-adjustment to 5-6 was done with 5 N NaOH once a day.

Growth of the cultures could be monitored at various time points by assessing OD ₆ oo- At day 4 of fermentation harvest of the cellfree supernatant was performed by centrifugation (15 min at 6000 rpm in JA14-rotor).

Example 12:

Determination of Expression Level by Trypsin Assay

Expressionlevels from media to large scale fermentations where assayed as trypsin inhibition activity basically as described in Example 10, with slight modifications. 100 μl of trypsin (from pore pancrease, Merck; 5 μg/ml in 0,001 N HCI/0,05% Tween 80) were mixed with 100 μl of serially diluted conditioned medium pre-incubated for 30 min at RT. The reaction was then started by the addition of L-BAPA (1 mg/ml in H ₂ O) and followed for 60 min in a Tecan reader at 405 nm. Control medium was spiked with defined amounts of aprotinin and served to calculate a standard curve that allowed determination of the unknown trypsin inhibitory activity in the secretions.

Example 13:

Purification of a Protein defined by SEO ID NO:1:

Recombinant protein defined by SEQ ID NO:1 produced in the fermentation process was purified from 10 liter yeast medium. The pH of the medium was adjusted to pH 7.8 with IM NaOH. The medium was cleared by centrifugation at 2,000 rpm (4 ⁰ C; 15 min; Beckmann-Allegra 6KR

centrifuge). The supernatant was applied to a 10 ml Trypsin agarose column (Sigma-T1763) at 1 ml/min. The column was washed with 70 ml 50 mM Tris pH 7.8, 250 mM NaCl and with 50 ml 50 mM Tris pH 7.8, 600 mM NaCl. The protein was eluted with 100 ml 50 mM KCl/ 10 mM HCl pH 2.0. The samples were collected in 2 ml fractions which contained 500 μl 200 mM Tris/Cl pH 7.6, 2 M NaCl each to neutralize the eluate. The protein was detected by its Trypsin-inhibiting activity.

Fractions containing Trypsin-inhibiting activity were pooled and applied to a Source 15 RPC column (GE Healthcare). The column was washed with 6 ml 0.1% TFA (buffer HPLC-A), and the protein was subsequently eluted with a 25 ml gradient from 0% to 50% buffer HPLC-B (0.1% TFA, 60% acetonitril) and a 5 ml gradient from 50% to 100% buffer HPLC-B. Samples containing the protein were lyophilized, dissolved in 50 mM Tris pH 7.5 and stored at -20 ⁰ C.

Example 14:

Determination of IC50-Values against Trypsin, Plasmin and Plasmakallikrein

IC50 values of non-native hBikD2 variants against trypsin, plasmin and plasmakallikrein were determined in biochemical assays perfomed in white 384-well microtiterplates employing defined fluorogenic substrates. The assay buffer was composed of 50 mM Tris /Cl, pH 7.4, 100 mM NaCI, 5 mM CaCI ₂ , 0,08 % (w/v) BSA.

The detailed assay conditions were as follows:

10 μl of the serially diluted test compound per well were mixed with 20 μl of enzyme solution and preincubated for 5 min at room temperature. Subsequently, the reaction was started by the addition of 20 μl of substrate solution. The reaction was followed for 60 to 90 min in a Tecan reader at an excitation of 360 nm and an emission of 465 nm. Dose-response curves and IC50 values were determined using the software GraphPad Prism (4.02).